(Salmo Salar) with Amoebic Gill Disease (AGD) Chlo

Home , Amoebic gill disease, Dactylopodida, Lingulamoeba, Neoparamoeba, Null result, Paramoeba

The diversity and pathogenicity of amoebae on the gills of

Atlantic salmon (Salmo salar) with amoebic gill disease (AGD)

Chloe Jessica English

BMarSt (Hons I)

A thesis submitted for the degree of Doctor of Philosophy at

The University of Queensland in 2019

Abstract

Amoebic gill disease (AGD) is an ectoparasitic condition affecting many teleost fish globally, and it is one of the main health issues impacting farmed Atlantic salmon, Salmo salar in Tasmania’s expanding aquaculture industry. To date, Neoparamoeba perurans is considered the only aetiological agent of AGD, based on laboratory trials that confirmed its pathogenicity, and its frequent presence on the gills of farmed Atlantic salmon with branchitis. However, the development of gill disease in salmonid aquaculture is complex and multifactorial and is not always closely associated with the presence and abundance of N. perurans. Moreover, multiple other amoeba species colonise the gills and their role in AGD is unknown.

In this study we profiled the Amoebozoa community on the gills of AGD-affected and healthy farmed Atlantic salmon and performed in vivo challenge trials to investigate the possible role these accompanying amoebae play alongside N. perurans in AGD onset and severity. Significantly, it was shown that despite N. perurans being the primary aetiological agent, it is possible AGD has a multi-amoeba aetiology.

Specifically, the diversity of amoebae colonising the gills of AGD-affected farmed Atlantic salmon was documented by culturing the gill community in vitro, then identifying amoebae using a combination of morphological and sequence-based taxonomic methods. In addition to N. perurans, 11 other Amoebozoa were isolated from the gills, and were classified within the genera Neoparamoeba, Paramoeba, Vexillifera, Pseudoparamoeba, Vannella and Nolandella. This work highlighted that there is a far greater diversity of amoebae colonising AGD-affected gills than previously established.

Drawing on this culture-based study, five new TaqMan quantitative PCR (qPCR) assays were developed and applied to more accurately determine the prevalence and abundance of multiple amoeba species colonising the gills of Atlantic salmon held at two Tasmanian farm sites over a one- year period. The presence of N. perurans was also assessed in parallel using a previously established qPCR method. N. perurans was the dominant species in the Amoebozoa community on gills, and its abundance positively correlated with the progression of gross gill pathology. Only sporadic detections of Pseudoparamoeba sp. and Vannellida species were observed across the sampling period at either farm site. Nolandella spp., however, were highly prevalent at one site at ii one sample time when N. perurans were not detected on gills presenting low levels of gross gill pathology.

To investigate the pathogenic potential of Nolandella sp. and a more closely related amoeba to N. perurans, Pseudoparamoeba sp., in vivo challenges of naïve Atlantic salmon were performed. Additionally, to elucidate how Nolandella sp. and Pseudoparamoeba sp. influence the onset or severity of N. perurans-induced AGD, the gill condition of fish challenged with N. perurans alone was compared to fish challenged with a mix of all three amoeba strains. Immersion challenge of all three species resulted in minor gill lesions, with the most severe epithelial hyperplasia documented in the N. perurans treatments, while lesions with infiltrating lymphocytes were the predominate pathology observed in fish challenged by Nolandella sp. and Pseudoparamoeba sp. The presence of individual Nolandella or Pseudoparamoeba cells were not linked with lesion sites, so the precise cause of pathology remains inconclusive. Moreover, the presence of these non-N. perurans species did not significantly increase the severity of N. perurans-induced branchitis.

Overall this investigation supports N. perurans being the primary agent of AGD, yet also shows that other species of amoebae which colonise the gills of Tasmanian Atlantic salmon can dominate the gill community and may be capable of causing some gill pathology under specific conditions. Thus, the involvement of non-N. perurans amoebae in AGD should not yet be discounted. The increasing list of putative gill pathogens, and the complexity of disease expression, provides supportive rationale to the consideration of gill disease in the context of dysbiosis of microbial community structure, rather than a pathological response to a single agent.

iii

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.

Chloe English 08/10/2019

Publications included in this thesis

English, C., Tyml, T., Botwright, N., Barnes, A., Wynne, J., Lima, P., Cook, M., 2019. A diversity of amoebae colonise the gills of farmed Atlantic salmon (Salmo salar) with amoebic gill disease (AGD). Eur. J. Protistol. 67, 27–45. https://doi.org/10.1016/j.ejop.2018.10.003

English, C., Swords, F., Downes, J., Ruane, N., Botwright, N., Taylor, R., Barnes, A., Wynne, J., Lima, P., Cook, M., 2019. Prevalence of six amoeba species colonising the gills of farmed Atlantic salmon with amoebic gill disease (AGD) using qPCR. Aquac. Environ. Interact. 11, 405–415. https://doi.org/10.3354/aei00325

Submitted manuscripts included in this thesis

English, C., Lima, P., 2018. A review of amoebic disease in aquatic animals: insights into defining aetiological agents. J. Fish Dis.

Other publications during candidature

Conference abstracts

English, C.J., Botwright, N.A., Barnes, A.C., Wynne, J.W., Lima, P.C., Cook, M.T (2018). Possible multi-amoeba aetiology of amoebic gill disease (AGD) of farmed Atlantic salmon. (Oral presentation) 8th International Symposium on Aquatic Animal Health, PEI, Canada, 2nd-14th Sep 2018

English, C.J., Botwright, N.A., Barnes, A.C., Wynne, J.W., Lima, P.C., Cook, M.T (2017). The role of non-N. perurans amoeba in amoebic gill disease (AGD) of Atlantic salmon. (Oral presentation) 2017 Australian Society for Parasitology Annual Conference, Leura, Australia, 26th- 29th Jul 2017

English, C.J., Botwright, N.A., Barnes, A.C., Wynne, J.W., Lima, P.C., Cook, M.T (2017). Possible multi-amoeba aetiology of amoebic gill disease (AGD) of farmed Atlantic salmon. (Oral presentation) 4th Australian Scientific Conference on Aquatic Animal Health and Biosecurity, Cairns, Australia, 10th -14th Jul 2017. Awarded Best Student Presentation.

English, C.J., Botwright, N.A., Barnes, A.C., Wynne, J.W., Lima, P.C., Cook, M.T (2016). Investigating the role of non-N. perurans amoeba in AGD of Atlantic salmon. (Oral presentation) 4th Gill Health Initiative Meeting, University of Stirling, Scotland, 9th-10th Jun 2016.

vii

Contributions by others to the thesis

Andrew Barnes, Paula Lima, James Wynne and Mat Cook contributed to the conception of this project, advised on methods and analyses, and provided comments on the thesis and the associated publications.

Kathryn Green and Erica Lovas provided transmission electron microscopy training and helped extensively during the imaging process for Chapter 3.

Tomas Tyml provided training on phylogenetics analysis in Chapter 3.

Fiona Sword and Jamie Downes advised on how to design and validate qPCR assays for Chapter 4.

Ben Maynard and Richard Taylor sampled 50 % of salmon gill swabs for the field survey in Chapter 4. Natasha Botwright assisted with molecular biology training, resource management and sampling during the in vivo trial in Chapter 5.

Russel McCulloch provided microscopy and histology training for Chapter 3 and 5.

Joel Slinger and Chris Stratford managed the Atlantic salmon husbandry at Bribie Island Research Centre.

Moira Menzies provided molecular biology laboratory training.

Mark Adams interpreted the histopathology, scored different gill lesion morphotypes and imaged relevant pathology for Chapter 5.

viii

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

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

Research involving animal subjects

All Atlantic salmon used for the research in this thesis were approved for sampling by CSIRO Queensland Animal Ethics Committee (AEC). A copy of the ethics approval letter for AEC application A13/2015, A9/2016 and A16/2017 is in Appendix F.

Acknowledgement

I would like to acknowledge my supervisors, Mat Cook, Paula Lima, Andrew Barnes and James Wynne. I am grateful for the opportunity to undertake this research project and for their repeated support of my participation in extra activities such as attending many training courses and laboratory visits. I would particularly like to thank James Wynne for his active supervision of this project and my professional development. I will be forever grateful for the time he carved out for me. Also, thanks to Tash for her unwavering guidance; her compassion and generosity does not go unnoticed. Thanks also to my fishy friends in the Salmon Health Team. I am so proud we made the complicated task of salmon research in the sub-tropics a reality. Special mention to Joel Slinger and Chris Stratford who gracefully fulfilled the dual role of technical support crew and comedic entertainment. The Bribie crew kept work fun. The endless activities – beach trips, dress up golf days, pub crawls, ping pong tournaments and the Monodons – made me feel like I had found ‘my people’. I would also like to gratefully acknowledge those who facilitated my laboratory visits, which were by far the highlight of my candidature. In no particular order, Neil Ruane, Fiona Swords and Jamie Downes from the Marine Institute, Ireland; Iva Dyková and Tomáš Tyml from Masaryk University, Czech Republic; and Jean Payne from Australian Animal Health Laboratory. On a personal note, my family and friends are the beautiful balance in my life that makes these achievements so much sweeter. I cannot thank them enough for their love and support. Particularly Dan, for the hugs that squeeze the worries out of me and for his seemingly endless patience with my long science stories.

Financial support

This research was primarily supported by an Australian Postgraduate Award and CSIRO Agriculture and Food. Other awards received during the candidature which supported the research and my professional development include a Crawford-in-Queensland 2016 Post Graduate Research Award, FRDC Aquatic Animal Health Training Award (Grant number 2017.02), The University of Queensland School of Biological Sciences Vacation Scholarship (2017), Australian Society for Parasitology Conference Travel Grant (2017) and The University of Queensland Candidate Development Award (2019).

xii

Keywords

Amoebozoa, Salmo salar, gill disease, aquaculture, Neoparamoeba, Nolandella, Pseudoparamoeba, aetiology, dysbiosis.

Australian and New Zealand Standard Research Classification (ANZSRC)

ANZSRC code: 060307 Host-Parasite Interactions, 60 %

ANZSRC code: 070401 Aquaculture, 20 %

ANZSRC code: 060301 Animal Systematics and Taxonomy, 20 %

Fields of Research (FoR) Classification

FoR code: 0605 Microbiology, 40 %

FoR code: 0707 Veterinary Sciences, 30 %

FoR code: 0602 Ecology, 30 %

xiii

Dedication

To my Mum, Linda, the strongest yet most compassionate person I know. Who was denied her opportunity for education and instead poured every part of herself into her three kids. That is why she now has a masters, two doctorates and a whole lot of love in her life.

xiv

Table of Contents

Chapter 1: General introduction ...... 1 1.1 Expanding aquaculture and its animal health challenges ...... 1 1.2 Amoebozoans ...... 1 1.3 Amoebic gill disease (AGD) ...... 2 1.3.1 Aetiology ...... 2 1.3.2 Clinical features ...... 3 1.3.3 Diagnosis ...... 4 1.3.4 Treatment ...... 6 1.3.5 Environmental risk factors ...... 6 1.4 The challenge of understanding AGD causation ...... 7 1.5 Aims of this thesis ...... 8 1.5 Thesis significance ...... 9 Chapter 2: A review of amoebic disease in aquatic animals: insights into defining aetiological agents ...... 10 2.1 Abstract ...... 10 2.2 Introduction ...... 11 2.3 Main amoebic diseases of aquatic organisms: impact and current aetiological status ...... 12 2.4 The early days: the contribution of morphology to our understanding of amoebic disease in aquatic animals ...... 18 2.4.1 The plasticity of amoeba morphology ...... 19 2.4.2 Electron microscopy to support light microscopy-based characterisation ...... 20 2.5 Resolving amoeba taxonomy through genetics ...... 21 2.5.1 Amoebozoa barcoding genes ...... 21 2.5.2 Sanger sequencing versus next generation sequencing (NGS) when barcoding amoebae 23 2.5.3 Molecular-based taxonomy to complement morphological-based identification ...... 23 2.5.4 Molecular detection of amoebae ...... 25 2.6 Determining pathogenic amoebae ...... 26 2.6.1 The application of Koch’s postulates in aquatic amoebic disease ...... 26 2.6.2 The histological and ISH approach to define causation ...... 31 2.7 Conclusion ...... 33 Chapter 3: A diversity of amoebae colonise the gills of farmed Atlantic salmon with amoebic gill disease (AGD) ...... 34 3.1 Abstract ...... 34 3.2 Introduction ...... 35

3.3 Material and Methods ...... 36 3.3.1 Sampling AGD-affected gills to establish amoeba cultures ...... 36 3.3.2 Primary isolation and maintenance of cultures ...... 37 3.3.3 Establishing monocultures ...... 37 3.3.4 Morphological characterisation of amoebae grown in monoculture ...... 37 3.3.5 Identification of amoebae by sequencing ...... 38 3.3.6 Sequence analysis ...... 39 3.3.7 Phylogenetic analysis of 18S rRNA amoeba sequences ...... 39 3.3.8 Phylogenetic analysis of amoeba COI sequences ...... 41 3.4 Results ...... 41 3.4.1 Morphological characterisation of amoeba monocultures ...... 41 3.4.2 Genetic characterisation of mixed and monocultures ...... 48 3.4.3 Phylogenetic analysis of all amoebae detected ...... 48 3.4.4 Diversity of amoebae detected on AGD-affected gills ...... 54 3.5 Discussion ...... 56 3.6 Conclusions ...... 59 Chapter 4: Prevalence of six amoeba species colonising the gills of farmed Atlantic salmon with amoebic gill disease (AGD) using qPCR ...... 61 4.1 Abstract ...... 61 4.2 Introduction ...... 63 4.3 Material and Methods ...... 64 4.3.1 Gill swab sample collection ...... 64 4.3.2 DNA extraction from gill swabs ...... 66 4.3.3 TaqMan qPCR assay design ...... 66 4.3.4 Validation of reaction efficiency, sensitivity, specificity and reproducibility ...... 66 4.3.5 Inhibition ...... 67 4.3.6 qPCR analysis of gill swab survey ...... 68 4.3.7 Statistical analysis ...... 68 4.4 Results ...... 69 4.1 qPCR assays design and optimisation ...... 69 4.2 qPCR assays validation...... 69 4.3 Survey of amoeba prevalence and abundance ...... 71 4.5 Discussion ...... 74 4.6 Conclusions ...... 77 Chapter 5: Immersion challenge of naïve Atlantic salmon with cultured Nolandella sp. and Pseudoparamoeba sp. did not increase the severity of N. perurans-induced amoebic gill disease (AGD) ...... 79 xvi

5.1 Abstract ...... 79 5.2 Introduction ...... 80 5.3 Materials and Method ...... 82 5.3.1 Amoeba cultures ...... 82 5.3.2 Experimental design ...... 82 5.3.3 Fish stocking and system setup...... 83 5.3.4 Amoeba harvest ...... 83 5.3.5 Amoeba challenge ...... 83 5.3.6 Sampling ...... 84 5.3.7 Gill histopathology ...... 84 5.3.8 Amoeba detection on gills post-challenge by qPCR ...... 85 5.3.9 Statistical analyses ...... 86 5.4 Results ...... 86 5.4.1 Confirmation of the identity of amoeba inoculum ...... 86 5.4.2 Survival during infection trial ...... 87 5.4.3 Gill pathology ...... 87 5.4.4 Amoeba detection on gills post-challenge ...... 92 5.5 Discussion ...... 94 Chapter 6: General discussion...... 98 References ...... 107 Appendices ...... 127 Appendix A: Gross morphology dimensions of amoeba strains grown in monoculture ...... 127 Appendix B: qPCR assay optimisation and validation ...... 128 Appendix C: In situ hybridisation of N. perurans ...... 136 Appendix D: Next generation sequencing attempt...... 140 Appendix E: Salinity tolerance of Vannella sp...... 142 Appendix F: Ethics approval ...... 143

xvii

List of tables

Table 1.1 Gross gill score system used by salmon industry to assess the severity AGD pathology (Taylor et al., 2009)...... 5 Table 2.1 Amoebic diseases reported in aquatic animals...... 13 Table 3.1 Universal Eukaryotic 18S rRNA and COI primer sets used to amplify amoebae...... 38 Table 3.2 Summary of all amoeba species/strains detected on the gills of Atlantic salmon Salmo salar farmed in Tasmania with signs of AGD, and their method of detection and characterisation. 55 Table 4.1 Case history of sampled Atlantic salmon. NA is not available because no samples could be collected at that sampling event...... 65 Table 4.2 Probe and primer sequences and assay parameters for detecting various amoeba taxa. ... 69 Table 4.3 Summary of validation metrics for the six amoeba Taqman qPCR assays. LOD is limit of detection and CV is coefficient of variation...... 70 Table 4.4 Specificity of the six qPCR assays (Nolandella spp. (Nol), Pseudoparamoeba sp. (Pse), P. eilhardi (ParE), Vannellida species (VanC2 and VanC3) and N. perurans (NPJ)) determined by testing each assay against various amoeba DNA samples listed as taxa (strain). Positive detections are indicated in bold with a ‡ symbol. The – symbol is undetected...... 71 Table A.1 Gross morphology dimensions of amoeba strains grown in monoculture. Presented as mean values (M) with corresponding range (R) and standard deviation (SD). All values in µm. .. 127 Table B.1 Optimisation of primer concentration for qPCR assays designed to detect Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2) and Vannellida C3 (VanC3). Results presented as mean Ct value and standard deviation (SD). Values in bold with a ‡ symbol indicate the chosen concentration...... 128 Table B.2 Optimisation of probe concentration for qPCR assays designed to detect Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2) and Vannellida C3 (VanC3). Results presented as mean Ct value and standard deviation (SD). Values in bold with a ‡ symbol indicate the chosen concentration...... 129 Table B.3 Limit of detection (LOD) of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). The Ct LOD is in bold, which was then converted to its equivalent number of 18S rRNA copies. und is undetermined...... 131 Table B.4 Reproducibility of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). Ct values of ten positive gill swabs were tested in triplicate on

xviii three separate days and the intra- and inter-assay variance was evaluated via the mean Ct ± standard deviation (SD) and the coefficient of variation (CV). The p-value value from a one-way ANOVA shows there was no significant difference in Ct between the assays conducted on different days. 133 Table D.1 Universal Eukaryotic 18S rRNA primer sets used for NGS...... 140

xix

List of figures

Figure 1.1 Gill pathology associated with AGD of Atlantic salmon. (a) Visual assessment of Atlantic salmon gills displaying white mucoid AGD lesions. (b) Typical Atlantic salmon gill histology. Scale bar = 50 µm. (c) AGD histopathology showing fusion of several secondary lamellae and attached N. perurans (arrow). Scale bar = 50 µm...... 4 Figure 1.2 Gill arches from AGD-infected Atlantic salmon and the corresponding gill score...... 5 Figure 1.3 Gross morphology of amoeba trophozoites viewed under light microscopy (a) The three main forms of amoeba trophozoites (attached, floating, pseudocyst) of two genera, Vannella and Neoparamoeba. All scale bars are 20 µm. (b) Similarity between the gross morphology of attached trophozoites from two different amoeba species, N. pemaquidensis (SEDCT1/I) and N. branchiphila (ST4N/I). Images from Dyková et al. (2005b)...... 8 Figure 3.1 Light microscopy of cultured amoeba strains isolated from gills of farmed Atlantic salmon Salmo salar with signs of AGD, including (a) attached trophozoite and (b) floating form. Strain MP1 and MP2 is Neoparamoeba perurans, MX6 is Vexillifera sp., MX1 is Pseudoparamoeba sp., MV5, MV2, MV3 and MV4 is Vannella sp., MX4 is Tubulinea-like Amoebozoa, MX3 and MX5 is Nolandella sp. All scale bars = 20 µm...... 42 Figure 3.2 Ultrastructure of Neoparamoeba perurans (strain MP1) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n) adjacent to endosymbiont (en), mitochondria (m), golgi apparatus (g), vesicles (v), vacuole (va), phagosome (p). (b) Ultrastructure of Perkinsela amoebae-like endosymbiont (en) adjacent to trophozoite nucleus (n) and golgi apparatus (g): nucleus of endosymbiont (nu) and mitochondrion (m) with darkly stained kinetoplast DNA. (c) Cell surface with a very thin amorphous glycocalyx (arrows)...... 43 Figure 3.3 Ultrastructure of Vexillifera sp. (strain MX6) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n), mitochondria (m), phagosome (p), vacuoles (va) vesicle (v). (b) Cell surface with approximately 60 nm thick glycocalyx made up of cylinder-like glycostyles (arrows). (c) Mitochondria with tubular cristae...... 44 Figure 3.4 Ultrastructure of Pseudoparamoeba sp. (strain MX1) isolated from gills of Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructures: nucleus (n), mitochondria (m), golgi apparatus (g) vesicles (v), phagosome (p). (b) Trophozoite cell surface lined with domed scales (arrows). (c) Golgi apparatus with parallel arrangement of cisternae. (d) Mitochondria with branching tubular cristae...... 45

Figure 3.5 Ultrastructure of Vannella sp. (strain MV3) isolated from gills of Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n), mitochondria (m), vesicles (v), early-stage phagosome (p), late-stage budding phagosome (bp). (b) Mitochondria with tubular branching cristae. (c) Golgi apparatus with parallel arrangement of cisternae. (d) Cell surface with amorphous glycocalyx (arrows)...... 46 Figure 3.6 Ultrastructure of Vannella sp. (strain MV4) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: mitochondria (m), vacuoles (va) vesicle (v). (b) Cell surface with amorphous glycocalyx (arrows). (c) Mitochondria with tubular branching cristae. (d) Vesicular nucleus...... 47 Figure 3.7 Ultrastructure of Nolandella sp. (strain MX5) isolated from gills of Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructures: vesicular nucleus (n), mitochondria (m), vacuoles (va), vesicles with unknown content (v), endoplasmic reticulum (e). (b) Granular endoplasmic reticulum encircles mitochondria with tubular cristae. (c) Trophozoite cell surface with a 25 nm thick glycocalyx (arrows) covering the cell membrane...... 48 Figure 3.8 Maximum likelihood analysis of taxa from Tubulinea 18S rRNA gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Echinamoebidia and Leptomyxida serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures...... 49 Figure 3.9 Maximum likelihood analysis of taxa from Discosea 18S rRNA gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Centramoebia serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures...... 51 Figure 3.10 Maximum likelihood analysis of taxa from Amoebozoa COI gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Fungi serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures...... 53 Figure 4.1 Atlantic salmon sea pen sites, Killala and Tasman, sampled during the survey...... 65 Figure 4.2 (a-g) Prevalence of Atlantic salmon gills that were colonised by N. perurans (NPJ), Nolandella spp. (Nol), Pseudoparamoeba sp. (Pse), and Vannellida (VanC2, VanC3) at each farm xxi site (Killala, Tasman) over a year. P. eilhardi was not included because no Atlantic salmon tested positive for the amoeba species. The average gill score (Gill Index (GI)) for each site and each time point is marked in the respective panels. The Feb-18, Killala sampling event marked with ‡ is blank because no samples could be collected...... 72 Figure 4.3 Relative abundance of N. perurans on Atlantic salmon gills expressed as the mean ± standard error of 18S rRNA copies with respect to (a) AGD gross gill pathology (gill score) quantified according to Taylor et al. (2009) and (b) farm site, and sample time. The Feb-18, Killala sampling is blank because no samples could be collected...... 73 Figure 5.1 Confirmation of the identity of the amoeba cultures used to experimentally infect naïve Atlantic salmon. (a) Light microscopy images of Nolandella sp. (Nol) strain MX5, Pseudoparamoeba sp. (Pse) strain MX1 and N. perurans (Neo) strain MP2. All scale bars = 20 µm. (b) qPCR amplification of cultured Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse) and N. perurans (Neo) DNA using the three specific qPCR assays...... 87 Figure 5.3 H&E stained gill histopathology from Atlantic salmon after exposure to cultured amoebae. (a) Anatomically normal primary lamella (PL) and secondary lamellae (SL), scale bar = 100 µm. Insert border (dashed lines) corresponds to (b) anatomically normal secondary lamellae at 60x magnification showing pavement cells (pv), mucus cells (m), erythrocytes (e), chloride cells (c) and pillar cells (p). Scale bar = 30 µm. (c) Epithelial hyperplasia with fusion of secondary lamellae (white arrows) closely associated with N. perurans trophozoites (t). Also note presence of leucocytes within the central venous sinus (csv), scale bar = 100 µm. Insert border (dashed lines) corresponds to (d) hyperplastic lesion at 60x magnification showing N. perurans trophozoite (t) adjacent to fused secondary lamellae (white arrow). Mucus cells (m), lymphocytes (L), macrophages (ma) and undifferentiated epithelial cells (e) also present, scale bar = 30 µm. (e) Interlamellar nodule showing two lamellae fused predominately with lymphocytes (L), scale bar = 30 µm. (f) Distal lamellar nodule showing infiltration of predominately lymphocytes (L) at the outer end of the secondary lamellae, scale bar = 50 µm...... 90 Figure 5.4 Gill histopathology from Atlantic salmon after exposure to cultured Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), N. perurans (Neo) and a mix of all three amoeba species (Mix). These treatments were compared to a no amoeba control (Ctl). The mean percentage of lesion- affected filaments in Atlantic salmon before (Time 0) and after exposure to amoebae at 0.5, 7, 14 and 21 days post infection (DPI) were graphed for (a) all lesion morphotypes and then split into the most common lesion morphotypes, including (b) epithelial hyperplasia with lamellar fusion, (c) distal lamellar nodules and (d) interlamellar nodules and plaques. Level of statistical significance indicated by p value * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001...... 91

xxii

Figure 5.5 Prevalence (mean ± SEM) of Atlantic salmon with positive qPCR detections of either (a) Nolandella sp. in the Nol treatment, (b) Pseudoparamoeba sp. in the Pse treatment, (c) N. perurans in the Neo treatment and (d) all three amoeba species in the Mix treatment...... 93 Figure B.1 Standard curve involving amplification of 10-fold dilution of plasmid DNA specific to (a) Nolandella sp. (Nol), (b) Pseudoparamoeba sp. (Pse), (c) Paramoeba eilhardi (ParE), (d) Vannellida C2 (VanC2), (e) Vannellida C3 (VanC3) and (f) Neoparamoeba perurans (NPJ). E is amplification efficiency...... 130 Figure B.2 Precision of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). The final dilution, which determined the LOD of each assay, was tested a further 20 times to determine the precision of the assay at a 95% confidence level. The mean ± standard deviation (SD) is listed above each plot. VanC2 mean±SD was calculated with the outlier excluded...... 132 Figure B.3 Test for the presence of PCR inhibitors using the Neoparamoeba perurans (NPJ) assay. Five gill swab DNA samples were diluted two-fold then spiked with the same amount of N. perurans plasmid, then tested in triplicate. The PCRs with the undiluted sample contained 150 ng of gill swab DNA within a 5 µl reaction. If PCR inhibitors were present the Ct value would be higher in lower gill swab DNA dilutions...... 135 Figure C.1 Oligonucleotide probes that hybridise to 18S rRNA of N. perurans attached to AGD- affected Atlantic salmon gills. (a, b) Results from the Young et al. (2007) ISH assay. (c, d) Results from the adapted Downes et al. (2015) ISH assay. Scale bars; a, c = 100 µm, b, d = 50 µm...... 138 Figure E.1 Morphology of Vannella sp. trophozoites after five days in malt yeast broth with salinity ranging from 0 - 42 parts per thousand (ppt) ...... 142

xxiii

List of abbreviations

AEC Animal ethics committee AGD Amoebic gill disease AGRF Australian Genome Research facility ANOVA Analysis of variance ANZSRC Australian and New Zealand Standard Research Classification AQUI-S Aquatic anaesthetic ATCC American Type Culture Collection BGD Bacterial gill disease BLAST Basic local alignment search tool BLASTn Basic local alignment search tool nucleotide BP Bayesian probability CCAP Culture Collection of Algae and Protozoa cm Centimetre COI Cytochrome oxidase subunit I CSIRO Commonwealth Scientific and Industrial Research Organisation Ct Cycle threshold CV Coefficient of variation DDBJ DNA Data Bank of Japan DNA Deoxyribonucleic acid DPI Days post infection E Amplification efficiency EMBL European Molecular Biology Laboratory e-value Expect value FAO The Food and Agriculture Organization FoR Fields of Research FRDC Fisheries Research and Development Corporation g Gram GCD Grey crab disease GI Gill Index h Hour IQ Intelligence quotient ISH In situ hybridisation

xxiv

ITS Internal transcribed spacer kV Kilovolt l Litres LB Luria-Bertani LM Light microscopy LOD Limit of detection LSU Large subunit MAFFT Multiple alignment using fast fourier transform mg Milligrams MGB Minor groove binder min Minute ML Maximum likelihood mm Millimetre MYB Malt yeast broth NA Not available NCBI National Center for Biotechnology Informatio ng Nanogram NGD Nodular gill disease NGS Next Generation Sequencing nm Nanometre nM Nanomolar Nol Nolandella NPJ Neoparamoba perurans qPCR assay ParE Paramoeba elhardi PCR Polymerase chain reaction PEI Prince Edward Island ppt Parts per thousand Pse Pseudoparamoeba p-value Calculated probability qPCR quantitative PCR R2 Coefficient of determination rRNA Ribosomal ribonucleic acid SD Standard deviation SILVA From Latin silva, forest

xxv

SSU Small subunit µl Microliter µm Micrometre

xxvi

Chapter 1: General introduction

1.1 Expanding aquaculture and its animal health challenges

Seafood consumption is increasing as the global human population grows and consumers seek higher value, protein rich food (World Bank, 2013). With many fisheries in decline, the growing demand for seafood is being met by aquaculture (FAO, 2018). Aquaculture has expanded quickly, increasing from 1.6 million tonnes of annual production in 1960 to 80 million tonnes in 2016 (FAO, 2018). As aquaculture production expands and intensifies, the industry faces many challenges that impede its growth and sustainability. Disease is a major cause of lost production. It also results in the use of potentially harmful treatments within the industry. Reviews by Meyer, (1991) and Bondad-Reantaso et al. (2005) depict how detrimental and widespread the impact of bacterial, viral, fungal and parasitic diseases have been across a variety of farmed aquatic animals.

Amoebic gill disease (AGD) is an expanding parasitic condition severely impacting several species of farmed fish (Oldham et al., 2016). Current treatment and prevention strategies for amoebic gill disease are not adequate if prevalence continues to rise. Greater understanding of the causes of infestation and the associated pathology is needed to help manage this ongoing health risk.

1.2 Amoebozoans

Amoebozoans are a diverse and abundant group of protozoans that are ubiquitously distributed throughout land and water (Rodríguez-Zaragoza, 1994; Smith et al., 2008). While Amoebozoans are generally free-living, several species of amoebae are amphizoic, i.e. capable of being free-living protozoa or parasites (Martinez and Visvesvara, 1997). Several amoebae have been described as agents of disease in marine animals. The most widely recognised pathogenic species belong to the genus Neoparamoeba (Page, 1987) (Nowak and Archibald, 2018). However, these amoebae have been isolated from diseased, asymptomatic and healthy hosts making it difficult to ascertain the parasitic, mutualistic or symbiotic relationship each amoeba species has with its hosts (Bermingham and Mulcahy, 2007; Dyková and Lom, 2004; Dyková and Tyml, 2015).

Therefore, determining the causative agent of amoebic diseases is often an important, unresolved question within aquaculture and fisheries.

1.3 Amoebic gill disease (AGD)

AGD is a form of branchitis which affects a diversity of marine teleosts across a large geographical range. Farmed salmonids are the most susceptible species to the disease, which has led to extensive commercial impact on salmon aquaculture (Oldham et al., 2016). AGD was first reported in farmed coho salmon in Washington in 1985 and soon after in farmed Atlantic salmon in Tasmania, Australia (Kent et al., 1988; Munday, 1986). Since then, the disease has been found in all major salmon producing countries, except Iceland. Crude mortality rates typically range from 10- 50%, but loss of up to 80% has been reported in Norway (Douglas-Helders et al., 2001; Rodger and Mcardle, 1996; Steinum et al., 2008). AGD is the most significant health problem in the marine phase of Atlantic salmon production in Australia.

1.3.1 Aetiology

AGD is attributed to Neoparamoeba perurans, Young et al., 2007, a marine, amphizoic ectoparasite belonging to the family Paramoebidae (Crosbie et al., 2012). A defining feature of this family is its obligate nucleus-associated endosymbiont that is curiously related to a eukaryotic ectoparasite of fish (Dyková et al., 2003). Neoparamoeba perurans is also referred to as Paramoeba perurans because phylogenetic analysis based on small subumit ribosomal rRNA gene (SSU rRNA) showed that the genera Neoparamoeba and Paramoeba were nested and, therefore, Neoparamoeba should be treated as a junior synonym of Paramoeba (Feehan et al., 2013). This proposed name change was deemed premature by Young et al. (2014), who advised there is an insufficient number of sequences from Paramoeba sp. and that the highly conserved natured of SSU rRNA makes it an unsuitable region to separate two genera that are differentiated by only the presence (Paramoeba Schaudinn, 1896) or absence (Neoparamoeba Page, 1987) of microscales. There has been continuous disagreement within the AGD research community about the most appropriate nomenclature, based on either molecular or morphological characteristics, and a consensus is yet to be reached. As a result, both genera are still employed by research groups worldwide. Since I share the same opinion as Young et al. (2014), that more evidence is required before the genera are synonymised, this thesis uses Neoparamoeba perurans.

Determining the aetiological agent of AGD took almost two decades of research, with the agent originally considered to be a Paramoeba pemaquidensis (Kent et al., 1988), which was later recognised as Neoparamoeba pemaquidensis (Dyková et al., 2000). After N. pemaquidensis failed to elicit gill lesions in a controlled fish trial, it was proposed that AGD could in fact have multiple aetiological agents, including closely related Neoparamoeba branchiphilia (Morrison et al., 2005), or a diversity of other amoeba genera isolated from AGD-affected Atlantic salmon in Ireland (Bermingham and Mulcahy, 2007). Likewise, cultured N. branchiphilia did not elicit AGD pathology when experimentally exposed to Atlantic salmon (Vincent et al., 2007). During the same era of research, a new species of amoeba, N. perurans, was found associated with AGD lesions via species-specific in situ hybridisation (ISH) (Young et al., 2007). A clonal culture of this new species was established, and Koch’s postulates were fulfilled via infection resulting in AGD pathology, then re-isolation and identification of N. perurans from diseased Atlantic salmon (Crosbie et al., 2012). Since this focal study, N. perurans has frequently been associated with amoebic gill disease across a variety of locations and fish species (Bustos et al., 2011; Crosbie et al., 2010b; Mouton et al., 2014).

Even though AGD has long been considered to be caused by a single agent, multiple other amoeba species such as N. branchiphila, N. pemaquidensis, Vannella sp., Flabellula sp., Vexillifera sp., Mayorella sp. and Nolandella sp. can be found alongside N. perurans on the gills of Atlantic salmon (Bermingham and Mulcahy, 2007; Dyková et al., 2005; Howard, 2001). The pathogenic potential of these amoebae has not yet been determined. However, considering some of the amoeba species listed above have been associated with other aquatic amoebic diseases (Leiro et al., 1998; Sawyer et al., 1978; Taylor, 1977; Webb et al., 2002) the possibility of AGD having a mixed aetiology should not be ruled out. Despite this, there has been little consideration of the role other Amoebozoa colonising Atlantic salmon gills play in AGD onset or progression.

1.3.2 Clinical features

Signs of AGD include respiratory distress, flared opercula, reduced appetite, lethargy and ultimately mortality if left untreated (Rodger and Mcardle, 1996; Steinum et al., 2008). Gross AGD gill lesions appear as raised, white mucoid spots or more extensive patches (Figure 1.1a) (Adams et al., 2004; Steinum et al., 2008). Through histology, diseased gills present as epithelial hyperplasia, lamellar fusion, excess mucous cells and the presence of N. perurans (Figure 1.1c) (Adams and Nowak, 2001). 3

1.3.3 Diagnosis

AGD is clinically diagnosed by identifying the appropriate gill lesions and Neoparamoeba spp. through histology (Bustos et al., 2011). Histology is considered the gold standard diagnosis for AGD; however, it is time consuming, labour intensive and requires sacrificial sampling of fish. These disadvantages make histology an impractical method for on-farm, mass diagnosis. Numerous other diagnostic methods have been developed to aid AGD management on-farm.

Gill scoring is the classification of AGD-like gross gill pathology and is routinely used by several salmon farming companies for rapid, population-based diagnostics and monitoring (Taylor et al., 2009). Gills are scored on a scale of zero (no lesions) to five (lesions covering > 50 % of gill surface) (Table 1.2, Figure 1.2). The average gill score is used by industry to indicate when AGD treatment is needed. The method is a presumptive diagnosis as it only measures host pathology

4 response, so some gross gill lesions quantified could be caused by issues other than AGD. However, AGD is the primary gill pathogen in Tasmanian Atlantic salmon aquaculture and macroscopic clinical signs have moderate to good conformity with histological lesions (Adams et al., 2004) and N. perurans load (Bridle et al., 2010) so gill scoring is a viable tool for industry and research in Tasmania.

Table 1.1 Gross gill score system used by salmon industry to assess the severity AGD pathology (Taylor et al., 2009). Infection Gill score Gross description clear 0 no sign of infection and healthy red colour very light 1 1 white spot, light scarring or undefined necrotic streaking light 2 2–3 spots/small mucus patch

established thickened mucus patch / spot groupings up to 20% moderate 3 of gill area advanced 4 established lesions covering up to 50% of gill area heavy 5 extensive lesions covering most of the gill surface

Figure 1.2 Gill arches from AGD-infected Atlantic salmon and the corresponding gill score.

Molecular detection of N. perurans in conjunction with associated gross gill pathology is increasingly used to confirm AGD. Molecular detection of N. perurans include conventional PCR (Young et al., 2008a), quantitative PCR (qPCR) (Bridle et al., 2010; Downes et al., 2015; Fringuelli et al., 2012) and ISH (Young et al., 2007). These methods are based on the 18S rRNA gene, a sequence region routinely used for determining eukaryote identity (Hugerth et al., 2014; Schroeder

5 et al., 2001; Wong et al., 2004). Quantitative PCR, which is a highly sensitive and specific technique, is starting to be used as an early warning and monitoring tool, particularly in the initial stages of infection when gross gill pathology is absent (Downes et al., 2015).

1.3.4 Treatment

There are two commercially available treatments for AGD, freshwater bathing and hydrogen peroxide (H2O2). Freshwater bathing for two to four hours is the main treatment used in Atlantic salmon aquaculture in Tasmania (Parsons et al., 2001). The osmotic effect of bathing removes amoebae from the gills and promotes healing of lesions (Clark et al., 2003). It is currently the most effective AGD management strategy. However, reinfection is inevitable, with amoeba numbers returning to pre-bath levels sometimes within 10 days post-bath (Clark et al., 2003). This practice is also expensive, labour intensive and limited by scarce freshwater availability (Adams et al., 2012). Consequently, freshwater bathing is not considered a viable long-term solution to AGD, particularly as sites for rearing salmon move further offshore.

Submersion in hydrogen peroxide (H2O2) is used in several finfish aquaculture industries to treat a range of parasitic and bacterial infections, and some commercial salmon producers have had success using it to treat AGD (Mansell et al., 2005; Rach et al., 2000; Treasurer et al., 2000). H2O2 is primarily used in Scotland, Ireland and Norway, where it is routinely used to treat sea lice, a major health issue for these aquaculture regions. Typical dosage is 1000-1400 mg/l for 18-22 min

(Oldham et al., 2016). H2O2 is highly reactive with organic matter, then breaks down to harmless by-products, water and oxygen. H2O2 treatment is only a safe and effective treatment for salmon at temperatures below 13.5°C on fish with moderate AGD (Adams et al., 2012; Oldham et al., 2016).

H2O2 is therefore not an appropriate AGD treatment in Tasmania where water tempertures often exceed 14°C.

1.3.5 Environmental risk factors

High temperature and salinity are considered the main environmental risk factors contributing to AGD prevalence and severity (Clark and Nowak, 1999; Oldham et al., 2016). AGD prevalence in Tasmania has been shown to peak in summer, followed by a second spike in autumn, and these periods correlate with higher water temperature and salinity (Clark and Nowak, 1999). In

6 vitro experiments also support this observation, as peak N. perurans growth occurred at 15°C and a salinity of 35 ppt (Collins et al., 2019), which is broadly reflective of ocean conditions during summer in Tasmania. However, N. perurans is highly adaptable to changes in environmental conditions (Collins et al., 2019; Lima et al., 2016), and is infectious over a range of temperatures and salinities (Clark and Nowak, 1999). Other factors proposed to influence AGD onset and severity are salmon stocking density (Crosbie et al., 2010a; Wright et al., 2017), and the bacterial community on salmon gills (Embar-Gopinath et al., 2005). While water temperature and salinity play an important role in AGD outbreaks, all the drivers of infestation and variables associated with gill pathology are not yet understood.

1.4 The challenge of understanding AGD causation

AGD is multifactorial and there is great complexity around establishing its aetiology. Past difficulties in defining N. perurans as the causal agent of AGD (section 1.3.1) reflect some of the practical challenges faced when studying amoebic disease in the aquatic environment. For example, amoeba biology has a long history of conflicting taxonomic and phylogenetic classification, mostly due to the inaccuracy of identifying amoeba based on morphology alone (Foissner and Korganova, 1995; Smirnov, 2002). Amoeba morphology is inherently plastic, as a single species can assume many forms (Figure 1.3a) and different species (Figure 1.3b), even genera, often look very similar. Other challenges include difficulty isolating and culturing amoebae, particularly as monocultures. Monocultures are useful for accurately identifying amoebae, as they provide a sure way of linking both morphological and molecular taxonomic information. In vitro monocultures can also be used to develop molecular detection and diagnostic tools which aid on-farm and experimental disease investigation. In vitro cultures are particularly important for investigating the pathogenicity of single putative agents through controlled challenge experiments. Prior to this project, the non- Neoparamoeba species associated with AGD had only been identified through morphology (Bermingham and Mulcahy, 2007; Howard, 2001). Moreover, in vitro cultures had not been maintained long-term without antibiotics and molecular detection tools had not been developed for the diversity of other amoeba species which colonise the gills of AGD-affected Atlantic salmon. Hence, the exact identity and role these accompanying species play in AGD is unknown.

Figure 1.3 Gross morphology of amoeba trophozoites viewed under light microscopy (a) The three main forms of amoeba trophozoites (attached, floating, pseudocyst) of two genera, Vannella and Neoparamoeba. All scale bars are 20 µm. (b) Similarity between the gross morphology of attached trophozoites from two different amoeba species, N. pemaquidensis (SEDCT1/I) and N. branchiphila (ST4N/I). Images from Dyková et al. (2005b).

1.5 Aims of this thesis

The main purpose of this PhD thesis was to investigate the role different amoeba species play alongside N. perurans on the gills of Atlantic salmon with AGD. This overarching goal was broken down into four separate aims as follows.

1. Understand the knowledge gaps and methodology of previous investigations into the aetiology of amoebic disease in aquatic animals by conducting a literature review. 2. Document the diversity of amoebae colonising the gills of AGD-affected farmed Atlantic salmon in Tasmania using a combination of morphological and sequence-based taxonomic methods. 8

3. Develop and apply multiple species-specific qPCR assays to profile amoeba prevalence and abundance on the gills of farmed Atlantic salmon and identify relationships between specific amoeba taxa and farm sites, time points and gill pathology. 4. Experimentally assess the pathogenicity of select amoeba species against the gills of Atlantic salmon.

1.5 Thesis significance

This research contributes to general amoeba systematics and biology. It also explores host- pathogen interactions of Atlantic salmon naturally infected with amoebae on-farm and experimentally infected through challenge trials. In the case that non-N. perurans species are found to be pathogenic, or play a role in AGD development, this finding may alter AGD treatment and management strategies. Considering many species show tolerance to salinity lower than 35 ppt (Smirnov, 2007), it is possible many of the associated species are unaffected by the current freshwater AGD treatment. Alternatively, if no non-N. perurans species prove to be parasitic, this raises interesting questions around the specific pathogenic mechanisms of N. perurans.

Notes on chapter style

Chapters two to five of this thesis present original findings that have been written in a format suitable for publication in scientific journals. Hence, each of these chapters can be read as an individual paper, but collectively they form the thesis. A statement describing the authors contributions is included at the start of each chapter. All tables and figures are embedded in the text of relevant chapters and cited literature is consolidated and listed at the end of the thesis.

Chapter 2: A review of amoebic disease in aquatic animals: insights into defining aetiological agents

Statement of author contributions

C.E. conceptualised the review aims and wrote the manuscript under the close guidance of P.L. Both authors contributed to reviewing and editing the final manuscript.

This work was submitted to the Journal of Fish Diseases on the 20th of March 2018 and was denied publication on the 26th April 2018. This version incorporates many of the reviewer’s comments and we intend to further amend the manuscript and resubmit later.

2.1 Abstract

Disease caused by parasitic amoebae impacts a range of aquatic organisms including finfish, crustaceans, echinoderms and molluscs. Despite the significant economic impact caused in both aquaculture and fisheries, the aetiology of most aquatic amoebic diseases is uncertain, which then affects diagnosis, treatment and prevention. The main factors hampering research effort in this area are the confusion around amoeba taxonomy and the difficulty proving a particular species causes specific lesions. These issues stem from morphological and genetic similarities between cryptic species and technical challenges such as establishing and maintaining pure amoeba cultures, scarcity of Amoebozoa sequence data, and the inability to trigger pathogenesis under experimental conditions. This review provides a critical analysis of how amoebae are identified and defined as aetiological agents of disease in aquatic animals and highlights gaps in the available knowledge regarding determining pathogenic Amoebozoa.

Keywords paramoebiasis; aetiological agent; amoebae; aquaculture; fisheries

2.2 Introduction

Amoebozoa Lühe 1913 are a diverse and abundant group of protozoans that are widely distributed throughout land and water (Rodríguez-Zaragoza, 1994; Smith et al., 2008). While most amoebae are free living, some species are parasitic, and several are classified as amphizoic due to their ability to exist either as free-living organisms or as facultative parasites (Martinez and Visvesvara, 1997).

Many species of Amoebozoa have been described as agents of disease in aquatic vertebrates and invertebrates of economic relevance to both aquaculture and fisheries (Bower et al., 1992; Feehan et al., 2013; Mullen et al., 2004; Oldham et al., 2016, Powell et al. 2008). However, morphological plasticity and genetic similarity between amoeba species has contributed to the difficulty in accurately identifying causative agents to species level. Moreover, the ability of particular amoebae to colonise diseased, asymptomatic and healthy hosts has further confused which agent is linked to certain amoebic diseases (Bermingham and Mulcahy, 2007; Dyková and Lom, 2004; Dyková and Tyml, 2015). As a result, the true aetiology of most aquatic amoebic diseases remains unresolved (Nowak and Archibald, 2018).

Identification of the true aetiological agent(s) is crucial for commercially relevant species, as understanding the biology of the parasite guides the development of effective management strategies. Unfortunately, most reports on aquatic amoebic diseases employ outdated methods for identifying the causative amoeba, and the aetiology of the disease is often based on inadequate evidence. Despite these problems, research efforts are often devoted to developing treatment strategies while the correct causative agent(s) is yet to be determined. Therefore, the adoption of appropriate amoeba identification techniques followed by validation of causative relationships between amoeba and host must be prioritised, so that diagnosis, treatment and prevention is based on the precise agent(s).

The current chapter critically reviews the methods used to define causation and the current level of proof supporting the aetiology of the five main aquatic amoebic diseases. This analysis highlights knowledge gaps, conflicts and practical limitations related to determining whether an amoeba is pathogenic. This will help guide researchers to overcome aetiology-associated issues relevant to economically significant aquatic amoebic diseases.

2.3 Main amoebic diseases of aquatic organisms: impact and current aetiological status

Amoebae have evolved as highly successful pathogens in a variety of marine and freshwater animals. The five most documented aquatic amoebic diseases are amoebic gill disease (AGD), nodular gill disease (NGD), sea urchin paramoebiasis, grey crab disease (GCD) and lobster paramoebiasis. AGD followed by NGD are by far the most researched aquatic amoebic diseases, presumably due to their significant economic impact on the salmonid aquaculture industry worldwide (Dyková and Tyml, 2015; Munday et al., 2001; Oldham et al., 2016; Quaglio et al., 2016; Speare, 1999). In contrast, fundamental knowledge on amoebic disease affecting other, less industry-relevant organisms such as sea urchins, crabs and lobsters is still scarce. Examples around aetiology will be drawn primarily from these five main diseases, but also from the less common cases of amoebae parasitising the internal organs of fish (Athanassopoulou et al., 2002; Dyková et al., 1997; Taylor, 1977) and molluscs (Bower et al., 1992; Cheng, 1970; Suhnel et al., 2014). A summary of amoeba-related conditions reported in aquatic organisms can be found in Table 2.1.

Table 2.1 Amoebic diseases reported in aquatic animals.

Koch’s postulates Infected tissue Pathogen Reference Disease Host species Amoeba pathogen Attempted Location or organ identification 1 2 3 4 Atlantic salmon, Neoparamoeba Gill Morphology, Australia, Bustos et al., 2011; P. Crosbie et al., 2012; Salmo salar perurans † Genetic Chile, Oldham et al., 2016; Norway, Rodger and Mcardle, Ireland, 1996; Steinum et al., 2008; Young et al., Scotland, 2008b Canada, x x x x USA, Spain, South Africa, Faroe Islands Neoparamoeba sp., Gill Morphology Ireland Bermingham and Mulcahy, 2007 Platyamoeba sp.

(now Vannella sp.), Flabellula sp., x x Vexillifer sp., Nolandella sp., Mayorella sp.

Mitchell and Rodger, Amoebozoa Gill Not specified x France 2011 Rainbow trout, Paramoeba sp. Gill Morphology Australia Munday et al., 1990 Oncorhynchus x Amoebic mykiss gill disease Chinook salmon, Neoparamoeba Gill Genetic New Young et al., 2008b Oncorhynchus perurans x Zealand tshawytscha Coho salmon, Neoparamoeba Gill Morphology USA Kent et al., 1988 x § Oncorhynchus pemaquidensis kisutch Neoparamoeba Gill Genetic Chile Rozas-Serri et al., x 2012 perurans Ayu, Plecoglossus Neoparamoeba Gill Morphology, Japan Crosbie et al., 2010b x altivelis perurans Genetic Turbot, Neoparamoeba Gill Genetic South Mouton et al., 2014 x Scophthalmus perurans Africa maximus Paramoeba sp. Gill Morphology x x Spain Dyková et al., 1998a Ballan Wrasse, Neoparamoeba Gill Morphology, Norway Karlsbakk et al., 2013 x § Labrus bergylta perurans Genetic European Seabass, Amoebozoa Gill Morphology Portugal Santos et al., 2010 Dicentrarchus x labrax Sharpsnout Neoparamoeba sp. Gill Morphology Spain Dyková and Novoa, 2001 seabream, x x Diplodus puntazzo Lumpfish, Neoparamoeba Gill Morphology, Norway Haugland et al., 2017 Cyclopterus perurans Genetic x § x x lumpus

Horse mackerel, Neoparamoeba Gill Genetic Scotland Stagg et al., 2015 Trachurus perurans x trachurus Black seabream, Neoparamoeba Gill Morphology, Korea Kim et al., 2017 Acanthopagrus perurans Genetic x schlegelii Rock bream, Neoparamoeba Gill Morphology, Korea Kim et al., 2017 Oplegnathus perurans Genetic x fasciatus Olive flounder, Neoparamoeba sp. Gill Morphology, Korea Kim et al., 2005 Paralichthys Genetic x olivaceus Halibut, Neoparamoeba Gill Morphology, Scotland Roger et al., 2019 Hippoglossus perurans Genetic x hippoglossus Blue warehou, Neoparamoeba sp. Gill Morphology Australia Adams et al., 2008 x Seriolella brama Rainbow trout, Rhogostoma minus Gill Morphology, Czech Dyková and Tyml, x x 2015 Oncorhynchus Genetic Republic mykiss Naegleria sp. Gill Morphology, Germany Dyková et al., 2010 x x Genetic Cochliopodium sp. Gill Morphology x USA Noble et al., 1997 Amoebozoa Gill Morphology Italy, Antychowicz, 2007; Buchmann et al., Poland, x 2004; Quaglio et al., Denmark, 2016; Speare, 1999 Canada Rainbow trout, Thecamoeba Gill Morphology USA Sawyer et al., 1974 Nodular x Salmo gairdneri hoffmani gill disease Daoust and Ferguson, Cochliopodiidae Gill Morphology x Canada 1985 Speare and Ferguson, Amoebozoa Gill Morphology x Canada 1989 Chinook salmon, Cochliopodium sp. Gill Morphology New Tubbs et al., 2010 Oncorhynchus x Zealand tshawytscha Arctic Char, Amoebozoa Gill Morphology Canada Speare, 1999 x Salvelinus alpinus Brown trout, Amoebozoa Gill Morphology Italy Perolo et al., 2019 x Salmo trutta Turbot, Platyamoeba sp. Gill Morphology Spain Leiro et al., 1998 Scophthalmus (now Vannella sp.) x x maximus European carp, Hartmannella Pancreas Morphology Czech Dyková et al., 1997 Cyprinus carpio vermiformis (now Republic Other Tilapia, Vermamoeba x x x amoeba Orechromis vermiformis) related fish niloticus diseases Cyprinids, Acanthamoeba sp. Epidermis, Morphology, USA Webb et al., 2002 Catostomids, Naegleria sp. dermis Genetic Ictalurids, x Centrarchids, Moronids

Goldfish, Amoebozoa Kidney, liver, Morphology USA Dyková et al., 1996; x Voelker et al., 1977 Carassius auratus spleen, pancreas Blue Tilapia, Acanthamoeba sp. Gill, spleen, Morphology USA Taylor, 1977 Sarotherodon liver, intestine, x x x aureus brain Pompano, Amoebozoa Gill, kidney, Morphology, Singapore Athanassopoulou et al., 2002 Trachinotus intestine, Genetic x falcatus pancreas, spleen European catfish, Amoebozoa Gill, kidney, Morphology Germany Nash and Nash, 1988 Silurus glanis liver, intestines, x spleen Peacock wrasse, Amoebozoa Gill Morphology France Chatton 1909 ‡

Symphodus tinc Rainbow trout Vexillifera Kidney Morphology Italy Sawyer et al., 1978 x § incertae sedis bacillipedes Green Sea Urchin, Paramoeba Radial nerve Morphology, Canada Buchwald, 2016; Feehan et al., 2013; Strongylocentrotu invadens Genetic x § x x Jones, 1985 s droebachiensis Sea urchin Long-spine Sea Neoparamoeba Coelomic fluid Morphology, Spain Dyková et al., 2011b Hernández et al., paramoebia Urchin, Diadema branchiphila Genetic 2020 sis africanum x x (synonymous with Diadema aff. antillarum) Blue crab, Paramoeba Haemolymph, Morphology USA Johnson, 1977; Grey crab Sprague et al., 1969 Callinectes perniciosa connective tissue x ¶ disease sapidus Lobster American lobster, Neoparamoeba Eyes, antenna, Morphology, USA Mullen et al., 2005, 2004 paramoebia Homarus pemaquidensis gills and nervous Genetic x ¶ sis americanus tissue Pacific white Paramoeba spp. Gill Morphology, North Han, 2019 Undefined shrimp, Genetic America gill x Litopenaeus syndrome vannamei Mangrove oystrer, Dactylopodida Female gonads Morphology, Brazil Suhnel et al., 2014 x Crassostrea gasar Genetic Sydney rock Hartmannella Alimentary tract, Morphology Tahiti Cheng, 1970 Amoebic oyster, tahitiensis gonads, x disease in Crassostrea ctenidium Molluscs commercialis Japanese Scollops, Amoebozoa Connective Morphology Canada Bower et al., 1992 Patinopecten tissue, haemal x yessoensis space † Agent was formally considered to be Neoparamoeba pemaquidensis (Page, 1970) in publications pre-2007.

‡ Information obtained from Buchmann et al. (2004) because primary literature could not be accessed.

§ Indicates the author established amoeba culture, but authors did not confirm they were pure cultures.

¶ Conducted experimental infection trials with animals infected by cohabitation or injected with infected hemolymph.

AGD affects a diversity of marine, temperate teleosts across a large geographical range (Table 2.1). Salmonids are the most susceptible species to the disease, which has led to extensive commercial impact on salmonid aquaculture (Oldham et al., 2016). AGD was first reported in farmed coho salmon Oncorhynchus kisutch in Washington 1985, and soon after in farmed Atlantic salmon Salmo salar in Tasmania (Kent et al., 1988; Munday, 1986). Since then, the disease has been found in all major salmon producing countries, except Iceland (Table 2.1). AGD is the most significant health problem in the marine phase of production in Australia and it is an increasing concern for European producers.

NGD primarily impacts farmed juvenile freshwater salmonids, such as rainbow trout Oncorhynchus mykiss (Quaglio et al., 2016), chinook salmon Oncorhynchus tshawytscha (Tubbs et al., 2010) and Arctic char Salvelinus alpinus (Speare, 1999). The disease was first described in three species of hatchery-reared salmonids in Michigan, Oregon and Washington (USA) but, at that time, the typical amoeba-derived lesions were not diagnosed as NGD (Sawyer et al., 1974). Since then, NGD has been detected in North America (Noble et al., 1997), Canada (Speare, 1999), New Zealand (Tubbs et al., 2010) and Europe (Denmark, Germany, Poland, Italy and Czech Republic) (Antychowicz, 2007; Buchmann et al., 2004; Dyková et al., 2010; Dyková and Tyml, 2015; Quaglio et al., 2016), but the true prevalence and geographic distribution of NGD is still largely unknown. Although not as problematic as AGD, NGD is considered an ongoing health issue for freshwater salmonid aquaculture worldwide (Dyková and Tyml, 2015).

Paramoebiasis in sea urchins primarily impacts the Nova Scotia green sea urchin Strongylocentrotus droebachiensis fisheries located along the Canadian Atlantic coast (Scheibling et al., 2010). The index case occurred in 1980, followed by periodic outbreaks with mortality rates reaching up to 85 % (Feehan et al., 2013; Jones and Scheibling, 1985). This dramatically impacted catch rates, with landings reducing from 1300 tonnes in 1997 to around 300 tonnes per year from 2002 onwards (Miller, 2008). An amoeba infection was also reported in diseased long-spine sea urchin Diadema aff. antillarum from Tenerife Island, Spain, but in this instance the amoebae coexisted with a bacterial infection, termed sea urchin bald disease (Dyková et al., 2011b).

GCD devastated the blue crab Callinectes sapidus commercial fishing industry of the United States of America’s mid-East coast from 1965 to 1977 (Mahood et al., 1970; Sprague and Beckett, 1966). The disease was first detected in coastal shedding tanks located along the Delmarva

Peninsula of Maryland and Virginia in 1965 (Sprague and Beckett, 1966), and a year later in wild stocks in Georgia and South Carolina (Newman and Ward, 1973). Disease prevalence was estimated to fluctuate between 5-55 % throughout the year, and mass mortality occurred sporadically throughout spring and summer (Johnson, 1977; Mahood et al., 1970; Newman and Ward, 1973). The high death rate caused commercial landings to drop from 18.2 million kg in 1964 to 11.1 million kg in 1968, and the abundance of blue crabs within Chesapeake Bay has remained depressed (Mahood et al., 1970; Morado, 2011). GCD appears to have a limited geographic range and no significant outbreaks have been reported post 1977 (Morado, 2011). However since then, amoebae displaying similar morphology to the agent described during the early outbreaks have been detected in C. sapidus off the East coast of USA (Dyková et al., 2007). This suggests low endemic levels of grey crab disease may persist.

Lobster paramoebiasis impacts American lobster Homarus americanus in Western Long Island Sound, USA (Mullen et al., 2004). The disease caused the collapse of the traditionally fertile fishery, with an estimated 90-99 % stock reduction reported in 1999 (CDEEP, 2000; Mullen et al., 2004). During this outbreak, mortalities in sea urchins (unspecified) and crabs (C. sapidus, Libinia sp., Cancer irroratus, Limulus polyphemus) also occurred in lower numbers (CDEEP, 2000). However, the crabs and urchins were not examined for pathology, therefore it is inconclusive whether the deaths were also associated with amoebae. Less severe and more localised episodes of the disease occurred in the following years, possibly due to lower disease transmission rates caused by the significant reduction in lobster population (Pearce and Balcom, 2005).

The aetiology of these five diseases has been assigned to different species, each with a varied level of evidence supporting the amoeba’s identity and the causal link. Neoparamoeba perurans has been extensively shown to cause AGD, via both the fulfilment of Koch’s postulates (Crosbie et al., 2012) and its consistent association with gill disease across a variety of locations and fish species (Table 2.1). However, it took almost two decades of research to reach this conclusion, with the agent falsely considered to be Neoparamoeba pemaquidensis until 2007 (Munday et al., 2001; Young et al., 2007). The aetiological agent(s) of NGD, on the other hand, has not yet been conclusively determined. Over four decades, NGD has been linked to Thecamoeba sp. (Buchmann et al., 2004; Sawyer et al., 1974), Cochliopodium sp. (Daoust and Ferguson, 1985; Noble et al., 1997; Tubbs et al., 2010), Naegleria sp. (Dyková et al., 2010) and Rhogostoma minus (Dyková and Tyml, 2015). While there are some conflicting reports on the causal agent of NGD, most research

17 supports the hypothesis that the disease has a multi-amoeba-species aetiology (Buchmann et al., 2004; Daoust and Ferguson, 1985; Dyková et al., 2010; Dyková and Tyml, 2015).

Paramoeba invadens has been proposed as the only causal agent of sea urchin paramoebiasis via laboratory infection trials, and detection of the agent in field samples. However the role of other species cannot be ruled out (Feehan et al., 2013; Jones, 1985). Much less research is available to support the aetiology of GCD and lobster paramoebiasis, which are currently assigned to Paramoeba sp. (Sprague et al., 1969) and N. pemaquidensis (Mullen et al., 2005) based on just the presence of the amoebae in the diseased host. The broad range of confidence around aetiology arises because there are many ways to approach evidencing the host-pathogen relationship, with some more robust than others. The following section will discuss how the aquatic animal health community assesses aetiological agents of amoebic disease and will highlight effective methods and areas where further research is required.

2.4 The early days: the contribution of morphology to our understanding of amoebic disease in aquatic animals

Like most parasitic investigations, morphology provides the backbone to describing the agents of amoebic disease, with light and electron microscopy often delivering critical insights into aetiology. Characterising gross morphology and locomotive patterns involves observing wet-mount preparations under a light microscope. The main features used to identify amoebae are cell morphotype (e.g fan-shaped, lingulate, dactylopodial), pseudopodia (e.g. lobose, filiform, lamellar projections), uroid structure (e.g. bulbous, plicate, villous) and floating-form structure (Page, 1983; Smirnov and Goodkov, 1999). An example of gross morphology provided some of the earliest aetiological evidence for AGD is the description of Paramoeba pemaquidensis isolated from diseased Oncorhynchus kisutch gills in 1985 (Kent et al., 1988). This study described typical Paramoebae trophozoites with a parasome adjacent to the nucleus and a floating form with digitiform pseudopodia. Although P. pemaquidensis is no longer considered the causal agent of AGD, this initial gross characterisation by Kent et al. (1988) supports the consistent link of amoebae from family Paramoebae with branchitis of salmonids.

2.4.1 The plasticity of amoeba morphology

Despite morphology being the foundation for amoeba identification, light microscopy-based characterisation is not always reliable due to the inherent morphological plasticity of amoebae. Under a light microscope, a single species can be polymorphic (Dyková and Tyml, 2015) and different species, or even genera, are often indistinguishable (Dyková et al., 2005b; Smirnov et al., 2007). Examples of genera that cannot be easily differentiated by their gross morphology are Neoparamoeba/Paramoeba, Copromyxa/Nollandella and Vannella/Lingulamoeba (Dyková and Kostka, 2013). Some of these genera can be distinguished by observing ultrastructure through transmission electron microscopy (TEM), such as Paramoeba Schaudinn 1896, which has cell surface microscales, and Neoparamoeba Page 1987, which does not. Though not all cryptic species can be easily separated through TEM. For instance trophozoites of the freshwater amoebae Naegleria pagei, and Naegleria clarki are almost indistinguishable in both light microscopy and TEM images (Dyková & Kostka, 2013). Hence, traditional taxonomy and nomenclature based on morphology alone is insufficient for Amoebozoa.

The exclusive use of morphology for identification raises doubt over the aetiological agents of several aquatic amoebic-diseases (Bermingham, 2004; Cheng, 1970; Sawyer et al., 1974; Sprague et al., 1969; Taylor, 1977; Tubbs et al., 2010). For example, the proposed causative amoeba of GCD, when isolated from moribund crabs, was classified as a new species, Paramoeba perniciosa based on a comparative morphological study with Paramoeba eilhardi, the only recognised species from this genus at the time (Sprague et al., 1969). P. perniciosa was differentiated from P. eilhardi based on its smaller trophozoites, linguiform pseudopodia, inability to survive in the media which grew P. eilhardi, and pathogenicity to C. sapidus (Sprague et al., 1969). In this instance P. eilhardi parasitic ability was defined by its presence in a diseased host and not by in vivo trials. A large size range of P. perniciosa trophozoites (3-25 µm) was observed in this study, and again reported by Johnson, (1977), where amoebae were grouped into two size classes ranging from 3-7 µm and 10-20 µm. Interestingly, the small form of P. perniciosa was found to be predominately abundant in the hemolymph of heavily infected individuals, while the large class was in the connective tissues of the antennal gland, in the endothelia of the blood vessels, and within the nervous system. Hemolymph containing the small form of amoebae also successfully induced disease in two naive crabs. Despite the obvious size variation and differences in tissue tropism these sub-classes were not considered different species, rather proposed as different life stages (Johnson, 1977). 19

P. perniciosa is not recognised in major taxonomic databases, such as the World Register of Marine Species (WoRMS) or The NCBI Taxonomy Database. Since its original description this species has not been cultured as a clonal line and no sequence data have been reported. Species characterisation was performed using light microscopy alone, which cannot differentiate the eight currently recognised Neoparamoeba and Paramoeba species (Sprague et al., 1969). Therefore, the possibility of P. perniciosa being conspecific with another well characterised, perhaps disease- causing amoeba within the genus Paramoeba or Neoparamoeba is likely (Dyková et al., 2011b; Feehan et al., 2013; Mullen et al., 2005; Young et al., 2007).

For instance, in 1974 P. perniciosa was reported in C. sapidus sampled at Long Island Sound (Johnson, 1977), the same location of the 1999 mass mortality of American lobsters linked to N. pemaquidensis (Mullen et al., 2005, 2004). Additionally, moribund C. sapidus were observed, but unfortunately not sampled, during the lobster paramoebiasis epidemic (CDEEP, 2000). Moribund sea urchins were also observed in the same epidemic (CDEEP, 2000), and the geographic range of sea urchin paramoebiasis, linked to P. invadens, is only 500 km south of Long Island Sound (Sprague and Beckett, 1966). Outbreaks of GCD, lobster paramoebiasis and sea urchin paramoebiasis were reported at different times (index case; 1965 (Sprague and Beckett, 1966), 1998 (CDEEP, 2000) and 1980 (Jones and Scheibling, 1985) respectively), but they all occurred along the mid-East coast of North America and are linked to phenotypically similar agents. Therefore, more research beyond morphological characterisation is needed to determine if the P. perniciosa is conspecific with the agents of these currently differentiated diseases, and to determine whether multiple amoeba species can cause disease in C. sapidus.

2.4.2 Electron microscopy to support light microscopy-based characterisation

Morphological identification of amoebae can be strengthened by studying fine cellular features via electron microscopy. This involves describing the structure of organelles documented in TEM images, most commonly mitochondrial cristae (e.g. tubular, discoid, flattened), nucleus (e.g. nuclear inclusions, bi-nucleated) and cell membranes (e.g. scales, glycocalyx structure), but a few other traits are also considered on a strain-by-strain basis (Dyková and Kostka, 2013; Dyková and Lom, 2004). A notable taxonomic feature discernible via TEM is the kinetoplastid endosymbiont (or parasome) described in species belonging to the genera Paramoeba Schaudinn, 1896, Neoparamoeba Page, 1987 and Janickina Chatton, 1953 (Dyková et al., 2003). The presence 20 of one or multiple parasomes found within the cytoplasm close to the nucleus has been used as a distinguishing feature to identify parasitic amoebae to genus level in the earliest reports of amoebic disease in crabs (Perkins and Castagna, 1971), sea urchins (Jones, 1985), lobsters (Mullen et al., 2004), and fish (Dyková et al., 1998a).

There are cases where fine-scale cellular structures are no longer considered adequate for differentiating taxa. Initially the genera Vannella Bovee, 1965 and Platyamoeba Page, 1969 were distinguished by the presence (Vannella) or absence (Platyamoeba) of glycostyles in the cell surface coat viewed in TEM images. The separation of these genera was disputed based on molecular evidence which showed that species of Vannella and Platyamoeba do not form distinct clades in phylogenetic analysis, and hence both genera are now synonymised (Sims et al., 2002; Smirnov et al., 2007). While gross and fine-scale morphological studies have provided valuable insights into amoebic disease aetiology, the plasticity of gross amoeba morphology combined with the limited number of fine-scale distinguishing features has necessitated development of genetic- based approaches for species identification.

2.5 Resolving amoeba taxonomy through genetics

Genetic-based identification and taxonomy has the potential to overcome the issues of morphological characterisation and to greatly improve the accuracy, speed and consistency of species recognition in Amoebozoa (Nassonova, et al 2010; Pawlowski, et al 2012; Tekle 2014). Molecular barcoding is based on the comparison of short sequences from candidate genes that are theoretically conserved within species but divergent between closely related species. Once an appropriate barcoding gene is identified and sequenced, this region can be used to differentiate new or cryptic species through phylogenetic analysis, or to detect and quantify species within the host or environment using molecular assays designed specific to the barcoding gene.

2.5.1 Amoebozoa barcoding genes

Commonly employed candidate genes for Amoebozoa are small subunit ribosomal RNA (SSU rRNA), mainly the 18S rRNA gene, the internal transcribed spacer (ITS) and cytochrome oxidase subunit I (COI) (Nassonova et al., 2010). Originally SSU rRNA was used to determine phylogenetic relationships of Amoebozoa, and later adapted for species identification (Bolivar et 21 al., 2001; Pawlowski and Burki, 2009). This gene sequence, however, can be problematic for species distinction as it is relatively well conserved between closely related strains and can contain intra-strain polymorphism (Nassonova et al., 2010; Smirnov et al., 2007; Young et al., 2014). Despite these issues, most Amoebozoa reference sequences on public sequence databases (i.e. Genbank) are SSU rRNA, making it the most informative amoeba barcoding region for broad scale multispecies comparative phylogenetics.

ITS has been shown to provide higher sequence variability than SSU rRNA gene sequences within Acanthamoeba sp. (Köhsler et al., 2006) and better resolution of relationships amongst Vannella sp. (Dyková et al., 2005a; Nassonova et al., 2010). However, similarly to SSU rRNA, intra-strain variability within the ITS gene was reported for the genus Neoparamoeba (Young et al., 2014) and Vannella (Nassonova et al., 2010) which may obscure species distinction. Additionally, no significant contradiction between the ITS and SSU rRNA phylogenetic trees was observed in Vannella sp. (Nassonova et al., 2010), but this was not congruent with analyses by Dyková et al. (2005a). Although only a few Amoebozoa genera have been barcoded with ITS, overall evidence suggests ITS resolves similar but slightly higher phylogenetic signals compared to SSU rRNA.

COI is currently considered a promising marker for Amoebozoa barcoding as it provides higher phylogenetic resolution between closely related strains compared to SSU rRNA and ITS- based analyses (Nassonova et al., 2010). However, even though COI is well utilised for barcoding other protist groups (Pawlowski et al., 2012), this region has only been comprehensively validated as an appropriate barcode for two Amoebozoa genera, Cochliopodium and Vannella (Nassonova et al., 2010; Tekle, 2014). COI has also been used to assess the taxonomic value of morphological traits within the genus Nebelid, which uncovered extensive cryptic diversity (Kosakyan et al., 2012). Due to the reduced number of sequences available for this gene in public databases it is difficult to match the identity of newly sequenced COI regions from amoebae. Therefore, further research is needed to increase the number of characterised COI amoeba sequences, and to confirm its accuracy for amoeba barcoding and taxonomy. As a result, DNA barcoding of amoebae is still centred on SSU rRNA.

Owing to the vast diversity and complex evolution of groups within Amoebozoa, it is unlikely one universal barcoding region will be suitable for all groups (Kang et al., 2017; Pawlowski et al., 2012). Following an in-depth review of barcoding eukaryotes, the Protist Working

Group (ProWG) recommended a two-step barcoding approach (Pawlowski et al., 2012). This method involves preliminary identification using 18S rRNA gene, then a subsequent assessment of species/strain-level using a more informative gene, potentially COI for Amoebozoa.

2.5.2 Sanger sequencing versus next generation sequencing (NGS) when barcoding amoebae

Studies using genetic-based identification of Amoebozoa normally sequence the chosen barcoding gene by means of Sanger sequencing of PCR-amplified DNA from a homogenous amoeba sample (Dyková et al., 2005b; Volkova and Kudryavtsev, 2017; Young et al., 2014). The main benefits to this approach are sequence accuracy and simple sequence analysis. If the PCR is performed with DNA from either heterogeneous amoeba samples or environmental samples the approach will likely yield a high proportion of sequences that may not be the target species (López- García et al., 2001). Without a pure isolate the use of Sanger sequencing is inefficient and expensive due to the high cost per base-pair compared to other sequencing techniques (Schuster, 2008). Obtaining a pure amoeba sample is normally achieved by establishing clonal amoeba cultures, which can be impractical to maintain or not possible for some species (discussed further in 2.6.1). In the case where a pure isolate is not available, NGS provides an alternative approach to amoeba identification because the method is more suitable for sequencing trace amounts of DNA in mixed samples. There are some limitations to NGS, including software and analysis packages not keeping pace with advancing sequencing technology. Additionally, there is a possibility that a high proportion of non-target sequences generated during PCR masks the true sequence diversity, and fails to detect less abundant amoebae (Gofton et al., 2015). Despite these technical challenges, NGS still has the ability to recover greater species diversity compared to the traditional Sanger sequence approach (Edgcomb et al., 2011).

2.5.3 Molecular-based taxonomy to complement morphological-based identification

Amoebozoa has a long history of conflicting taxonomy using both morphological and genetic methods (Foissner and Korganova, 1995; Kang et al., 2017; Smirnov, 2002; Smirnov et al., 2011). There has been constant reclassification of amoebae at all taxonomic levels as methods based on morphology are not always suitable, and effective barcoding genes are still being refined. A good example of conflicting taxonomy within aquatic amoebic disease is found in AGD. Even though AGD is well recognised to be caused by amoebae from the genus Neoparamoeba (Young et al., 2007), a recent manuscript used SSU rRNA-based analysis to suggest that Neoparamoeba and 23

Paramoeba are phylogenetically inseparable and, therefore, Neoparamoeba should be treated as a junior synonym of Paramoeba (Feehan et al., 2013). This proposed name change was deemed premature by Young et al. (2014), who advised that SSU rRNA is an unsuitable region to separate two genera that are defined by the presence (Paramoeba Schaudinn, 1896) or absence (Neoparamoeba Page, 1987) of microscales. The use of the most appropriate nomenclature, based on either molecular or morphological characteristics, has sparked continuous disagreement within the AGD research community, and a consensus is yet to be reached. As a result, both genera are still employed by research groups worldwide.

Despite the lack of current consensus in the Neoparamoeba and Paramoeba debate, there are multiple cases where phylogenetics has circumvented difficulties associated with microscopic characterisation. The causative agent of sea urchin paramoebiasis was characterised and named P. invadens, based on cell dimensions, locomotion patterns, presence of parasomes, and ultrastructure of trophozoites isolated from diseased S. droebachiensis (Jones, 1985). Two decades later, the identity of the parasite was readdressed by Feehan et al. (2013) due to speculation that P. invadens was conspecific with N. pemaquidensis, a morphologically similar species associated with diseased lobsters H. americanus in USA (Mullen et al., 2005). This research provided the first genetic information on P. invadens and found that it was not conspecific, but a distinct species more closely related to Neoparamoeba branchiphila.

Likewise, the aetiological agent of lobster Paramoebaiasis was found to be an amoeba consistent with the genus Paramoeba, based on the presences of a parasome in histological and microscopic analysis of dead and moribund lobsters (Mullen et al., 2004). Attempts to isolate and culture the target amoeba from infected lobsters failed, leaving Koch’s postulates unfulfilled. Almost six years after the major outbreak, sequencing of the SSU rRNA gene of amoebae in archived samples identified the most likely causative agent of lobster paramoebiasis as N. pemaquidensis (Mullen et al., 2005).

Phylogenetic analysis has been a crucial tool for supporting the differentiation of the agents of aquatic amoebic disease, most commonly either Neoparamoeba or Paramoeba, genera which have species that are morphologically almost identical. A good example of Neoparamoeba strains being differentiated by phylogenetic analyses is shown by Dyková et al. (2005b). Here, 18 Neoparamoeba strains were characterised via light-microscopy and phylogenetic analyses based on SSU rRNA gene sequences. The 18 strains were almost indistinguishable by microscopy, yet they 24 were found to be genetically diverse and clearly formed separate clades of either N. pemaquidensis or N. branchiphila. Thus, research is increasingly relying on sequence homology supported by phylogenetic analysis to resolve the taxonomy of new and cryptic species, and to validate morphological-based identification.

Phylogenetic methods are also becoming more sophisticated. Rather than forming taxonomic relationships based on one gene, commonly 18S rRNA, sequenced genomes and associated transcriptome and proteome data are enabling multigene phylogenetics (Cavalier-Smith et al., 2016; Kang et al., 2017). In Amoebozoa for example, a revised phylogeny that contested many previous studies was constructed by Kang et al. (2017) using culture-based and single cell transcriptomics from 325 protein-coding genes in 61 Amoebozoa taxa. Despite these new methods offering a more robust taxonomy, it is hard to completely discount the benefit of cross-validating these relationships with morphology. Again, sequence homology and phylogenetics should be employed as a complementary tool rather than entirely replacing traditional morphology-based taxonomy.

2.5.4 Molecular detection of amoebae

Once a suitable barcoding gene is sequenced, species-specific probes can be developed to identify and quantify the target amoeba on affected tissue or within the environment by techniques such as in situ hybridisation (ISH) and quantitative real time PCR (qPCR). Molecular-based assays have become a common tool for detecting aquatic pathogens owing to their sensitivity, specificity, reproducibility and high-throughput (Purcell et al., 2011). Currently there is a scarcity of assays available for pathogenic aquatic amoebae, but this is expected to change as researchers and industry increasingly rely on genetic approaches to study parasitic amoebae. The amoeba assays that have been developed are powerful tools to answer a range of aetiological questions, such as determining host specificity, as well as agent identity, distribution and abundance. For example, three species- specific ISH probes that bind with the 18S rRNA of N. perurans, N. pemaquidensis and N. branchiphila on diseased gill sections were used to show N. perurans was the main aetiological agent of AGD in six countries and across four fish hosts (Young et al., 2008b). This shows the agent of AGD is globally distributed and not host specific, which has significant implications for both industry and research.

There are three published qPCR assays based on 18S rRNA gene to detect the main agent of AGD, N. perurans (Bridle et al., 2010; Downes et al., 2015; Fringuelli et al., 2012), and one that detects the agent of sea urchin paramoebaiasis, P. invadens (Buchwald, 2016). These assays have been used to monitor disease progression (Downes et al., 2015), determine host specificity (Stagg et al., 2015), and identify environmental reservoirs of the agent (Hellebø et al., 2017; Rolin et al., 2016). However, due to sequence similarity, designing high quality qPCR and ISH assays based on SSU rRNA requires careful optimisation and validation to achieve suitable specificity and sensitivity (Broeders et al., 2014; Wolff et al., 2007). After a probe sequence has been chosen, assay specificity should be checked against closely related species, which are either impractical to isolate from the wild, or expensive to purchase from culture collections. Consequently, many of the amoeba assays relevant to aquatic disease are not species-specific, or are not adequately validated for specificity (Bridle et al., 2010; Buchwald, 2016; Downes et al., 2015). Despite these design issues, molecular detection assays undoubtedly have an important role in aetiological investigations of amoebic disease. Nevertheless, qPCR and ISH assays should not entirely replace the need to visually confirm the presence of the pathogen with the host through either wet mounts under light microscopy or histopathology.

2.6 Determining pathogenic amoebae

Accurate identification of the amoebae isolated from the diseased host is an important first step, but further evidence is necessary to prove the causal relationship between the amoebae and disease. For this purpose, two general approaches are commonly employed to define disease causation; the first is to fulfil Koch’s Postulates, and the second is an investigation based on histological and ISH analysis. The following will discuss Koch’s Postulates in the context of facultative parasitic amoebae. Then, in the cases where Koch’s postulates cannot be fulfilled, a revision of the histological/ISH approach to causation is proposed in section 2.6.2.

2.6.1 The application of Koch’s postulates in aquatic amoebic disease

Koch’s postulates, known as the four rules of proof, originally developed for bacterial disease, are still considered a robust approach to determining the pathogenicity of microorganisms (Koch, 1884). However, it is well recognised that these criteria often fail to reflect the complexity of multifactorial disease and, therefore, should be treated as a guideline for collecting evidence, 26 rather than as stringent rules. In-depth discussion of the limitations and various interpretations of Koch’s Postulates can be found in Prescott et al. (2017), Ross and Woodward, (2016) and Falkow, (2004). This manuscript considers the four postulates to be;

1. The pathogen should be constantly associated with the disease. 2. The pathogen must be isolated from a diseased organism and grown in pure culture. 3. The cultured pathogen should cause disease when experimentally exposed to healthy, naive hosts. 4. The pathogen must be re-isolated from the experimentally infected hosts and identified as being identical to the original inoculated pathogen.

With respect to this interpretation of Koch’s Postulates, Table 2.1 shows the level of proof in which each diseased host has been linked to a particular aetiological agent, and whether morphological and/or molecular-based techniques were used to identify the proposed pathogen. From this list of disease cases Crosbie et al. (2012) is the only published study to suitably fulfil all four criteria, confirming N. perurans as an agent of AGD in Atlantic salmon. Similar in vivo experiments later proved N. perurans is also able to cause AGD in lumpfish and ballan wrasse, which are cleaner fish used by the European salmon industry to delouse farmed salmon (Haugland et al., 2017; Kvinnsland 2017).

The four postulates were also addressed by Feehan et al., (2013) in which the aetiology of paramoebiasis in green sea urchins was investigated, however a few limitations are evident. It is unclear whether the amoebae cultures used to infect the urchins were derived from a single cell and the amoebae that were re-isolated from the experimentally infected urchins were not identified by molecular methods. This is particularly important as P. invadens shares a similar morphology with several species from the genus Neoparamoeba and Paramoeba, with N. branchiphila linked to disease in long-spine sea urchin (Dyková et al., 2011; Hernández et al., 2020) and N. pemaquidensis linked to moribund lobsters (Mullen et al., 2005). This makes it difficult to definitively conclude that P. invadens was the only species associated with the observed lesions.

Most other published reports of amoeba-related diseases in aquatic animals have based their aetiological conclusions on one, sometimes two, of the four criteria, which may only be interpreted

27 as an assumption of causation. To understand why only a few studies have fulfilled Koch’s postulates, each criterion is examined in more detail using specific examples of aquatic amoebic disease.

Koch’s first postulate

The original interpretation of the first postulate defined that the causative agent should be constantly associated with the disease, and not with healthy hosts (Koch, 1884). This definition should be relaxed for amoebic diseases of aquatic organisms because pathogenic amoebae have been detected in asymptomatic animals (Bermingham and Mulcahy, 2007; Franke and Mackiewicz, 1982; Taylor, 1977). The first postulate is the criterion most frequently fulfilled by studies on aquatic amoebic diseases and is commonly achieved by microscopic and/or molecular identification of amoebae isolated from the diseased host. The presence of the suspected agent in diseased tissue forms the majority of evidence linking Hartmannella tahitiensis Cheng, 1970 to an unspecified disease in Sydney rock oysters (Cheng, 1970), Cochliopodia sp. to NGD of chinook salmon (Tubbs et al., 2010), N. pemaquidensis to lobster paramoebiasis (Mullen et al., 2005) and P. perniciosa to GCD (Sprague et al., 1969). Although this criterion provides strong evidence that a particular amoeba may be involved in disease development, it does not conclusively prove it caused the condition.

The first postulate can be misleading if employed alone. Firstly, non-pathogenic amoebae are often co-isolated with virulent species. For example Vannella sp. are generally considered avirulent based on their frequent isolation from healthy fish (Dyková et al., 2005a), yet this genera is regularly isolated alongside Neoparamoeba sp. from AGD infected gills (Bermingham and Mulcahy, 2007; Howard, 2001). This criterion also overlooks the possibility that early onset lesions could be triggered by other organisms, such as bacteria, which are then colonised by opportunistic, normally non-pathogenic amoebae. This scenario that has been hypothesised for NGD (Bullock et al., 1994; Speare, 1999) and one case of diseased long-spine sea urchins D. antillarum infected with N. branchiphila (Dyková et al., 2011b). Additionally, it is difficult to establish a constant parasite- lesion-host association when disease occurs infrequently, which is the case for wild animals such as C. sapidus with GCD (Mahood et al., 1970), lobster paramoebiasis (Mullen et al., 2004) and amoebic disease in molluscs (Bower et al., 1992; Cheng, 1970; Suhnel et al., 2014). Still, much of

28 the available literature has relied on the first postulate to propose the aetiology of aquatic amoebic diseases, and therefore should be carefully interpreted.

Koch’s second postulate

The second postulate entails isolating the pathogen from a diseased host then growing it in pure culture (Koch, 1884). In the case of amoebic disease of aquatic organisms, it is crucial that the primary isolate is taken from the very focal point of the lesion, as many amoeba species colonise multiple organs of healthy fish and invertebrates (Dyková et al., 2007, 1999, 1998b). It is then recommended that clonal lines are established from this primary isolate because monocultures offer two main advantages to conducting aetiological research. Firstly, clonal cultures guarantee that molecular and morphological information used to identify the amoebae are from a single strain, and secondly they ensure that only one species is acting as the pathogen in future in vivo disease challenge trials. This is particularly relevant for amoebic diseases that have been found to be associated with multiple amoeba species, such as NGD where up to seven genera have been isolated from the gills of infected trout but it is uncertain if one or many species are involved in disease progression (Dyková et al., 2010; Dyková and Tyml, 2015). In fact, this is also the case for AGD, as up to nine species has been found accompanying N. perurans on infected Atlantic salmon gills (Bermingham and Mulcahy, 2007; Howard, 2001) but only three have been grown in pure culture and used in in vivo disease challenge trials (Crosbie et al., 2012; Morrison et al., 2005; Vincent et al., 2007). Hence, it is unclear whether amoeba other than N. perurans also influence the onset or disease progression of AGD. Establishing clonal lines provides the means to assess the pathogenicity of individual species, and to determine if a disease has a multispecies aetiology.

Despite the benefits of using pure amoeba cultures in aetiological research, they can be impractical to establish and maintain, and sometimes challenging to grow enough for large-scale experiments. Culturing amoeba is labour-intensive, and it is often difficult to eliminate co-isolated microflora, especially, fungus, ciliates and bacteria, which can rapidly overgrow target trophozoites. The use of antibiotics to control bacteria is not recommended as it could eliminate “cooperating” strains potentially involved in amoeba pathogenicity (Bowman and Nowak, 2004; Embar-Gopinath et al., 2005). It is also likely many amoeba species are not capable of thriving in vitro, despite the variety of culturing techniques currently available. Hence, there are multiple disease investigations that report their attempt to culture amoebae were unsuccessful, such as Vexillifera bacillipedes from

29 trout (Sawyer et al., 1978), Cochliopodidae-like species from chinook salmon (Noble et al., 1997), and Neoparamoeba sp. from lobster (Mullen et al., 2004). Due to the challenges of culturing it is not surprising only a few published studies, listed in Table 2.1, draw their aetiological conclusions based on clonal lines (Bermingham, 2004; Crosbie et al., 2012; Dyková et al., 2010).

Koch’s third postulate

The third postulate states that the cultured microorganism must cause disease when introduced into a healthy, naive host (Koch, 1884). However, given that aquatic amoebic diseases are multifactorial and, therefore, not necessarily triggered under controlled conditions, the employment of “should” instead of “must” is recommended when interpreting this criterion. For example, the one-off mass mortality of American lobsters in Long Island Sound was linked to multiple environmental factors, such as warmer than average water temperature, low oxygen, pesticides, and high lobster population density, rather than simply the presence of N. pemaquidensis (Pearce and Balcom, 2005). Likewise, NGD has been reported in Arctic char as a secondary infection to bacterial gill disease (BGD) caused by Flavobacterium branchiophilum (Speare, 1999), and outbreaks of BGD have been associated with high stocking densities, fish handling, and poor water quality (Daoust and Ferguson, 1985). The complex relationship between the environment, host, and parasite may make it difficult to trigger a facultative parasite to express its virulence towards a host under experimental conditions.

Another exception that could preclude postulate three from being fulfilled is whether the cultured amoebae accurately reflect the pathogenicity of the wild type. Different isolates of the same amoeba species often show varied virulence (Biller et al., 2009), meaning that the inability of a given monoculture to trigger infection could be due to the culture being established from a cell of low virulence. Additionally, long-term maintenance under in vitro conditions can result in reduced virulence (Bridle et al., 2015; Jellet and Scheibling, 1988; Wong et al., 1977). For instance, a clonal line of N. perurans which induced AGD in salmon just over two months-post isolation lost virulence after three years in culture (Bridle et al., 2015). Cryopreservation of amoeba cultures could be considered for preventing loss of virulence in long-term culture.

A practical limitation evident in some past attempts to fulfil the third postulate is the requirement of an experimental facility that can hold aquatic animals long-term. This is particularly 30 relevant for non-domesticated animals, such as sea urchin and crabs, which may be difficult to hold in optimal health and thus adversely impact trial results. Facilities should also have a reliable influent/effluent disinfection system to avoid introducing non-target pathogens into the experimental system, and to prevent environmental contamination. The high financial investment to build and maintain such infrastructure is perhaps the reason why some in vivo trials have low replication, pseudo-replication, or a low number of challenged hosts (Bermingham, 2004; Dyková et al., 1997; Johnson, 1977; Morrison et al., 2005; Vincent et al., 2007). Despite the limitations of these in vivo trials, they do provide relevant insight into aetiology, and a platform for future research.

Koch’s fourth postulate

In the fourth postulate the agent must be re-isolated from the newly infected host and identified as the same inoculated pathogen. Some interpretations of the fourth criterion also imply that the re-isolated amoebae needs to be cultured again (Engelkirk et al., 2011), but we find this step unnecessary as long as the re-isolated agent is properly identified using both microscopic and genetic-based methods. Another aspect that is sometimes overlooked in this criterion is that the re- isolated organism should be shown to be closely linked to the observed pathology. This is of particular importance given that lesions could be caused by other agents, such as bacteria present in the cultures. Therefore, the use of histopathology to confirm whether the expected cellular changes are linked to the target amoeba and rules out the influence of other agents would be beneficial.

It is evident several technical barriers have meant Koch’s Postulates have not been fulfilled for most amoebic diseases affecting aquatic animals. Where practical issues render Koch’s approach unfeasible, an alternative method to determine aetiology is proposed.

2.6.2 The histological and ISH approach to define causation

Histology is a valuable tool for field-based aetiological studies as it is the only technique that captures the association between agent(s) and lesion, making it a better approach for documenting multifactorial diseases. For instance, many agents can cause gill lesions, including protozoans (e.g. ciliates (Bowater and O’Donoghue, 2015), bacteria (e.g. F. branchiophilum (Bullock et al., 1994)) and viruses (e.g. gill poxvirus (Gjessing et al. 2017)), and histology can 31 assist with diagnosis of whether amoebae alone, or amoebae in conjunction with other agents are involved in the pathology. Unlike manipulated disease challenge trials, histology samples from naturally occurring disease outbreaks can capture numerous factors involved in disease onset.

One downfall of using histology to diagnose amoebic disease is that it cannot easily identify amoebae to species or even genus level, because the fixation process alters the morphology of trophozoites. Despite this, some studies have attempted to classify parasitic amoeba using only histology samples, and hence these taxonomic assessments are not definitive (Sawyer et al., 1974; Tubbs et al., 2010). The only exceptions to this are Neoparamoeba and Paramoeba which can be identified by the presence of the parasome (Dyková et al., 2003). Even so, the two genera cannot be distinguished from each other without molecular tools. Thus, species-specific ISH assays should be employed when the specific aetiological agent needs to be identified.

Histology and ISH provided strong aetiological evidence of the causation of an NGD outbreak in rainbow trout in Germany that had not been otherwise experimentally investigated (Dyková et al., 2010). Seven different genera were isolated from the gills of infected rainbow trout, including Vannella sp., Naegleria sp., Acanthamoeba sp., Vermamoeba sp. (formerly Hartmannella sp.) and Protacanthamoeba sp. Of these isolates, Naegleria sp. was the only one found attached to hyperplastic gill epithelium in histological sections via a species-specific ISH probe designed to hybridise with SSU rRNA gene. Despite this compelling evidence, Naegleria sp. should not be considered the sole aetiological agent of NGD as a later study by Dyková and Tyml, (2015) found Rhogostoma minus was the main agent colonising NGD-affected gills of rainbow trout in Czech Republic. Additionally, Cochliopodium sp. has also been linked with several different NGD outbreaks (Daoust and Ferguson, 1985; Noble et al., 1997; Tubbs et al., 2010). Collectively this supports the hypothesis that NGD has a multi-species aetiology and provides evidence for the need to conduct several histology/ISH-based studies to clearly define the aetiology of a given amoebic disease.

When using histology and ISH in aetiological research it is recommend that material that from multiple disease outbreaks is compared with ‘healthy’ samples from the same locations before the cause of a disease is reliably determined. Despite this, most aquatic amoebic disease studies which used histology-based methods concluded causation on either a low number of samples or single disease outbreaks and, therefore more research would be beneficial (Dyková et al., 2010;

Dyková and Tyml, 2015; Jones et al., 1985; Mullen et al., 2004; Sawyer et al., 1974). With careful experimental design the histology/ISH-based approach can be adopted as an alternative to Koch’s postulates when the disease cannot be simulated under controlled conditions, when an appropriate research facility for in vivo trails is not available, or when clonal amoeba cultures cannot be grown.

2.7 Conclusion

The current chapter reviewed how amoebae have been identified and defined as the aetiological agents of disease in aquatic animals. From this analysis it is evident the aetiology of most aquatic amoebic diseases remains unresolved because many reports used outdated methods for identifying amoebae, and insufficient evidence for defining the aetiological relationship. There is an opportunity to strengthen the understanding of pathogenic aquatic amoeba as techniques for studying aetiology are evolving. Genetic-based approaches are improving identification of cryptic amoebae species and act complementarily to the traditional morphological-based characterisation. There is no ‘one-approach fits all’ to proving aetiology, rather a diversity of suitable methods that provide a conglomerate of supporting evidence. This importantly underpins developing therapeutics and the management strategies necessary to minimise the impact of amoebic diseases on aquaculture and fisheries.

Chapter 3: A diversity of amoebae colonise the gills of farmed Atlantic salmon with amoebic gill disease (AGD)

Statement of author contributions M.C., P.L. and A.B. conceptualised the broader project idea. C.E. refined the project design and carried out the research investigation, with training and resource administration provided by N.B., P.L. and J.W.. C.E. wrote the manuscript under the supervision of J.W., N. B., A. B., P. L. and M.C. T.T. and C.E. carried out the phylogenetic analysis and T.T. advised on how to write the methods and results in relation to this work. All authors contributed to reviewing and editing the final manuscript.

3.1 Abstract

Neoparamoeba perurans is the aetiological agent of amoebic gill disease (AGD) in salmonids, however multiple other amoeba species colonise the gills and their role in AGD is unknown. Taxonomic assessments of these accompanying amoebae on AGD-affected salmon have previously been based on gross morphology alone. The aim of the present study was to document the diversity of amoebae colonising the gills of AGD-affected farmed Atlantic salmon using a combination of morphological and sequence-based taxonomic methods. Amoebae were characterised morphologically via light microscopy and transmission electron microscopy, and by phylogenetic analyses based on the 18S rRNA gene and cytochrome oxidase subunit I (COI) gene. In addition to N. perurans, 11 other Amoebozoans were isolated from the gills of farmed Tasmanian salmon, and were classified within the genera Neoparamoeba, Paramoeba, Vexillifera, Pseudoparamoeba, Vannella and Nolandella. In some cases, such as Paramoeba eilhardi, this is the first time this species has been isolated from the gills of teleost fish. Furthermore, sequencing of both the 18S rRNA and COI gene revealed significant genetic variation within genera. We highlight that there is a far greater diversity of amoebae colonising AGD-affected gills than previously established.

Keywords: Amoebozoa; AGD; Aquaculture; Atlantic salmon; Discosea; Tubulinea 34

3.2 Introduction

The gills of teleost fish play a vital role in a number of essential physiological processes including respiration, osmoregulation, ammonia secretion and acid-base regulation (Evans, 2005). However, by virtue of their position and physical structure the gills represent an important yet vulnerable barrier that is in constant and intimate contact with the external environment. As such the gills can be subjected to a variety of environmental and pathogenic insults which, under the appropriate conditions, may result in gill pathology and or injury. While in some cases a single pathological agent can be responsible, a number of more complex gill diseases with apparent mixed aetiologies have emerged in farmed teleosts (Gjessing et al., 2017; Herrero et al., 2018). The increasing list of proven and putative gill pathogens, and the complexity of disease expression gives reason to consider gill disease in the context of dysbiosis of microbial community structure, rather than focusing on a single agent (Downes et al., 2018; Egan and Gardiner, 2016; Gjessing et al., 2017; Herrero et al., 2018).

Amoebic gill disease remains one of the most important diseases affecting Atlantic salmon aquaculture in Tasmania, Australia. Caused by the free-living protozoan parasite, Neoparamoeba perurans, AGD affects Atlantic salmon during the marine grow-out phase (Crosbie et al., 2012; Young et al., 2007). Attachment of N. perurans to gill epithelium causes epithelial hyperplasia, oedema and lamellar fusion and, ultimately if not treated, mortality (Adams and Nowak, 2001). N. perurans however is not the only amoeba species capable of colonising the gills of marine cultured Atlantic salmon. Early reports, based on gross morphology, identified five different genera accompanying Neoparamoeba spp. on the gills of farmed Atlantic salmon in Tasmania, Australia, including Acanthamoeba, Flabellula, Heteroamoeba, Vannella and Vexillifera (Howard, 2001). Similarly, five genera were isolated from the gills of farmed Atlantic salmon in Ireland, including Flabellula, Mayorella, Nolandella, Vannella and Vexillifera (Bermingham and Mulcahy, 2007). While it is evident that the gills of Atlantic salmon may be colonised by a variety of Amoebozoa, the role that non-N. perurans amoebae play in AGD, as either a secondary invader, primary pathogen or commensal bystander, remains unclear (Morrison et al., 2005; Nowak and Archibald, 2018).

Dysbiosis describes a microbial community shift that has a negative impact on the host (Petersen and Round, 2014). In the context of microbial communities in teleost gills, recent studies

35 have shown pronounced shifts in bacterial communities associated with disease status (Legrand et al., 2018). While it is clear that N. perurans alone is capable of causing AGD, it remains uncertain whether AGD promotes a more global Amoebozoan dysbiosis and ultimately how such a community shift contributes to disease onset or progression. In some cases of gill disease in salmonids, for example nodular gill disease (NGD), a single pathogen has not been attributed as the aetiological agent, rather a multi-amoeba aetiology is proposed (Dyková et al., 2010; Dyková and Tyml, 2015).

Traditionally the identification of Amoebozoan communities through gross morphological features alone has been difficult, largely due to the inherent plastic morphology of Amoebozoa (Dyková and Lom, 2004). More recently however, the application of genetic approaches, improved culture practices and advanced microscopy has facilitated more extensive profiling of the diversity of Amoebozoan communities. Using these methodological improvements, the goal of the present study was to document the diversity of Amoebozoa capable of colonising the gills of AGD-affected Atlantic salmon using both morphological and molecular taxonomic approaches. Specifically, we isolated and established mixed-cultures and monocultures from AGD-affected gills, then recorded the amoebae morphology via light microscopy and, where possible, transmission electron microscopy (TEM). The diversity of amoebae was further assessed by sequencing the 18S ribosomal RNA (18S rRNA) gene and cytochrome oxidase subunit I (COI) gene, and each newly obtained amoeba sequence was identified by sequence homology and supported by phylogenetic analysis.

3.3 Material and Methods

3.3.1 Sampling AGD-affected gills to establish amoeba cultures

The gill basket of five farmed Tasmanian Atlantic salmon that displayed clinical signs of AGD were collected from the Dover lease in South Eastern Tasmania during each of the four sampling events: June and October 2015, May 2016, and August 2017. For this purpose, gills with a gill score greater than three (Taylor et al., 2009) were dissected and transported in chilled, filtered seawater to the laboratory. All fish used in this study were approved for sampling by CSIRO Queensland Animal Ethics Committee (AEC number A13/2015 and A9/2016).

3.3.2 Primary isolation and maintenance of cultures

Amoebae were isolated from the gill baskets by inoculating culture flasks with either mucus scrapes or small tissue samples. These primary isolates were cultured in 0.2 µm filtered, sterile 33 ppt seawater at 14°C, and regularly observed for two weeks for the presence of amoebae using an inverted microscope (Olympus CK2, Japan) at 200 x magnification. Once the amoebae attached to the bottom of the culture flask, the overlayed seawater was carefully removed by pipetting, followed by gentle rinsing with filtered, sterile seawater to remove excess bacteria and tissue debris, then replaced with an aliquot of 1 % malt yeast broth (MYB; 0.01 % (w/v) malt extract and 0.01 % (w/v) yeast extract in filtered, sterile seawater). Successfully established mixed-amoeba-cultures were maintained weekly, which involved media exchange, contaminant checks and splitting cultures as necessary. The amoeba cultures were sampled for DNA extraction when they were approximately 70 % confluent.

3.3.3 Establishing monocultures

Single amoeba cells were isolated using an adapted pipette and dilution technique from Smirnov (1999). The presence of one cell in 0.1 µl of seawater was confirmed using a light microscope at 200 x magnification. The single amoeba cell was then transferred to a 96-well culture plate and grown in 1 % MYB at 14°C. Once 70 % confluent, each monoculture was sampled for DNA extraction. To prevent the potential loss of virulence, antibiotics were not used to control bacterial overgrowth (Bowman and Nowak, 2004; Embar-Gopinath et al., 2005). As a result, the monocultures were not axenic and contained associated bacterial communities.

3.3.4 Morphological characterisation of amoebae grown in monoculture

Gross morphology of 20 amoebae from each monoculture was documented using an inverted microscope (Olympus CK2, Japan) under 200 x magnification. Images were obtained with a Luminoptic camera and processed using ISCapter (Tucsen) software. The dimensions of the attached and floating forms of each amoeba strain were measured and expressed as mean values, with range and standard deviation indicating variation of intra-strain size. A key to marine gymnamoebae by Page (1983) was employed to assist with morphology-based identification.

For transmission electron microscopy (TEM), amoebae were cultured on carbon-coated 3 mm sapphire discs (Wohlwend, Switzerland) and frozen in a cryoprotectant (20 % BSA in artificial seawater) in an HPM 010 high pressure freezer (Baltec, Liechtenstein) before undergoing freeze- substitution in 1 % osmium tetroxide and 0.5 % uranyl acetate in acetone for 48 h at -90°C in a Leica AFS2 automated freeze substitution machine (Leica, Austria). Samples were then embedded in Epon, polymerised at 60°C for 48 h and sectioned at 80 nm using a Leica UC6 ultramicrotome (Leica Microsystems, Austria) before viewing at 80 kV on a Jeol JSM 1011 transmission electron microscope (Jeol, Japan).

3.3.5 Identification of amoebae by sequencing

DNA extractions from both mixed and monocultures was performed using a DNeasy Blood and Tissue Kit (QIAGEN) according to the manufacturer’s instructions. Extracted DNA was quantified with a Nanodrop ND-1000 spectrophotometer (Life Technologies, USA) and stored at - 20°C. A 669 to 950 bp fragment of the 18S rRNA gene or COI gene was amplified by PCR using one of five universal eukaryotic primers (Table 3.1). PCR reactions followed the Kapa Taq PCR Protocol (KapaBiosystems) for a 25 µl reaction, with a thermal profile of denaturation at 95°C for 3 min, followed by 35 cycles at 95°C for 30 s, 2 min with an annealing temperature specific to each primer set (Table 3.1), 72°C for 2 min, with a final extension at 72°C for 2 min. Successful amplification was confirmed by visualising the amplified products in a 1.2 % agarose gel.

Table 3.1 Universal Eukaryotic 18S rRNA and COI primer sets used to amplify amoebae. Gene Primer Sequence (5’-3’) Annealing Expected Reference Temperature product size (°C) 18S rRNA RibB TGATCCATCTGCAGGTTCACCTAC 50 800 - 900 Smirnov et S12.2 GATYAGATACCGTCG TAGTC al., 2007 18S rRNA Ami6F1 CCAGCTCCAATAGCGTATATT 60 700 - 900 Thomas et Ami9R GTTGAGTCGAATTAAGCCGC al., 2006 18S rRNA 570C GTAATTCCAGCTCCAATAGC 58 700 - 900 Schroeder et 1137R GTGCCCTTCCGTCAAT al., 2001 COI Eucox1F GAYATGGCKTTNCCAAGATTAAA 50 800 - 1000 Heger et al., Euglycox1R AGCACCCATTGAHAAAACRTAATG 2011 COI LCO1490 GGTCAACAAATCATAAAGATATTGG 50 600 - 700 Folmer et HCO2198 TAAACTTCAGGGTGACCAAAAAATCA al., 1994 38

Amplified DNA was purified with the QIAGEN PCR purification kit following the manufacturer’s instructions. The purified DNA was ligated into a pGEM-T Easy Vector (Promega) then transformed using Alpha-Select Silver Efficiency Competent Cells (Bioline) according to the manufacturer’s protocol. The transformed cells were plated onto LB Agar plates containing ampicillin (100 mg/L) and incubated overnight at 37°C. Successful ligation and transformation was confirmed by PCR of multiple colonies using T7 and SP6 vector specific primers. Colonies with inserts were grown in LB broth overnight at 37°C. Corresponding glycerol (20 %) stocks were prepared for archiving and plasmid was purified using the QIAprep Spin Miniprep Kit (QIAGEN). Plasmids were sequenced using the BigDye® Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, country) and purified with the Agencourt CleanSEQ (Beckman Coulter) as per the manufacturer’s instructions. Sequences were generated by the ABI 3130xl genetic analyser (Applied Biosystems). The sequences were edited and aligned using ChromasPro (Version 1.5, Technelysium Pty Ltd) and CLC Main Workbench 7 (Version 7.6.4, QIAGEN Aarhus A/S).

3.3.6 Sequence analysis

Each newly sequenced 18S rRNA and COI gene fragment was subjected to a BLASTn (NCBI) (https://blast.ncbi.nlm.nih.gov/Blast.cgi) search to find the top hit. The 18S rRNA sequences were only accepted as Amoebozoa sequences if they had an e-value of zero. In contrast, to determine if the COI sequences were amoebae or contaminants such as fungi or bacteria, preliminary taxonomic positions of all COI sequences were determined by phylogenetic analysis. This is because there is a paucity of Amoebozoa COI reference sequences on public databases to infer their identity based on BLAST analysis alone. The sequences found to be Amoebozoa were then cloned and sequenced twice to account for potential mixed base calls in some positions in the original sequence. In total, 19 18S rRNA gene fragments and 17 COI gene fragments obtained from mixed-cultures and monocultures were used in the following phylogenetic analysis.

3.3.7 Phylogenetic analysis of 18S rRNA amoeba sequences

Initially we aimed to construct a phylogenetic analysis based on 18S rRNA that comprised all Amoebozoa groups so that the most likely identity and phylogenetic position of each of the newly sequenced strains were shown within one analysis. Despite trialling a number of different sequence datasets, alignments, and trimming methods this approach proved unsuitable because it 39 did not resolve the currently accepted Amoebozoa phylogeny which has been established in more sophisticated, multi-gene analyses (Kang et al., 2017; Tekle et al., 2008). As our aim was to determine the most closely related species to our newly sequenced strains, not reconstruct the full Amoebozoan phylogeny, we focused our analysis on lower taxonomic groups of interest.

All newly obtained 18S rRNA sequences were either Tubulinea (suborder) or Discosea (class) based on the closest BLASTn hit. Thus, a separate phylogenetic analysis was performed for each group. The most recent phylogenetic models were used as the backbone to the analysis, which included those of Tyml and Dyková (2017) for Tubulinea, Sibbald et al. (2017), Kudryavtsev and Gladkikh, (2017) and Udalov et al. (2016) for Discosea, in addition to an overall Amoebozoa guide from Kang et al. (2017). Several different sequence datasets were assembled using publicly available databases (NCBI; www.ncbi.nlm.nih.gov/, EMBL; www.ebi.ac.uk/ena, DDBJ; www.ddbj.nig.ac.jp/index-e.html) to varying taxonomic extents. Most of the reference sequences chosen from the databases were as long as possible, around 2000 bp, and were derived from well- characterised strains from recognised culture collections such as the American Type Culture Collection (ATCC; www.atcc.org/), the Culture Collection of Algae and Protozoa (CCAP; www.ccap.ac.uk/) or the Institute of Parasitology - Biology Centre, Czech Republic (Dyková and Kostka, 2013).

The sequence datasets were aligned in MAFFT v. 7 (Katoh et al., 2017) using the G-INS-I algorithm for the Tubulinea dataset, and the E-INS-I algorithm for the Discosea dataset. The alignments were trimmed using trimAl v.12 (Capella-Gutiérrez et al., 2009) with -gt 0.3 -st 0.001 restrictions for both the Tubulinea and Discosea analysis. All alignments were inspected in AliView v. 1.18.1 (Larsson, 2014). The tree was constructed in IQ Tree online version (http://iqtree.cibiv.univie.ac.at) (Trifinopoulos et al., 2016) with a model selected in the built-in Model Finder (Kalyaanamoorthy et al., 2017), standard bootstrapping (100 bootstrap alignment; (Felsenstein, 1985)) and an approximate Bayes test (Anisimova et al., 2011). The labelling of higher taxa on the final two analyses was based on Kang et al. (2017). The focus group of the Tubulinea tree was Elardia, and the outgroup was Echinamoebidia and Leptomyxida. The focus group of the Discosea tree was Vannellida and Dactylopodida, and the outgroup included Acanthamoebidae, Dermamoebida and Thecamoebida.

3.3.8 Phylogenetic analysis of amoeba COI sequences

To determine the most likely identity of each newly obtained amoeba COI sequence the analysis was constructed using all major Amoebozoa groups currently characterised by COI. The analysis was carried out as for the Tubulinea dataset with one variation in that trimming was with the –automated1 restriction. The labelling of higher taxa of interest on the final tree was based on Kang et al. (2017), with the main focus group being Vannellida and Dactylopodida; the outgroup comprised fungi.

3.4 Results

3.4.1 Morphological characterisation of amoeba monocultures

The attached and floating forms of 11 monocultures established from the primary mixed- cultures were documented under light microscopy (Figure 3.1) to support assigning each strain to its lowest practical taxonomic level. The amoeba strains displayed morphological characteristics consistent with either Dactylopodida, Vannellida or Tubulinea (Smirnov et al., 2011). The Dactylopodida strains, MP1, MP2, MX6 and MX1, were all laterally flattened with finger-like pseudopodia. Meanwhile, strains MV2, MV3, MV4 and MV5 had the distinct fan-shaped morphotype of Vannellida. Strains MX4, MX3 and MX5 were classed as Tubulinea based on their tubular, elongated morphotype and hyaline cap at the anterior during locomotion. Cyst-like forms were observed in all 11 strains, most commonly one week post sub-culture. Vast inter- and intra- strain size variation was evidenced by trophozoite dimensions available in Appendix A, Table A.1

Figure 3.1 Light microscopy of cultured amoeba strains isolated from gills of farmed Atlantic salmon Salmo salar with signs of AGD, including (a) attached trophozoite and (b) floating form. Strain MP1 and MP2 is Neoparamoeba perurans, MX6 is Vexillifera sp., MX1 is Pseudoparamoeba sp., MV5, MV2, MV3 and MV4 is Vannella sp., MX4 is Tubulinea-like Amoebozoa, MX3 and MX5 is Nolandella sp. All scale bars = 20 µm.

The attached forms of six of the 11 monocultures were further characterised by imaging ultrastructure via TEM. Not all monocultures were documented, partially to avoid repeat processing of the same species, but also due to loss of cultures. Despite not all monocultures being documented, each genus is represented.

The general ultrastructure of MP1 corresponded to previous description of N. perurans (Wiik-Nielsen et al., 2016) and other Neoparamoeba species (Dyková et al., 2005b). The most conspicuous characteristic of MP1 was the Perkinsela amoebae-like endosymbiont lying adjacent to the nucleus (Figure 3.2a, 3.2b). Other defining characteristics were the lack of scales, a rather thin 10 nm amorphous glycocalyx (Figure 3.2c), the relatively small spherical mitochondria (Figure 3.2a), and the golgi apparatus located in the perinuclear zone (Figure 3.2b).

Figure 3.2 Ultrastructure of Neoparamoeba perurans (strain MP1) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n) adjacent to endosymbiont (en), mitochondria (m), golgi apparatus (g), vesicles (v), vacuole (va), phagosome (p). (b) Ultrastructure of Perkinsela amoebae-like endosymbiont (en) adjacent to trophozoite nucleus (n) and golgi apparatus (g): nucleus of endosymbiont (nu) and mitochondrion (m) with darkly stained kinetoplast DNA. (c) Cell surface with a very thin amorphous glycocalyx (arrows).

MX6, did not respond well to freeze fixation, which was evident by the cytoplasmic degradation and dilation of nuclear membranes (Figure 3.3a). We included these images alongside the higher quality figures because they show the defining ultrastructure features, such as structure of mitochondria cristae and cell surface glycocalyx. These ultrastructures were congruent with previous descriptions of Vexillifera sp. (Dyková et al., 2011a; Page, 1983). For instance the mitochondria had tubular branching cristae (Figure 3.3c) and the nuclei had a slightly darker stained nucleolus (Figure 3.3a) (Dyková et al., 2011a). The glycocalyx (Figure 3.3b) was approximately 60 nm thick and comprised of glycostyles (arrows) that looked like cylinders in longitudinal sections. This glycocalyx thickness and form aligned with the descriptions of Vexillifera sp. by Page, (1983) (60-70 nm), however in this instance the freeze fixation was not adequate to determine whether each glycostyle had a hexagonal form in cross section, which is also a key feature of the Vexillifera genus (Page, 1983).

Figure 3.3 Ultrastructure of Vexillifera sp. (strain MX6) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n), mitochondria (m), phagosome (p), vacuoles (va) vesicle (v). (b) Cell surface with approximately 60 nm thick glycocalyx made up of cylinder-like glycostyles (arrows). (c) Mitochondria with tubular cristae.

MX1 had several pseudopodia extending from the hyaloplasm (Figure 3.4a), indicating the amoeba was likely fixed during locomotion, and belongs to the Dactylopodida taxa. The MX1 trophozoite cell surface was lined with domed scales (arrows Figure 3.4b) similar to those described in Pseudoparamoeba microlepis by Udalov, (2016) and for the genus Pseudoparamoeba by Page, (1983). However the scales of MX1 were not as elevated from the cell membrane compared with those imaged for Pseudoparamoeba microlepis (Udalov, 2016). The cytoplasm contained a prominent vesicular nucleus and many darkly stained mitochondria (Figure 3.4a) that appeared to have branched tubular cristae (Figure 3.4d). The golgi apparatus featured a parallel arrangement of cisternae (Figure 3.4c).

Figure 3.4 Ultrastructure of Pseudoparamoeba sp. (strain MX1) isolated from gills of Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructures: nucleus (n), mitochondria (m), golgi apparatus (g) vesicles (v), phagosome (p). (b) Trophozoite cell surface lined with domed scales (arrows). (c) Golgi apparatus with parallel arrangement of cisternae. (d) Mitochondria with branching tubular cristae.

The ultrastructures of MV3 (Figure 3.5) and MV4 (Figure 3.6) had features consistent with the genus Vannella, including an extensive hyaloplasm and distinct glycocalyx (Bovee, 1965; Page, 1983). The main differentiating features of these Vannella strains are the trophozoite size, with 45

MV3 approximately two times larger than MV4, and the cell surface structure. MV3 cell surfaces had a thin amorphous glycocalyx, approximately 10 nm thick (Figure 3.5d), while MV4 had a thick glycocalyx differentiated into distinct glycostyles projecting approximately 60 nm from the cell membrane (Figure 3.6b). Both Vannella strains had mitochondria with tubular branching cristae (Figure 3.5b, 3.6c). We were unsure whether the darkly stained inclusions in the mitochondria of MV3 (Figure 3.5a, 3.5b) were an unusual and yet to be described ultrastructure, or a processing artefact. However, the dark round structures appeared to be site-specific, as they were seen in a number of different mitochondria and not in any other organelle.

Figure 3.5 Ultrastructure of Vannella sp. (strain MV3) isolated from gills of Atlantic salmon,

Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: nucleus (n), mitochondria (m), vesicles (v), early-stage phagosome (p), late-stage budding phagosome (bp). (b) Mitochondria with tubular branching cristae. (c) Golgi apparatus with parallel arrangement of cisternae. (d) Cell surface with amorphous glycocalyx (arrows).

Figure 3.6 Ultrastructure of Vannella sp. (strain MV4) isolated from gills of Atlantic salmon, Salmo salar farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructure: mitochondria (m), vacuoles (va) vesicle (v). (b) Cell surface with amorphous glycocalyx (arrows). (c) Mitochondria with tubular branching cristae. (d) Vesicular nucleus.

MX5 looked similar to images of Nolandella sp. published in Dyková and Kostka, (2013) and Bermingham and Mulcahy, (2007). Similarities included the relatively large mitochondria in relation to the size of the nucleus (Figure 3.7a), and the presence of granular endoplasmic reticulum curled around each mitochondrium with tubular branching cristae (Figure 3.7b). According to Page, (1983) the cell surface of Nolandella comprises tightly packed hexagonal elements rising 30 nm above the membrane. The thickness of this strain’s glycocalyx was consistent with the Page, (1983) description (25 nm), however we did not manage to section a plane that confirmed or ruled-out the presence of hexagonal glycostyles (Figure 3.7c). Other notable features include the vesicular nuclei with a prominent bi-layered nuclear membrane (Figure 3.7a).

Figure 3.7 Ultrastructure of Nolandella sp. (strain MX5) isolated from gills of Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. (a) Overview of trophozoite ultrastructures: vesicular nucleus (n), mitochondria (m), vacuoles (va), vesicles with unknown content (v), endoplasmic reticulum (e). (b) Granular endoplasmic reticulum encircles mitochondria with tubular cristae. (c) Trophozoite cell surface with a 25 nm thick glycocalyx (arrows) covering the cell membrane.

3.4.2 Genetic characterisation of mixed and monocultures

18S rRNA and COI amplicons of the expected size were obtained from all five primer sets used. The average amplicon length was 841 bp for 18S rRNA and 874 bp for COI sequences. The molecular clones were very similar to the original sequences, with 1-3 different bases detected at the tail ends of 18 out of the 36 sequences cloned. Based on the high similarity between the molecular clones, only one of each was used in the final phylogenetic analysis.

3.4.3 Phylogenetic analysis of all amoebae detected

Both the Tubulinea and Discosea 18S rRNA analysis were congruent with former Amoebozoa analysis (Kang et al., 2017; Kudryavtsev and Gladkikh, 2017; Tyml and Dyková, 48

2017), as many typical higher taxa were found grouped within the trees presented. All the newly sequenced strains within Tubulinea had high homology with sequences representative of genus Nolandella, (Clade 1 (C1) in Figure 3.8). They were not identical to any described species, but were nested as a separate, well-supported clade within Nolandella (ML bootstrap; ML 100, Bayesian probability; BP 1).

Figure 3.8 Maximum likelihood analysis of taxa from Tubulinea 18S rRNA gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Echinamoebidia and Leptomyxida serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures.

The newly sequenced strains from Discosea were nested in either Dactylopodida or Vannellida (Figure 3.9). The sequences identified as Dactylopodida represented four genera, Neoparamoeba, Vexillifera, Paramoeba and Pseudoparamoeba. As expected, N. perurans was recovered in mixed-cultures and monocultures, shown by the cluster including strain 183MP1, 249- 1MP2 and 279SVA (ML 100, BP 1). Despite this grouping, the three N. perurans sequences were not identical, indicating possible intra-species variation within the same geographic region. Of the Dactylopodida monocultures, strain 322MX6 was closely related to Vexillifera tasmaniana (ML 74, BP 0.99) and strain 333-1MX1 was nested with the genus Pseudoparamoeba (ML 100, BP 1) but was not identical to any currently described species. Finally, strain 106KRT and 107-1HRT obtained from two different mixed-cultures corresponded to the marine amoeba Paramoeba eilhardi in a well-supported clade (ML 100, BP 1).

Figure 3.9 Maximum likelihood analysis of taxa from Discosea 18S rRNA gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Centramoebia serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures.

Within Vannellida, strain 285MV5 was closely related to Vannella australis (ML 94, BP 0.98), and strain 282-2MV4 as homologous with Vannella sp. strain PMCH-II (ML100, BP 1). Additionally, strains 205MV3 and 243MV2 were grouped with Vannella septentrionalis, however this relationship was weakly supported (ML 33, BP 0.38). Interestingly, two groups with unclear positions formed part of the basal Vannellida genera, including Paravannella, Ripella, Lingulamoeba and Clydonella. The first clade (C2), comprising strain 149-1SVA, 147SVA and 151-1SVA, was quite separate to all other genera within Vannellida. The second clade (C3), including 136SVA and 153-2SVA, was nested within the basal Vannellida genera, however this had a poorly supported position (ML 42, BP 0.59).

The analysis derived from COI sequences (Figure 3.10) was similar to, but not completely congruent with, former 18S rRNA and multi-gene-based Amoebozoa phylogenies (Kang et al., 2017; Kudryavtsev and Gladkikh, 2017; Tyml and Dyková, 2017). For instance, the main focus groups of this study, Dactylopodida and Vannellida, clustered separately. However, they were not as closely related to each other in terms of phylogenetic distance compared with findings derived from 18S rRNA-based analysis by Udalov et al. (2016) and multi-gene-based analysis by Kang et al. (2017), which cluster these groups together in Discosea. An additional difference is evident by the position of the reference sequences of Parvamoeba rugata and Cochliopodium minus which were grouped within Himatismenida in Kang et al. (2017) but were quite distant in this analysis. Considering this is one of the few published Amoebozoa phylogenies based on COI rather than 18S rRNA, some disparities should be expected and should not render the analysis redundant, rather an alternative perspective on Amoebozoa evolutionary relationships.

Figure 3.10 Maximum likelihood analysis of taxa from Amoebozoa COI gene sequences. Numbers at the nodes represent ML bootstraps (ML) and Bayesian posterior probability (BP). Only values higher than 80 and 0.8 are presented. Black dots indicate 100/1 support values. Fungi serves as the outgroup. Taxon and strain names are listed before GenBank accession numbers. Strains in bold are the newly obtained sequences, and ‘mon’ refers to sequences obtained from monocultures and ‘mix’ refers to sequences from mixed-cultures.

Similar to the 18S rRNA analysis, this COI phylogeny resolved that most newly sequenced amoeba strains on AGD-affected gills were closely related to either Dactylopodida or Vannellida. Due to the lower resolution of the COI phylogeny, the identities of the monoculture COI sequences were based on the 18S rRNA tree in Figure 3.9. Four Dactylopodida species were recovered by sequencing COI, including N. perurans which formed a well-supported clade labelled as C4 (ML 100, BP 1), another unidentifiable Neoparamoeba species clustered in C5 (ML 100, BP 1), Pseudoparamoeba sp. (306MX1) and Vexillifera sp. (336MX6). Of the Vannellida species, the four Vannella monocultures (297MV3, 302MV2, 296MV5, 346MV4) fell into a well-supported clade (ML 100, BP 1). Finally, there were three additional strains sequenced from mixed-cultures (22IXB, 49TGR, and 30SVA) that appeared to be neither Dactylopodida nor Vannellida but could not be confidently identified to genus level. Of these three unidentified strains, 22IXB and 49TGR clustered with Squamamoeba japonica, however this position was not well supported by ML bootstraps (ML 33, BP 0.98), and the other remaining strain (30SVA) did not group with any currently recognised Amoebozoa COI reference sequence. These three strains may represent additional novel species.

3.4.4 Diversity of amoebae detected on AGD-affected gills

In total, 12 different species were detected on the gills of AGD-affected Atlantic salmon farmed in Tasmania. There were three additional amoeba strains sequenced with COI primers which could not be identified (22IXB, 49TGR, 30SVA), but they could represent additional novel species. Each of the Amoebozoa listed in Table 3.2 were characterised to varying taxonomic extents. The most well characterised species were: N. perurans; Pseudoparamoeba strain MX1; Vexillifera strain MX6; and Vannella strain MV3 and strain MV4. Gross morphology and ultrastructure were documented and fragments of both the 18S rRNA and COI genes were sequenced. Table 3.2 also shows the frequency each species was recovered from the gill samples. This information is unlikely to accurately reflect true species abundance on the gills, rather it indicates that some species were recovered more than once at different sample times, in particular N. perurans and Nolandella sp., and therefore were more likely to frequently colonise AGD-affected gills.

Table 3.2 Summary of all amoeba species/strains detected on the gills of Atlantic salmon Salmo salar farmed in Tasmania with signs of AGD, and their method of detection and characterisation.

Amoeba Frequency of recovery Sample used characterisa from gills Gene GenBank Taxonomic for amoeba Strain tion fragment Primers accession assignment detection Jun Oct May Aug (LM, TEM, sequenced number (Mix, Mon)† 2015 2015 2016 2017 SS)‡ Neoparamoeba MP1, MP2, Mix, Mon LM, TEM, 1 4 1 18S rRNA, S12.2Y-RibB MH535932 perurans 279SVA, SS COI Ami6F1- MH535934 82HRT, 4IXB, Ami9R MH535940 26SVA Eucox1F- MH535946 Euglycox1R MH535948 MH535959 MH535962 MH535963 Vexillifera sp. MX6 Mon LM, TEM, 1 18S rRNA, Ami6F1- MH535945 SS COI Ami9R MH535966 Eucox1F- Euglycox1R Pseudoparamoeba MX1 Mon LM, TEM, 1 18S rRNA, S12.2Y-RibB MH535944 sp. SS COI Eucox1F- MH535967 Euglycox1R Vannella australis MV5 Mon LM, SS 1 18S rRNA, S12.2Y-RibB MH535941 COI Eucox1F- MH535965 Euglycox1R Vannella sp. MV2 Mon LM, TEM, 2 18S rRNA, S12.2Y-RibB MH535942 MV3 SS COI Ami6F1- MH535943 Ami9R MH535960 Eucox1F- MH535961 Euglycox1R Vannella sp. MV4 Mon LM, TEM, 1 18S rRNA, 570C - MH535947 SS COI 1137R MH535964 LCO1490- HCO2198 Tubulinea MX4 Mon LM 1 N/A N/A N/A Nolandella sp. (C1) MX3, MX5, Mix, Mon LM, TEM, 1 1 1 18S rRNA S12.2Y-RibB MH535949 105KRT SS MH535950 MH535951 Paramoeba eilhardi 106KRT, 107- Mix SS 2 18S rRNA S12.2Y-RibB MH535952 1HRT MH535953 Vannellida (C2) 149-1SVA, Mix SS 3 18S rRNA S12.2Y-RibB MH535955 147SVA, 151- MH535956 1SVA MH535957 Vannellida (C3) 136SVA, 153- Mix SS 2 18S rRNA S12.2Y-RibB MH535954 2SVA MH535958 Neoparamoeba sp. 73BVA, Mix SS 2 1 COI Eucox1F- MH535937 (C5) 58NEB, 66KRT Euglycox1R MH535938 MH535939 COI strain 22 22IXB Mix SS 1 COI Eucox1F- MH535933 Euglycox1R COI strain 49 49TGR Mix SS 1 COI Eucox1F- MH535936 Euglycox1R COI strain 30 30SVA Mix SS 1 COI Eucox1F- MH535935 Euglycox1R † Mix = mixed-culture, Mon = monoculture ‡ LM = light microscopy, TEM = transmission electron microscopy, SS = Sanger sequencing

3.5 Discussion

A diversity of amoebae was found accompanying N. perurans on the gills of farmed Tasmanian Atlantic salmon with characteristic AGD pathology. Six species were differentiated by both the 18S rRNA and COI gene, four by 18S rRNA gene alone, one by the COI gene only, and one species based on its distinct morphotype compared to all other genera sequenced in this study. The sequence-based taxonomic assessment of the strains grown in monoculture was well supported by trophozoite gross morphology and ultrastructure. Using a combined genetic and morphological approach effectively characterised a variety of amoeba species, in addition to N. perurans, colonising the gills of farmed Atlantic salmon with AGD. Significantly, we provide the first report of P. eilhardi being isolated from the gills of fish, and of Pseudoparamoeba sp. being isolated from the gills of Atlantic salmon, to our knowledge. Furthermore, considerable genetic variability within genera was observed across both the COI and 18S rRNA loci, revealing new phylogenetic clades and lineages not previously described. It is important to note the diversity of amoebae captured in this Tasmanian based study may not reflect the diversity of amoebae colonising the gills of AGD- affected fish elsewhere in the world.

Six genera of amoeba were detected on the gills and, of these, Vannella offered the greatest species diversity in this study. Vannella spp. are frequently isolated from both freshwater and marine environments, and have been cited as the most common amoeba isolated from various teleost organs, in particular the gills (Dyková et al., 2005a; Dyková and Lom, 2004; Smirnov et al., 2007). Vannella spp. are generally considered non-pathogenic due to their frequent detection on healthy gills (Dyková et al., 2005a), and a challenge trial which found one Vannella strain (previously referred to as Platyamoeba sp.) isolated from Atlantic salmon in Ireland was not associated with gill lesions (Nowak et al., 2004). Indeed, while a number of studies have isolated Vannella spp. from the gills of both asymptomatic and diseased fish, including Atlantic salmon with AGD and rainbow trout with NGD (Bermingham and Mulcahy, 2007; Dyková et al., 2010; Dyková et al., 2005a; Dyková and Tyml, 2015), the true role Vannella spp. may play in these diseases remains unknown.

Within the Vannellida phylogeny presented here, five newly obtained 18S rRNA sequences formed two well separated clades (referred to as C2 and C3) positioned within the basal Vannellida genera (i.e. Paravannella, Ripella, Lingulamoeba and Clydonella). Although their position is unstable, C2 and C3 represent two new lineages, possibly genera, because they are quite different 56 from any other currently recognised Vannellida genera. Whether C2 and C3 are in fact new genera cannot be established with certainty from the data currently provided. To confirm this finding the complete 18S rRNA gene sequence is needed, as well as gross and fine-scale morphological features. Unfortunately, the strains that make up C2 and C3 were all sequenced from mixed- cultures, hence type material is not available to carry out the necessary work required to fully characterise these lineages.

Of the Dactylopodida strains isolated, as expected N. perurans was detected during multiple sampling events, and from both mixed- and monocultures. This finding likely indicates that N. perurans is the most abundant species on salmon gills. Vexilliffera tasmaniana has been previously detected on Atlantic salmon gills, as it was first isolated from Atlantic salmon held within an AGD experimental facility in Tasmania (Dyková et al., 2011a). Vexilliffera sp. has also been found on the gills of farmed Atlantic salmon with AGD in Ireland, though this taxonomic assessment was based only on morphology (Bermingham and Mulcahy, 2007). While there are a few instances of Vexilliffera spp. isolated from AGD-affected fish, there have also been multiple detections from various other asymptomatic fish gills, including bitterling Rhodeus sericeus, banded leporinus Leporinus fasciatus, vimba bream Vimba vimba, turbot Scophthalmus maximus, and Nile tilapia Oreochromis niloticus (Dyková and Kostka, 2013).

Interestingly, the N. perurans strain (MP1) examined under TEM in this study repeatedly showed ultrastructural features of the endosymbiont which looked different to all other currently published images (Dyková et al., 2003; Nowak and Archibald, 2018; Wiik-Nielsen et al., 2016). In particular, the darkly stained arrangement of kinetoplast DNA appear to be in distinct elongated structures which were not as tightly associated with each other compared to previous detailed characterisation by Dyková et al. (2003). This difference was predicted to have been caused by methodological approach. High pressure freeze fixation and freeze substitution was used, as opposed to the common approach of using glutaraldehyde-based chemical fixation (Bermingham and Mulcahy, 2007; Dyková et al., 2005a; Udalov, 2016). We also trialled chemical fixation but found freeze fixation generated images with far greater detail and fewer artefacts. Other studies which compared these methods in detail also found that freeze fixation and freeze substitution improved the preservation of biological specimens, with less evidence of shrinkage or extraction artefacts (Harahush et al., 2012; Kurth et al., 2012). To our knowledge, this is the first study to document these particular amoebae in their attached form and preserved by high pressure freeze

57 fixation. For these reasons the TEM images published in this study present an alternative perspective of some Amoebozoa ultrastructures, and possibly represent a more accurate depiction.

Nolandella sp. was the only well-characterised Tubulinea species isolated and was the second most frequently isolated species in this study. Nolandella sp. has also previously been found on the gills of Atlantic salmon farmed in Ireland (Bermingham and Mulcahy, 2007), but again this taxonomic assessment was based on morphology alone. Additionally, two well characterised Nolandella strains have colonised the gills of turbot, Scophthalmus maximus and rainbow trout, Oncorhynchus mykiss (Dyková and Kostka, 2013; Dyková and Novoa, 2001). Nolandella spp. are not considered to cause pathology because several strains closely related to N. abertawensis (species formerly attributed to Hartmannella genus, transferred by (Smirnov et al., 2011)) has been isolated from the liver, spleen, kidneys and gills of various healthy teleost species (Dyková and Kostka, 2013).

Whilst the results of our study represent the most comprehensive list of amoeba species colonising AGD-affected salmon gills to date, it is likely the diversity of amoebae is higher. For instance, two species that have been isolated from salmon gills in the past, N. pemaquidensis and N. branchiphila, were not detected in this study (Dyková et al., 2011b; Young et al., 2007). Additionally, there were three COI gene fragments that could not be identified to genus level hence it is unclear whether they represent novel species, thereby elevating species diversity. The reason for missed species is probably due to the low abundance of each species on the gills, and the methodological bias as highlighted below.

Two methods were used to address the aim of this study, including sequencing mixed- cultures, and sequencing alongside morphological characterisation of monocultures. Each approach generated a different community of amoebae species as summarised in Table 3.2. Establishing monocultures was ideal for comprehensively characterising species, as they provided a robust way of linking the gross and fine-scale morphological features with two variable genetic regions. However, only eight of the total 12 species were established as monocultures, which supports the premise that not all amoebae are easily grown in pure culture. Sequencing mixed-primary isolates from the gills lead to the detection of four additional strains. This approach likely yielded different species because it increased the chance of sequencing species that cannot grow in pure culture, and less abundant cryptic species such as N. pemaquidensis and N. branchiphila (Young et al., 2007). However, the two methods used to document species diversity in this study do not account for 58 species that cannot grow in vitro and are biased towards selecting amoebae that are more capable of growing in the chosen culture conditions.

In this study two Amoebozoa barcoding regions, 18S rRNA and COI, were employed to examine the sequence diversity of amoebae on the gills of Atlantic salmon. 18S rRNA is the most commonly used gene to decipher Amoebozoa phylogenetic relationships and identify species (Tyml and Dyková, 2017; Udalov et al., 2016; Volkova and Kudryavtsev, 2017). Accordingly, 18S rRNA was the most taxonomically informative gene in this study due to the many reference sequences available on public databases. However 18S rRNA can be problematic for barcoding amoebae and developing molecular detection assays because it is too conserved between species and can contain intra-strain polymorphism (Nassonova et al., 2010; Smirnov et al., 2007; Young et al., 2014). Indeed, large conserved regions were evident throughout our newly sequenced amplicon of 18S rRNA representing multiple distinct species. For instance, 58% of base pairs were conserved between the 15 strains sequenced with the RibB-S12.2 primers. Intra-species polymorphisms were also detected within the 18S rRNA fragments representing three N. perurans strains and three Nolandella strains sequenced in this study.

COI is currently considered a promising gene for amoeba species barcoding because it can provide higher phylogenetic resolution between closely related species compared to 18S rRNA (Nassonova et al., 2010). However this finding has only been evidenced in a few amoeba genera, Vannella (Nassonova et al., 2010), Cochliopodium (Tekle, 2014) and nebelid testate amoebae (Hyalosphenia, Nebela, Quadrulella, Padaungiella) (Kosakyan et al., 2012). This indicates that more amoeba COI sequences are needed to further validate the utility of this gene for amoeba barcoding and taxonomy. We have contributed 17 new Amoebozoa COI fragments to GenBank, all of which represent amoeba that have not been previously characterised with the COI gene sequence. These sequences will contribute to future Amoebozoa sequenced-based taxonomy.

3.6 Conclusions

N. perurans and up to 11 other amoeba species colonise the gills of farmed Tasmanian Atlantic salmon with AGD. While N. perurans is the primary pathogen of AGD, the role of these accompanying amoebae in AGD remains unclear. Future research should investigate whether

59 amoeba species other than N. perurans are parasitic or commensal, and whether AGD is associated with Amoebozoa dysbiosis which potentially influences disease onset or progression.

Chapter 4: Prevalence of six amoeba species colonising the gills of farmed Atlantic salmon with amoebic gill disease (AGD) using qPCR

Statement of author contributions M.C., P.L. and A.B. conceptualised the broader project idea. C.E. refined the project design and carried out the research investigation, with training and resource administration provided by N.B., P.L. and J.W.. C.E. wrote the manuscript under the supervision of J.W., N. B., A. B., P. L. and M.C. R.T helped design the survey and collected 50 % of the gill swab samples. C.E., F.S., J.D., and N.R. designed and validated the qPCR assays. All authors contributed to reviewing and editing the final manuscript.

4.1 Abstract

Amoebic gill disease (AGD) is the primary health concern for Atlantic salmon, Salmo salar, farmed in Tasmania, Australia. Neoparamoeba perurans is the aetiological agent of AGD, however a diversity of other amoebae colonise the gills and their role in AGD is unknown. Previous studies that document these accompanying amoebae on AGD-affected farmed Atlantic salmon relied on culture-based techniques which do not accurately determine the prevalence and abundance of these species, nor whether they correlate with AGD pathology. Drawing on our previous culture-based study, here we develop and apply five new Taqman quantitative PCR (qPCR) assays to profile the prevalence of multiple amoeba species on the gills of AGD-affected Atlantic salmon held at two Tasmanian farm sites over a one year period. The prevalence and abundance of N. perurans was also assessed using a previously established qPCR method. N. perurans was the dominant species, and its abundance positively correlated with the progression of gross gill pathology. Only a small number of sporadic detections of Pseudoparamoeba and Vannellida species were observed. Nolandella spp. was the notable exception, as it was the most prevalent amoeba (92 %) at one site at one sample time, during which no N. perurans were detected on gills but low levels of gross gill

61 pathology were observed. N. perurans is the predominant species and primary pathogen of AGD, however there were instances when they were not detected on diseased gills and Nolandella spp. were highly prevalent. The significance of Nolandella spp. in relation to AGD is not yet understood.

Keywords Neoparamoeba, Nolandella, Amoebozoa, AGD, Taqman qPCR, Atlantic salmon

4.2 Introduction

Gill health is fundamental to the success of finfish aquaculture. These delicate organs are responsible for gas exchange, acid-base balance, waste excretion and osmoregulation, yet are constantly exposed to a variety of microorganisms and environmental insults which can result in various gill diseases (Evans 2005, Mitchell & Rodger 2011). While the aetiologies of many disease pathologies are attributed to single pathological agents, the drivers of gill disease can be more complex (Gjessing et al. 2017, Herrero et al. 2018). A multifactorial approach to disease occurrence that considers the impact of environmental factors and microbial community structure, rather than focusing on a single agent, may therefore be a rational approach to gain a better understanding of disease in marine ecosystems (Egan and Gardiner, 2016) and, in this case, gill disease.

Amoebic gill disease (AGD) is a globally significant disease in marine aquaculture and remains the main health issue challenging Atlantic salmon, Salmo salar, aquaculture in Tasmania, Australia. AGD is caused by Neoparamoeba perurans (syn. Paramoeba perurans; Feehan et al. 2013), a free-living protozoan ectoparasite which colonises the gills of Atlantic salmon in the marine phase of production (Young et al. 2007, Crosbie et al. 2012). Attachment of N. perurans causes lamellar fusion, epithelial hyperplasia, oedema and ultimately mortality when left untreated (Adams and Nowak, 2001). While it is understood that salinity, water temperature and stocking density play a role in AGD prevalence and severity (Oldham et al. 2016), all the drivers of infestation and correlates of the associated gill pathology are not yet understood. This could contribute to inefficiencies in disease prediction and control during Atlantic salmon marine grow- out.

In addition to N. perurans, a diversity of other amoeba species colonise the gills of AGD- affected Atlantic salmon, with up to 11 other species having been isolated from farmed salmon in Tasmania, including species of the genera Paramoeba, Vexillifera, Pseudoparamoeba, Vannella and Nolandella (English et al., 2019b; Howard, 2001). However, it is not understood whether these accompanying amoeba species are parasitic or commensal, or if changes to the Amoebozoan community structure influence disease onset or progression. Moreover, the seasonal abundance of these amoebae and whether their presence correlates with AGD prevalence and severity of pathology is also unknown. Previous studies of gill-associated Amoebozoa identified the amoebae after isolating and culturing in vitro (Bermingham and Mulcahy, 2007; English et al., 2019b; Howard, 2001). These methods likely generate a skewed assessment of infection prevalence and abundance because they are biased towards amoebae that are more capable of growing in the 63 chosen culture conditions. A more accurate, sensitive and high-throughput technique for detecting and quantifying particular microorganisms is quantitative PCR (qPCR) (Purcell et al. 2011). There are three published qPCR assays for species-specific detection of N. perurans (Bridle et al. 2010, Fringuelli et al. 2012, Downes et al. 2015). These assays are versatile tools for research, being used to answer a variety of AGD-related questions, including retrospective validation that N. perurans was the main causative agent in the earliest AGD outbreaks in Ireland (Downes et al. 2018) and confirmation that Atlantic salmon can be experimentally infected from lumpfish Cyclopterus lumpus, a cleaner fish which is often cohabitated with farmed salmon to control sea lice (Haugland et al. 2017). Despite these assays being highly informative and adaptable tools for disease and ecological research, there are no qPCR assays designed for the variety of other marine amoebae which colonise AGD-affected fish.

In a recent study, we described 18S rRNA sequences from many of the amoebae which colonise Atlantic salmon gills in Tasmania (English et al., 2019b). These sequences are ideal for developing molecular detection assays which can be used to explore the potential role the Amoebozoan community plays in AGD onset and severity. Here, we design and validate five new amoeba qPCR assays using these 18S rRNA sequences. These assays, along with an existing N. perurans assay (Downes et al., 2015), were then used to assess amoeba prevalence and abundance on the gills of farmed Atlantic salmon over a one year period, from two different farm sites. This longitudinal survey aimed to identify relationships between specific amoeba taxa and farm sites, time points and gross gill pathology.

4.3 Material and Methods

4.3.1 Gill swab sample collection

Every three months, from May 2017 to February 2018, during routine commercial gill checks at Tassal Operations Pty Ltd Tasman and Killala leases (Figure 4.1), large groups of Atlantic salmon were seined and crowded to ensure the fish were haphazardly mixed, then 40 fish were sampled from the crowd. The two farm sites have contrasting environmental conditions, with the Tasman lease being more oceanic, located closer to open water, and the Killala lease more brackish, located in the lower Huon Estuary. The average monthly water temperature and salinity at 5 m before each sampling event was obtained from Tassal Operations Pty Ltd.

Figure 4.1 Atlantic salmon sea pen sites, Killala and Tasman, sampled during the survey.

During sampling, fish were anaesthetised with 15 ppm of AQUI-S then their gross gill pathology was scored according to Taylor et al. (2009) on a scale of zero (no gross lesions) to five (lesions covering > 50 % of gill surface). All 16 gill surfaces were then swabbed (Westlabs) in situ, with one swab used for the left gill arches and one for the right arches to maximize the DNA yield. The left and right swabs for each fish were pooled and stored in RNALater (Ambion) at room temperature for a minimum of two days and then at -20°C until processed for DNA extraction. Other information collected included months since the fish were transferred to sea, the average fish weight and the number of days since the fish were last treated for AGD (2-4 h freshwater bath). The case history of each sampling event is summarised in Table 4.1. All sampling events occurred as planned, apart from February 2018 at Killala as the site was de-stocked as part of standard commercial operations. Sampling of 280 salmon in this study was approved by the CSIRO Queensland Animal Ethics Committee (AEC number 2017-16).

Table 4.1 Case history of sampled Atlantic salmon. NA is not available because no samples could be collected at that sampling event. Farm Sample Months since Average fish Days since last Water temp at Salinity at 5 m site time ocean transfer weight (g) freshwater bath 5 m (°C) (ppt) May-17 1 225 no bath 13.6 31.7 Aug-17 3 719 25 11.9 31.9 Killala Nov-17 7 2530 64 14.7 31.1 Feb-18 NA NA NA NA NA May-17 11 2365 48 12.1 34.5 Aug-17 3 639 35 10.2 31.1 Tasman Nov-17 7 1618 37 15.5 32 Feb-18 10 2734 50 17.5 31.8 65

4.3.2 DNA extraction from gill swabs

DNA was extracted from the preserved gill swab samples using a plate based DNeasy Blood and Tissue Kit (QIAGEN) according to the manufacturer’s instructions. Extracted DNA was quantified with a Nanodrop ND-1000 spectrophotometer (Life Technologies, USA), then a working stock (30 ng/µl) of each sample was prepared using an epMotion 5070 liquid handling robot (Eppendorf, Germany) and stored at -20°C.

4.3.3 TaqMan qPCR assay design

Five TaqMan MGB probes and PCR primer sets were designed based on the 18S rRNA gene sequences of amoebae previously isolated from AGD-affected Atlantic salmon (English et al., 2019b). The chosen amoebae for assay design included Nolandella sp. (strain MX5, accession number MH535951), Pseudoparamoeba sp. (MX1, MH535967), Paramoeba eilhardi (106KRT, MH535952), Vannellida C2 (149SVA, MH535956) and Vannellida C3 (136SVA, MH535954). From here on, these assays, in the order listed above, will be referred to as Nol, Pse, ParE, VanC2 and VanC3. Based on an alignment of 18S rRNA gene sequences from these species and Neoparamoeba perurans (MH535959), the primer and probe sets were designed on Primer Express 3.0.1 (Life Technologies) according to the guidelines. Before ordering the probes, the amplicon derived from PCR using each primer set was sequenced according to English et al. (2019b) to confirm the correct sequence was being amplified. Primers were ordered from GeneWorks and probes from Thermo Fisher Scientific. The optimum probe and primer concentration for a single- plex assay was then determined with a series of experiments following the Thermo Fisher Scientific standard protocols using amplicon-specific plasmid DNA as template (Thermo Fisher, 2010). The amplicon-specific plasmid DNA was generated as described previously (English et al., 2019b).

4.3.4 Validation of reaction efficiency, sensitivity, specificity and reproducibility

Assay validation was performed according to Downes et al. (2015) and Bustin et al. (2009). The following procedures were conducted separately on each of the five qPCR assays. To determine the efficiency of the assays, amplicon-specific plasmid DNA was serially diluted 10-fold and amplified in quadruplicate qPCR reactions. Only dilutions that provided a cycle threshold (Ct) value in all replicates were used to generate a standard curve. Amplification efficiency was calculated based on the Ct slope method (Efficiency = [10(-1/slope)]-1) and the linearity was 66 demonstrated with the coefficient of determination (R2). The equivalent number of 18S rRNA copies corresponding to each standard curve was then determined by measuring the concentration of the plasmid DNA with a NanodropND-1000 spectrophotometer (LifeTechnologies) and then submitting this value and the length of the amplicon insert and vector to the DNA copy number calculator (Staroscik, 2004). The 10-fold dilution series was then used to determine the limit of detection (LOD) for each assay. For this purpose, the lowest dilution that provided a Ct value in all replicates underwent a two-fold dilution and was tested in quadruplicate. The mean of the lowest dilution of the two-fold dilution that provided a Ct value in all replicates was deemed the LOD. This sample was further analysed 20 times to assess the precision of the assay with 95 % confidence.

The specificity of primers and probes was theoretically assessed using NCBI nucleotide Basic Local Alignment Search Tool (BLASTn) to identify potential non-target amplification of other amoebae and protozoans that colonise the gills of marine fish. The actual specificity was assessed by testing the assays against seven in-house amoeba cultures previously isolated from AGD-affected Atlantic salmon gills (N. perurans strain MP2, Vexillifera sp. strain MX6, Pseudoparamoeba sp. strain MX1, Vannella sp. strains MV3, MV4 and MV5 and Nolandella sp. strain MX5) (English et al., 2019b) and four amoeba cultures obtained from American Type Culture Collection (ATCC) (N. pemaquidensis ATCC® 50172 TM, Acanthamoeba jacobsi ATCC® 30732 TM, Nolandella sp. ATCC® PRA-27 TM and Hartmannella vermiformis ATCC® 30967 TM).

The reproducibility of each assay was assessed by testing ten gill swab DNA samples in triplicate on three consecutive days. These samples had previously tested positive for the target amoeba in preliminary experiments. The reproducibility was analysed by the coefficient of variation for intra- and inter-assay variation. The inter-assay differences were also assessed with a one-way ANOVA. All qPCR validation experiments had the appropriate controls, including positive control and no-template negative control.

4.3.5 Inhibition

To determine whether any amplification inhibitors were present within the gill swab DNA, five gill swabs were diluted two-fold seven times such that the highest concentration was 150 ng/µl 1 (undiluted) and the lowest was 2.34 ng/µl ( /64 dilution). All dilutions had a working stock volume of 10 µl. Each of the diluted samples was then spiked with 0.008 ng of plasmid DNA specific to the existing N. perurans assay (NPJ) (Downes et al. 2015). Template (1.8 µl in 5 µl reaction) from each

67 of the spiked dilution series were then tested in triplicate according to the qPCR protocol outlined by Downes et al. (2015).

4.3.6 qPCR analysis of gill swab survey

To quantify the relative abundance of several amoeba taxa on the gills of farmed Atlantic salmon, the five newly designed qPCR assays and one previously published qPCR assay were performed on the gill swabs collected from farm sites. Template (30 ng) DNA was tested in 5 µl reactions using MyTaqTM DNA Polymerase (Bioline) with three technical replicates across a 384- well PCR plate. DNA was amplified using a ViiATM 7 Real-Time PCR Machine (Applied Biosystems, USA) with the thermal cycling conditions specific for each assay (Table 4.2). A 10- fold dilution series of plasmid DNA specific to the amplicon of each assay, with a known copy number, was amplified on each plate to generate a standard curve used to determine the copy number of each sample. Each plate also included a positive control, a negative control and a negative-process control (a blank sample extracted along with the gill swab samples). An external process control (salmon elongation factor-1α) was also run against each sample (Bruno et al. 2007). Samples were determined as positive detections if the average Ct of two or three technical replicates was lower than the limit of detection specific to each assay.

4.3.7 Statistical analysis

All graphics and analysis were performed in R version 3.5.2 (R Core Team, 2018). To evaluate how spatial, temporal and pathological variables influenced the N. perurans infection load, the N. perurans 18S rRNA copy number was firstly log10 transformed after increasing all values by one to account for the zero values. To investigate the correlation between gill score and N. perurans abundance, a linear model was fitted. The mean N. perurans abundance with respect to time and site was compared using a two-way ANOVA and then a Tukey post-hoc test to determine significant pairwise differences among sampling events. A paired t-test was used to compare the mean N. perurans abundance by site, the N. perurans prevalence by site and the mean gill score by site.

4.4 Results

4.1 qPCR assays design and optimisation

After the design and optimisation of TaqMan qPCR assays specific to the chosen amoebae, the probe and primer sequences and reaction parameters were established as listed in Table 4.2. Results related to the primer and probe optimisation are available in Appendix B (Table B1, B2).

Table 4.2 Probe and primer sequences and assay parameters for detecting various amoeba taxa. Probe (Pr) Reaction Annealing Amplicon Amoebae Primer Sequence (5’-3’) conc. (nM) temp. (°C) length (bp) (F/R) Nolandella spp. Nol-Pr AGAGCTTTAGCTTGCCC 250 Nol-F CCGGTGAGGATTCAGGAT 400 63 61 Nol-R TGGCTGAACACGCTTACCCT 400 Pseudoparamoeba Pse-Pr CATCCTGTCCTGACTTGT 250 sp. Pse-F CCAGCAATGGAACGCTTTGT 300 60 67 Pse-R CACCAAGTGTCCCTCTAAGAAGTTAA 300 Paramoeba ParE-Pr TCGTTTTCATTCGTCAGAAT 125 eilhardi ParE-F CATCCTTTATGGGGAGGGTTCTA 300 60 111 ParE-R GAACAGTTTTAACCGAAGTTGCAAC 900 Vannellida C2 VanC2-Pr ATCGTGAGGATATAGTTGCTT 100 VanC2-F CGCTCCTACCGATTGGATGT 300 60 116 VanC2-R CGACTTCTCCTTCCTCTAGATGTTATG 300 Vannellida C3 VanC3-Pr ATATCCTCTTTAGCACCTTTGA 150 VanC3-F CCAGGGATTAGAGGGAGAAACA 300 60 91 VanC3-R CCCCCCAGAATTTATCTCAATG 900

4.2 qPCR assays validation

Each newly designed qPCR assay underwent a series of validation experiments to determine its performance using both plasmid DNA and amoeba DNA as template. The results of these experiments are summarised in Table 4.3 with reference to the supplementary material containing the raw data (Appendix B). Despite NPJ being a previously published assay (Downes et al., 2015), a new standard curve and limit of detection was generated so that it was specific to our qPCR conditions (i.e. machine, taq, plasmid DNA template etc.). Additionally, the precision, reproducibility and specificity experiments were repeated for NPJ, to act as a benchmark assay.

Table 4.3 Summary of validation metrics for the six amoeba Taqman qPCR assays. LOD is limit of detection and CV is coefficient of variation.

Supplementary Nol Pse ParE VanC2 VanC3 NPJ reference Standard curve R2 0.998 0.990 0.996 0.996 0.998 0.998 Figure B.1 Amplification efficiency (%) 82.23 99.98 65.41 72.82 98.28 99.21 Figure B.1 Limit of detection (Ct) 36.37 38.36 36.53 37.21 37.73 38.73 Table B.3 Limit of detection (18S rRNA copies/µl) 13.50 1.35 7871 39.25 1.09 0.66 Table B.3 36.44 39.42 35.96 37.98 35.98 39.48 Precision of LOD (mean±SD) Figure B.2 ±0.57 ±0.58 ±0.57 ±0.62 ±0.62 ±0.54 Intra-assay variance (mean CV %) 0.81 1.02 2.09 1.01 1.45 0.53 Table B.4 Inter-assay variance (mean CV %) 1.06 1.35 2.19 1.31 1.72 0.88 Table B.4

All six assays generated a linear standard curve, indicated by the coefficient of determination (R2) value close to 1. The amplification efficiency varied between the assays, with the Pse being the most efficient at 99.98 % and the ParE being the least at 65.41 %. The limit of detection is first presented in Table 4.3 as the mean Ct of the dilution, which provided a Ct value for all four technical replicates. This value was then converted to 18S rRNA copies and shows that NPJ was the most sensitive assay, detecting up to 0.66 18S rRNA copies/µl, while ParE was the least sensitive with a limit of detection equivalent to 7871 18S rRNA copies/µl. The final dilution, which determined the limit of detection, was tested a further 20 times to determine the precision of the assay at a 95 % confidence level. All precision data, apart from two outliers (Appendix B: Figure B.2), were close to the mean value for each assay, indicating low variability (all SD < 0.7).

The final two values in Table 4.3 refer to the reproducibility of each assay between technical replicates (intra-assay variance) and between different days (inter-assay variance) and are expressed as the coefficient of variation (CV). The mean intra-assay variation of all assays ranged between 0.53 % for the NPJ assay and 2.09 % for the ParE assay. Whereas, the mean inter-assay variation of all assays ranged between 0.88 % for NPJ and 2.19 % for ParE, which were again the best and worst performers respectively. A one-way ANOVA further indicated there was no significant difference between the assays’ results conducted on three different days, as all p-values were greater than 0.05 (Nol: 0.90, Pse: 0.99, ParE: 0.96, VanC2: 0.96, VanC3: 0.99, NPJ: 0.72).

Initially, all the newly designed probes and primers were deemed theoretically specific as they had no complete matches to existing reference sequences in the BLASTn database. The assays specificity was further tested against DNA samples from clonal amoeba cultures, and specific plasmid DNA in the cases where no amoeba DNA-positive control was available. A positive

70 detection was defined by a mean Ct value lower than the limit of detection and is marked with a ‡ in Table 4.4. All assays appeared to be species-specific in terms of the amoeba strains against which they were tested, except for NPJ and Nol. NPJ had some cross-reactivity with Pseudoparamoeba sp., while Nol amplified two Nolandella species, the ATCC Nolandella sp. strain as well as the Nolandella culture it was designed to detect. The Nol assay was therefore defined as genus-specific, rather than species-specific.

Table 4.4 Specificity of the six qPCR assays (Nolandella spp. (Nol), Pseudoparamoeba sp. (Pse), P. eilhardi (ParE), Vannellida species (VanC2 and VanC3) and N. perurans (NPJ)) determined by testing each assay against various amoeba DNA samples listed as taxa (strain). Positive detections are indicated in bold with a ‡ symbol. The – symbol is undetected. DNA samples Nol Pse ParE VanC2 VanC3 NPJ Neoparamoeba perurans (MP2) 41.42 – 38.36 37.51 – 23.03 ‡ Vexillifera sp. (MX6) 39.66 – – – 41.78 – Pseudoparamoeba sp. (MX1) 37.66 22.20 ‡ – 37.42 39.42 35.88 ‡ Vannella sp. (MV3) 36.74 – – 37.62 – – Vannella sp. (MV4) – – – – – – Vannella sp. (MV5) 38.68 – – – – – Nolandella sp. (MX5) 24.50 ‡ – – – – – Neoparamoeba pemaquidensis (ATCC – – 50172) – – – – Acanthamoeba jacobsi (ATCC 30732) – – – – – – Nolandella sp. (ATCC PRA-27) 24.15 ‡ – – – – – Hartmannella vermiformis (ATCC 30967) – – – – – – ParE plasmid – – 15.27 ‡ – – – VanC2 plasmid – – – 13.53 ‡ – – VanC3 plasmid – – – – 6.90 ‡ – No template – – – – – –

To assess the presence of amplification inhibitors in gill swab DNA sourced from Tasmanian salmon farms, five samples were diluted, spiked with the same amount of plasmid DNA, and then tested in triplicate. No inhibition was detected because amplification remained consistent across the dilution series (Appendix B: Figure B.3).

4.3 Survey of amoeba prevalence and abundance

Of the six amoeba species surveyed on the gills of farmed Atlantic salmon, N. perurans was the most prevalent (Figure 4.2). N. perurans was dominant across both temporal and spatial factors,

71 except at one site at a single time point. Atlantic salmon held at Killala during May 2017 were not colonised by detectable levels of N. perurans, rather Nolandella spp. was highly prevalent, infecting 92 % of salmon which had an average gill score of 0.225 (Figure 4.2a). Except for this sampling event, Nolandella spp. was not detected in any other gill samples. Pseudoparamoeba sp. and the two Vannellida species were detected on the gills of a relatively small proportion of Atlantic salmon compared to N. perurans and Nolandella spp., and these few instances were not correlated with temporal or pathological factors. The remaining species, P. eilhardi, was not detected on any Atlantic salmon gills. In terms of the amoeba community on individual fish, non-N. perurans amoeba species were detected on 41 Atlantic salmon throughout the entire survey, and, of these, eight individual fish tested positive for N. perurans and one other amoeba species. There were no individual fish that tested positive for three or more amoeba species colonising the gills.

Figure 4.2 (a-g) Prevalence of Atlantic salmon gills that were colonised by N. perurans (NPJ), Nolandella spp. (Nol), Pseudoparamoeba sp. (Pse), and Vannellida (VanC2, VanC3) at each farm site (Killala, Tasman) over a year. P. eilhardi was not included because no Atlantic salmon tested positive for the amoeba species. The average gill score (Gill Index (GI)) for each site and each time point is marked in the respective panels. The Feb-18, Killala sampling event marked with ‡ is blank because no samples could be collected.

The relative abundance of N. perurans was further analysed to explore correlations between gross gill pathology, farm site and time points. The relative abundance of N. perurans increased with increasing gill score, indicated by the upward trend of the regression line fitted to Figure 4.3a.

Although this model was significant (p = 2.2 x 10 -16), only 25 % of the variability in N. perurans abundance could be explained by gross gill pathology (R2 = 0.247). In terms of temporal and spatial factors (Figure 4.3b), there were differences by site (p = 9.57 x 10 -5), but when considering both there was a significant difference between May-17 and Aug-17 (p = 8 x 10 -7), Nov-17 and Aug-17 (p = 5 x 10 -6) and May-17 and Feb-18 (p = 0.004). The trend in N. perurans abundance across site and time (Figure 4.3b) also reflected the fluctuations in N. perurans prevalence (Figure 4.2), in that the number of Atlantic salmon infected with N. perurans increased with the N. perurans load on the gills. Abundance also aligned with the severity of gross gill pathology, shown by the average gill scores marked on Figure 4.2.

Figure 4.3 Relative abundance of N. perurans on Atlantic salmon gills expressed as the mean ± standard error of 18S rRNA copies with respect to (a) AGD gross gill pathology (gill score) quantified according to Taylor et al. (2009) and (b) farm site, and sample time. The Feb-18, Killala sampling is blank because no samples could be collected.

When considering site-based differences alone, the overall N. perurans infection prevalence significantly differed (p = 1.374 x 10 -6) with 47.4 % of salmon sampled at Killala infected compared to 76.1 % at Tasman. The abundance of N. perurans on the gills was also significantly different between the sites (p = 2.232 x 10 -5), as salmon at Killala had a lower average N. perurans load of 1.213 log10 18S rRNA copies compared to Tasman at 1.871 log10 18S rRNA copies. There was no significant difference in severity of gross gill pathology with respect to site (mean gill score Killala = 0.6, Tasman = 0.69, p = 0.366).

4.5 Discussion

Five qPCR assays were designed to detect amoeba species previously isolated from AGD- affected Atlantic salmon, and then used in a spatial-temporal survey to infer the potential role these species play alongside Neoparamoeba perurans in AGD of farmed Atlantic salmon. N. perurans was by far the dominant species colonising Atlantic salmon gills (Figure 4.2), and their abundance was related to gross pathology (Figure 4.3a), farm sites and time points throughout the year (Figure 4.3b). There were very few detections of the five non-N. perurans species apart from Nolandella spp. which was dominant at one site (Killala) during one sample time (May 2017) (Figure 4.2a). Significantly, this is the first molecular-based study to investigate the ecology of multiple amoeba species in association with AGD and shows that, although the gills of farmed seawater Atlantic salmon subjected to a freshwater treatment regime are colonised with a diverse Amoebozoan community (English et al., 2019b), N. perurans dominates this ecological niche, consistent with its documented role in AGD pathology (Crosbie et al. 2012).

As gross gill pathology became more severe, N. perurans abundance was positively correlated (Figure 4.3a). Although this linear relationship was expected, the amount of variation in N. perurans abundance, expressed as 18S rRNA copies, across each gill score was not anticipated. A positive relationship between agent abundance and gross gill pathology (gill score one to four) has previously been demonstrated with a small number of gill swabs from farmed Tasmanian salmon (n = 32) (Bridle et al. 2010). This previous experiment retrieved a much tighter association with an R2 of 0.998 compared to the model from the present study where only 25 % of the variability in N. perurans abundance could be explained by gill pathology (R2 = 0.247). The increased unexplained variability could be reflective of the use of a different N. perurans-specific assay or variation in swabbing technique at each sampling event. It could also be because our study swabbed salmon across a number of time points at two different sites, as opposed to a singular sampling event, showing that N. perurans abundance and AGD progression was responsive to some site-based and seasonal variables.

From a commercial perspective, it is useful to understand whether particular farm sites produce fish with less severe AGD. The two farm sites in this survey were chosen based on hypothetical differences in terms of N. perurans infection prevalence and load due to the contrasting environmental conditions and the observation that lower salinity reduces N. perurans abundance (Wright et al. 2017). The Killala site is exposed to more freshwater runoff, attributed to its location in the Huon Estuary, while the Tasman farm is more oceanic (Figure 4.1). Indeed, we 74 did find Killala had an overall lower percentage of Atlantic salmon infected with N. perurans and a lower mean N. perurans load compared to Tasman. However, there was no difference in severity of gross gill pathology between the two sites, with an overall mean gill score of 0.6 at Killala and 0.69 at Tasman (p-value = 0.366). It appears that fish stocked at the Tasman lease are more likely to be infected with N. perurans, but these differences do not impact the severity of gross gill pathology at the two sites when considering gross pathology alone. However, it is likely a significant difference in AGD severity would become evident over more sampling events and with more sensitive methods for scoring pathology, such as histopathology (Adams and Nowak, 2003).

Temporal differences in N. perurans infection prevalence and load also differed by location. At the Tasman lease, the highest N. perurans infection prevalence and load was recorded in May 2017 and Feb 2018, and the lowest in Nov 2017 (Figure 4.3b). At the Killala lease, the highest was recorded in Aug 2017, and the lowest in May 2017. Although there were too few sampling events to draw conclusions on seasonal variation from our survey, the temporal trends at the Tasman lease broadly reflect past seasonal trends recorded in AGD-affected farmed Tasmania Atlantic salmon (Clark and Nowak, 1999). AGD prevalence has been shown to peak in summer, followed by a second spike in autumn, and this also correlated with higher water temperature and salinity (Clark and Nowak, 1999). Similarly, an in vitro growth optimisation study found peak N. perurans growth occurs at 15°C and a salinity of 35 ppt (Collins et al., 2019). However, the Killala lease did not reflect that higher Tasmanian water temperatures support the highest N. perurans infection prevalence and load. Instead, the time point with the coolest water temperature of 11.9°C, Aug 2017, had the highest N. perurans infection prevalence (92.5 %) and load at this lease. This observation provides support that N. perurans are very adaptable to environmental parameters (Collins et al., 2019; Lima et al., 2016) and remain infectious over a range of temperatures and salinities (Clark and Nowak, 1999).

The reasons for the dichotomy between Nolandella spp. and N. perurans prevalence in May 2017 (Figure 4.2a and 4.2d) is speculative. One explanation could relate to exposure time to N. perurans. The Atlantic salmon sampled at Killala in May 2017 were transferred to sea less than one month before sampling and had never been treated for AGD (2-4 h freshwater bath), while all other fish swabbed during the survey were at sea for three months plus and had undergone freshwater treatments. The Killala site had also been fallowed for seven months prior to restocking and it is possible that N. perurans density within the environment decreased over this fallowing period. Then, in the absence of N. perurans other amoeba species filled this niche. Moreover, Nolandella sp. strain MX5 (English et al., 2019b) showed tolerance to salinity lower than 35 ppt during our in- 75 house culturing. Further research could investigate whether Nolandella spp. are specific to sites with more estuarine conditions, or their presence is attributable to the absence of N. perurans.

Further research should also investigate whether the presence of Nolandella spp. is reflective of healthy Atlantic salmon gills, simply a commensal bystander or a potential player in the early stages of gill disease. For instance, a similar amoebic disease, nodular gill disease (NGD), which affects freshwater salmonids, is proposed to have a multi-amoeba aetiology, and different amoeba species are hypothesised to influence different phases of disease development (Dyková et al. 2010, Dyková & Tyml 2015). The Atlantic salmon that tested positive for Nolandella spp. and negative for N. perurans had an average gill score of 0.225, which was a relatively low level of AGD-like gross gill pathology compared to the other sampling events. The gross AGD-like lesions were recorded with the Taylor et al. (2009) scoring method and were not confirmed by histology because destructive sampling did not suit working alongside the commercial farm. Since histology was not used to confirm an AGD diagnosis, it is possible some gross gill lesions could be unrelated to AGD. The use of histology in conjunction with qPCR should be considered in future surveys aiming to identify relationships between specific amoeba taxa and gill pathology, or more specifically, in future investigation into the significance of Nolandella spp. in relation to AGD.

Pseudoparamoeba sp., the two Vannellida species and Paramoeba elhardi, are minor players in the farmed salmon gill Amoebozoa community on the east coast of Tasmania. With so few detections no strong spatial, temporal or pathological correlations could be established for the majority of the non-Neoparamoeba species. However, the amoebae identified in chapter three which were used to develop the qPCR assays for this study were isolated from salmon with a gill score greater than three (reasonably severe pathology) and the overall average gill score of the salmon surveyed in this study was 0.67 (low level of pathology). It is possible different amoeba species are involved in different stages of disease development, as hypothesized for NGD in salmonids (Dyková et al. 2010, Dyková & Tyml 2015). Differences in community composition in relation to AGD severity could be one explanation for the low level of non-N. perurans detection in the gills presenting low level of AGD pathology in this survey. Ideally these assays should be run against all stages of naturally occurring AGD. Unfortunately, this would be difficult to carry out in Tasmanian farms because they generally manage their AGD bathing treatment regime so that gill scores remain below three.

Our findings do not corroborate an earlier study which is the only other published survey that attempted to associate AGD pathology with a variety of protozoans (Bermingham and 76

Mulcahy, 2006). The Ireland-based survey used histological observations to conclude that abundance of amoebae other than Neoparamoeba spp., Ichthyobodo-like flagellates and trichodinid ciliates correlated with gill pathology, while the abundance of Neoparamoeba spp. did not. This differs from our findings, where the abundance of Neoparamoeba sp. correlated with increased gross gill pathology, while the abundance of non-Neoparamoeba did not. Indeed, the contrasting results could be due to different locality, the possible mistaken documentation of complex gill disease rather than AGD (Herrero et al., 2018), or the study being conducted before species-specific molecular tools were available, relying instead on histology of gills. This technique is vulnerable to inaccurate documentation of amoeba community composition due to loss of amoebae during fixation, difficulty identifying protozoans in histology sections, and the limited sample size from each gill basket (one section from one gill arch). Since this early study, molecular diagnostics identified N. perurans as the causative agent in the earliest outbreaks of AGD in Ireland which were preserved in histological samples (Downes et al., 2018).

The absence of P. eilhardi detections suggests either the isolation of this species documented by English et al. (2019b) was a one-off occurrence, or the qPCR assay was not sensitive enough. The latter reason is most likely, as the P. eilhardi assay performed relatively poorly in terms of amplification efficiency, limit of detections and intra- and inter-assay variance (Table 4.3). In comparison, the other four newly designed qPCR assays performed markedly better in the validation experiments and were of similar efficiency to the previously published N. perurans (NPJ) assay (Downes et al., 2015). The only concern with the remaining assays was that NPJ and Nol had some cross reactivity with non-target species. However, the NPJ assay only narrowly returned a positive result when tested against a high concentration of Pseudoparamoeba DNA, and the Nol assay was cross reactive with a species from the same genus and was thus deemed genus- specific, not species-specific. We therefore believe this lower specificity would not have affected the final conclusions. Hence the qPCR assays, excluding perhaps the P. eilhardi assay, were considered fit for purpose, accurately quantifying their respective amoeba in the survey gill swabs.

4.6 Conclusions

While a diversity of amoebae colonise the gills of farmed Atlantic salmon with AGD, N. perurans is the dominant species, and increasing N. perurans load on the gills correlates with increasing AGD pathology. However, despite N. perurans being the dominant species, there were times when they could not be detected and Nolandella spp. were highly prevalent on gills with low

77 levels of gross gill pathology. Further research should investigate whether Nolandella spp. frequently colonises farmed Atlantic salmon gills and whether they are parasitic or commensal.

Chapter 5: Immersion challenge of naïve Atlantic salmon with cultured Nolandella sp. and Pseudoparamoeba sp. did not increase the severity of N. perurans-induced amoebic gill disease (AGD)

5.1 Abstract

Amoebic gill disease (AGD) is one of the main health issues impacting farmed Atlantic salmon in Tasmania, Australia. Neoparamoeba perurans causes AGD, however a diversity of other amoeba species colonise the gills and their role in the gill environment is unknown. There is little understanding of whether these other amoebae are commensal or potentially involved in different stages of gill disease development. Here we conducted in vivo challenges of naïve Atlantic salmon with cultured, gill-derived Nolandella sp. and Pseudoparamoeba sp. to determine their pathogenicity to Atlantic salmon gills. Additionally, we assessed whether Nolandella sp. and Pseudoparamoeba sp. could influence the onset or severity of N. perurans-induced AGD by comparing the gill condition of fish challenged with N. perurans alone with the gills of fish challenged with a mix of all three amoeba strains. Nolandella sp., Pseudoparamoeba sp. and N. perurans infected and multiplied on the gills according to qPCR analysis of gill samples and a small number of amoeba cultures established post infection. Minor gross gill lesions and histological changes were also observed after exposure to all three amoeba strains. While N. perurans was found associated with classical AGD lesions, the lesions in both Nolandella sp. and Pseudoparamoeba sp. treatments did not met the expected composite of histopathological lesions for AGD. Additionally, there was no histological evidence of Nolandella sp. and Pseudoparamoeba sp. associated with these lesion sites, so the precise cause of pathology remains inconclusive. Moreover, the presence of these non-N. perurans species did not significantly increase the severity of N. perurans-induced gill pathology. This trial supports N. perurans as being the primary agent of AGD and provides evidence that Nolandella sp. and Pseudoparamoeba sp. do not influence or cause AGD under the given experimental conditions.

Keywords Gill disease, Nolandella, Pseudoparamoeba, histopathology, Atlantic salmon, Aquaculture

5.2 Introduction

The protozoan ectoparasite, Neoparamoeba perurans, causes amoebic gill disease (AGD) which increasingly impedes the marine phase of salmonid aquaculture globally (Oldham et al., 2016). This aetiological link is supported by both the fulfilment of Koch’s postulates (Crosbie et al., 2012) and the consistent association of N. perurans with branchitis across various fish hosts and locations (Karlsbakk et al., 2013; Kim et al., 2017; Young et al., 2008b). Attachment of N. perurans to gills causes epithelial hyperplasia, lamellar fusion, oedema and ultimately mortality if not treated (Munday et al., 2001). While it is understood that N. perurans causes branchitis, the pathogenic mechanism of this amoeba is not yet understood, nor whether these mechanisms are highly specific to this species in the context of gill disease (Nowak and Archibald, 2018).

Defining the aetiological agent of AGD took almost two decades of research, with the agent initially considered to be Neoparamoeba pemaquidensis (syn. Paramoeba pemaquidensis) until 2007 (Munday et al., 2001; Young et al., 2007). Under laboratory-based conditions, Neoparamoeba spp. repeatedly induced branchitis in Atlantic salmon and there was a linear relationship between exposure concentration of Neoparamoeba spp. and severity of AGD (Morrison et al., 2004; Zilberg et al., 2001). However, after N. pemaquidensis did not elicit gill lesions in a controlled fish trial (Morrison et al., 2005), it was proposed that AGD could have multiple aetiological agents, including closely related Neoparamoeba branchiphilia or a diversity of other amoeba genera isolated from AGD-affected Atlantic salmon in Ireland (Bermingham and Mulcahy, 2007). Likewise, cultured N. branchiphilia did not induce branchitis when experimentally exposed to Atlantic salmon (Vincent et al., 2007). During the same era of research, a new species of amoeba, N. perurans, was found associated with AGD lesions via species-specific in situ hybridisation (Young et al., 2007). A clonal culture of this new species was established, and Koch’s postulates were fulfilled via infection resulting in gill lesions associated with N. perurans (confirmed by in situ hybridization), then re-isolation and identification of N. perurans from diseased Atlantic salmon (Crosbie et al., 2012). Since this focal study, there has been little consideration of the role other Amoebozoa colonising Atlantic salmon gills may play in AGD onset or progression.

In addition to N. perurans, up to 11 other amoeba species were isolated from the gills of AGD-affected Atlantic salmon in Tasmania, and comprise the genera Paramoeba, Vexillifera, Pseudoparamoeba, Vannella and Nolandella (English et al., 2019b). While N. perurans is the dominant species colonising the gills of farmed Atlantic salmon in Tasmania, our recent survey 80 found they could not be detected on the gills of fish presenting a low level of AGD-like lesions at one site at one sample time, and instead Nolandella spp. were highly prevalent (92 %) (English et al., 2019a). There is no understanding of whether Nolandella, or the other 10 plus amoeba species, are commensal bystanders, support healthy gills or are involved in different stages of disease development.

Nolandella sp. (strain MX5) and Pseudoparamoeba sp. (strain MX1) were previously isolated from AGD-affected gills and maintained to explore whether these species could play a role in the development of AGD (English et al., 2019a, 2019b). Nolandella spp. were the second most prevalent species colonising the gills of farmed Atlantic salmon following N. perurans infection in Tasmania (English et al., 2019a). The genus has also been isolated from AGD-affected gills of farmed Atlantic salmon in Ireland, then a small scale challenge trial detected nodules on gill filaments after seven days post infection (Bermingham, 2004; Bermingham and Mulcahy, 2007). In contrast, Pseudoparamoeba sp. was not frequently detected on Atlantic salmon gills in Tasmania (English et al., 2019a), however it was the most closely related species to N. perurans available to us in monoculture acquired from our previous study (English et al., 2019b). This led to the hypothesis that, of the seven species isolated from AGD-affected gills and established as clonal lines, Pseudoparamoeba sp. would be the most likely to have evolved similar pathogenic mechanism to N. perurans. Nolandella sp. frequently colonised Atlantic salmon gills but was the most phylogenetically distant monoculture from N. perurans (English et al., 2019a, 2019b). These two species also grew well in culture conditions and were therefore practical candidates for experimentally investigation.

Here we conduct in vivo challenges of naïve Atlantic salmon with Nolandella sp. and Pseudoparamoeba sp. to determine whether they are parasitic towards Atlantic salmon gills. We also investigate how Nolandella sp. and Pseudoparamoeba sp. influence the onset or severity of N. perurans-induced AGD by comparing the gill condition of fish challenged with N. perurans alone to fish challenged with a mix of all three amoeba strains.

5.3 Materials and Method

5.3.1 Amoeba cultures

The three amoeba monocultures, Nolandella sp. (strain MX5), Pseudoparamoeba sp. (strain MX1) and N. perurans (MP2) were established from a single cell isolated from AGD-affected gills of farmed Tasmania Atlantic salmon (English et al., 2019b). The strain’s identity was established through gross and ultrastructure morphological characterisation, and by molecular taxonomy in our previous study (English et al., 2019b). The cultures were grown in 1 % seawater malt yeast broth (MYB) at 14°C, and were maintained weekly, involving media exchange, contaminant checks and splitting cultures as necessary. Antibiotics were not used to control bacterial overgrowth, instead amoebae were transferred out of MYB culture medium and grown in 0.2 µm-filtered, autoclaved seawater until bacteria growth re-established equilibrium with amoeba density. The age of the three cultures at the beginning of the trial was 23 months since isolation from Atlantic salmon gills. Amoeba cultures were imaged one week before the trial commenced with an Axiovert 25 microscope and AxioCam MRc camera (Germany) at 40x magnification.

5.3.2 Experimental design

To determine the pathogenicity of three amoeba strains independently and in combination, five groups of AGD-naïve Atlantic salmon were exposed to different challenge scenarios. Each treatment (in triplicate) involved exposing 30 Atlantic salmon per tank to either Nolandella sp. (Nol) (5000 cells/l), Pseudoparamoeba sp. (Pse) (5000 cells/l), N. perurans (Neo) (200 cells/l), a mix of the three strains (Mix) (5000 cells/l of Nol + 5000 cells/l of Pse + 200 cells/l of Neo) or a control containing no amoebae (Ctl). A lower N. perurans dose was used compared to Nolandella sp. and Pseudoparamoeba sp. because the aim was to induce AGD which gradually progressed over the three weeks, rather than rapid high mortality. Ideally this design would provide a three-week series of data to determine whether the addition of different amoeba species can increase the severity of pathology caused by N. perurans alone (i.e. Neo vs Mix treatment). An N. perurans dose of 200 cells/l was deemed appropriate based on data from past immersion trials at our facility. A high dose of Nolandella sp. and Pseudoparamoeba sp. was used because the aim was to investigate whether these species could cause pathology, even at high levels of exposure (i.e. Nol vs Ctl and Pse vs Ctl).

5.3.3 Fish stocking and system setup

A total of 450 Atlantic salmon were taken through smoltification by manipulating the photoperiod and gradually increasing the salinity over five weeks at the Bribie Island Research Centre in a purpose-built salmon rearing facility using industry standard husbandry practices. The fish were then transferred to 15 x 500 l flow-through seawater experimental tanks and their average weight and length was 224 g (±62 g standard deviation) and 267 cm (±26 cm standard deviation) with a stocking density of 13.44 kg/m3 at the beginning of the trial. The fish had four days of habituation in the trial tanks before the trial commenced. Water quality was monitored daily, including pH, salinity, oxygen and temperature (YSI Professional Pro Plus multiparameter water quality analyser). Water temperature was maintained at 15°C and dissolved oxygen above 80 % saturation. The fish were fed 1 % of their body weight daily by auto feeder. All procedures in this trial were approved by CSIRO Queensland Animal Ethics Committee (AEC number A9/2016).

5.3.4 Amoeba harvest

The cultured amoebae were collected by gently tapping culture flasks and repeatedly flushing with a small amount of 0.2 µm-filtered and autoclaved seawater, followed by cell scrapping any remaining attached trophozoites. Amoebae density was calculated from the average of 10 cell counts using a hemocytometer. A sub-sample of each inoculum was pelleted by centrifugation and stored in 70 % ethanol to later re-confirm the identity of each culture using TaqMan quantitative polymerase chain reaction (qPCR) assays previously designed to detect these species (Downes et al., 2015; English et al., 2019a). The inoculum was agitated every ~5 min to prevent amoebae adhering to the beaker pre-challenge and it was used to infect fish within two hours of collection.

5.3.5 Amoeba challenge

Fish were infected with amoebae by immersion. The water flow was stopped, and the water level of each tank reduced to 150 l prior to distributing the amoeba inoculum according to the treatments and dose rates detailed above. Two hours after the addition of amoebae the water flow was restored to 4 l/min and the tanks refilled to 500 l. The dissolved oxygen content was monitored during inoculation, but addition of pure oxygen boost was not required.

5.3.6 Sampling

Five sampling events occurred throughout the 21-day trial. To establish the baseline condition of the gills, six fish from each tank were sampled one day before the fish were challenged with amoebae (Time 0 samples). After amoeba inoculation, six fish from each tank were sampled 12 h, seven, 14, and 21 days post infection (DPI). Fish were destructively sampled by sedating in 20 ppm AQUI-S followed by pithing. The gross gill pathology of every sampled fish was scored according to Taylor et al. (2009) on a ordinal scale of zero (no gross lesions) to five (lesions covering > 50 % of gill surface).

Of the six fish sampled from each tank, gills from two fish were dissected for histology, gills from three were swabbed for qPCR detection of amoebae, and one for a wet mount to re- isolate amoeba post-infection. For histology, whole gill baskets were excised and fixed in seawater Davidson’s fixative for 24 h. For qPCR analysis, all 16 gill surfaces were swabbed (Westlabs), with one swab used for the left gill arches and one for the right arches to maximize the DNA yield. The gill basket of every swabbed fish was then dissected. These swabs and gill baskets were stored in 70 % ethanol at -80°C until processed for DNA extraction. The gills of the remaining fish were swabbed, and this sample was placed in a 15 ml falcon tube containing 3 ml of 0.2 µm-filtered and autoclaved seawater, agitated for 15 s then plated on a six well NunclonTm Delta Surface culture plate (ThermoScientific) and stored in an incubator at 14°C. These seeded culture plates were viewed 24 h post-inoculation under an inverted microscope (Olympus CK2, Japan) at 20x magnification, then monitored for amoeba growth for seven days.

Fish were checked daily for mortality and, where fish were removed, the gills were scored and then sampled for histology if the tissue was still mostly pink and intact. If the gills were not intact, a gill swab was taken for qPCR analysis.

5.3.7 Gill histopathology

After the gill baskets were fixed in seawater Davidson’s fixative, individual gill arches were dissected, and all 16 gill surfaces were photographed (Canon EOS 400D, Japan). The gill arches

84 were transferred to 70 % ethanol then dehydrated and set in paraffin wax using Leica Tissue Processor TP1020 (Germany) within a week of fixation. The second right gill arch from each fish was sectioned at 4 µm with a Leica RM2235 microtome (Germany) and stained with haematoxylin and eosin.

To assess the severity of branchial pathology over the course of the trial, the total number of viewable filaments were counted in one gill section from the second right hemibranch of each fish. The total number of viewable filaments and the number of these filaments containing lamellae with lymphocytic nodules/plaques and/or epithelial hyperplasia with lamellar fusion were counted based on previous descriptions (Adams and Nowak, 2003; Nowak and Munday, 1994). Gill sections were analysed using a light microscope (Olympus BH2, Germany) and then photographed digitally (Nikon Eclipse Ni-U/DS-Ri2, Nikon Instruments, Japan).

5.3.8 Amoeba detection on gills post-challenge by qPCR

To determine the amoeba infection prevalence and check for cross-contamination between treatments, both gill swabs and gill tissue were analysed by qPCR with Taqman assays previously designed to detect the three amoeba strains (Downes et al., 2015; English et al., 2019a). Before qPCR analysis, 25 mg of gill tissue was dissected from the ventral region of every left, third arch of each gill basket that had been swabbed. DNA from the gill tissue and gill swabs were purified with a DNeasy Blood and Tissue Kit (QIAGEN) according to the manufacturer’s instructions. Extracted DNA was quantified with a Nanodrop ND-1000 spectrophotometer (Life Technologies, USA), then a 30 ng/µl working stock of each sample was prepared using an epMotion 5070 liquid handling robot (Eppendorf, Germany) and stored at -20°C. The gill tissue and gill swab DNA (30 ng) were then tested in triplicated 5 µl reactions across a 384-well PCR plate using MyTaq TM DNA Polymerases (Bioline, Australia) and a ViiATM 7 Real-Time PCR Machine (Applied Biosystems, USA). Each plate included a positive control, a negative control and a negative-process control (a blank sample extracted alongside the gill samples). An external process control (salmon elongation factor-1α) was also run against each sample (Bruno et al., 2007). Samples were deemed positive when the average cycle threshold (Ct) of two or three technical replicates was less than the limit of detection specific to each assay detailed in English et al. (2019a). If any samples were positive for the incorrect amoeba species with respect to its treatment, three additional gill samples from that fish underwent qPCR analysis.

5.3.9 Statistical analyses

All statistical analyses and graphics were produced in R version 3.5.2 (R Core Team, 2018). The mean gill score and mean percentage lesion-affected filaments for each treatment at each sample time were compared using a Kruskal-Wallis test then Dunn’s post-hoc test to determine significant pairwise differences. The p.adjust.method option was used to correct the p-values for multiple comparison. These nonparametric analyses were used because assumptions of normality (Shapiro–Wilk test) and homogeneity of variance (Levene’s test) were not met. Additionally, the data was independent since fish were destructively sampled at each time point and each tank were discrete units separated by the flow-through system.

5.4 Results

5.4.1 Confirmation of the identity of amoeba inoculum

The amoeba cultures used to challenge naïve Atlantic salmon displayed morphological characteristics consistent with Nolandella sp., Pseudoparamoeba sp. and N. perurans. Figure 5.1a shows the tubular, elongated morphotype of Nolandella sp., and the laterally flattened, and finger- like pseudopodia of Pseudoparamoeba sp. and N. perurans (English et al., 2019b; Smirnov et al., 2011). Pseudoparamoeba sp. and N. perurans trophozoites were easily differentiated based on trophozoite size and the Perkinsela amoebae-like endosymbiont characteristic of Neoparamoeba (Dyková et al., 2003). The identity of the amoeba inocula was further confirmed using specific qPCR assays (Figure 5.1b). Amplification of N. perurans, Nolandella sp. and Pseudoparamoeba sp. DNA only occurred in their corresponding qPCR assays, and the absence of cross-reactivity verified the cultures were not contaminated with the other species.

Figure 5.1 Confirmation of the identity of the amoeba cultures used to experimentally infect naïve Atlantic salmon. (a) Light microscopy images of Nolandella sp. (Nol) strain MX5, Pseudoparamoeba sp. (Pse) strain MX1 and N. perurans (Neo) strain MP2. All scale bars = 20 µm. (b) qPCR amplification of cultured Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse) and N. perurans (Neo) DNA using the three specific qPCR assays.

5.4.2 Survival during infection trial

Two mortalities occurred from the total 450 fish, one from the control treatment five DPI and the other from the Pse treatment 19 DPI. The gills of both fish were too degraded to accurately gill score or sample for histology. Both mortalities had no obvious exterior lesions or abnormalities to provide a potential explanation for cause of death.

5.4.3 Gill pathology

The gross gill pathology scored according to Taylor et al. (2009) appeared as raised white spots or more prominent patches (Figure 5.2a). Typical lesions in the Nol and Pse treatment fish were small nodular spots, while most lesions in the Neo and Mix treatments were slightly larger streaked filaments. Lesions in all amoeba-containing treatments were most commonly located in the proximal and central area of the filaments in the dorsal region of the second arch. The average gross gill pathology, scored on a scale of zero (no lesions) to five (lesions covering > 50 % of gill

87 surface), of each treatment over the three-week trial is displayed in Figure 5.2b. The average gill score of both Mix and Neo at 14 and 21 DPI were significantly different to their equivalent Time 0 samples (p values; Mix-T0 v 14DPI = 0.00246, Mix-T0 v 21DPI = 0.00196, Neo-T0 v 14DPI = 2.2e-05, Neo-T0 v 21DPI = 9.2e-06), and to the uninfected control at those time points (p values; 21DPI-Ctl v Mix = 0.02690, 21DPI-Ctl v Neo = 0.00292, 14DPI-Ctl v Mix = 0.00246, 14DPI-Ctl v Neo = 0.00031). There was no significant difference between the gross gill score of the Mix and Neo treatments at any timepoint. When considering the gross gill pathology of the non-N. perurans treatments, Nol at 21 DPI was significantly different to Nol at Time 0 (p = 0.02301), however it was not significantly different to the no amoeba control sampled at 21 DPI. The only significant difference in the Pse treatment was between 0.5 and 14 DPI (p = 0.00717). The trends in gross pathology over time suggest Nol progressively increased throughout the trial, whereas the gross pathology of Pse peaked at 14 DPI then reduced at 21 DPI.

Figure 5.2 (a) Gross gill pathology in Atlantic salmon after exposure to cultured Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), N. perurans (Neo) and a mix of all three amoeba species (Mix) at 14 days post infection (DPI). These treatments were compared to a no amoeba control (Ctl). (b) Mean gross gill pathology score in Atlantic salmon before (Time 0) and after exposure to amoebae at 0.5, 7, 14 and 21 DPI. Bars display standard error. Level of statistical significance indicated by p value * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001. 88

Fish that were not exposed to amoebae (Ctl) remained relatively lesion free throughout the trial (Figure 5.3 a, b and 5.4 a). The control gills predominately showed anatomically normal secondary lamellae covered by a thin layer of epithelial cells projecting from the primary filaments (Figure 5.3 a, b). One fish had 4.3% of filaments affected with epithelial hyperplastic lesions (Figure 5.4 b, c, d). A third of the control fish (33.33 %) had lymphocytic nodules/plaques. Nodules/plaques typically had a single layer of squamous epithelium encasing lymphocytes with infrequent inflammation within the central venous sinus (Figure 5.3 e, f). Several distal lamellar nodules had some basal infiltration of lymphocytes evident. Interlamellar nodules were typically two to three lamellae fused predominately with lyphocytes (Figure 5.3 e), while plaques were essentially the same but involved more than three lamellae.

Epithelial hyperplasia with some inflammation was more common in the treatments with fish exposed to N. perurans, Neo and Mix, but there were no significant differences in the mean percentage of lesion-affected filaments for these treatments during the three-week trial (Figure 5.4). Epithelial hyperplasia of the primary and secondary lamellae showed fusion of secondary lamellae with undifferentiated epithelial cells, a high density of mucus cells and various leucocytes throughout proliferative tissue (Figure 5.3 c, d). Various inflammatory leucocytes were also noted within the central venous sinus (Figure 5.3 c). For the Mix and Neo treatments, a single fish had a small number of N. perurans trophozoites observed in close proximity to the hyperplastic regions (Figure 5.3 c, d) which is indicative of light AGD.

Significant histological changes were observed in fish challenged with Nolandella sp. at 14 DPI compared to Time 0 when considering all lesion morphotypes (p values; Nol-T0 vs Nol-14 DPI = 0.025) (Figure 5.4 a). However, the increase in mean percentage of lesion-affected filaments for the Nol treatment was not significant when analysing epithelial hyperplastic lesions, distal lamellar nodules and interlamellar nodules/plaques separately. Some hyperplastic lesions and nodules were also observed in fish challenged with Pseudoparamoeba sp., but these changes were not significant when considering the mean percentage of lesion-affected filaments. There was no histological evidence of amoeboid-like cells observed within the Nol and Pse treatments.

Figure 5.3 H&E stained gill histopathology from Atlantic salmon after exposure to cultured amoebae. (a) Anatomically normal primary lamella (PL) and secondary lamellae (SL), scale bar = 100 µm. Insert border (dashed lines) corresponds to (b) anatomically normal secondary lamellae at 60x magnification showing pavement cells (pv), mucus cells (m), erythrocytes (e), chloride cells (c) and pillar cells (p). Scale bar = 30 µm. (c) Epithelial hyperplasia with fusion of secondary lamellae (white arrows) closely associated with N. perurans trophozoites (t). Also note presence of leucocytes within the central venous sinus (csv), scale bar = 100 µm. Insert border (dashed lines) corresponds to (d) hyperplastic lesion at 60x magnification showing N. perurans trophozoite (t) adjacent to fused secondary lamellae (white arrow). Mucus cells (m), lymphocytes (L), macrophages (ma) and undifferentiated epithelial cells (e) also present, scale bar = 30 µm. (e) Interlamellar nodule showing two lamellae fused predominately with lymphocytes (L), scale bar = 30 µm. (f) Distal lamellar nodule showing infiltration of predominately lymphocytes (L) at the outer end of the secondary lamellae, scale bar = 50 µm.

Figure 5.4 Gill histopathology from Atlantic salmon after exposure to cultured Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), N. perurans (Neo) and a mix of all three amoeba species (Mix). These treatments were compared to a no amoeba control (Ctl). The mean percentage of lesion- affected filaments in Atlantic salmon before (Time 0) and after exposure to amoebae at 0.5, 7, 14 and 21 days post infection (DPI) were graphed for (a) all lesion morphotypes and then split into the most common lesion morphotypes, including (b) epithelial hyperplasia with lamellar fusion, (c) distal lamellar nodules and (d) interlamellar nodules and plaques. Level of statistical significance indicated by p value * ≤ 0.05, ** ≤ 0.01, *** ≤ 0.001.

5.4.4 Amoeba detection on gills post-challenge

From the wet mounts sampled post amoeba challenge both Nolandella sp. and Pseudoparamoeba sp. were reisolated. There were two Nolandella sp. cultures, one from the Nol treatment 0.5 DPI and the other from the Mix treatment 7 DPI. Pseudoparamoeba sp. was also isolated from Pse treatment 0.5 DPI. No N. perurans growth was recorded in the wet mount samples. These results were based on visual assessment only, in that trophozoite morphology and size matched the images in Figure 5.1a.

The prevalence of amoeba infection was established by qPCR analysis of the gill swab and gill tissue samples (Figure 5.5). Fish were defined to be colonised by one of the three amoeba strains if either or both the swab and tissue sample was positive, and inversely deemed uncolonised if both samples were negative. The prevalence of Nolandella sp. (Figure 5.5a) and N. perurans (Figure 5.5c) infection in their respective treatments both gradually increased throughout the trial, while Pseudoparamoeba sp. infection in the Pse treatment increased until 14 DPI then rapidly decreased by 21 DPI (Figure 5.5b). The prevalence of the three amoeba species appeared to fluctuate in the Mix treatment fish (Figure 5.5d). Overall there was an increased prevalence from 0.5 DPI to 7 DPI, then a decrease at 14 DPI, followed by another increase by 21 DPI which did not surpass 7 DPI levels. The N. perurans infection prevalence in the Mix treatment remained lower than the other two species and did not gradually increase over the course of the trial, as seen in the Neo treatment containing only N. perurans. The no-amoeba control treatment remained negative for all three amoeba strains.

No cross-contamination between the treatments occurred. However, during the initial qPCR analysis there was some non-specific amplification that required further investigation. Two tissue samples, one from Pse Time 0 and one from Ctl 7 DPI, narrowly returned a positive (within one cycle of the limit of detection) for N. perurans. Three more tissue samples from each of these gills were analysed and all were negative for N. perurans. Similarly, six non-target tissue samples were narrowly positive for Pseudoparamoeba sp., but again further qPCR analysis defined these fish as negative. Additionally, during the first round of qPCR analysis the Nol assay had multiple cross reactivity with non-target fish in the Time 0 (before amoeba exposure) and 0.5 DPI samples. After raising the annealing temperature of the assay from 63°C to 64°C to increase assay specificity, the non-target amplification was eliminated.

5.5 Discussion

AGD is a longstanding and costly health issue for the farmed Atlantic salmon industry in Australia. AGD is caused by Neoparamoeba perurans (Crosbie et al., 2012), but several other amoeba species have been isolated from AGD-affected farmed Tasmania Atlantic salmon (English et al., 2019b) and their role in the gill environment is not well understood. These accompanying Amoebozoa could be commensal bystanders, support healthy gills or be involved in different stages of AGD development. Examining the pathogenicity of these recently isolated strains (English et al., 2019b), separately and in combination with N. perurans, towards Atlantic salmon will offer insight into whether non-N. perurans amoebae influence AGD development on-farm.

In this study, we assessed the pathogenicity of two non-Paramoebidae species, Nolandella sp. and Pseudoparamoeba sp., previously isolated from AGD-affected farmed Atlantic salmon (English et al., 2019b), through an immersion challenge trial of AGD-naïve Atlantic salmon. Additionally, the affect that these species had on the onset and severity of AGD caused by N. perurans was assessed in parallel. Cultured Nolandella sp. and Pseudoparamoeba sp. successfully colonised the gills according to qPCR data, and minor gross and histopathological changes were detected over time. However, these pathological changes were not consistent with AGD pathology (Adams and Nowak, 2003) and were not linked with the presence of amoebae in histology. Furthermore, a relatively high dose of Nolandella sp. and Pseudoparamoeba sp. did not increase the severity of AGD induced by cultured N. perurans. This study provides evidence that amoebae other than N. perurans do not cause or influence AGD under the experimental conditions described here, and this may also be the case on-farm. This finding aligns with other in vivo challenge trials which found N. perurans induces AGD in Atlantic salmon (Crosbie et al., 2012), while N. pemaquidensis and N. branchiphila did not (Morrison et al., 2005; Vincent et al., 2007).

Fish exposed to N. perurans alone and a mix of the three species developed typical AGD pathology (Adams and Nowak, 2003), but was very light disease expression when considering the mean percentage of filaments with hyperplastic lesions (< 5 %) compared to previous AGD studies (between 10 and 39 %) (Adams et al., 2012; Embar-Gopinath et al., 2005; Pennacchi et al., 2016). The relatively high exposure of Nolandella sp. and Pseudoparamoeba sp. compared to earlier in vivo amoeba challenge trials (Haugland et al., 2017; Marcos-López et al., 2017; Vincent et al., 2007; Wynne et al., 2020) alongside a 25 x lower dose of N. perurans (200 cell/) did not 94 significantly increase the severity of gill pathology compared to fish exposed to just 200 N. perurans/l. The presence of N. perurans did however correlate with higher colonisation of Nolandella sp. and Pseudoparamoeba sp. compared to their respective single-species treatments (Figure 5.5). This could reflect an opportunistic colonisation by the non-Paramoebidae species after N. perurans compromised the host’s immune system, like Balamuthia mandrillaris infections occurring mainly in immunocompromised people (Schuster and Visvesvara, 2004). The differences in prevalence of the three strains could also reflect the amoebae were competing for resources on the gills or directly interfering with a competitor’s survival or their access to an ecological niche (Rendueles and Ghigo, 2015). It was possible the presence of high numbers of non-Paramoebidae species hindered N. perurans from replicating on the gills, resulting in slightly fewer lesions in the Mix treatment fish compared to the fish exposed to only N. perurans.

Lesions found in fish exposed to the non-Paramoebidae species were not consistent with AGD pathology, rather it comprised mostly focal lymphatic inflammation similar to the nodules and plaques described in salmon after transferring from the freshwater hatchery to marine sites (Nowak and Munday, 1994). Similar gill lesions were also documented in a more recent survey of Atlantic salmon on marine farms, where peak abundance of nodules and plaques were recorded 5-6 weeks post-transfer and appeared to occur separately to the AGD outbreak a further six weeks later (Adams and Nowak, 2003). Likewise, the histopathology of the non-Paramoebidae challenged gills in this study revealed no association between the lesion sites and the presence of amoebae, contrasting with frequent associations of N. perurans with epithelial hyperplasia and lamellae fusion (Munday et al., 2001, Adams and Nowak, 2003, Bustos et al., 2011). It is possible the lesions in the Nol and Pse treatments were caused by other parasites or bacteria unintendedly present in the experimental system or by the bacterial community associated with the non-axenic amoeba monocultures used. For instance, the bacteria Tenacibaculum maritimum (syn. Flexibacter maritimus (Suzuki et al., 2001)) has been reported in Tasmanian Atlantic salmon causing gill disease (Handlinger et al., 1997; Powell et al., 2004). While no filamentous bacteria were seen on the gills indicating tenacibaculosis, there are a diversity of other pathogens which are associated with gill disease in salmonids (Rozas-Serri, 2019) and it is difficult to determine the causation of gill lesions observed in this trial in the non-Paramoebidae groups.

A clear disparity in this trial was that Nolandella sp. and Pseudoparamoeba sp. were readily detected on the gills by qPCR of gill swabs but were not seen in the histology, whereas N. perurans

95 was detected by both methods. Tissue fixation for histology can remove much of the mucus layer and glycocalyx where Nolandella sp. and Pseudoparamoeba sp. could have inhabited without being attached to the epithelial surface (Blick et al., 2019; Merrifield and Rodiles, 2015), while N. perurans have been shown tightly associated with gill epithelium via transmission electron microscopy (Wiik-Nielsen et al., 2016). The reduced density of non-Paramoebidae species, possibly due to fixation, would have made it difficult to identify these amoeba species which are much smaller than N. perurans (Figure 5.1a). Species-specific in situ hybridization (ISH) assays, that are yet to be designed, would help understand how and where the non-Paramoebidae species were colonising the gills (refer to Appendix C for progress on ISH). The discrepancy between histology and qPCR data also highlights a limitation in the experimental design, in that different fish were sampled for histology and for qPCR analysis which could explain the lack of agreement and also reduced the sample size. Ideally each fish should have had their gross pathology photographed and been sampled for qPCR, histology and wet mounts to better align pathogen and host response from the entire population and to increase the statistical power of the study.

Another limitation of this experiment was the low level of disease expression induced by cultured N. perurans in the Neo and Mix treatments made it difficult to discriminate pathological response in each treatment. The low level of disease expression and low prevalence of N. perurans on the gills likely illustrates very few viable cells successfully infected the fish, and future trials should increase the dose rate or use younger cultures because amoebae can lose virulence during long-term clonal culture (Bridle et al., 2015; Jellet and Scheibling, 1988; Wong et al., 1977). Alternatively, the cultures used in this trial may have been low virulent strains compared to the other available strains, as clonal N. perurans cultures can display differences in virulence (Collins et al., 2017) and this could also be the case for the Nolandella sp. and Pseudoparamoeba sp. strains. Differences in virulence between the available strains could have been assessed by screening for cytopathic effect in vito using a salmonid gill cell line (Cano et al., 2019), and establishing the most appropriate dose rate could have been resolved with a preliminary in vivo challenge trial.

Future in vivo challenge trials could investigate the pathogenicity of the other non- Paramoebidae species isolated from AGD-affected salmon (English et al., 2019b) or how they influence a more severe or late stage AGD. For instance nodular gill disease, an amoebic disease that affects freshwater salmonids, has been ascribed to multiple species and is hypothesised to have multi-amoeba aetiology with different species playing more dominant role in different stages of

96 disease development (Dyková et al., 2010; Dyková and Tyml, 2015). The secondary effects to disease expression of these alternative species could be investigated by first infecting naïve Atlantic salmon with N. perurans, then rechallenging the fish with non-Paramoebidae species once the AGD gill response has been well established. Another useful addition to similar in vivo trials would be a no-amoeba bacterial control to determine whether the bacterial community growing alongside the amoeba monocultures were involved in the gross and histological gill changes, particularly for Nolandella sp. culture which was the only treatment that recorded a significant increase in the mean percentage of lesion affected filaments.

Gill disease in the marine phase of Atlantic salmon aquaculture has become more significant in recent years (Boerlage et al., 2020), increasing the need to understand the complexity of disease expression. This trial provides evidence that cultured Nolandella sp. and Pseudoparamoeba sp. do not induce gill changes consistent with AGD pathology and did not influence the severity of AGD during the early stage of development.

Chapter 6: General discussion

Amoebic gill disease (AGD) is an important ectoparasitic condition impacting farmed Atlantic salmon and it has been ascribed to one agent, Neoparamoeba perurans (Crosbie et al., 2012). While N. perurans alone does elicit AGD, all the variables of infestation on-farm are not yet understood. Moreover, in recent years several complex gill diseases with apparent mixed aetiology have emerged alongside AGD (Gjessing et al., 2017; Herrero et al., 2018). This complexity has led to the consideration of gill disease in the context of dysbiosis of microbial community structure (Egan and Gardiner, 2016), rather than focusing on a single agent.

Multiple species of amoebae colonise the gills of AGD-affected salmon (Bermingham and Mulcahy, 2007; Howard, 2001) and prior to this project, little attention had been given to their precise identity and to the role they could play in the Atlantic salmon gill environment. Considering some of the genera detected on salmon gills, including Vexillifera, Vannella, and Acanthamoeba, have been linked to other aquatic amoebic diseases (Leiro et al., 1998; Sawyer et al., 1978; Taylor, 1977; Webb et al., 2002), it was hypothesised the accompanying amoebae species could be influencing the onset and severity of AGD alongside N. perurans.

The main purpose of this thesis was to examine whether multiple amoebae species influence AGD of Atlantic salmon farmed in Tasmania. This goal was approached by profiling the amoeba community on the gills of AGD-affected and healthy farmed Atlantic salmon, followed by an in vivo challenge trial to investigate the pathogenicity of specific species in isolation and in combination with N. perurans. Significantly, it was shown that despite N. perurans being the primary aetiological agent, it is possible the accompanying species could influence disease onset and progression. The possibility of AGD having a multi-amoeba aetiology is discussed.

After reviewing the knowledge gaps and methodology related to investigating the aetiology of amoebic diseases in aquatic animals in Chapter 2, it was evident the causative agent of most aquatic amoebic diseases remains largely unresolved. The uncertainty stems from many reports using outdated methods for identifying amoebae (e.g. using morphology only), and insufficient evidence for defining the aetiological relationship (e.g. based on a single case of amoebae isolated from diseased hosts). Of the five main aquatic amoebic diseases, AGD, nodular gill disease (NGD), sea urchin paramoebiasis, grey crab disease (GCD) and lobster paramoebiasis, AGD research provides the most comprehensive aetiological evidence (Table 2.1). While questions remain, the

98 decades of AGD research forms a multifaceted approach, involving on-farm surveys (Downes et al., 2015; Wright et al., 2015) and laboratory experiments (Crosbie et al., 2012; Haugland et al., 2017) using a variety of molecular assays (Bridle et al., 2010; Young et al., 2007) and microscopy techniques (Wiik-Nielsen et al., 2016). The review in Chapter 2 shaped the approach of this project towards investigating the role of the accompanying amoeba species in AGD. Great effort was taken to identify the amoebae associated with AGD using a combination of gross and fine-scale morphology and molecular taxonomy based on two candidate sequence regions. The use of in vitro monocultures was also central to this project, as they aided the development of molecular detection assays, and were utilised in subsequent in vivo trials. Once amoebae were cultured and accurately identified, the aim was to further examine these species in relation to AGD by surveying naturally infected farmed Atlantic salmon, and through controlled challenge experiments. This approach uncovered novel findings, however, as discussed below, each element of this thesis could be extended to strengthen the discoveries.

In Chapter 3, I documented the amoebae colonising the gills of Atlantic salmon farmed in Tasmania and found a far greater diversity of amoebae than previously recognised. In addition to N. perurans, 11 other amoebae were isolated and were classified within the genera Paramoeba, Vexillifera, Pseudoparamoeba, Vannella and Nolandella (English et al., 2019b). Some of these isolations were unique, such as the first case of Paramoeba eilhardi being isolated from teleost gills and the discovery of two new lineages, possibly genera, within the Vannellida phylogeny. Overall, this chapter forms the most comprehensive and precisely identified list of AGD-associated amoebae compared to previous reports that relied on morphology alone for taxonomic assessment (Bermingham and Mulcahy, 2007; Howard, 2001). Ideally this method should be repeated on gills at all stages of AGD development from different farm sites during different seasons or years to potentially uncover new species and assemblages. This research emphasises the diversity of amoebae accompanying N. perurans on gills, findings that may help shift AGD research from a single-agent focus to more awareness of the broader microbial community potentially influencing disease expression.

The diversity of amoebae colonising salmon gills is expected to be greater than documented in Chapter 3. For instance, two species that have previously been isolated from Atlantic salmon gills in Tasmania, N. pemaquidensis and N. branchiphilia, were not detected in this study (Dyková et al., 2005). The reason for these missed species is likely due to low species abundance in the 20 gill baskets sampled (from one farm site across four timepoints) and methodological bias. Two methods

99 were used to characterise amoebae, Sanger sequencing mixed-cultures and Sanger sequencing together with morphological characterisation of monocultures, and each method uncovered different species. However, both methods were biased towards detecting amoebae that were more capable of growing in the chosen in vitro conditions. Next generation sequencing (NGS) is an obvious alternative approach to profiling the amoeba community on gills because it is considered suitable for sequencing trace amounts of DNA in heterogeneous samples and would be less biased towards detecting amoebae that grow in vitro. For example, a comparison between NGS and Sanger sequencing found a far great diversity of marine protist were sequenced by NGS (Edgcomb et al., 2011). Therefore, NGS would likely recover greater species diversity compared to what was recovered in Chapter 3, which relied on cloning and Sanger sequencing.

NGS to profile the amoeba community on gills was attempted during this project but was unfortunately unsuccessful due to methodological issues that required further optimisation. As outlined in Appendix D, gill swabs were amplified with two universal 18S rRNA primers, chosen based on their success in Chapter 3, and then sent for 100 bp paired-end sequencing using the Illumina 2500 AGRF platform. Both primer sets recovered only N. perurans and Atlantic salmon sequences and no novel amoeba species were confidently identified. This null result is likely due to the highly conserved nature of 18S rRNA across amoeba species (Nassonova et al., 2010; Young et al., 2014) which made it difficult to assemble and determine novel taxa. It is possible some reads were species other than N. perurans but there was no way to confirm that. If NGS of amoebae was repeated a more variable region, such as COI, should be sequenced (Hansen et al., 2019; Nassonova et al., 2010). It would also be beneficial to sequence several gill samples representing different stages of AGD development, taken from a range of farm sites and seasons to compare healthy and diseased gill microbiota with respect to temporal-spatial variables.

The NGS results from this project did however reflect the dominance of N. perurans compared to non-Paramoebidae species colonising farmed Atlantic salmon shown in English et al. (2019a). If this high proportion of N. perurans continued to impede the detection of less abundant species, one option could be to block the amplification of N. perurans DNA during PCR, similar to the process described by Gofton et al. (2015). In this study, a blocking primer was used to decrease unwanted sequences by 96 % and significantly increase the total detectable bacterial diversity found in a medically important species of tick. Although our attempts at NGS may not have accurately profiled the amoeba community, with careful optimisation this approach would still be an ideal next step for comparing the diversity of amoebae colonising AGD-affected and healthy Atlantic salmon gills on-farm and during in vivo challenge trials. Importantly, this method could have helped 100 mitigated the possibility of missing critical amoeba species which should have been further investigated in Chapter 4 and 5.

In Chapter 4, I designed and applied taxa-specific qPCR assays in a spatial and temporal survey of farmed Atlantic salmon to determine the prevalence and abundance of six amoeba species previously isolated from AGD-affected Atlantic salmon (Chapter 3). This survey aimed to identify relationships between amoebae and farm sites, time points and gill pathology to infer the significance each species may have to AGD. This was the first molecular-based study to investigate the ecology of multiple amoeba species in association with AGD. This chapter potentially provided a more robust approach compared to the only other published survey that attempted to associate AGD pathology of farmed Atlantic salmon with a variety of protozoans in gill histology because amoebae are difficult to accurately identify by morphology post-fixation (Bermingham and Mulcahy, 2006). The results from Chapter 4 showed that although a diversity of amoebae colonised farmed Atlantic salmon gills (English et al., 2019b), N. perurans was highly dominant and its abundance positively correlated with the progression of gross gill pathology (English et al., 2019a), which aligns with its documented role in AGD pathology (Crosbie et al., 2012).

The reasons for N. perurans dominance in Tasmania’s marine salmon farms is currently speculative but could be due to a variety of mechanisms relating to the how the parasite interacts with its host. It is possible N. perurans is better at attaching to the gill epithelium, better at obtaining food resources from the gill environment, can resist the host’s immune response, or can divide more quickly compared to other competing microbes. N. perurans may also have mechanisms for cell-to-cell communication or quorum sensing that coordinated activity within a population of amoebae depending on nutrient availability or the presence of other microorganisms. For example, the social amoeba Dictyostelium discoideum secretes small signalling molecules into the extracellular environment which then guide cell aggregation and commitment to form fruiting bodies (Gregor et al., 2010). While it is clear N. perurans is the predominant amoebae on farmed Tasmanian Atlantic salmon gills, it would be interesting to understand whether this is also the case throughout its 16 known teleost host species which are global distribution (Oldham et al., 2016).

Of the remaining species surveyed, Pseudoparamoeba sp., the two Vannellida species and P. elhardi, were infrequent members of the farmed salmon gill Amoebozoa community on the East coast of Tasmania. Interestingly Nolandella spp. was the most prevalent amoeba (92 %) at one site

101 at one sample time, during which no N. perurans were detected on gills presenting low levels of gross gill pathology. Given this population of fish were sampled less than one month after transfer to sea, the pathology could have originated in the freshwater hatcheries. Hence it could be useful to screen for Nolandella spp. and associated gill pathology in hatchery stock prior to being transferred to the marine farm sites. Overall, this survey reduced the hypothesised significance of Pseudoparamoeba sp., P. elhardi and the Vannellida species with respect to their involvement in AGD onset and severity due to low prevalence compared to N. perurans and no association with AGD-like pathology. However, the results did not completely discount the hypothesis that other amoebae affect AGD when considering N. perurans was undetectable and Nolandella spp. were highly prevalent on gills with some AGD-like gross pathology. More research into the significance of Nolandella with respect to AGD is needed.

The findings of Chapter 4 should be developed further through more in-depth surveys of naturally infected farmed Atlantic salmon. The present design was limited as it could not draw well supported conclusions about seasonal and environmental effects, despite detecting temporal and spatial trends in N. perurans prevalence and abundance (Figure 4.2, 4.3b). This was because only one sampling event occurred per season across two sites in a one-year survey. The design would be improved with more frequent sample events (i.e. fortnightly or monthly) taken over multiple years from additional farm sites. A farm site of particular interest that could not be accessed in this study is Macquarie Harbour, located on the West coast of Tasmania, which reports much lower incidence of AGD compared to East coast sites (Christine Huynh, personal communication). The reason for this lower AGD incidence is speculative at present but could be due to water chemistry influencing the abundance of N. perurans and the other accompanying amoebae. For example, the water hardness of freshwater can impact Neoparamoeba spp. survival in vitro (Powell and Clark, 2003), so the concentration of Ca2+ and Mg2+ of the Gordon River feeding Macquarie Harbour could be influencing amoeba community composition. Environmental variables such as hydrography, weather, organic load, tides and currents likely influence regional distribution of amoebae (Charman et al., 2002). A more in-depth survey of contrasting farms could reveal further insights into the role of N. perurans and the accompanying amoeba species in AGD and how that is influenced by environmental variables.

Importantly, a more thorough survey could also shed light on whether Nolandella spp. colonise salmon over multiple years, and whether they correlate with high or low levels of gill pathology. Moreover, it could investigate whether seven months of fallowing influenced the 102 absence of N. perurans during the sampling event when Nolandella was highly prevalent and this could evidence the potential benefit of fallowing as an industry management practice to lessen AGD severity. Fallowing has been proposed as a beneficial management practice in farmed Atlantic salmon in Ireland where an epidemiological review of 11 farm sites over 34 years of production data found total Atlantic salmon mortality was significantly higher in years when sites were not fallowed, alongside other management practices (Wheatley et al., 1995).

Future surveys should also consider pathogen load and gill pathology with respect to environmental variables, such as water chemistry, temperature, and dissolved oxygen. It would be novel to consider environmental variables with respect to host response, or parasite virulence, rather than just parasite numbers and pathology, which have been repeatedly considered (Bridle et al., 2010; Downes et al., 2017; English et al., 2019a). It is possible environmental factors influence amphizoic amoebae to shift between their free-living and parasitic states, similar to how encystment of free-living amoeba trophozoites occurs in response to osmotic stress, nutrient deprivation and after exposure to bacterial toxins (Dudley et al., 2005; Lee et al., 2012; Neff et al., 1964). In short, there are many new avenues to further investigate the relationship these species have with host and the environment regarding AGD, particularly now that five non-N. perurans qPCR assays are designed and validated (English et al., 2019a).

In Chapter 5, I drew from the culture-based study and on-farm survey by identifying candidate amoebae to experimentally investigate the pathogenicity of non-N. perurans species against the gills of Atlantic salmon. Nolandella sp. (strain MX5) was chosen because it was the second most prevalent species following N. perurans on Atlantic salmon gills in Tasmania (English et al., 2019a). Pseudoparamoeba sp. (strain MX1) was selected because it was the most closely related to N. perurans of the monocultures isolated in this study and therefore hypothesised to be the species most likely to have evolved similar pathogenic mechanisms to N. perurans (English et al., 2019b). Cultured isolates of both strains infected challenged AGD-naïve Atlantic salmon gills, multiplied over time, then subtle changes in gross gill pathology and histopathology were observed post-exposure. Interestingly, lesion morphotype differed in fish exposed to the non-N. perurans amoebae compared to those exposed to N. perurans, suggesting the means of pathogenesis may differ between the species. Unfortunately, there was no histological evidence of Nolandella sp. and Pseudoparamoeba sp. associated with lesions, while typically AGD histopathology frequently has Neoparamoeba flanking epithelial hyperplasia (Adams and Nowak, 2003; Munday et al., 2001;

103

Zilberg et al., 2001), so the precise cause of pathology in the non-N. perurans treatments remains uncertain.

It is possible that N. perurans has unique pathogenic mechanisms compared to the other two species examined in vivo. Nolandella sp. and Pseudoparamoeba sp. may release toxins such as serine protease identified in Acanthamoeba pathogenesis in human corneas (Lorenzo-Morales et al., 2015) which could accumulate making their physical presence not essential for lesions to progress. Alternatively, N. perurans may have unique attachment mechanisms making it more able to adapt at remaining on lesion sites, such as a carbohydrate-mediated system that allows recognition and binding to specific sugars on the host gill epithelial tissue. Acanthamoeba castellanii and Entamoeba hystolitica, the causative agents of amoebic-related condition in humans, are known to attach and invade the host cells via highly specific carbohydrate-binding proteins (Frederick and Petri, 2005; Yang et al., 1997). It has been suggested that tissue invasion following carbohydrate- mediated adhesion is considered a characteristic that distinguishes pathogenic from non-pathogenic amoebae (Da Rocha-Azevedo et al., 2009; Jamerson et al., 2012). While the pathogenic mechanisms of gill-associated amoebae are unclear, this study suggests N. perurans is the primary cause of the composite of lesions that make up AGD pathology, but may not be the only amoeba capable of causing gill pathology in marine phased of Atlantic salmon aquaculture. While N. perurans is clearly the main agent of AGD, the hypothesised multi-amoeba aetiology cannot be completely dismissed form the evidence provided in Chapter 4 and 5.

In chapter 5, I also investigated how Nolandella sp. and Pseudoparamoeba sp. influence the onset and severity of N. perurans-induced AGD by comparing the gill condition of fish challenged with N. perurans alone to fish challenged with a mix of all three strains. The original hypothesis was that the additional amoebae alongside N. perurans would increase branchial lesions, however the presence of Nolandella sp. and Pseudoparamoeba sp. did not have that effect. Considering some minor changes in gill pathology were observed in fish exposed to Nolandella sp. and Pseudoparamoeba sp. alone, this null result was not anticipated. A possible explanation is that these species were competing for resources within the gill environment (Rendueles and Ghigo, 2015) and high numbers of Nolandella sp. and Pseudoparamoeba sp. hindered N. perurans, the most pathogenic species, from colonising the gills and inducing extensive branchial lesions. Interestingly however, the presence of N. perurans appeared to increase the colonisation of Nolandella sp. and Pseudoparamoeba sp. over the first week post infection compared to their corresponding single- species treatment. Some pathogenic amoebic infections in humans, such as Acanthamoeba and 104

Balamuthia encephalitides, have been described as opportunistic infections occurring primarily in immunocompromised hosts (Schuster and Visvesvara, 2004). Increased colonisation of non-N. perurans amoebae could have been an opportunistic secondary infection following an initial N. perurans infection which impaired the hosts immune system.

While Nolandella sp. and Pseudoparamoeba sp. did not increase the severity of an experimentally induced, light AGD infection, they, and other species could still be involved in different stages of AGD on-farm. Nodular gill disease (NGD), an amoebic disease that affects freshwater salmonids, has been ascribed to more than four species of amoebae and has been hypothesised to have a multi-amoeba aetiology (Dyková et al., 2010; Dyková and Tyml, 2015; Noble et al., 1997; Sawyer et al., 1974; Tubbs et al., 2010). For example, NGD lesions in farmed rainbow trout have been strongly associated Naegleria sp. (Dyková et al., 2010) and then Rhogostoma minus (Dyková and Tyml, 2015). The later study hypothesised R. minus may have played a dominant role in late stages of NGD development while a mixed population of naked amoebae were accountable for triggering a primary fast-course infection. This research advocates investigating gill disease over the whole course of infection, as well as pre- and post-treatment, so component causes of the disease can be understood. Future trials should explore this secondary infection theory and investigate whether non-N. perurans species influence late-stage AGD by first inoculating with N. perurans alone, then inducing a secondary infection with non-N. perurans species once gills have been compromised by N. perurans.

Along with further investigation into the pathogenicity of Nolandella sp. and Pseudoparamoeba sp., research should also be directed into investigating whether the nine other species influence AGD. Rather than conducting multiple, largescale and costly in vivo trials, the other associated amoebae species could be screened in vitro using a salmonid gill-derived cell line. A recent study successfully reproduced in vivo Atlantic salmon host response to N. perurans in an in vitro gill cell monolayer (Cano et al., 2019). This model could be used to screen the pathogenicity of alternative species and to compare the host response induced by N. perurans to the other putative amoebic pathogens. Additionally, this model could be used to compare multiple different clonal lines of the same species, as it is known different clonal cultures of N. perurans display differences in virulence (Collins et al. 2017) and could also be the case for non-N. perurans amoebae. This would reduce the risk of conducting large scale in vivo trials with non-virulent clones. Importantly these comparisons could shed light on the pathogenic mechanisms that appear specific to amoeba within the family Paramoebae (Nowak and Archibald, 2018). 105

Another approach to investigating the pathogenesis of amoebae on Atlantic salmon gills and how amoebae interact with each other and the host is through in situ hybridisation (ISH). Sequences of the diversity of amoebae on healthy and diseased fish could be obtained through NGS and then used to develop species-specific ISH (see appendix C for progress on ISH). These assays can be applied to naturally infected fish on-farm to provide information on prevalence of species colonising gills and whether they are linked with lesions. For example, before N. perurans had been described and cultured, species-specific ISH probes were used to show N. perurans was associated with AGD-lesions while N. branchiphila, and N. pemaquidensis were not (Young et al., 2007). Therefore, the use of NGS and ISH could circumnavigate the need to isolate strains that may be difficult to maintain in culture but still provide robust aetiological evidence.

The Tasmanian Atlantic salmon industry may need to review its current freshwater AGD treatment regime if Nolandella sp. and the other 10 plus non-N. perurans species significantly increase AGD severity. Freshwater bathing of salmon for two to four hours may not clear the gills of Nolandella, because this strain appears to be euryhaline. Nolandella strain MX5 tolerated salinity lower than 35 ppt and was detected in 92 % of salmon sampled from the brackish lease located in the lower Huon Estuary. Additionally, an in-house culture of Vannella sp., similarly isolated from AGD-affect gills, also grew in a wide range of salinity, surviving for five days in zero, 10, 35, 38 and 42 ppt MYB media (see images in Appendix E). Euryhaline amoebae are not overly rare, as strains from the genera Mayorella, Cochliopodium, Kortnevella, Vannella, Flabellula and

Stygamoeba have also survived large shifts in salinity (Smirnov, 2007). The common approach for investigating new AGD treatments is to first screen the effect a wide variety of antiparasitics have on the viability of strains of N. perurans in vitro (Botwright et al., in press; Powell and Clark, 2003) before trialing candidate chemical in vivo. If non-N. perurans amoebae are not clearing the gills after freshwater bathing and lesions continue to progress, research developing new AGD treatments should consider how the antiparasitic chemicals impact all amoebae, not just N. perurans.

My alternate hypothesis that AGD has a multi-amoeba aetiology was neither proven nor refuted by this project. This thesis supports N. perurans being the primary agent of AGD because it is highly dominant on the gills of farmed Atlantic salmon in Tasmania and the most pathogenic compared to the two amoebae investigated to date. However, other Amoebozoans colonise farmed Atlantic salmon and they are possibly capable of causing gill pathology under certain conditions. Thus, the involvement of non-N. perurans amoebae in AGD should not yet be discounted. 106

References

Adams, M., Ellard, K., Nowak, B., 2004. Gross pathology and its relationship with histopatholgy of amoebic gill disease (AGD) in farmed Atlantic salmon, Salmo salar L. J. Fish Dis. 27, 151– 161.

Adams, M., Nowak, B., 2003. Amoebic gill disease: sequential pathology in Atlantic salmon, Salmo salar L. J. Fish Dis 26, 601–614.

Adams, M., Crosbie, P., Nowak, B., 2012. Preliminary success using hydrogen peroxide to treat Atlantic salmon, Salmo salar L., affected with experimentally induced amoebic gill disease (AGD). J. Fish Dis. 35, 839–848.

Adams, M., Villavedra, M., Nowak, B., 2008. An opportunistic detection of amoebic gill disease in blue warehou, Seriolella brama Günther, collected from an Atlantic salmon, Salmo salar L., production cage in south eastern Tasmania. J. Fish Dis. 31, 713–717.

Anisimova, M., Gil, M., Dufayard, J., Dessimoz, C., Gascuel, O., 2011. Survey of branch support methods demonstrates accuracy, power, and robustness of fast likelihood-based approximation schemes. Syst. Biol. 60, 685–699.

Antychowicz, J., 2007. Study on rainbow trout nodular gill disease detected in Poland. Bull. Vet. Inst. Pulawy 51, 547–551.

Athanassopoulou, F., Cawthorn, R., Lytra, K., 2002. Amoeba-like infections in cultured marine fishes: systemic infection in pompano Trachinotus falcatus L. from Singapore and gill disease associated with Paramoeba sp. in sea bream Sparus aurata L. from Greece. J. Vet. Med. Ser. B 49, 411–412.

Bermingham, M., 2004. The multifactorial aetiology of amoebic gill disease (AGD) in Ireland. PhD. National University of Ireland.

Bermingham, M., Mulcahy, M., 2007. Neoparamoeba sp. and other protozoans on the gills of Atlantic salmon Salmo salar smolts in seawater. Dis. Aquat. Organ. 76, 234–240.

Bermingham, M., Mulcahy, M., 2006. Microfauna associated with amoebic gill disease in sea- farmed Atlantic salmon, Salmo salar L., smolts. J. Fish Dis. 29, 455–465.

Bermingham, M., Mulcahy, M., 2007. Neoparamoeba sp. and other protozoans on the gills of Atlantic salmon Salmo salar smolts in seawater. Dis. Aquat. Organ. 76, 234–240.

107

Biller, L., Schmidt, H., Krause, E., Gelhaus, C., Matthiesen, J., Handal, G., Lotter, H., Janssen, O., Tannich, E., Bruchhaus, I., 2009. Comparison of two genetically related Entamoeba histolytica cell lines derived from the same isolate with different pathogenic properties. Proteomics 9, 4107–4120.

Blick, A., Giaretta, P., Sprayberry, S., Bush-Vadala, C., Paulk, C., Boeckman, J., Callaway, T., Gill, J., Rech, R., 2019. Comparison of two fixatives in the porcine colon for in situ microbiota studies. J. Anim. Sci. 97, 4803–4809

Boerlage, A., Ashby, A., Herrero, A., Reeves, A., Gunn, G., Rodger, H., 2020. Epidemiology of marine gill diseases in Atlantic salmon (Salmo salar) aquaculture: a review. Rev. Aquac. 1–20.

Bolivar, I., Fahrni, J., Smirnov, A., Pawlowski, J., 2001. SSU rRNA-based phylogenetic position of the genera Amoeba and Chaos (Lobosea, Gymnamoebia): the origin of gymnamoebae revisited. Mol. Biol. Evol. 18, 2306–2314.

Bondad-Reantaso, M., Subasinghe, R., Arthur, J., Ogawa, K., Chinabut, S., Adlard, R., Tan, Z., Shariff, M., 2005. Disease and health management in Asian aquaculture. Vet. Parasitol. 132, 249–272.

Botwright, N., Rusu, A., English, C., Hutt, O., Wynne, J., (in press). A high throughput viability screening method for the marine ectoparasite Neoparamoeba perurans. Protist.

Bovee, E., 1965. An emendation of the ameba genus Flabellula and a description of Vannella gen. nov. Trans. Am. Microsc. Soc. 84, 217–227.

Bowater, R., O’Donoghue, P., 2015. An epizootic of chilodonelliasis in farmed barramundi Lates calcarifer (Bloch), a case report. J. Fish Dis. 38, 931–936.

Bower, S., Blackbourn, J., Meyer, G., Nishimura, D., 1992. Diseases of cultured Japanese scallops (Patinupecten yessoewis) in British Columbia, Canada. Aquaculture 107, 201–210.

Bowman, J., Nowak, B., 2004. Salmonid gill bacteria and their relationship to amoebic gill disease. J. Fish Dis. 27, 483–492.

Bridle, A., Crosbie, P., Cadoret, K., Nowak, B., 2010. Rapid detection and quantification of Neoparamoeba perurans in the marine environment. Aquaculture 309, 56–61.

Bridle, A., Davenport, D., Crosbie, P., Polinski, M., Nowak, B., 2015. Neoparamoeba perurans loses virulence during clonal culture. Int. J. Parasitol. 45, 575–578.

Broeders, S., Huber, I., Grohmann, L., Berben, G., Taverniers, I., Mazzara, M., Roosens, N.,

108

Morisset, D., 2014. Guidelines for validation of qualitative real-time PCR methods. Trends Food Sci. Technol. 37, 115–126.

Bruno, D., Collet, B., Turnbull, A., Kilburn, R., Walker, A., Pendrey, D., McIntosh, A., Urquhart, K., Taylor, G., 2007. Evaluation and development of diagnostic methods for Renibacterium salmoninarum causing bacterial kidney disease (BKD) in the UK. Aquaculture 269, 114–122.

Buchmann, K., Nielsen, T., Sigh, J., Bresciani, J., 2004. Amoebic gill infections of rainbow trout in freshwater ponds. Bull. Eur. Assoc. Fish Pathol. 24, 87–91.

Buchwald, R., 2016. Detection and quantification of an ecologically important marine pathogen, Paramoeba invadens. MSc. Dalhousie University.

Bullock, G., Herman, R., Heinen, J., Noble, A., Weber, A., Hankins, J., 1994. Observations on the occurrence of bacterial gill disease and amoeba gill infestation in rainbow trout cultured in a water recirculation system. J. Aquat. Anim. Health 6, 310–317.

Bustin, S., Benes, V., Garson, J., Hellemans, J., Huggett, J., Kubista, M., Mueller, R., Nolan, T., Pfaffl, M., Shipley, G., Vandesompele, J., Wittwer, C., 2009. The MIQE guidelines: minimum information for publication of quantitative real-time PCR experiments. Clin. Chem. 55, 611– 622.

Bustos, P., Young, N., Rozas, M., Bohle, H., Ildefonso, R., Morrison, R., Nowak, B., 2011. Amoebic gill disease (AGD) in Atlantic salmon (Salmo salar) farmed in Chile. Aquaculture 310, 281–288.

Charman, D., Roe, H., Roland Gehrels, W., 2002. Modern distribution of saltmarsh testate amoebae: Regional variability of zonation and response to environmental variables. J. Quat. Sci. 17, 387–409.

Cano, I., Taylor, N., Bayley, A., Gunning, S., McCullough, R., Bateman, K., Nowak, B., Paley, R., 2019. In vitro gill cell monolayer successfully reproduces in vivo Atlantic salmon host responses to Neoparamoeba perurans infection. Fish Shellfish Immunol. 86, 287–300.

Capella-Gutiérrez, S., Silla-Martínez, J.M., Gabaldón, T., 2009. trimAl: A tool for automated alignment trimming in large-scale phylogenetic analyses. Bioinformatics 25, 1972–1973.

Cavalier-Smith, T., Chao, E., Lewis, R., 2016. 187-gene phylogeny of protozoan phylum Amoebozoa reveals a new class (Cutosea) of deep-branching, ultrastructurally unique, enveloped marine Lobosa and clarifies amoeba evolution. Mol. Phylogenet. Evol. 99, 275– 296. 109

CDEEP, 2000. Information regarding the impact of 1999 lobster mortalities in Long Island Sound. Connecticut Department of Energy and Environmental Protection. http://www.ct.gov/deep/lib/deep/fishing/fisheries_management/lobmor.pdf

Cheng, T., 1970. Hartmannella tahitiensis sp. n., an amoeba associated with mass mortalities of the oyster Crassostrea commercialis in Tahiti, French Polynesia. J. Invertebr. Pathol. 15, 405–419.

Clark, A., Nowak, B., 1999. Field investigations of amoebic gill disease in Atlantic salmon, Salmo salar L., in Tasmania. J. Fish Dis. 22, 433–443.

Clark, G., Powell, M., Nowak, B., 2003. Effects of commercial freshwater bathing on reinfection of Atlantic salmon, Salmo salar, with Amoebic Gill Disease. Aquaculture 219, 135–142.

Collins, C., Hall, M., Bruno, D., Sokolowska, J., Duncan, L., Yuecel, R., McCarthy, U., Fordyce, M.J., Pert, C.C., McIntosh, R., MacKay, Z., 2017. Generation of Paramoeba perurans clonal cultures using flow cytometry and confirmation of virulence. J. Fish Dis. 40, 351–365.

Collins, C., Hall, M., Fordyce, M., White, P., 2019. Survival and growth in vitro of Paramoeba perurans populations cultured under different salinities and temperatures. Protist 170, 153– 167.

Crosbie, P., Bridle, A., Cadoret, K., Nowak, B., 2012. In vitro cultured Neoparamoeba perurans causes amoebic gill disease in Atlantic salmon and fulfils Koch’s postulates. Int. J. Parasitol. 42, 511–515.

Crosbie, P., Bridle, A., Leef, M., Nowak, B., 2010a. Effects of different batches of Neoparamoeba perurans and fish stocking densities on the severity of amoebic gill disease in experimental infection of Atlantic salmon, Salmo salar L. Aquac. Res. 41, 505–516.

Crosbie, P., Ogawa, K., Nakano, D., Nowak, B., 2010b. Amoebic gill disease in hatchery-reared ayu, Plecoglossus altivelis (Temminck & Schlegel), in Japan is caused by Neoparamoeba perurans. J. Fish Dis. 33, 455–458.

Da Rocha-Azevedo, B., Jamerson, M., Cabral, G., Silva-Filho, F., Marciano-Cabral, F., 2009. Acanthamoeba interaction with extracellular matrix glycoproteins: biological and biochemical characterization and role in cytotoxicity and invasiveness. J. Eukaryot. Microbiol. 56, 270– 278.

Daoust, P., Ferguson, H., 1985. Nodular gill disease: a unique form of proliferative gill disease in rainbow trout, Salmo gairdneri Richardson. J. Fish Dis. 8, 511–522.

110

Douglas-Helders, M., Saksida, S., Nowak, B., 2001. Temperature as a risk factor for outbreaks of Amoebic Gill Disease in farmed Atlantic salmon (Salmo salar). Bull. Eur. Assoc. Fish Pathol. 21, 114–116.

Downes, J., Collins, E., Morrissey, T., Connor, I., Rodger, H., Maccarthy, E., Palmer, R., Ruttledge, M., Ruane, N., 2018. Confirmation of Neoparamoeba perurans on the gills of Atlantic salmon during the earliest outbreaks of amoebic gill disease in Ireland. Bull. Eur. Assoc. Fish Pathol. 38, 42–48.

Downes, J., Henshilwood, K., Collins, E., Ryan, A., O’Connor, I., Rodger, H., MacCarthy, E., Ruane, N., 2015. A longitudinal study of amoebic gill disease on a marine Atlantic salmon farm utilising a real-time PCR assay for the detection of Neoparamoeba perurans. Aquac. Environ. Interact. 7, 239–251.

Downes, J., Rigby, M., Taylor, R., Maynard, B., MacCarthy, E., O’Connor, I., Marcos-Lopez, M., Rodger, H., Collins, E., Ruane, N., Cook, M., 2017. Evaluation of non-destructive molecular diagnostics for the detection of Neoparamoeba perurans. Front. Mar. Sci. 4, 1–9.

Dudley, R., Matin, A., Alsam, S., Sissons, J., Maghsood, A., Khan, N., 2005. Acanthamoeba isolates belonging to T1, T2, T3, T4 but not T7 encyst in response to increased osmolarity and cysts do not bind to human corneal epithelial cells. Acta Trop. 95, 100–108.

Dyková, I, Boháčová, L., Fiala, I., Macháčková, B., Pecková, H., Dvořáková, H., 2005a. Amoebae of the genera Vannella Bovee, 1965 and Platyamoeba Page, 1969 isolated from fish and their phylogeny inferred from SSU rRNA gene and ITS sequences. Eur. J. Protistol. 41, 219–230.

Dyková, I., Fiala, I., Lom, J., Lukeš, J., 2003. Perkinsiella amoebae-like encdosymbionts of Neoparamoeba spp., relatives of the kinetoplastid Ichthyobodo. Eur. J. Protistol. 39, 37–52.

Dyková, I., Figueras, A., Novoa, B., Casal, J., 1998a. Paramoeba sp., an agent of amoebic gill disease of turbot Scophthalmus maximus. Dis. Aquat. Organ. 33, 137–141.

Dyková, I., Figueras, A., Peric, Z., 2000. Neoparamoeba Page, 1987: light and electron microscopic observations on six strains of different origin. Dis. Aquat. Organ. 43, 217–223.

Dyková, I., Kostka, M., 2013. Illustrated guide to culture collection of free-living amoebae. Academia, Prague.

Dyková, I., Kostka, M., Pecková, H., 2011a. Three new species of Amoebozoan genus Vexillifera Schaeffer, 1926. Protozoologica 50, 57–65.

111

Dyková, I., Kostka, M., Wortberg, F., Nardy, E., Pecková, H., 2010. New data on aetiology of nodular gill disease in rainbow trout, Oncorhynchus mykiss. Folia Parasitol. (Praha). 57, 157– 163.

Dyková, I., Lom, J., 2004. Advances in the knowledge of amphizoic amoebae infecting fish. Folia Parasitol. (Praha). 51, 81–97.

Dyková, I., Lom, J., Macháčková, B., Pecková, H., 1998b. Vexillifera expectata sp. n. and other non-encysting amoebae isolated from organs of freshwater fish. Folia Parasitol. (Praha). 45, 17–26.

Dyková, I., Lom, J., Machackova, B., Sawyer, T., 1996. Amoebic infections in goldfishes and granulomatous lesions. Folia Parasitol. (Praha). 43, 81–90.

Dyková, I., Lom, J., Schroeder-Diedrich, J., Booton, G., Byers, T., 1999. Acanthamoeba strains isolated from organs of freshwater fishes. J. Parasitol. 85, 1106–1113.

Dyková, I., Lorenzo-Morales, J., Kostka, M., Valladares, B., Pecková, H., 2011b. Neoparamoeba branchiphila infections in moribund sea urchins Diadema aff. antillarum in Tenerife, Canary Islands, Spain. Dis. Aquat. Organ. 95, 225–231.

Dyková, I., Macháčková, B., Pecková, H., 1997. Amoebae isolated from organs of farmed tilapias, Oreochromis niloticus. Folia Parasitol. 44, 81–90.

Dyková, I., Novoa, B., 2001. Comments on diagnosis of amoebic gill disease (AGD) in turbot, Scophthalmus maximus. Bull. Eur. Assoc. Fish Pathol. 21, 40–44.

Dyková, I, Nowak, B., Crosbie, P., Fiala, I., Pecková, H., Adams, M., Macháčková, B., Dvořáková, H., 2005b. Neoparamoeba branchiphila n. sp., and related species of the genus Neoparamoeba Page, 1987: morphological and molecular characterization of selected strains. J. Fish Dis. 28, 49–64.

Dyková, I., Nowak, B., Pecková, H., Fiala, I., Crosbie, P., Dvořáková, H., 2007. Phylogeny of Neoparamoeba strains isolated from marine fish and invertebrates as inferred from SSU rDNA sequences. Dis. Aquat. Organ. 74, 57–65.

Dyková, I., Tyml, T., 2015. Testate amoeba Rhogostoma minus Belar, 1921, associated with nodular gill disease of rainbow trout, Oncorhynchus mykiss (Walbaum). J. Fish Dis. 39, 539– 546.

Edgcomb, V., Orsi, W., Bunge, J., Jeon, S., Christen, R., Leslin, C., Holder, M., Taylor, G., Suarez,

112

P., Varela, R., Epstein, S., 2011. Protistan microbial observatory in the Cariaco Basin, Caribbean. I. pyrosequencing vs Sanger insights into species richness. ISME J. 5, 1344–1356.

Egan, S., Gardiner, M., 2016. Microbial dysbiosis: rethinking disease in marine ecosystems. Front. Microbiol. 7, 1–8.

Eisel, D., Seth, O., Grunewald-Janho, S., Kruchen, B., 2008. DIG application manual for nonradioactive in situ hybridization. Roche Diagnostics.

Embar-Gopinath, S., 2006. Role of bacteria in amoebic gill disease. PhD. University of Tasmania.

Embar-Gopinath, S., Butler, R., Nowak, B., 2005. Influence of salmonid gill bacteria on development and severity of amoebic gill disease. Dis. Aquat. Organ. 67, 55–60.

Engelkirk, P., Duben-Engelkirk, J., Burton, G., 2011. Burton’s microbiology for the health sciences. 9th ed. Lippincott Williams & Wilkins. 238.

English, C., Sword, F., Downes, J., Ruane, N., Botwright, N., Taylor, R., Barnes, A., Wynne, J., Lima, P., Cook, M., 2019a. Prevalence of six amoeba species colonising the gills of farmed Atlantic salmon with amoebic gill disease (AGD) using qPCR. Aquac. Environ. Interact. 11, 405–415.

English, C., Tyml, T., Botwright, N., Barnes, A., Wynne, J., Lima, P., Cook, M., 2019b. A diversity of amoebae colonise the gills of farmed Atlantic salmon (Salmo salar) with amoebic gill disease (AGD). Eur. J. Protistol. 67.

Evans, D., 2005. The multifunctional fish gill: dominant site of gas exchange, osmoregulation, acid- base regulation, and excretion of nitrogenous waste. Physiol. Rev. 85, 97–177.

Falkow, S., 2004. Molecular Koch’s postulates applied to bacterial pathogenicity-a personal recollection 15 years later. Nat. Rev. Microbiol. 2, 67–72.

FAO, 2018. Fisheries and Aquaculture Statistics. Food and Agriculture Organisation. United Nations. http://www.fao.org/fishery/statistics/en.

Feehan, C., Johnson-Mackinnon, J., Scheibling, R., Lauzon-Guay, J., Simpson, A., 2013. Validating the identity of Paramoeba invadens, the causative agent of recurrent mass mortality of sea urchins in Nova Scotia, Canada. Dis. Aquat. Organ. 103, 209–227.

Felsenstein, J., 1985. Confidence limits on phylogenies: an approach using the bootstrap. Soc. Study Evol. 39, 1–15.

Foissner, W., Korganova, G., 1995. Redescription of three testate amoebae (Protozoa, Rhizopoda) 113

from a caucasian soil: Centropyxis plagiostoma Bonnet & Thomas, Cyclopyxis kahli (Deflandre) and C. intermedia Kufferath. Arch. fur Protistenkd. 146, 13–28.

Folmer, O., Black, M., Hoeh, W., Lutz, R., Vrijenhoek, R., 1994. DNA primers for amplification of mitochondrial cytochrome c oxidase subunit I from diverse metazoan invertebrates. Mol. Mar. Biol. Biotechnol. 3, 294–299.

Franke, E., Mackiewicz, J., 1982. Isolation of Acanthamoeba and Naegleria from the intestinal contents of freshwater fishes and their potential pathogenicity. J. Parasitol. 68, 164–166.

Frederick, J., Petri, W., 2005. Roles for the galactose-/N-acetylgalactosamine-binding lectin of Entamoeba in parasite virulence and differentiation. Glycobiology 15, 53–59.

Fringuelli, E., Gordon, A., Rodger, H., Welsh, M., Graham, D., 2012. Detection of Neoparamoeba perurans by duplex quantitative Taqman real-time PCR in formalin-fixed, paraffin-embedded atlantic salmonid gill tissues. J. Fish Dis. 35, 711–724.

Gjessing, M., Thoen, E., Tengs, T., Skotheim, S., Dale, O., 2017. Salmon gill poxvirus, a recently characterized infectious agent of multifactorial gill disease in freshwater- and seawater-reared Atlantic salmon. J. Fish Dis. 40, 1253–1265.

Gofton, A., Oskam, C., Lo, N., Beninati, T., Wei, H., McCarl, V., Murray, D., Paparini, A., Greay, T., Holmes, A., Bunce, M., Ryan, U., Irwin, P., 2015. Inhibition of the endosymbiont “Candidatus Midichloria mitochondrii” during 16S rRNA gene profiling reveals potential pathogens in Ixodes ticks from Australia. Parasites and Vectors. 8, 1–11.

Gregor, T., Fujimoto, K., Masaki, N., Sawai, S., 2010. The onset of collective behavior in social amoebae. Science 80, 1021–1025.

Han, J., 2019. Detection of the amoebic parasite (order Dactylopodida) in cultured Pacific white shrimp (Litopenaeus vannamei). Aquaculture 507, 246–250.

Handlinger, J., Soltani, M., Percival, S., 1997. The pathology of Flexibacter maritimus in aquaculture species in Tasmania, Australia. J. Fish Dis. 20, 159–168.

Hansen, H., Botwright, N., Cook, M., Douglas, A., Downes, J., Gallagher, M., Ruane, N., Matejusova, I., 2019. Genetic diversity among geographically distant isolates of Neoparamoeba perurans. Dis. Aquat. Organ. 137, 81–87.

Harahush, B., Green, K., Webb, R., Hart, N.S., Collin, S., 2012. Optimal preservation of the shark retina for ultrastructural analysis: An assessment of chemical, microwave, and high-pressure

114

freezing fixation techniques. Microsc. Res. Tech. 75, 1218–1228.

Haugland, G., Olsen, A., Rønneseth, A., Andersen, L., 2017. Lumpfish (Cyclopterus lumpus L.) develop amoebic gill disease (AGD) after experimental challenge with Paramoeba perurans and can transfer amoebae to Atlantic salmon (Salmo salar L.). Aquaculture 478, 48–55.

Heger, T., Pawlowski, J., Lara, E., Leander, B., Todorov, M., Golemansky, V., Mitchell, E., 2011. Comparing potential COI and SSU rDNA barcodes for assessing the diversity and phylogenetic relationships of cyphoderiid testate amoebae (Rhizaria: Euglyphida). Protist 162, 131–141.

Hellebø, A., Stene, A., Aspehaug, V., 2017. PCR survey for Paramoeba perurans in fauna, environmental samples and fish associated with marine farming sites for Atlantic salmon (Salmo salar L.). J. Fish Dis. 40, 661–670.

Hernández, J.C., Sangil, C., Lorenzo-Morales, J., 2020. Uncommon southwest swells trigger sea urchin disease outbreaks in Eastern Atlantic archipelagos. Ecol. Evol. 1–8.

Herrero, A., Thompson, K., Ashby, A., Rodger, H., Dagleish, M., 2018. Complex gill disease: an emerging syndrome in farmed Atlantic salmon (Salmo salar L.). J. Comp. Pathol. 163, 23–28.

Howard, T., 2001. Paramoebiasis of sea-farmed salmonids in Tasmania - a study of its aetiology, pathogenicity, and control. PhD. University of Tasmania.

Hugerth, L., Muller, E., Hu, Y., Lebrun, L., Roume, H., Lundin, D., Wilmes, P., Andersson, A., 2014. Systematic design of 18S rRNA gene primers for determining eukaryotic diversity in microbial consortia. PLoS One. 9, 1–11.

Jamerson, M., da Rocha-Azevedo, B., Cabral, G., Marciano-Cabral, F., 2012. Pathogenic Naegleria fowleri and non-pathogenic Naegleria lovaniensis exhibit differential adhesion to, and invasion of, extracellular matrix proteins. Microbiology 158, 791–803.

Jellet, J., Scheibling, R., 1988. Virulence of Paramoeba invadens Jones (Amoebida, Paramoebidae) from monoxenic and polyxenic culture. J. Protozool. 35, 422–424.

Johnson, P., 1977. Paramoebiasis in the blue crab, Callinectes sapidus. J. Invertebr. Pathol. 29, 308–320.

Jones, G., 1985. Paramoeba invadens n. sp. (Amoebida, Paramoebidae), a pathogenic amoeba from the sea urchin, Strongylocentrotus droebachiensis, in Eastern Canada. J. Protozool. 32, 564– 569.

115

Jones, G., Hebda, A., Scheibling, R., Miller, R., 1985. Histopathology of the disease causing mass mortality of sea urchins (Strongylocentrotus droebachiensis) in Nova Scotia. J. Invertebr. Pathol. 45, 260–271.

Jones, G., Scheibling, R., 1985. Paramoeba sp. (Amoebida, Paramoebidae) as the possible causative agent of sea urchin mass mortality in Nova Scotia. J. Parasitol. 71, 559–565.

Kalyaanamoorthy, S., Minh, B., Wong, T., Von Haeseler, A., Jermiin, L., 2017. ModelFinder: Fast model selection for accurate phylogenetic estimates. Nat. Methods 14, 587–589.

Kang, S., Tice, A., Spiegel, F., Silberman, J., Pánek, T., Čepička, I., Kostka, M., Kosakyan, A., Alcântara, D., Roger, A., Shadwick, L., Smirnov, A., Kudryavtsev, A., Lahr, D., Brown, M., 2017. Between a pod and a hard test: the deep evolution of amoebae. Mol. Biol. Evol. 34, 2258–2270.

Karlsbakk, E., Olsen, A., Einen, A., Mo, T., Fiksdal, I., Aase, H., Kalgraff, C., Skår, S., Hansen, H., 2013. Amoebic gill disease due to Paramoeba perurans in ballan wrasse (Labrus bergylta). Aquaculture 412–413, 41–44.

Katoh, K., Rozewicki, J., Yamada, K., 2017. MAFFT online service: multiple sequence alignment, interactive sequence choice and visualization. Brief. Bioinform. 1–7.

Kent, M., Sawyer, T., Hedrick, R., 1988. Paramoeba pemaquidensis (Sarcomastigophora: Paramoebidae) infestation of the gills of coho salmon Oncorhynchus kisutch reared in sea water. Dis. Aquat. Organ. 5, 163–169.

Kim, H., Cho, J., Lee, M., Huh, M. Do, Kim, K., 2005. Neoparamoeba sp. infection on gills of olive flounder, Paralichthys olivaceus in Korea. J. Fish Pathol. 18, 125–131.

Kim, W., Kong, K., Kim, J., Jung, S., Kim, J., Oh, M., 2017. Amoebic gill disease outbreak in marine fish cultured in Korea. J. Vet. Diagnostic Investig. 29, 357–361.

Koch, R., 1884. Die Aetiologie der Tuberkulose. Mitt Kaiser Gesundh 2, 1–88.

Köhsler, M., Leitner, B., Blaschitz, M., Michel, R., Aspöck, H., Walochnik, J., 2006. ITS1 sequence variabilities correlate with 18S rDNA sequence types in the genus Acanthamoeba (Protozoa: Amoebozoa). Parasitol. Res. 98, 86–93.

Kosakyan, A., Heger, T., Leander, B., Todorov, M., Mitchell, E., Lara, E., 2012. COI barcoding of Nebelid testate amoebae (Amoebozoa: Arcellinida): extensive cryptic diversity and redefinition of the Hyalospheniidae Schultze. Protist 163, 415–434.

116

Kudryavtsev, A., Gladkikh, A., 2017. Two new species of Ripella (Amoebozoa, Vannellida) and unusual intragenomic variability in the SSU rRNA gene of this genus. Eur. J. Protistol. 61, 92– 106.

Kurth, T., Wiedmer, S., Entzeroth, R., 2012. Improvement of ultrastructural preservation of Eimeria oocysts by microwave-assisted chemical fixation or by high pressure freezing and freeze substitution. Protist 163, 296–305.

Kvinnsland, h., Histochemical characterization of AGD lesions in ballan wrasse (Labrus bergylta). MSc. University of Bergen.

Larsson, A., 2014. AliView: A fast and lightweight alignment viewer and editor for large datasets. Bioinformatics 30, 3276–3278.

Lee, X., Reimmann, C., Greub, G., Sufrin, J., Croxatto, A., 2012. The Pseudomonas aeruginosa toxin L-2-amino-4-methoxy-trans-3-butenoic acid inhibits growth and induces encystment in Acanthamoeba castellanii. Microbes Infect. 14, 268–272.

Legrand, T., Catalano, S., Wos-Oxley, M., Stephens, F., Landos, M., Bansemer, M., Stone, D., Qin, J., Oxley, A., 2018. The inner workings of the outer surface: skin and gill microbiota as indicators of changing gut health in yellowtail kingfish. Front. Microbiol. 8, 1–17.

Leiro, J., Paniagua, E., Ortega, M., Parama, A., Ferna, J., 1998. An amoeba associated with gill disease in turbot , Scophthalmus maximus (L .). J. Fish Dis. 21, 281–288.

Lima, P., Taylor, R., Cook, M., 2016. Involvement of contractile vacuoles in the osmoregulation process of the marine parasitic amoeba Neoparamoeba perurans. J. Fish Dis. 39, 629–633.

López-García, P., Rodríguez-Valera, F., Pedrós-Alió, C., Moreira, D., 2001. Unexpected diversity of small eukaryotes in deep-sea Antarctic plankton. Nature 409, 603–607.

Lorenzo-Morales, J., Khan, N., Walochnik, J., 2015. An update on Acanthamoeba keratitis: diagnosis, pathogenesis and treatment. Parasite 22, 10.

Mahood, R., McKenzie, M., Middaugh, D., Bollar, S., Davis, J., 1970. A report on the cooperative blue crab study - South Atlantic States. US Bur. Commer. Fish.

Mansell, B., Powell, M., Ernst, I., Nowak, B., 2005. Effects of the gill monogenean Zeuxapta seriolae (Meserve, 1938) and treatment with hydrogen peroxide on pathophysiology of kingfish, Seriola lalandi Valenciennes, 1833. J. Fish Dis. 28, 253–262.

Marcos-López, M., Rodger, H., O’Connor, I., Braceland, M., Burchmore, R., Eckersall, P.,

117

MacCarthy, E., 2017. A proteomic approach to assess the host response in gills of farmed Atlantic salmon Salmo salar L. affected by amoebic gill disease. Aquaculture 470, 1–10.

Martinez, A., Visvesvara, G., 1997. Free-living, amphizoic and opportunistic amebas. Brain Pathol. 7, 583–598.

Merrifield, D., Rodiles, A., 2015. The fish microbiome and its interactions with mucosal tissues, Mucosal Health in Aquaculture. Elsevier Inc. Alabama.

Meyer, F., 1991. Aquaculture disease and health management. J. Anim. Sci. 69, 4201–4208.

Miller, R., 2008. A sea urchin diver fishery managed by exclusive fishing areas. Food and Agriculture Organisation. United Nations.

Mitchell, S., Rodger, H., 2011. A review of infectious gill disease in marine salmonid fish. J. Fish Dis. 34, 411–432.

Morado, J., 2011. Protistan diseases of commercially important crabs: a review. J. Invertebr. Pathol. 106, 27–53.

Morrison, R., Crosbie, P., Cook, M., Adams, M., Nowak, B., 2005. Cultured gill-derived Neoparamoeba pemaquidensis fails to elicit amoebic gill disease (AGD) in Atlantic salmon Salmo salar. Dis. Aquat. Organ. 66, 135–144.

Morrison, R., Crosbie, P., Nowak, B., 2004. The induction of laboratory-based amoebic gill disease revisited. J. Fish Dis. 27, 445–449.

Mouton, A., Crosbie, P., Cadoret, K., Nowak, B., 2014. First record of amoebic gill disease caused by Neoparamoeba perurans in South Africa. J. Fish Dis. 37, 407–409.

Mullen, T., Nevis, K., O’Kelly, C., Gast, R., Frasca, S., 2005. Nuclear small-subunit ribosomal RNA gene-based characterization, molecular phylogeny and PCR detection of the Neoparamoeba from Western Long Island Sound lobster. J. Shellfish Res. 24, 719–731.

Mullen, T., Russell, S., Tucker, M., Maratea, J., Koerting, C., Hinckley, L., De Guise, S., Frasca, S., French, R., Burrage, T., Perkins, C., 2004. Paramoebiasis associated with mass mortality of American lobster Homarus americanus in Long Island Sound, USA. J. Aquat. Anim. Health 16, 29–38.

Munday, B., 1986. Disease of salmonids. In: Humphrey, J., Langdon, J. (Eds.), Workshop on Disease of Australian Fish and Shellfish. Department of Agriculture and Rural Affairs. Benalla, Australia.

118

Munday, B., Foster, C., Roubal, F., Lester, R., 1990. Paramoebic gill infection and associated pathology of Atlantic salmon, Salmo salar, and rainbow trout, Salmo gairdneri, in Tasmania. Pathol. Mar. Sci. 215–222.

Munday, B., Zilberg, D., Findlay, V., 2001. Gill disease of marine fish caused by infection with Neoparamoeba pemaquidensis. J. Fish Dis. 24, 497–507.

Nash, G., Nash, M., 1988. Systemic amebiasis in cultured European catfish, Silurus glanis L. J. Fish Dis. 11, 57–71.

Nassonova, E., Smirnov, A., Fahrni, J., Pawlowski, J., 2010. Barcoding amoebae: comparison of SSU, ITS and COI genes as tools for molecular identification of naked lobose amoebae. Protist 161, 102–115.

Neff, R., Ray, S., Benton, W., Wilborn, M., 1964. Induction of synchronous encystment (differentiation) in Acanthamoeba sp. Methods Cell Biol. 1, 55–83.

Newman, M., Ward, G., 1973. An epizootic of blue crabs, Callinectes sapidus, caused by Paramoeba perniciosa. J. Invertebr. Pathol. 22, 329–334.

Noble, A., Herman, R., Noga, E., Bullock, G., 1997. Recurrent amoebic gill infestation in rainbow trout cultured in a semiclosed water recirculation system. J. Aquat. Anim. Health 9, 64–69.

Nowak, B., Archibald, J., 2018. Opportunistic but lethal: the mystery of paramoebae. Trends Parasitol. 34, 404–419.

Nowak, B., Morrison, R., Crosbie, P., Butler, R., Bridle, A., Gross, K., Embar-gopinath, S., Carson, J., Raison, R., Villavedra, M., McCarthy, K., Broady, K., Wallach, M., 2004. Host-pathogen interactions in amoebic gill disease. Aquafin CRC FRDC Final Project Report.

Nowak, B., Munday, B., 1994. Histology of gills of Atlantic salmon during the first few months following transfer to sea water. Bull. Eur. Assoc. Fish Pathol. 14, 77–81.

Oldham, T., Rodger, H., Nowak, B., 2016. Incidence and distribution of amoebic gill disease (AGD) - An epidemiological review. Aquaculture 457, 35–42.

Page, F., 1983. Marine gymnamoeba. The Lavenham Press, Great Britain.

Parsons, H., Nowak, B., Fisk, D., Powell, M., 2001. Effectiveness of commercial freshwater bathing as a treatment against amoebic gill disease in Atlantic salmon. Aquaculture 195, 205– 210.

Pawlowski, J., Audic, S., Adl, S., Bass, D., Belbahri, L., Berney, C., Bowser, S., Cepicka, I., 119

Decelle, J., Dunthorn, M., Fiore-Donno, A., Gile, G., Holzmann, M., Jahn, R., Jirků, M., Keeling, P., Kostka, M., Kudryavtsev, A., Lara, E., Lukeš, J., Mann, D., Mitchell, E., Nitsche, F., Romeralo, M., Saunders, G., Simpson, A., Smirnov, A., Spouge, J., Stern, R., Stoeck, T., Zimmermann, J., Schindel, D., de Vargas, C., 2012. CBOL Protist Working Group: barcoding eukaryotic richness beyond the animal, plant, and fungal kingdoms. PLoS Biol. 10, 1–5.

Pawlowski, J., Burki, F., 2009. Untangling the phylogeny of amoeboid protists. J. Eukaryot. Microbiol. 56, 16–25.

Pearce, J., Balcom, N., 2005. The 1999 Long Island Sound lobster mortality event: findings of the comprehensive research initiative. J. Shellfish Res. 24, 691–697.

Pennacchi, Y., Adams, M., Nowak, B., Bridle, A., 2016. Immune gene expression in the gills of Atlantic salmon (Salmo salar L.) following experimental reinfection with Neoparamoeba perurans. Aquaculture 464, 410–419.

Perkins, F., Castagna, M., 1971. Ultrastructure of the Nebenkörper or "secondary nucleus: of the parasitic amoeba Paramoeba perniciosa (Amoebida, Paramoebidae). J. Invertabrate Pathol. 17, 186–193.

Perolo, A., Gustinelli, A., Fioravanti, M.L., Manfrin, A., Dalla Pozza, M., Lunelli, F., Accini, A., Quaglio, F., 2019. Occurrence of nodular gill disease in farmed brown trout (Salmo trutta L.). J. Fish Dis. 42, 1315–1320.

Petersen, C., Round, J., 2014. Defining dysbiosis and its influence on host immunity and disease. Cell. Microbiol. 16, 1024–1033.

Powell, M., Carson, J., Van Gelderen, R., 2004. Experimental induction of gill disease in Atlantic salmon Salmo salar smolts with Tenacibaculum maritimum. Dis. Aquat. Organ. 61, 179–185.

Powell, M., Clark, G., 2003. In vitro survival and the effect of water chemistry and oxidative chemical treatments on isolated gill amoebae from AGD-affected Atlantic salmon. Aquaculture 220, 135–144.

Powell, M., Leef, M., Roberts, S., Jones, M. 2008. Neoparamoebic gill infections: host response and physiology in salmonids. J. Fish Biol. 73, 2161–2183.

Prescott, J., Feldmann, H., Safronetz, D., 2017. Amending Koch’s postulates for viral disease: when “growth in pure culture” leads to a loss of virulence. Antiviral Res. 137, 1–5.

Purcell, M., Getchell, R., McClure, C., Garver, K., 2011. Quantitative polymerase chain reaction

120

(PCR) for detection of aquatic animal pathogens in a diagnostic laboratory setting. J. Aquat. Anim. Health 23, 148–161.

Quaglio, F., Perolo, A., Bronzatti, P., Gustinelli, A., Menconi, V., Cavazza, G., Caffara, M., Manfrin, A., Gallo, E., Fioravanti, M., 2016. Nodular gill disease in farmed rainbow trout (Oncorhynchus mykiss) in Italy. J. Fish Dis. 39, 1139–1142.

Quast, C., Pruesse, E., Yilmaz, P., Gerken, J., Schweer, T., Yarza, P., Peplies, J., Glöckner, F., 2013. The SILVA ribosomal RNA gene database project: improved data processing and web- based tools. Nucleic Acids Res. 41, 590–596.

R Core Team, 2018. R: a language and environment for statistical computing. R Foundation for Statistical Computing.

Rach, J., Gaikowski, M., Ramsay, R., 2000. Efficacy of hydrogen peroxide to control mortalities associated with bacterial gill disease infections on hatchery-reared salmonids. J. Aquat. Anim. Health 12, 119–127.

Rendueles, O., Ghigo, J., 2015. Mechanisms of competition in biofilm communities. Microbiol. Spectr. 3, 319–342.

Rodger, H., 2019. Amoebic gill disease in farmed halibut (Hippoglossus hippoglossus) in the United Kingdom. Vet. Rec. Case Reports 7, 2018–2020.

Rodger, H., Mcardle, J.F., 1996. An outbreak of amoebic gill disease in Ireland. Vet. Rec. 139, 348–349.

Rodríguez-Zaragoza, S., 1994. Ecology of free-living amoebae. Crit. Rev. Microbiol. 20, 225–241.

Rolin, C., Graham, J., McCarthy, U., Martin, S., Matejusova, I., 2016. Interactions between Paramoeba perurans, the causative agent of amoebic gill disease, and the blue mussel, Mytilus edulis. Aquaculture 456, 1–8.

Ross, L., Woodward, J., 2016. Koch’s postulates: an interventionist perspective. Stud. Hist. Philos. Biol. Biomed. Sci. 59, 35–46.

Rozas-Serri, M., 2019. Gill diseases in marine salmon aquaculture with an emphasis on amoebic gill disease. CAB Rev. 14.

Rozas-Serri, M., Bohle, H., Grothusen, H., Bustos, P., 2012. Epidemiology of amoebic gill disease (AGD) in Chilean salmon industry between 2007 and 2010. Bull. Eur. Assoc. Fish Pathol. 32, 181–188.

121

Santos, M., Cavaleiro, F., Campos, P., Sousa, A., Teixeira, F., Martins, M., 2010. Impact of amoeba and scuticociliatidia infections on the aquaculture European sea bass (Dicentrarchus labrax L.) in Portugal. Vet. Parasitol. 171, 15–21.

Sawyer, T., Ghittino, P., Andruetto, S., Pernin, P., Pussard, M., 1978. Vexillifera bacillipedes Page, 1969, an amphizoic amoeba of hatchery rainbow trout in Italy. Trans. Am. Microsc. Soc. 97, 596–600.

Sawyer, T., Hnath, J., Conrad, J., 1974. Thecamoeba hoffmani sp. n. (Amoebida: Thecamoebidae) from gills of fingerling salmonid fish. J. Parasitol. 60, 677–682.

Scheibling, R., Feehan, C., Lauzon-Guay, J., 2010. Disease outbreaks associated with recent hurricanes cause mass mortality of sea urchins in Nova Scotia. Mar. Ecol. Prog. Ser. 408, 109– 116.

Schroeder, J., Booton, G., Hay, J., Niszl, I., Seal, D., Markus, M., Fuerst, P., Byers, T., 2001. Use of subgenic 18S ribosomal DNA PCR and sequencing for genus and genotype identification of Acanthamoebae from humans with keratitis and from sewage sludge. J. Clin. Microbiol. 39, 1903–1911.

Schuster, F., Visvesvara, G., 2004. Free-living amoebae as opportunistic and non-opportunistic pathogens of humans and animals. Int. J. Parasitol. 34, 1001–1027.

Schuster, S., 2008. Next-generation sequencing transforms today’s biology. Nat. Methods 5, 16–18.

Sibbald, S., Cenci, U., Colp, M., Eglit, Y., O’Kelly, C., Archibald, J., 2017. Diversity and evolution of Paramoeba spp. and their kinetoplastid endosymbionts. J. Eukaryot. Microbiol. 64, 598– 607.

Sims, G., Aitken, R., Rogerson, A., 2002. Identification and phylogenetic analysis of morphologically similar naked amoebae using small subunit ribosomal RNA. J. Eukaryot. Microbiol. 49, 478–484.

Smirnov, A., 1999. Amoebae on the Web . St Petersbg. Univ. http://amoeba.ifmo.ru/start.htm.

Smirnov, A., 2002. Re-description of Vannella mira Schaeffer 1926 (Gymnamoebia, Vannellidae), an often mentioned but poorly known amoebae species. Protistology 2, 178–184.

Smirnov, A., 2007. Cryptic freshwater amoeba species in the bottom sediments of Nivå Bay (Øresund, Baltic Sea). Eur. J. Protistol. 43, 87–94.

Smirnov, A., Chao, E., Nassonova, E., Cavalier-Smith, T., 2011. A revised classification of naked

122

lobose amoebae (Amoebozoa: Lobosa). Protist 162, 545–570.

Smirnov, A., Goodkov, A., 1999. An illustrated list of basic morphotypes of Gymnamoebia (Rhizopoda, Lobosea). Protistology 1, 20–29.

Smirnov, A., Nassonova, E., Chao, E., Cavalier-Smith, T., 2007. Phylogeny, evolution, and taxonomy of Vannellid amoebae. Protist 158, 295–324.

Smith, H., Bobrov, A., Lara, E., 2008. Diversity and biogeography of testate amoebae. Biodivers. Conserv. 17, 329–343.

Smith, S., Newman, S., Coleman, M., Alex, C., 2018. Characterization of the histologic appearance of normal gill tissue using special staining techniques. J. Vet. Diagnostic Investig. 30, 688– 698.

Speare, D., 1999. Nodular gill disease (amoebic gill infestation) in Arctic char, Salvelinus alpinus. J. Comp. Pathol. 121, 277–282.

Speare, D., Ferguson, H., 1989. Clinical and pathological features of common gill diseases of cultured salmonids in Ontario. Can. Vet. J. 30, 882–887.

Sprague, V., Beckett, R., Sawyer, T., 1969. A new species of Paramoeba (Amoebida, Paramoebidae) parasitic in the crab Callinectes sapidus. J. Invertebr. Pathol. 14, 167–174.

Sprague, V., Beckett, R., 1966. A disease of blue crabs (Callinectes sapidus) in Maryland and Virginia. J. Invertebr. Pathol. 8, 287–289.

Stagg, H., Hall, M., Wallace, I., Pert, C., Garcia Perez, S., Collins, C., 2015. Detection of Paramoeba perurans in Scottish marine wild fish populations. Bull. Eur. Assoc. Fish Pathol. 35, 217–226.

Staroscik, A., 2004. Calculator for determining the number of copies of a template. URI Genomics Seq. Cent.

Steinum, T., Kvellestad, A., Rønneberg, L., Nilsen, H., Asheim, A., Fjell, K., Nygård, S., Olsen, A., Dale, O., 2008. First cases of amoebic gill disease (AGD) in Norwegian seawater farmed Atlantic salmon, Salmo salar L., and phylogeny of the causative amoeba using 18S cDNA sequences. J. Fish Dis. 31, 205–214.

Suhnel, S., Da S. Ivachuk, C., Schaefer, A., Pontinha, V., Martins, M., Figueras, A., Meyer, G., Jones, S., Stewart, J., Gurney-Smith, H., Magalhaes, A., Bower, S., 2014. Detection of a parasitic amoeba (Order Dactylopodida) in the female gonads of oysters in Brazil. Dis. Aquat.

123

Organ. 109, 241–250.

Suzuki, M., Nakagawa, Y., Harayama, S., Yamamoto, S., 2001. Phylogenetic analysis and taxonomic study of marine Cytophaga-like bacteria: Proposal for Tenacibaculum gen. nov. with Tenacibaculum maritimum comb. nov. and Tenacibaculum ovolyticum comb. nov., and description of Tenacibaculum mesophilum sp. nov. and Tenacibaculum amylolyticum sp. nov. Int. J. Syst. Evol. Microbiol. 51, 1639–1652.

Taylor, P., 1977. Isolation and experimental infection of free-living amoebae in freshwater fishes. J. Parasitol. 63, 232–237.

Taylor, R., Muller, W., Cook, M., Kube, P., Elliott, N., 2009. Gill observations in Atlantic salmon (Salmo salar, L.) during repeated amoebic gill disease (AGD) field exposure and survival challenge. Aquaculture 290, 1–8.

Tekle, Y., Grant, J., Anderson, R., Nerad, T., Cole, J., Patterson, D., Katz, L., 2008. Phylogenetic placement of diverse amoebae inferred from multigene analyses and assessment of clade stability within “Amoebozoa” upon removal of varying rate classes of SSU-rDNA. Mol. Phylogenet. Evol. 47, 339–352.

Tekle, Y., 2014. DNA barcoding in Amoebozoa and challenges: the example of Cochliopodium. Protist 165, 473–484.

Thermo Fisher, 2010. Applied Biosystems Custom Primer & TaqMan Probes.

Thomas, V., Herrera-Rimann, K., Blanc, D., Greub, G., 2006. Biodiversity of amoebae and amoebae-resistant bacteria in a hospital water network. Appl. Environ. Microbiol. 72, 2428– 2438.

Treasurer, J., Wadsworth, S., Grant, A., 2000. Resistance of sea lice, Lepeophtheirus salmonis (Kroyer), to hydrogen peroxide on farmed Atlantic salmon, Salmo salar L. Aquac. Res. 31, 855–860.

Trifinopoulos, J., Nguyen, L., von Haeseler, A., Minh, B., 2016. W-IQ-TREE: a fast online phylogenetic tool for maximum likelihood analysis. Nucleic Acids Res. 44, W232–W235.

Tubbs, L., Wybourne, B., Lumsden, J., Lumsden, J., 2010. Nodular gill disease causing proliferative branchitis and mortality in chinook salmon (Oncorhynchus tshawytscha). N. Z. Vet. J. 58, 59–61.

Tyml, T., Dyková, I., 2017. Phylogeny and taxonomy of new and re-examined strains of Tubulinea

124

(Amoebozoa). Eur. J. Protistol. 61, 41–47.

Udalov, I., 2016. Pseudoparamoeba microlepis n. sp., Korotnevella fousta n. sp. (Amoebozoa, Dactylopodida), with notes on the evolution of scales among dactylopodid amoebae. Eur. J. Protistol. 54, 33–46.

Udalov, I., Zlatogursky, V., Smirnov, A., 2016. A new freshwater naked lobose amoeba Korotnevella venosa n. sp. (Amoebozoa, Discosea). J. Eukaryot. Microbiol. 63, 834–840.

Vincent, B., Adams, M., Crosbie, P., Nowak, B., Morrison, R., 2007. Atlantic salmon (Salmo salar L .) exposed to cultured gill-derived Neoparamoeba branchiphila fail to develop amoebic gill disease ( AGD ). Bull. Eur. Assoc. Fish Pathol. 27, 112–115.

Voelker, F., Anver, M., Mckee, A., Casey, H., Brenniman, G., 1977. Amebiasis in goldfish. Vet. Pathol. 14, 247–255.

Volkova, E., Kudryavtsev, A., 2017. Description of Neoparamoeba longipodia n. sp. and a new strain of Neoparamoeba aestuarina (Page, 1970) (Amoebozoa, Dactylopodida) from deep-sea habitats. Eur. J. Protistol. 61, 107–121.

Webb, S., Garman, G., McIninch, S., Brown, B., 2002. Amoebae associated with ulcerative lesions of fish from tidal freshwater of the James River, Virginia. J. Aquat. Anim. Health 14, 68–76.

Wheatley, S., McLoughlin, M., Menzies, F., Goodall, E., 1995. Site management factors influencing mortality rates in Atlantic salmon (Salmo salar L.) during marine production. Aquaculture 136, 195–207.

Wiik-Nielsen, J., Mo, T., Kolstad, H., Mohammad, S., Hytterød, S., Powell, M., 2016. Morphological diversity of Paramoeba perurans trophozoites and their interaction with Atlantic salmon, Salmo salar L., gills. J. Fish Dis. 39, 1113–1123.

Wolff, D., Bagg, A., Cooley, L., Dewald, G., Hirsch, B., Jacky, P., Rao, K., Rao, N., 2007. Guidance for fluorescence in situ hybridization testing in hematologic disorders. J. Mol. Diagnostics 9, 134–143.

Wong, F., Carson, J., Elliott, N., 2004. 18S ribosomal DNA-based PCR identification of Neoparamoeba pemaquidensis, the agent of amoebic gill disease in sea-farmed salmonids. Dis. Aquat. Organ. 60, 65–76.

Wong, M., Karr, S., Chow, C., 1977. Changes in the virulence of Naegleria fowleri maintained in vitro. J. Parasitol. 63, 872–878.

125

World Bank, 2013. Fish to 2030: prospects for fisheries and aquaculture. Food and Agriculture Organisation. Washington. http://www.fao.org/3/i3640e/i3640e.pdf

Wright, D., Nowak, B., Oppedal, F., Bridle, A., Dempster, T., 2015. Depth distribution of the amoebic gill disease agent, Neoparamoeba perurans, in salmon sea-cages. Aquac. Environ. Interact. 7, 67–74.

Wright, D., Nowak, B., Oppedal, F., Bridle, A., Dempster, T., 2017. Free-living Neoparamoeba perurans depth distribution is mostly uniform in salmon cages, but reshaped by stratification and potentially extreme fish crowding. Aquac. Environ. Interact. 9, 269–279.

Wynne, J., Stratford, C., Slinger, J., Samsing, F., Rigby, M., McCulloch, R., Quezada-Rodriguez, P., Taylor, R., 2020. The interaction between temperature and dose on the efficacy and biochemical response of Atlantic salmon to hydrogen peroxide treatment for amoebic gill disease. J. Fish Dis. 43, 39–48.

Yang, Z., Cao, Z., Panjwani, N., 1997. Pathogenesis of Acanthamoeba keratitis: Carbohydrate- mediated host- parasite interactions. Infect. Immun. 65, 439–445.

Young, N., Crosbie, P., Adams, M., Nowak, B., Morrison, R., 2007. Neoparamoeba perurans n. sp., an agent of amoebic gill disease of Atlantic salmon (Salmo salar). Int. J. Parasitol. 37, 1469–1481.

Young, N., Dyková, I., Crosbie, P., Wolf, M., Morrison, R., Bridle, A., Nowak, B., 2014. Support for the coevolution of Neoparamoeba and their endosymbionts, Perkinsela amoebae-like organisms. Eur. J. Protistol. 50, 509–523.

Young, N. , Dyková, I., Nowak, B., Morrison, R., 2008a. Development of a diagnostic PCR to detect Neoparamoeba perurans, agent of amoebic gill disease. J. Fish Dis. 31, 285–295.

Young, N., Dyková, I., Snekvik, K., Nowak, B., Morrison, R., 2008b. Neoparamoeba perurans is a cosmopolitan aetiological agent of amoebic gill disease. Dis. Aquat. Organ. 78, 217–223.

Zilberg, D., Gross, A., Munday, B., 2001. Production of salmonid amoebic gill disease by exposure to Paramoeba sp. harvested from the gills of infected fish. J. Fish Dis. 24, 79–82.

126

Appendices

Appendix A: Gross morphology dimensions of amoeba strains grown in monoculture

Table A.1 Gross morphology dimensions of amoeba strains grown in monoculture. Presented as mean values (M) with corresponding range (R) and standard deviation (SD). All values in µm.

Attached Floating

Length Breadth L:B ratio Diameter Pseudopod M R SD M R SD M R SD M R SD M R SD MP1 36.68 27.5- 6.33 24.7 17.8- 5.39 1.54 1.0- 0.40 17.0 11.7 2.61 11.6 5.2- 4.51 51.5 4 39.4 2.9 1 - 3 22.5 21.0 MP2 38.57 29.7- 5.02 21.0 10.1- 7.09 2.12 1.2- 1.00 16.0 11.5 2.68 11.9 3.6- 5.08 46.9 0 35.4 4.6 8 - 2 24.3 20.8 MX6 15.37 9.6- 4.35 10.8 7.5- 1.92 1.44 1.0- 0.41 6.26 3.4- 1.36 7.15 3.0- 3.32 25.3 1 15.7 2.4 9.6 14.9 MX1 11.67 7.7- 2.29 6.27 2.6- 1.88 2.11 1.2- 1.07 5.32 2.8- 1.23 2.97 1.6- 1.49 18.6 9.4 5.4 7.3 8.0 MV5 36.35 27.0- 6.50 21.5 15.6- 3.33 1.81 1.2- 0.36 14.6 12.1 2.17 10.0 4.7- 5.82 47.7 3 27.8 2.7 2 - 9 23.2 19.1 MV2 27.97 18.6- 4.95 31.1 20.5- 4.57 0.92 0.6- 2.47 8.04 4.6- 2.47 16.4 8.2- 5.53 43.8 9 40.7 1.5 13.0 7 25.1 MV3 28.06 19.5- 3.87 27.1 18.6- 5.71 0.92 0.6- 0.24 8.40 3.6- 2.14 15.4 7.3- 6.38 34.1 4 36.7 1.6 12.1 6 32.3 MV4 14.64 10.5- 3.43 12.9 8.9- 2.91 1.21 0.6- 0.28 5.26 3.4- 1.08 4.53 2.6- 2.65 27.7 2 20.7 2.0 7.3 14.8 MX4 27.41 20.2- 4.21 4.02 2.6- 0.98 7.26 3.8- 2.40 4.46 3.6- 0.57 8.02 4.0- 3.09 34.1 4.6 12.5 5.2 14.3 MX3 9.26 6.2- 1.33 4.11 2.1- 1.09 2.38 1.6- 0.66 † † † † † † 11.4 6.7 3.6 MX5 8.20 5.5- 1.91 3.53 2.1- 0.84 2.48 1.1- 0.93 † † † † † † 13.7 5.4 4.6 † Cells too small to accurately measure these features under 200 x magnification.

127

Appendix B: qPCR assay optimisation and validation

Table B.1 Optimisation of primer concentration for qPCR assays designed to detect Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2) and Vannellida C3 (VanC3). Results presented as mean Ct value and standard deviation (SD). Values in bold with a ‡ symbol indicate the chosen concentration.

Concentration (nM) of Fwd Nol Pse ParE VanC2 VanC3 and Rv primers Mean SD Mean SD Mean SD Mean SD Mean SD 100F_100R 24.75 0.10 25.52 0.47 20.92 0.18 NA NA 23.30 0.41 50F_300R 24.39 0.19 23.82 0.82 19.76 0.11 20.56 0.13 19.79 0.13 50F_900R 23.98 0.28 23.50 0.51 19.28 0.13 20.50 0.05 16.77 0.10 300F_50R 24.37 0.16 37.35 NA 21.06 0.19 20.88 0.13 28.83 1.09 300F_300R 23.78 0.10 22.59‡ 0.25 18.96 0.17 20.38‡ 0.08 19.30 0.21 400F_400R 23.39‡ 0.11 22.59 0.14 18.49 0.17 20.47 0.01 18.89 0.77 300F_900R 23.37 0.15 22.05 0.26 17.87‡ 0.24 20.29 0.06 16.76‡ 0.27 900F_50R 24.81 0.22 27.71 1.85 21.41 0.26 20.75 0.14 26.70 0.26 900F_300R 23.74 0.07 22.25 0.23 18.68 0.21 20.37 0.11 19.76 0.20

128

Table B.2 Optimisation of probe concentration for qPCR assays designed to detect Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2) and Vannellida C3 (VanC3). Results presented as mean Ct value and standard deviation (SD). Values in bold with a ‡ symbol indicate the chosen concentration. Nol Pse ParE VanC2 VanC3

Concentration (nM) of probe Mean SD Mean SD Mean SD Mean SD Mean SD 25 21.98 0.56 NA NA 21.37 0.04 21.26 NA 50 18.96 0.08 24.75 0.01 19.76 0.22 19.80 0.42 18.25 0.61 75 18.08 0.10 23.14 0.35 19.10 0.12 17.53 0.23 100 17.55 0.27 22.10 0.70 19.34 0.22 19.67‡ 0.03 17.17 0.23 125 17.27 0.16 21.02 0.31 19.38‡ 0.06 16.11 0.43 150 16.73 0.09 20.86 0.29 19.25 0.26 19.32 0.16 15.26‡ 0.30 175 16.48 0.11 19.84 0.32 19.19 0.14 16.49 0.06 200 16.22 0.05 19.62 0.07 19.17 0.04 19.29 0.01 16.33 0.14 225 16.18 0.03 19.45 0.03 19.22 0.11 16.26 0.41 250 15.90‡ 0.20 19.42‡ 0.53 18.99 0.05 19.35 0.05 16.42 0.27 275 15.83 0.08 18.68 0.11 18.87 0.18 16.95 1.26 300 15.67 0.01 18.56 0.20 19.12 0.05 16.49 0.30

129

Figure B.1 Standard curve involving amplification of 10-fold dilution of plasmid DNA specific to (a) Nolandella sp. (Nol), (b) Pseudoparamoeba sp. (Pse), (c) Paramoeba eilhardi (ParE), (d) Vannellida C2 (VanC2), (e) Vannellida C3 (VanC3) and (f) Neoparamoeba perurans (NPJ). E is amplification efficiency.

130

Table B.3 Limit of detection (LOD) of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). The Ct LOD is in bold, which was then converted to its equivalent number of 18S rRNA copies. und is undetermined. Nol Pse Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD 2-1 36.58 36.14 36.53 36.23 36.37 0.22 2-1 38.28 38.09 38.07 38.00 38.11 0.12 2-2 Und 37.50 37.93 und 37.72 0.30 2-2 40.51 und und 39.99 40.25 0.37 LOD = 13.50 18S rRNA copies/µl LOD = 1.35 18S rRNA copies/µl ParE VanC2 Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD 2-1 33.60 34.06 33.38 33.63 33.67 0.29 2-1 35.65 36.13 34.88 35.98 35.66 0.56 2-2 34.29 34.88 34.99 35.10 34.81 0.36 2-2 36.57 37.75 36.77 37.75 37.21 0.63 2-3 36.18 37.28 36.71 35.94 36.53 0.60 2-3 36.70 und 37.29 und 36.99 0.41 2-4 42.63 und 44.82 und 43.73 1.55

LOD = 7871 18S rRNA copies/µl LOD = 39.25 18S rRNA copies/µl VanC3 NPJ Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD Dilution Ct 1 Ct 2 Ct 3 Ct 4 Mean SD 2-1 36.22 36.35 36.94 36.93 36.61 0.38 2-1 34.41 34.77 34.90 35.25 34.83 0.35 2-2 38.02 36.77 38.20 37.93 37.73 0.65 2-2 36.25 36.53 35.28 36.44 36.12 0.58 2-3 36.82 37.21 40.41 und 38.15 1.97 2-3 36.88 37.05 36.95 38.35 37.31 0.70 2-4 39.45 38.97 38.37 37.95 38.69 0.66

2-5 37.34 38.51 38.67 40.41 38.73 1.27

2-6 39.79 40.19 und und 39.99 0.28

LOD = 1.09 18S rRNA copies/µl LOD = 0.66 18S rRNA copies/µl

131

39.48 39.42 ±0.54 ±0.58

37.79 ±0.62

36.45 ±0.57 35.96 35.98 ±0.62 ±0.57

Figure B.2 Precision of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). The final dilution, which determined the LOD of each assay, was tested a further 20 times to determine the precision of the assay at a 95% confidence level. The mean ± standard deviation (SD) is listed above each plot. VanC2 mean±SD was calculated with the outlier excluded.

132

Table B.4 Reproducibility of the qPCR assays specific to Nolandella sp. (Nol), Pseudoparamoeba sp. (Pse), Paramoeba eilhardi (ParE), Vannellida C2 (VanC2), Vannellida C3 (VanC3) and Neoparamoeba perurans (NPJ). Ct values of ten positive gill swabs were tested in triplicate on three separate days and the intra- and inter-assay variance was evaluated via the mean Ct ± standard deviation (SD) and the coefficient of variation (CV). The p-value value from a one-way ANOVA shows there was no significant difference in Ct between the assays conducted on different days. Nol qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 33.54 ± 0.43 1.28 33.18 ± 0.31 0.95 33.24 ± 0.11 0.34 33.32 ± 0.32 0.96 0.90 2 28.14 ± 0.22 0.77 27.69 ± 0.07 0.27 27.87 ± 0.19 0.67 27.90 ± 0.25 0.88 3 30.48 ± 0.30 0.99 30.07 ± 0.09 0.31 30.32 ± 0.32 1.05 30.29 ± 0.29 0.94 4 34.92 ± 0.23 0.67 33.67 ± 0.07 0.21 33.65 ± 0.83 2.47 34.08 ± 0.76 2.24 5 30.66 ± 0.09 0.31 30.27 ± 0.28 0.91 30.71 ± 0.06 0.19 30.55 ± 0.26 0.84 6 32.46 ± 0.41 1.26 31.96 ± 0.55 1.73 32.33 ± 0.09 0.28 32.24 ± 0.44 1.36 7 32.51 ± 0.24 0.73 32.23 ± 0.15 0.48 32.51 ± 0.25 0.76 32.42 ± 0.23 0.72 8 31.30 ± 0.22 0.70 31.26 ± 0.32 1.02 31.60 ± 0.29 0.92 31.39 ± 0.29 0.92 9 29.33 ± 0.23 0.80 29.02 ± 0.11 0.39 29.18 ± 0.06 0.22 29.18 ± 0.19 0.65 10 33.25 ± 0.29 0.87 33.24 ± 0.34 1.02 33.42 ± 0.55 1.66 33.31 ± 0.37 1.10 Pse qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV% Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 35.13 ± 0.33 0.95 34.61 ± 0.20 0.20 34.52 ± 0.58 1.69 34.75 ± 0.45 1.30 0.99 2 29.46 ± 0.54 1.83 29.39 ± 0.66 0.66 29.39 ± 0.73 2.47 29.41 ± 0.56 1.91 3 25.68 ± 0.46 1.80 24.90 ± 0.17 0.17 25.22 ± 0.76 3.02 25.27 ± 0.57 2.25 4 29.17 ± 0.46 1.59 28.75 ± 0.72 0.72 28.81 ± 0.55 1.89 28.91 ± 0.54 1.88 5 35.42 ± 0.35 0.99 35.86 ± 0.25 0.25 35.31 ± 0.17 0.49 35.53 ± 0.34 0.96 6 33.13 ± 0.14 0.43 33.03 ± 0.08 0.08 33.21 ± 0.14 0.43 33.12 ± 0.14 0.41 7 35.02 ± 0.31 0.87 35.24 ± 0.31 0.31 35.23 ± 0.57 1.62 35.16 ± 0.38 1.07 8 35.26 ± 0.58 1.63 35.25 ± 0.38 0.38 35.03 ± 0.23 0.65 35.18 ± 0.38 1.08 9 29.18 ± 0.34 1.18 29.11 ± 0.44 0.44 29.01 ± 0.26 0.89 29.10 ± 0.32 1.09 10 28.74 ± 0.59 2.05 28.42 ± 0.58 0.58 28.83 ± 0.10 0.34 28.66 ± 0.46 1.59 ParE qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV % Mean Ct ± SD CV% Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 37.42 ± 0.29 0.77 37.64 ± 0.75 2.00 38.31 ± 1.16 3.02 37.79 ± 0.81 2.15 0.96 2 36.54 ± 0.34 0.92 36.60 ± 1.75 4.78 36.96 ± 0.41 1.12 36.70 ±0.94 2.55 3 37.20 ± 0.80 2.16 37.92 ± 0.59 1.56 37.76 ±0.78 2.06 37.63 ±0.71 1.89 4 38.80 ± 0.19 0.49 37.89 ± 1.36 3.59 37.32 ± 1.69 4.53 38.00 ± 1.27 3.34 5 38.81 ± 0.63 1.64 37.93 ± 0.38 1.01 37.96 ± 0.77 2.02 38.23 ± 0.69 1.79 6 38.24 ± 0.64 1.67 39.21 ± 1.36 3.47 39.13 ± 0.65 1.67 38.86 ± 0.94 2.42 7 40.25 ± 1.24 3.07 39.13 ± 1.06 2.70 39.47 ± 1.51 3.83 39.62 ± 1.22 3.07

133

8 37.97 ± 1.18 3.09 37.46 ± 0.93 2.49 38.37 ± 1.76 4.60 37.93 ± 1.22 3.22 9 35.07 ± 0.20 0.58 35.04 ± 0.14 0.39 34.91 ± 0.05 0.14 35.01 ± 0.14 0.41 10 35.70 ± 0.37 1.03 35.49 ± 0.33 0.92 35.81 ± 0.50 1.40 35.67 ± 0.38 1.06 VanC2 qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 35.26 ± 0.76 2.15 35.07 ± 0.70 2.00 35.15 ± 0.25 0.71 35.16 ± 0.54 1.53 0.96 2 34.51 ± 0.31 0.88 35.22 ± 0.47 1.34 35.54 ± 0.62 1.75 35.09 ± 0.62 1.76 3 32.22 ± 0.10 0.30 32.36 ± 0.31 0.97 32.53 ± 0.11 0.32 32.37 ± 0.22 0.67 4 34.44 ± 0.21 0.61 34.32 ± 0.77 2.26 34.87 ± 0.58 1.67 34.54 ± 0.56 1.61 5 30.67 ± 0.40 1.32 31.29 ± 0.21 0.67 31.09 ± 0.34 1.08 31.01 ± 0.40 1.28 6 31.02 ± 0.10 0.32 30.69 ± 0.12 0.38 30.81 ± 0.71 2.30 30.84 ± 0.39 1.26 7 30.35 ± 0.43 1.42 30.73 ± 0.20 0.63 31.48 ± 0.11 0.36 30.85 ± 0.56 1.80 8 35.36 ± 0.18 0.52 35.05 ± 0.19 0.53 34.77 ± 0.15 0.43 35.06 ± 0.30 0.85 9 31.86 ± 0.15 0.47 31.88 ± 0.05 0.17 32.24 ± 0.17 0.53 31.00 ± 0.22 0.69 10 35.45 ± 0.49 1.37 36.10 ± 0.64 1.77 35.34 ± 0.39 1.09 35.63 ± 0.57 1.60 VanC3 qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 36.29 ± 0.21 0.56 37.35 ± 0.66 1.76 37.49 ± 0.36 0.97 37.04 ± 0.69 1.85 0.99 2 37.92 ± 0.49 1.30 38.52 ± 0.04 0.09 38.17 ± 0.41 1.08 38.20 ± 0.42 1.09 3 38.19 ± 0.69 1.80 38.11 ± 0.95 2.49 39.19 ± 0.76 1.94 38.50 ± 0.87 2.26 4 44.24 ± 0.79 1.78 44.31 ± 0.61 1.37 44.01 ± 0.98 2.22 44.19 ± 0.71 1.61 5 34.32 ± 1.29 3.76 34.56 ± 0.56 1.61 34.03 ± 1.14 3.34 34.30 ± 0.93 2.72 6 36.66 ± 0.05 0.13 36.72 ± 1.11 3.03 36.88 ± 0.27 0.74 36.75 ± 0.58 1.59 7 40.21 ± 0.23 0.58 41.28 ± 0.21 0.50 41.06 ± 0.84 2.04 40.85 ± 0.66 1.62 8 36.63 ± 0.21 0.57 36.06 ± 0.20 0.54 35.82 ± 0.20 0.56 36.17 ± 0.40 1.11 9 37.24 ± 1.07 2.88 37.22 ± 0.52 1.41 37.60 ± 0.64 1.70 37.35 ± 0.70 1.88 10 43.54 ± 0.13 0.30 42.41 ± 0.41 0.96 43.24 ± 0.60 1.38 43.06 ± 0.63 1.45 NPJ qPCR assay 1 qPCR assay 2 qPCR assay 3 Inter-assay variance Sample Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % Mean Ct ± SD CV % P-value 1 33.45 ± 0.54 1.62 33.19 ± 0.44 1.33 33.78 ± 0.10 0.30 33.47 ± 0.44 1.30 0.72 2 33.82 ± 0.29 0.86 33.44 ± 0.23 0.68 33.96 ± 0.11 0.32 33.74 ± 0.30 0.90 3 33.45 ± 0.08 0.23 33.04 ± 0.14 0.43 33.28 ± 0.21 0.63 33.26 ± 0.22 0.67 4 30.59 ± 0.08 0.25 30.20 ± 0.13 0.44 30.79 ± 0.08 0.26 30.53 ± 0.27 0.90 5 32.11 ± 0.12 0.37 31.49 ± 0.07 0.23 31.97 ± 0.17 0.54 31.86 ± 0.30 0.95 6 30.78 ± 0.16 0.51 30.26 ± 0.19 0.61 30.76 ± 0.13 0.42 30.60 ± 0.29 0.95 7 33.62 ± 0.28 0.84 33.12 ± 0.17 0.52 33.96 ± 0.29 0.84 33.57 ± 0.43 1.28 8 31.73 ± 0.10 0.31 31.39 ± 0.23 0.72 31.59 ± 0.08 0.25 31.57 ± 0.20 0.62 9 30.40 ± 0.04 0.14 30.04 ± 0.08 0.25 30.44 ± 0.11 0.36 30.30 ± 0.20 0.67 10 32.05 ± 0.03 0.08 31.94 ± 0.28 0.89 32.07 ± 0.20 0.62 32.02 ± 0.18 0.57

134

20 Gill swab

1 2 Ct value Ct 10 3 4

5 5

0 undiluted 2-12 -1 2-22 -2 22-3 -3 2-42 -4 22-5 -5 22-6 -6

Gill swab DNA dilution series

Figure B.3 Test for the presence of PCR inhibitors using the Neoparamoeba perurans (NPJ) assay. Five gill swab DNA samples were diluted two-fold then spiked with the same amount of N. perurans plasmid, then tested in triplicate. The PCRs with the undiluted sample contained 150 ng of gill swab DNA within a 5 µl reaction. If PCR inhibitors were present the Ct value would be higher in lower gill swab DNA dilutions.

135

Appendix C: In situ hybridisation of N. perurans

In situ hybridisation (ISH) of Neoparamoeba perurans attached to Atlantic salmon gills was attempted with the long-term plan of running this existing assay (Young et al., 2007) and newly designed assays specific to Nolandella sp. (strain MX5) and Pseudoparamoeba sp. (strain MX1) against the gill samples generated from the in vivo trial detailed in Chapter 5. ISH would determine whether each amoeba species was associated with lesions to provide additional evidence for disease causation. These assays could also be applied to Atlantic salmon gills sampled from Tasmanian farms to investigate how these species interact with each other and with hosts that have been naturally infected rather than experimentally challenged.

Two species-specific oligonucleotide probes that hybridise to 18S rRNA of N. perurans were trialled. The first was the probe designed by Young et al. (2007) and the second was an adaptation of the 18S rRNA N. perurans qPCR amplicon from Downes et al. (2015). The aim was to see if I could successfully perform ISH and to explore the option of adapting the qPCR assays designed in Chapter 4 as ISH assays.

Material and Methods

Preparation of DIG-labelled probe

The N. perurans probe (22 bp long) designed by Young et al. (2007) was ordered from Thermo Fisher Scientific. The second probe, specific to the N. perurans qPCR amplicon selected by Downes et al. (2015) (70 bp long), was prepared with a PCR DIG Probe Synthesis Kit (Roche) following the manufacturer’s instructions while annealing at 56°C. The product of two PCR reactions, one with and one without the DIG Probe Synthesis Mix, was visualised on a 1.2 % agarose gel to confirm the probe was successfully labelled.

Preparation of sections

Severely AGD-affected gills (gill score four) generated during an in-house in vivo trial (unrelated to this project) were fixed in Davidson fixative, dehydrated and set in paraffin wax using

136

Leica Tissue Processor TP1020 (Germany) within a week of fixation. Sections (4 µm) were cut with a Leica RM2235 microtome (Germany) and set onto Flex IHC microscope slides (Dako) at 65°C for 30 min. Slides were dewaxed in xylene and ethanol then rehydrated in distilled water.

Hybridisation procedure

The hybridisation procedure followed the protocol outlined in Young et al. (2007), except that sections were not acetylated in 0.1 M triethanolamine. Post-hybridisation, the sections were stained with Bismarck brown and coverslips were mounted with Faramount Aqueous Mounting Medium (Agilent). Images were taken with a Leica DM 2000 LED microscope with a Leica DFC 495 camera (Germany).

Results

Both assays successfully hybridised to N. perurans and were of equal efficacy. Some non- specific staining did occur in both assays, however it was always in the gill cartilage making it easily distinguishable from amoeba-specific staining.

137

Figure C.1 Oligonucleotide probes that hybridise to 18S rRNA of N. perurans attached to AGD- affected Atlantic salmon gills. (a, b) Results from the Young et al. (2007) ISH assay. (c, d) Results from the adapted Downes et al. (2015) ISH assay. Scale bars; a, c = 100 µm, b, d = 50 µm.

The successful application of both an existing and a modified N. perurans ISH assay on AGD-affected tissue suggests it would be feasible to try adapt the qPCR assay designed in Chapter 4 as ISH assays. However, specificity of these new ISH assays and the Young et al. (2007) assay would need to be tested to ensure there is no cross-reactivity between species before they are applied in future research projects. The two assays trialled here are comparatively short probes, 22 bp and 70 bp, and are designed around 18S rRNA which is a highly conserved sequence region of amoebae (Nassonova et al., 2010; Young et al., 2014). Generally, ISH probes are greater than 150 bp in length and longer probes can be considered more specific if they are designed to hybridize a longer, more variable region (Eisel et al., 2008). In saying that, there are other published examples of ISH probes ranging from 19-22 bp that target SSU rRNA sequences of amoebae (Naegleria sp. and Cochliopodium sp.) colonising rainbow trout gills (Dyková et al., 2010). However, non-specific hybridisation of these probes was only tested against one non-target species. Fortunately, specificity 138 of the new ISH assays described here can be easily tested against the monocultures isolated during this project (Chapter 3). If any assay hybridised with a non-target species, assay parameters such as stringency washes could be optimised or new assays targeting a more variable sequences may need to be designed.

139

Appendix D: Next generation sequencing attempt

Next generation sequencing (NGS) of 18S rRNA amplicons from the gills of AGD-affected farmed Atlantic salmon was attempted with the aim of profiling the amoeba community on the gills and potentially detecting novel amoeba species.

Material and Methods

Two universal 18S rRNA primer sets, RibB and Ami (Table D.1), that amplified multiple amoeba species in Chapter 3, were used to amplify 30 gill swab samples collected from farmed Tasmanian Atlantic salmon between July 2015 and April 2016 for deep sequencing on the Illumina platform. The gill swabs were chosen to represent the range of gill scores quantified according to the Taylor et al. (2009) method, including zero (no gross lesions) to five (lesions covering > 50 % of gill surface). Successful amplification of the gill samples was confirmed by visualising in a 1.2 % agarose gel. For each primer set, 14 of the best amplified DNA samples were purified with a QIAGEN PCR purification kit according to the manufacturer’s instructions. Purified DNA was quantified with a Nanodrop ND-1000 Spectrophotometer (Life Technologies, USA), then 150 ng of each DNA sample was again visualised on a 1.2 % agarose gel to confirm its quality. The 14 amplified DNA samples were pooled and subjected to the TruSeq® Nano DNA Library Prep construction (Illumina, SanDiego, CA, USA) for 100 bp paired end sequencing using the Illumina 2500 platform AGRF (Australian Genome Research facility, Melbourne, Australia). Kelpie (https://peerj.com/articles/6174/) was used to generate full-length amplicon sequences based on the 18S primer sets (Table D.1) from the Illumina generated metagenomic data. The taxonomic profile of the dataset was generated by clustering the amplicons and then comparison to the SILVA SSU/LSU reference database (Quast et al., 2013) to determine the species identification.

Table D.1 Universal Eukaryotic 18S rRNA primer sets used for NGS. Primer Sequence (5’-3’) Annealing Product Reference Temp (°C) size RibB TGATCCATCTGCAGGTTCACCTAC 50 800 - 900 Smirnov et S12.2 GATYAGATACCGTCG TAGTC al., 2007 Ami6F1 CCAGCTCCAATAGCGTATATT 60 700 - 900 Thomas et Ami9R GTTGAGTCGAATTAAGCCGC al., 2006

140

Results

The NGS of gill swabs recovered only Neoparamoeba perurans and Atlantic salmon sequences from both primer sets. The percentage of sequences recovered was 3 % N. perurans and 97 % Atlantic salmon for the Ami primers and 100 % Atlantic salmon for the RibB primers. No alternative amoeba species could be confidently identified from this data.

Discussion

NGS was hypothesised to provide an alternative approach to the methods used in Chapter 3 and 4 for profiling the diversity of amoeba on farmed Atlantic salmon gills. NGS is generally considered suitable for sequencing trace amounts of DNA in heterogeneous samples and therefore could recover greater species diversity compared to the traditional Sanger sequence approach (Edgcomb et al., 2011). The method was also considered less bias, documenting the frequency of all amoebae, rather than only species that can grow in vitro. However, our first attempt at this method did not produce the hypothesised results. Instead only N. perurans and Atlantic salmon DNA was recovered, which does not reflect the diversity of amoeba documented in Chapter 3.

One reason for this null result could be the highly conserved nature of the 18S rRNA region across amoeba species (Nassonova et al., 2010; Young et al., 2014). High sequence similarity made the data difficult to assemble and determine novel species. It is possible some of the many N. perurans reads may actually be different species, but there is no way of confirming this. If this method is repeated, we suggest sequencing a more variable barcoding region, such as COI (Nassonova et al., 2010)

Other methodological improvements include avoiding pooling samples because if the non- N. perurans amoebae are only found in a very small proportion of salmon tissue, then pooling will dilute the rare DNA. The depth of sequencing could also be reduced, as it was not required for this type of metagenomic analysis. In this case, more individual fish could have been analysed rather than sequencing at such depth. If high proportions of N. perurans continued to impede the detection of less dominant amoeba taxa, one option could be to block the amplification of N. perurans DNA during PCR, similar to what was achieved in Gofton et al. (2015) where a blocking primer decreased unwanted sequences by 96 % and significantly increased the total detectable bacterial diversity in a medically important species of tick.

141

Appendix E: Salinity tolerance of Vannella sp.

Figure E.1 Morphology of Vannella sp. trophozoites after five days in malt yeast broth with salinity ranging from 0 - 42 parts per thousand (ppt)

142

Appendix F: Ethics approval

Approval of Animal Ethics Research Activity – A13/2015

143

Approval of Animal Ethics Research Activity – A9/2016

144

Approval of Animal Ethics Research Activity – A17/2017

145