THE RISK TO HUMAN HEALTH FROM FREE-LIVING AMOEBAE INTERACTION WITH IN DRINKING AND RECYCLED WATER SYSTEMS

Dissertation submitted by  JACQUELINE MARIE THOMAS BACHELOR OF SCIENCE (HONOURS) AND BACHELOR OF ARTS, UNSW

In partial fulfillment of the requirements for the award of

DOCTOR OF PHILOSOPHY in ENVIRONMENTAL ENGINEERING

SCHOOL OF CIVIL AND ENVIRONMENTAL ENGINEERING FACULTY OF ENGINEERING

MAY 2012

SUPERVISORS

Professor Nicholas Ashbolt Office of Research and Development United States Environmental Protection Agency Cincinnati, Ohio USA and School of Civil and Environmental Engineering Faculty of Engineering The University of New South Wales Sydney, Australia

Professor Richard Stuetz School of Civil and Environmental Engineering Faculty of Engineering The University of New South Wales Sydney, Australia

Doctor Torsten Thomas School of Biotechnology and Biomolecular Sciences Faculty of Science The University of New South Wales Sydney, Australia

  ORIGINALITY STATEMENT

'1 hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom 1 have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.'

Signed ~ ......

2>/M M fJ-D( Date······················0··········~··· ..···

iii COPYRIGHT STATEMENT

'I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

1 also authorise University Microfilms to use the 350 word abstract of my thesis III Dissertation Abstract International. I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ~ .

Date ....•...... 3.e.M~..m.(Cb-....

AUTHENTICITY STATEMENT

'I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.'

Signed ~ .

Date ;; 1.~.t1.O'cJ(L..O!b.

iv ABSTRACT

Treated drinking water in developed countries still causes disease despite meeting current regulations and compliance monitoring needs that focus on faecal contamination. Of growing concern are the increasing rates of community acquired (CAP) infections which are in part due to inhalation of treated drinking water aerosols containing opportunistic pathogens including Legionella and Mycobacterium. CAP infections due to Legionella presently are estimated to account for one third of all reported drinking water outbreaks in the USA. If the number of Legionella infections are to be reduced then the means of replication in drinking water need to be better understood as it is apparent that current drinking water disinfection and control mechanism are ineffective. Free-living amoebae (FLA) are eukaryotic microorganisms which are known to feed on but some microorganisms, such as pathogenic Legionella, effectively avoid digestion and infect the FLA. Microorganisms which can infect FLA are collectively known as amoebae resistant microorganisms (ARM). As hosts FLA effectively facilitate the survival, growth and infectivity of their pathogenic ARM. Despite FLA being an accepted host for pathogenic Legionella and other ARM, very little is known about their density and diversity nor their interactions with Legionella in drinking and recycled water systems and end use applications. The aim of this research was to address the present knowledge gap about FLA by analysing FLA and their interactions with Legionella in a drinking and recycled water systems and applications and then use that information in a novel quantitative microbial risk assessment for FLA and Legionella. Through literature analysis of 29 published studies reporting on FLA in drinking water systems and applications it appears that FLA are ubiquitous. FLA were detected in drinking water tap samples at a mean frequency of 46 % (n = 18,  = 28). Overall, the density of FLA was poorly reported but there was a large diversity of FLA with 14 different genera of FLA identified. Already, six different genera of FLA have been isolated from drinking water systems and applications infected with 14 different of pathogenic ARM, thus highlighting the variety of FLA and ARM interaction occurring in drinking water. From water and samples FLA were isolated using traditional culture methods utilising non-nutrient agar plates with overlays. More sensitive molecular techniques were also applied with qPCR targeting Acanthamoeba spp., H. vermiformis and Naegleria spp. Detected FLA were identified by cloning and partial 18S rRNA gene sequencing. Legionella spp. were also isolated using standard culture methods and detected in samples and FLA using PCR and qPCR targeting the Legionella genus. Identification of detected Legionella

  was achieved by cloning and partial 16S rRNA gene sequencing. Water quality characteristics including biofilm quantity, heterotrophic plate counts and free and combined chlorine were also recorded during sampling and the course of the experiments to ascertain possible trends. To partly address the knowledge gap about FLA in recycled water systems a water recycling plant (WRP) and its recycled water distribution system were sampled for FLA and Legionella. By qPCR H. vermiformis was detected breaking through the WRP at densities of 2.7 amoebae.mL-1 and entering the distribution system. The water and biofilm for both the parallel recycled and drinking systems within the dual distribution system was sampled using a modified Robbins Device (MRD). Acanthamoeba spp. and H. vermiformis were detected by culture and qPCR in drinking and recycled water and biofilm samples with the highest mean densities (4.6 amoebae.cm-2 ) detected in drinking water biofilm. Pathogenic Legionella anisa were detected in low densities in one recycled water sample (40 cells.mL-1) and biofilm samples (14 cells.cm-2) but no Legionella was detected in the drinking water samples. Two of the 16 FLA isolated from drinking and recycled water were shown to be infected with Legionella sp. illuminating how common FLA infections with ARM were. Garden hoses were identified as an exposure pathway for Legionella infection that had been neglected in the literature to date. To evaluate FLA and Legionella density and diversity two garden hoses types (standard green and lilac) supplied with drinking water were sampled over 18 months. Over the three sampling periods the lilac hose type water consistently had higher mean density of Acanthamoeba sp. (324 amoebae.mL-1) and H. vermiformis (300 amoebae.mL-1) compared to the green hose water Acanthamoeba sp. (18 amoebae.mL-1) and H. vermiformis (31 amoebae.mL-1). Similarly, Legionella detection was significantly higher (1 way ANOVA, p < 0.0001) for the lilac hose type water (6.5  103 cells.mL-1) which were higher than all the other Legionella densities reported in the literature. Two thirds of the Legionella detected were identified as unclassified Legionella species with unknown pathogenicity that needs to be explored further. To replicate the conditions in heated applications of drinking water, such as showering, annular reactors were incubated at 42 °C for 13.5 months. Despite having comparable biofilm quantities to the ambient reactors there were consistently less FLA detected in the heated reactor biofilm by both culture and qPCR. A greater diversity of FLA were identified in the reactors including Echinamoeba exudans and Vahlkampfia sp. compared to the MRD drinking water which supplied the reactors. The highest mean detection of Legionella was for heated reactor water (80 cells.mL-1) while heated reactor biofilm had the lowest mean detection (13 cells.mL-1). During heated conditions biofilm appears to be less important in the proliferation of FLA and Legionella than under ambient conditions which was incorporated into the risk assessment.

  Acanthamoeba sp. and H. vermiformis isolates from the MRD, garden hoses and annular reactors were incubated with fluorescently labeled L. pneumophila at 22 and 37 °C and tracked by fluorescent microscopy over seven days to determine if the FLA would be infected. After seven days at 22 °C all of the Acanthamoeba sp. trophozoites were infected with L. pneumophila but only a maximum mean of 25 % of the H. vermiformis were. Accordingly, the numbers of L. pneumophila increased in the presence of Acanthamoeba sp. by 4.4 -   compared to only 58 - 108 times when incubated with H. vermiformis. The FLA isolated from the drinking water systems and applications were infected with L. pneumophila at different rates which was used in the risk assessment calculations. FLA density and diversities in water and biofilm were incorporated to form a quantitative microbial risk assessment (QMRA) for Legionella infection. The QMRA was unique in that the input data for FLA and Legionella came from real drinking water end use applications, showering and garden hose use. When the model was run with estimated mean densities of pathogenic L. pneumophila and FLA the probability of human infection was greater than the recommended risk benchmark of 1  10-4 for both garden hose use and showering. Hence, based on these novel probabilities of infection it is recommended that those people who are at risk of Legionella infection, such as the elderly, diabetics and people with suppressed immune systems, should make behavioral changes to reduce exposure. Innovative changes recommended include draining and hanging garden hoses and avoiding using them on spray settings. Furthermore, it is advised to take baths over showers or if that it not possible to use a point of use filter system. It must be remembered that FLA and Legionella are only two groups of microorganism capable of interaction in a drinking water system. There is a range of hosts and pathogenic ARM in drinking water whose density, diversity and interactions need to be further explored. New next generation sequencing platforms and fluorescent assisted cell sorting technologies present excellent opportunities to make advancement in this area of research. Combined with improved epidemiological surveying then the true health burden of pathogenic ARM in drinking water can be established and in turn controlled.

  PUBLICATIONS AND PRESENTATIONS

JOURNAL PUBLICATIONS  Thomas, J. M. and N. J. Ashbolt. 2011. Do free-living amoebae in treated drinking water systems present an emerging health risk? Environmental Science & Technology. 45(3): 860 - 869.  Thomas, J.M., T. Thomas, R. Stuetz and N. J. Ashbolt. (in preparation). Free-living amoebae diversity and density and interaction with Legionella in a dual (drinking and recycled) water distribution system. Submission to Water Research.  Thomas, J.M., T. Thomas, R. Stuetz and N. J. Ashbolt. (in preparation). Garden hose use presents an increased risk of exposure to free-living amoebae facilitated Legionella. Submission to Applied and Environmental .

Thomas, J.M., S. Schoen and N. J. Ashbolt. (in preparation). Quantifying the risks of infection after exposure during showering and garden hose use to Legionella facilitated by free- living amoebae. Submission to Journal of Water and Health.

INTERNATIONAL CONFERENCE PRESENTATIONS

Thomas, J.M., R. Stuetz, T. Thomas and N. J. Ashbolt. 2011. The risk of free-living amoebae interaction with bacterial pathogens in drinking water. International Water Association 16th Health Related Water Microbiology Symposium. Platform presentation. Rotorua, New Zealand. 18 - 23 September.

Thomas, J.M., M. Storey, R. Stuetz, T. Thomas and N. J. Ashbolt. 2010. Diversity of free- living amoebal pathogen hosts in a dual distribution (drinking and recycled) water system. American Water Works Association Water Quality and Technology Conference. Platform presentation. Savannah, USA. 14 -18 November.

Thomas, J.M., M. Storey, R. Stuetz, S. Kjelleberg and N. J. Ashbolt. 2009. Diversity of free- living amoebae in a dual distribution (potable and recycled) water system. International Water Association; 15th Health Related Water Microbiology Symposium. Poster presentation. Naxos, Greece. 31 May - 5 June.

  LIST OF CHAPTERS

1. INTRODUCTION AND AIMS 1

2. LITERATURE REVIEW 13

3. METHODOLOGY 63

4. FLA AND LEGIONELLA IN A RECYCLED WATER SCHEME 103

5. FLA AND LEGIONELLA IN GARDEN HOSES 139

6. FLA AND LEGIONELLA IN HEATED ANNULAR REACTORS 163

7. FLA INFECTION WITH 193

8. HUMAN HEALTH RISK ASSESSMENT 226

9. CONCLUSION 249

  ACRONYMS AND ABBREVIATIONS

ARB amoebae resistant bacteria ARM amoebae resistant microorganisms ATCC American Type Culture Collection™ AWQC Australian Water Quality Centre BABS School of Biotechnology and Biomedical Sciences BCYE buffer charcoal yeast extract BDNF brain-derived neurotrophic factor BLAST basic local alignment search tool BYE buffer yeast extract CAP community acquired pneumonia CDFA SE carboxyfluorescein diacetate succinimidyl ester CFU colony forming unit COI cytocrome c oxidase subunit I CRCWQT Cooperative Research Centre for Water Quality and Treatment CSLM confocal scanning laser microscope DAPI 4',6-diamidino-2-phenylindole DGGE denaturing gradient gel electrophoresis DIC differential interference contrast DNA deoxyribonucleic acid EDTA ethylenediaminetetraacetic acid EMA ethidium monoazide FACS fluorescence assisted cell sorting FISH fluorescent in situ hybridisation FITC        FLA free-living amoebae GVPC glycine vancomycin hydrochloride polymyxin B sulphate cycloheximide HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid HPC heterotrophic plate counts ITS internal transcribed spacer LLAP Legionella like amoebal pathogens MRD modified Robbins Device N total nitrogen NCBI National Center for Biotechnology Information NNA non-nutrient agar plates NTU Nephelometric turbidity units P total phosphorus PAM primary amoebic meningoencephalitis PBS phosphate buffer saline PCR polymerase chain reaction PFU plaque forming unit PMA propidium monoazide PVC     PYNFH         QMRA quantitative microbial risk assessment qPCR quantitative polymerase chain reaction rDNA ribosomal deoxyribonucleic acid RDP ribosomal database project RNA ribonucleic acid SD standard deviation

  SSU small subunit T-RFLP terminal restriction fragment length polymorphism TBE tris-boric EDTA buffer TOC total organic carbon UK United UNSW The University of New South Wales USA United States of America USEPA United States Environmental Protection Agency UV ultraviolet WRC Water Research Centre WRP water recycling plant 

  ACKNOWLEDGEMENTS  There are a great number of people who have enabled my PhD research and who I wish to acknowledge. Firstly my supervisors; Nicholas Ashbolt who generously took me on as a student and who invited me to work with him in the USA. Richard Stuetz who has helped me with all the administration requirements of PhD research and has been a voice of reason when things went wrong. Torsten Thomas who stepped up late in the stage and took over from Staffan Kjelleberg and gave me much needed technical input on the molecular biology aspects. Then there is the industry connection with Michael Storey, Peter Cox and Greg Kennedy facilitating access to sampling sites and data. In the Civil and Environmental Engineering PC2 lab the assistance from fellow microbiologist Christine Kaucner and Marcus Klein was much appreciated. So too were the casual chats with the other lab users which kept me sane: Leearna Brown, Marion Minouflet, Anna Leung and Ben Van den Akker. Thanks to Gautam Chattopadhyay and Kelvin Ong for their willingness to assist with ordering and equipment problems. There are a number of administration staff within the School who have helped with just about everything: Patricia McLaughlin, Les Brown, Angela Spano, Patricia Karwan and Robert Steel. Thanks should also go to Bret Robinson from the Australian Water Quality Centre who inspired and taught me with his passion for amoebae. Also the staff at the Ramaciotti Sequencing Centre and Biomedical Medical Imaging Facility for their work sequencing samples or helping use the microscopes. To my fellow PhD students, past and present, thank you for your friendship and support: Fiona Johnson, Nhat Le Minh, Eli Lai, Hazel Rowley, Rebecca Barnes, Adam Hambly, Gavin Parsci, Farj Elhadayri and all the other geotechnical PhD students in Rm 510. Then there is a whole other group of people in the USA. To the members of Nick Ashbolt and Jorge Santo Domingo US Environmental Protection Agency groups who welcomed me and helped answer my endless stream of questions: Brandon Iker, Mary Schoen, Jingrang Lu, Helen Buse, Hodon Ryu and Ian Struewing. Also the other people at the US EPA who I either worked with or who let me use their lab's equipment: Maura Donohue, Stacey Fowler, Randy Revetta, Claudine Curioso, Michael Ware and Scott Keely. Plus all the friends that I made in Cincinnati; Carlos Castillo, Samir Aziz, Rachelle, Tarsha Eason, Kaneatra Simmons, Molly Cropenbaker and Reagan Murray. PhD research was a long and lonely undertaking for me and my friends deserve a special mention for distracting me from my research when I needed it and listening to me when I was struggling: Tori Bolton, Laura Cook, Beth Suthers, Sarah Gaskin, Brendan Jacka, Anthony Nicholas and Christine Toxvaerd. My family is a huge part of my life and I'm so fortunate to have such a diverse one. Mum, Aidan, Roni, Elyssa, and Mimi thank you for being a loving family here in Sydney and Mum   special thanks for your editing work on my thesis. In Tasmania to Dad, Stuart and Abigail hearing how the horses were racing and visiting you was always welcome break. Then there are Michael's parents Anne and Peter who helped look after the girls when Michael couldn't so I could do more research. Finally, (I saved the best till last) to Michael, Jacqueline and Erica I love you all so much and I'm so grateful for all the help that you have given me. Michael your love, patience and understanding made all this so much more achievable.

  CHAPTER 1

INTRODUCTION AND AIMS 1. TABLE OF CONTENTS

1.1 DRINKING WATER AND DISEASE ______2 1.1.1 Community acquired pneumonia (CAP) ______2 1.1.2 Water-based pathogens ______3 1.2 FREE-LIVING AMOEBAE (FLA) INFECTED WITH AMOEBAE RESISTANT MICROORGANIMS (ARM)______5 1.2.1 The human health risk presented by FLA and ARM ______6 1.3 RESEARCH AIMS______7 1.3.1 Thesis structure ______7 1.4 REFERENCES ______9

LIST OF TABLES

Table 1.1 Selection of recent Legionnaires outbreaks worldwide ______3

LIST OF FIGURES

Figure 1.1 Environmental free-living amoebae (Acanthamoeba sp.) ______6

Chapter 1. Introduction

1.1 DRINKING WATER AND DISEASE Treated drinking water in developed countries still causes disease despite meeting current regulations and compliance monitoring needs that focus on faecal contamination (Hrudey and Hrudey, 2007; Rizak and Hrudey, 2008; Craun et al., 2010). Ensuring treated drinking water quality is of paramount importance as population growth and climate change places increasing demands on existing drinking water resources especially in urban areas (Vairavamoorthy, 2008; De Toni et al., 2009). Additionally, as recycled water is being utilised more extensively to supplement traditional drinking water uses (Asano et al., 2007) more research is needed to be confident of the safety of this essential alternative.

1.1.1 Community acquired pneumonia (CAP) Of growing concern are the increasing rates of community acquired pneumonia (CAP) infections which are in part due to exposure to treated drinking water (Breiman et al., 1990; Stout et al., 1992; Feazel et al., 2009). CAP is one of the top ten causes of death in the United States and hospitalisation rates have increased 20 % in recent decades despite improvements in treatment (Fry et al., 2005). CAP infection rates are higher compared with adults for children under five years (36 infections per 1000) and those aged over 74 years (34.2 infections per 1000) based on European infection frequencies (Woodhead, 2002). Between 8 - 51 % of CAP patients will require hospital treatment and 4 -15 % will die (Woodhead, 2002). CAP can be caused by inhalation of water droplets (aerosols) containing opportunistic water-based bacterial pathogens from the genera Legionella (Fields et al., 2002), Mycobacterium (Field and Cowie, 2006) and a host of emerging pathogens (Lamoth and Greub, 2010). In a single epidemiological study drinking water was identified as the environmental source of the infecting Legionella pneumophila in 40 % of cases (Stout et al., 1992). Similarly, for non-tuberculosis mycobacterium, Mycobacterium lentiflavum, drinking water was identified as the source of 11 % of infections (Marshall et al., 2011). Bathing using a shower has been identified as a significant exposure pathway due to the re-growth of these opportunistic pathogens in showerhead (Breiman et al., 1990; Falkinham et al., 2008; Feazel et al., 2009). Additionally complex hot water systems particularly in hospitals (Lin et al., 1998), spas (Okada et al., 2005) and cooling towers (Barbaree et al., 1986) are a significant source of Legionellosis outbreaks. With cooling towers alone accounting for an estimated 28 % of all Legionella infections in an urban population (Bhopal, 1995). CAP infections due to Legionella presently account for 29 % of all reported drinking water outbreaks in the USA (Craun et al., 2010). Furthermore, Legionella bacteria were the third most common etiological agent for all USA drinking water related outbreaks over the last 36 2 Chapter 1. Introduction years, despite only being added to the surveillance data in the last 6 years (Craun et al., 2010). There are frequent outbreaks of Legionnaires disease worldwide with hundreds of infections and scores of fatalities (Table 1.1). These reported infections represent only the tip of the total infections caused by water-based Legionella and other CAP causing pathogens. Table 1.1 Selection of Legionnaires outbreaks worldwide. Year Country Source of outbreak Cases Reference (deaths) 2008 Australia Carwash 7h (McArthur, 2008) 2002 Japan Bathhouse with spas 295 (7) (Okada et al., 2005) 2001 Spain Hospital cooling 650 (5) (García-Fulgueiras et al., 2003) 2000 Australia Cooling towers 87 (2) (Anonymous, 2000) 1999 Netherlands Whirlpools/sprinkler 188 (Den Boer et al., 2002) 1995 Australia Multiple cooling towers 11 (3) (Heath et al., 1998) 1994 USA Cruise ship whirlpool 50 (Jernigan et al., 1996) h hospitalised cases only with total cases likely to be higher but not reported

In Europe it is estimated that CAP infection caused by water-based pathogens make up over 8 % of the total CAP cases in intensive care units (Woodhead, 2002). Specifically in Australia, it is estimated that 3-5 % of admissions to hospital for CAP are caused by Legionella infection (Broadbent, 1996) but this is likely to be an under-estimate due the lack of systematic testing for the pathogen (Wilson and Ferguson, 2005). Also CAP infections caused by non-tuberculosis Mycobacterium have been on the increase for decades (Field and Cowie, 2006). Water-based pathogens may account for more infections than presently realised due to the fact that the agents of disease are consistently not identified in nearly half of all CAP cases (Fields et al., 2002; Woodhead, 2002; Wilson and Ferguson, 2005; Lamoth and Greub, 2010). In part this is due to limitations of epidemiological studies that are undertaken after an outbreak to indentify the environmental source (Heath et al., 1998). Controlling drinking water-based microbial pathogens capable of causing CAP infections is critical to reducing these rates of infection. This is especially pertinent in developed countries where aging demographics will result in a larger portion of the population being more vulnerable to CAP. Ultimately, as a society we must strive to ensure the safety of treated drinking water quality as population growth and climate change is placing increasing demands on existing drinking water resources especially in urban areas (Vairavamoorthy, 2008; De Toni et al., 2009).

1.1.2 Water-based pathogens Waterborne pathogens, being of faecal origin, in drinking water systems are normally controlled by treating source water via water treatment plants and maintaining a disinfection

3 Chapter 1. Introduction residual throughout the distribution system usually by chlorination. Short term failures in the control system allow faecal origin microbial pathogens to enter the water system, such as Cryptosporidium, pathogenic Escherichia coli and rotavirus, which have resulted in hundreds of thousands of gastroenteritis infections and scores of deaths (Hrudey and Hrudey, 2007). In contrast, CAP causing water-based pathogens, such as Legionella and Mycobacterium, are not of faecal origin and are naturally able to regrow in drinking water systems (Diederen et al., 2007; Feazel et al., 2009; Marciano-Cabral et al., 2010). Legionella is naturally found in fresh-water environments but rarely associated with infections, however the CAP outbreaks due to artificial man-made drinking water systems appear to provide favourable conditions such as elevated temperatures and altered microbial populations that may promote Legionella growth (Fields et al., 2002). Control of Legionella in treated drinking water systems and drinking water applications is attempted by maintaining a chlorine disinfection residual between 0.2 and 0.5 mg.L-1, as recommended by the Australian Drinking Water guidelines (Australian Government, 2004). However, this residual is not entirely effective as it has been shown that there is a direct linear relationship between tap water chlorine residuals below 0.5 mg.L-1 and increased Legionella infections (Kool et al., 1999). An alternative to chlorination used widely in the USA is chloramination; where hypochlorous acid reacts with ammonia to form monochloramine (Droste, 1997)., para Chloramines produce a longer lasting disinfection residual compared to chlorine and have been reported to be more effective than chlorine against CAP causing pathogens such as Legionella (Kool et al., 1999; Thomas et al., 2004). The high rates of water-based Legionella outbreaks in the USA, despite widespread chlorine and chloramine disinfection residuals, is strong evidence that current water treatment methods are not effective for this group of CAP causing pathogens. This presents a significant control problem as increasing disinfection residuals in water systems is generally not a viable control option due to the concerns over carcinogenic health risks of disinfection by-products (U.S. Federal Government, 2006). Further, the concerns about CAP causing pathogens are not exclusively associated with treated drinking water but also recycled water. Recycling waste water directly, for both drinking and non-drinking applications, is a viable solution to the worlds growing water scarcity problems (Asano, 2004; Vairavamoorthy, 2008; De Toni et al., 2009). Recycled water can have increased nutrient levels and higher microbial loads compared to treated surface or ground water most commonly used as drinking water, although this is entirely dependant on the treatment process used to produce the recycled water (Toze, 2006). Recycled water is already used in cooling towers (Levine, 2003), toilet flushing (Lazarova et al., 2003), irrigation, pools, drinking, cooking and washing (Asano, 2004). As recycled water is used as extensively as traditional treated

4 Chapter 1. Introduction drinking water it is essential that there is confidence that these CAP causing microbial pathogens are adequately controlled. Increasing disinfection residuals in drinking and recycled water systems is not a viable control option and it is currently unclear how best to control these opportunistic pathogens and if control mechanisms should be applied to distribution systems or in-premise plumbing or both. The answer lies in understanding more about the ecology of these pathogens and how they are able to grow in treated drinking water.

1.2 FREE-LIVING AMOEBAE (FLA) INFECTED WITH AMOEBAE RESISTANT MICROORGANIMS (ARM) It has been shown that some water-based pathogenic microorganisms can survive and grow inside larger eukaryotic organisms, particularly free-living amoebae (FLA) (Greub and Raoult, 2004). When FLA are in the active trophozoite form they prey on a range of bacteria and other microorganisms (Figure 1.1) (Rodriguez-Zaragoza, 1994; Bozue and Johnson, 1996; Scheid et al., 2008). Once inside the FLA the prey microorganisms are usually digested in acidic vacuoles but some microorganisms are able to survive by avoiding digestion and replicating successfully using a variety of mechanisms (Snelling et al., 2006; Isberg et al., 2009). Microorganisms with this ability are commonly referred to as amoebae-resisting microorganisms (ARM) and when specifically referring to bacteria as -resisting bacteria (ARB) (Greub and Raoult, 2004). FLA are able to facilitate the growth and survival of ARM in drinking water systems through an array of mechanisms. Although the interactions between FLA and ARM are generally species specific (Thom et al., 1992; Cirillo et al., 1997; Tezcan-Merdol et al., 2004). To date a number of water-based pathogenic ARM have been directly detected in FLA isolated from treated drinking water including; Legionella (Thomas et al., 2006; Thomas et al., 2008; Corsaro et al., 2010), Mycobacteria mucogenicum (Thomas et al., 2008) and Pseudomonas spp. (Michel et al., 1995; Hoffmann and Michel, 2001) as well as emerging pathogens from the genera Flavobacterium (Hoffmann and Michel, 2001; Thomas et al., 2008) and Neochlamydia (Thomas et al., 2008; Corsaro et al., 2009; Corsaro et al., 2010). Furthermore, research conducted under laboratory conditions reports additional water-based pathogens capable of being hosted by FLA including: Legionella spp., Mycobacterium spp., Parachlamydia spp., , Burkholderia ccepacia, , Listeria monocytogenes, Toxoplasma gondii; and even various virions, such as Adenovirus and Mimivirus (Greub and Raoult, 2004; Thomas et al., 2010). FLA are thought to be one of the key biological factor that allows water-based opportunistic pathogens to grow despite

5 Chapter 1. Introduction drinking water treatment processes (Fields et al., 2002; Greub and Raoult, 2004; Lau and Ashbolt, 2009; Thomas et al., 2010).

Figure 1.1 Environmental free-living amoebae (Acanthamoeba sp.) in dormant (cyst) and active (trophozoite) life cycle stages. Image taken using phase contrast microscopy at 1000  magnification with a 10 μm scale bar. 1.2.1 The human health risk presented by FLA and ARM FLA can facilitate the growth, survival, virulence and potentially transportation of a range of water-based pathogens. It is for these reasons that it is highly plausible that FLA are a primary mechanism used by human pathogenic water-based ARM to proliferate in treated drinking water systems. Alarmingly, we have very limited data on which FLA act as hosts in treated drinking water, what may control them or indeed the ecological conditions that promote ARM that may be human pathogens. To estimate the human health risks presented by FLA within treated drinking water systems and the increasingly more common recycled urban water systems, three aspects need to be considered. Firstly, FLA can be infected with a range of pathogenic ARM. Secondly, FLA may act as transport hosts within engineered systems effectively facilitating pathogenic ARM infection of humans. Thirdly, some FLA are human pathogens in their own right and cause both clinical and sub-clinical infections. There is presently a lack of knowledge about FLA that needs to be addressed given the recently recognized high health burden caused by just one ARM, Legionella pneumophila (Craun et al., 2010). As the population demographics changes to an older populace, a major research effort is needed to identify and manage the pathogenic ARM in our treated drinking water systems in order to control CAP infections in aged and other more sensitive sub-populations.

6 Chapter 1. Introduction 1.3 RESEARCH AIMS The overall aim of this research was to develop a novel comprehensive quantitative microbial risk assessment (QMRA) for human infections with Legionella from drinking and recycled water systems using biofilm and microbiological data from a full drinking and recycled water system. This aim was achieved through the following objectives: • Objective one - to identify current knowledge gaps about FLA populations in drinking and recycled water systems. by examining a dual drinking and recycled water system in detail. • Objective two - to fill the identified knowledge gaps by examining a dual drinking and recycled water system in detail. The system was examined for FLA and Legionella spp. from the waste water recycling plant (WRP) through to the mains distribution pipes, then to in-premise taps and finally the common applications of drinking water (garden hoses and heated water applications). • Objective three - to link the occurrence of FLA identified to water quality characteristics and engineering system factors as well as the presence and interaction with Legionella spp. as a significant group of ARM. • Objective four - to develop a comprehensive quantitative microbial risk assessment (QMRA) for Legionella infection and make important recommendations in order to reduce the risk of infection with Legionella and develop possible control strategies for FLA in drinking and recycled water systems.

1.3.1 Thesis structure Chapter 2 contains a detailed review of the available literature describing FLA and ARM interactions in drinking water systems. The literature detailling FLA and Legionella spp. density and diversity was also included in the scope. The literature review highlights the current knowledge gaps. The purpose of Chapter 3 was to review the existing methods for the detection of FLA and Legionella spp. in water samples. Both culture and molecular methods were examined and selected or developed. Additionally, methods for measuring water quality characteristics were also selected. In Chapter 4 the FLA and Legionella populations are described in the final stages of a water recycling plant (WRP) over six weeks and the dual distribution system that it supplied in what appears to be the first study of its type for recycled water. The distribution system was sampled once in the mid-section using a pipe biofilm sampler termed a modified Robbins Device (MRD). Water quality characteristics were also collected.

7 Chapter 1. Introduction The diversity of the FLA and Legionella populations present within in-premise taps and common garden hoses are described in Chapter 5. Garden hoses were identified as an application of water where users could be exposed during use to FLA and Legionella via aerosols but had been neglected in the literature. The research was conducted over an 18 month period in order to gain information about population dynamics over time. In Chapter 6 colonisation and growth rates of FLA populations within heated applications of drinking water that were similar to shower hot water systems were explored. The specific aim of this chapter was to use heated and ambient rotating drum reactors to determine how FLA colonised the biofilm and water of these systems and the effects that temperature may have had. The experiments were run over a 13.5 month period to determine the longer-term population trends. These trends were also linked to water quality characteristics. The infection rates of different FLA isolated from the drinking and recycled water system by L. pneumophila are presented in Chapter 7. The aim was to track the infection of the FLA by the L. pneumophila cells to determine the rates of infection and the effect on L. pneumophila numbers at different temperatures. This information was critical to the adapting of the QMRA to the existent water system. All data was consolidated and used to produce a quantitative microbial risk assessment (QMRA) for FLA and Legionella in drinking and recycled water systems (Chapter 8). The outcomes of the QMRA determined the recommendations for risk reduction and level of FLA control needed. In Chapter 9 the research outcomes were summarised and future research needs were identified.

8 Chapter 1. Introduction

1.4 REFERENCES 1. Anonymous. 2000. Outbreak of Legionnaires' disease associated with an aquarium in Australia. Communicable Disease Report Weekly. 10: 161. 2. Asano, T. 2004. Water recycling - A relevant solution? Workshop on Water Crisis - Myth or Reality, Santander, SPAIN, Taylor & Francis Ltd. 3. Asano, T., F. Burton, H. Leverenz, R. Tsuchihashi and G. Tchobanoglous. 2007. Water reuse issues, technologies and applications. New York, McGraw Hill 4. Australian Government. 2004. Australian Drinking Water Quality Guidelines. National Health And Medical Research Council. 5. Barbaree, J. M., B. S. Fields, J. C. Feeley, G. W. Gorman and W. T. Martin. 1986. Isolation of protozoa from water associated with a legionellosis outbreak and demonstration of intracellular multiplication of Legionella pneumophila. Applied and Environmental Microbiology. 51(2): 422-424. 6. Bhopal, R. 1995. Source of infection of sporadic Legionnaires' disease; a review. Journal of Infection. 30(1): 9-12. 7. Bozue, J. and W. Johnson. 1996. Interaction of Legionella pneumophila with Acanthamoeba castellanii: uptake by coiling phagocytosis and inhibition of phagosome-lysosome fusion. Infection and Immunity. 64(2): 668-673. 8. Breiman, R. F., B. Fields, G. N. Sanden, L. Volmer, A. Meier and J. Spika. 1990. Association of shower use with Legionnaires' disease. Possible role of amoebae. Journal of the American Medical Association. 263(21): 2924 - 2926. 9. Broadbent, C. 1996. Guidance for the control of Legionella Adelaide National Environmental Health Forum. 10. Cirillo, J., S. Falkow, L. Tompkins and L. Bermudez. 1997. Interaction of Mycobacterium avium with environmental amoebae enhances virulence. Infection and Immunity. 65(9): 3759-3767. 11. Corsaro, D., V. Feroldi, G. Saucedo, F. Ribas, J.-F. Loret and G. Greub. 2009. Novel Chlamydiales strains isolated from a water treatment plant. Environmental Microbiology. 11(1): 188-200. 12. Corsaro, D., G. S. Pages, V. Catalan, J.-F. Loret and G. Greub. 2010. Biodiversity of amoebae and amoeba-associated bacteria in water treatment plants. International Journal of Hygiene and Environmental Health. 213(3): 158-166. 13. Craun, G. F., J. M. Brunkard, J. S. Yoder, V. A. Roberts, J. Carpenter, T. Wade, R. L. Calderon, J. M. Roberts, M. J. Beach and S. L. Roy. 2010. Causes of outbreaks associated with drinking water in the United States from 1971 to 2006. Clinical Microbiology Reviews. 23(3): 507-528. 14. De Toni, A., A. Touron-Bodilis and F. Wallet. 2009. The impact of climate change on pathogenic aquatic microorganisms: some examples. Environnemental, Risque & Sante. 8(4): 311- 321. 15. Den Boer, J. W., P. F. Yzerman, J. Schellekens, K. D. Lettinga, H. C. Boshuizen, J. E. V. Steenbergen, A. Bosman, S. V. d. Hof, H. A. V. Vliet, M. F. Peeters, R. J. V. Ketel, P. Speelman, J. L. Kool, A. E. Marina and C. V. Spaendonck. 2002. A large outbreak of Legionnaires’ disease at a flower show, the Netherlands, 1999. Emerging Infectious Diseases. 8(1): 37-43. 16. Diederen, B., C. d. Jong, I. Aarts, M. F. Peeters and A. v. d. Zee. 2007. Molecular evidence for the ubiquitous presence of Legionella species in Dutch tap water installations. Journal of Water and Health. 5(3): 375 - 383. 17. Droste, R. 1997. The theory and practice of water and wastewater treatment. New York, John Wiley and Sons, Inc.

9 Chapter 1. Introduction 18. Falkinham, J., M. Iseman, P. de Haas and D. van Soolingen. 2008. Mycobacterium avium in a shower linked to pulmonary disease. Journal of Water and Health. 06(2): 209-213. 19. Feazel, L. M., L. K. Baumgartner, K. L. Peterson, D. N. Frank, J. K. Harris and N. R. Pace. 2009. Opportunistic pathogens enriched in showerhead biofilms. Proceedings of the National Academy of Sciences of the United States of America. 106: 16393- 16399. 20. Field, S. K. and R. L. Cowie. 2006. Lung disease due to the more common nontuberculous Mycobacteria. Chest. 129(6): 1653-1672. 21. Fields, B., R. Benson and E. Besser. 2002. Legionella and Legionnaires' disease: 25 years of investigation. Clinical Microbiology Reviews. 15(3): 506-526. 22. Fry, A. M., D. K. Shay, R. C. Holman, A. T. Curns and L. J. Anderson. 2005. Trends in hospitalizations for pneumonia among persons aged 65 years or older in the United States, 1988-2002. Journal of the American Medical Association. 294(21): 2712-2719. 23. García-Fulgueiras, A., C. Navarro, D. Fenoll, J. García, P. González-Diego, T. Jiménez-Buñuales, M. Rodriguez, R. Lopez, F. Pacheco, J. Ruiz, M. Segovia, B. Baladrón and C. Pelaz. 2003. Legionnaires’ disease outbreak in Murcia, Spain. Emerging Infectious Diseases. 9(8): 915-921. 24. Greub, G. and D. Raoult. 2004. Microorganisms resistant to free-living amoebae. Clinical Microbiology Reviews. 17(2): 413-433. 25. Heath, T. C., C. Roberts, B. Jalaludin, I. Goldthrope and A. G. Capon. 1998. Environmental investigation of a legionellosis outbreak in western Sydney: the role of molecular profiling. Australian and New Zealand Journal of Public Health. 22(4): 428-431. 26. Hoffmann, R. and R. Michel. 2001. Distribution of free-living amoebae (FLA) during preparation and supply of drinking water. International Journal of Hygiene and Environmental Health. 203(3): 215-219. 27. Hrudey, S. and E. Hrudey. 2007. Published case studies of waterborne disease outbreaks - evidence of a recurrent threat. Water Environment Research. 79(3): 233- 245. 28. Isberg, R. R., T. J. O'Connor and M. Heidtman. 2009. The Legionella pneumophila replication vacuole: making a cosy niche inside host cells. Nature Reviews Microbiology. 7(1): 13-24. 29. Jernigan, D. B., J. Hofmann, M. S. Cetron, J. P. Nuorti, B. S. Fields, R. F. Benson, R. F. Breiman, H. B. Lipman, R. J. Carter, C. A. Genese, S. M. Paul, P. H. Edelstein and I. C. Guerrero. 1996. Outbreak of Legionnaires' disease among cruise ship passengers exposed to a contaminated whirlpool spa. The Lancet. 347(9000): 494-499. 30. Kool, J. L. M. D., D. M. P. H. Bergmire-Sweat, J. C. M. D. Butler, E. W. Brown, D. J. M. D. Peabody, D. S. M. D. Massi, J. C. P. E. Carpenter, J. M. B. A. Pruckler, M. S. Robert F. Benson and B. S. P. Fields. 1999. Hospital characteristics associated with colonization of water systems by Legionella and risk of nosocomial Legionnaires' Disease: a cohort study of 15 hospitals. Infection Control and Hospital Epidemiology. 20(12): 798-805. 31. Lamoth, F. and G. Greub. 2010. Amoebal pathogens as emerging causal agents of pneumonia. FEMS Microbiology Reviews. 34(3): 260-280. 32. Lau, H. Y. and N. J. Ashbolt. 2009. The role of biofilms and protozoa in Legionella pathogenesis: implications for drinking water. Journal of Applied Microbiology. 207(2): 368-378. 33. Lazarova, V., S. Hills and R. Birks. 2003. Using recycled water for non-potable, urban uses: a review with particular reference to toilet flushing. Water Science & Technology: Water Supply. 3(4): 69-77.

10 Chapter 1. Introduction 34. Levine, A. D. 2003. Use of reclaimed wastewater for cooling tower applications. Membranes for Industrial Wastewater Recovery and Reuse, International Water Association, International Water Association. 35. Lin, Y.-s. E., R. D. Vidic, J. E. Stout and V. L. Yu. 1998. Legionella in water distribution systems. Journal American Water Works Association. 90(9): 11. 36. Marciano-Cabral, F., M. Jamerson and E. S. Kaneshiro. 2010. Free-living amoebae, Legionella and Mycobacterium in tap water supplied by a municipal drinking water utility in the USA. Journal of Water and Health. 8(1): 71-82. 37. Marshall, H. M., R. Carter, M. J. Torbey, S. Minion, C. Tolson, H. E. Sidjabat, F. Huygens, M. Hargreaves and R. Thomson. 2011. Mycobacterium lentiflavum in drinking water supplies, Australia. Emerging Infectious Diseases. 17(3): 395-402. 38. McArthur, G. 2008. Legionella outbreak puts seven in hospital Herald Sun. Melbourne. 39. Michel, R., H. Burghardt and H. Bergmann. 1995. Acanthamoebae isolated from a highly contaminated drinking water system of a hospital exhibited natural infections with . Zentralblatt Fur Hygiene Und Umweltmedizin. 196(6): 532-544. 40. Okada, M., K. Kawano, K. Fumiaki, J. Amemura-Maekawa, H. Watanabe, K. Yagita, T. Endo and S. Suzuki. 2005. The largest outbreak of legionellosis in Japan associated with spa baths : Epidemic curve and environmental investigation. Kansenshogaku zasshi 79(6): 365-374. 41. Rizak, S. and S. E. Hrudey. 2008. Drinking water safety; challenges for community managed systems. Journal of Water and Health. 6: 33-41. 42. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241. 43. Scheid, P., L. Zöller, S. Pressmar, G. Richard and R. Michel. 2008. An extraordinary endocytobiont in Acanthamoeba sp. isolated from a patient with keratitis. Parasitology Research. 102(5): 945-950. 44. Snelling, W., J. Moore, J. McKenna, D. Lecky and J. Dooley. 2006. Bacterial- protozoa interactions; an update on the role these phenomena play towards human illness. Microbes and Infection. 8: 578-587. 45. Stout, J. E., V. L. Yu, P. Muraca, J. Joly, N. Troup and L. S. Tompkins. 1992. Potable water as a cause of sporadic cases of community-acquired Legionnaires' Disease. New England Journal of Medicine. 326(3): 151-155. 46. Tezcan-Merdol, D., M. Ljungstrom, J. Winiecka-Krusnell, E. Linder, L. Engstrand and M. Rhen. 2004. Uptake and replication of in Acanthamoeba rhysodes. Applied and Environmental Microbiology. 70(6): 3706-3714. 47. Thom, S., D. Warhurst and B. S. Drasar. 1992. Association of Vibrio cholerae with fresh water amoebae. Journal of Medical Microbiology. 36(5): 303-306. 48. Thomas, V., T. Bouchez, V. Nicolas, S. Rober, J. F. Loret and Y. Lèvi. 2004. Amoebae in domestic water systems: resistance to disinfection treatments and implication in Legionella persistence. Journal of Applied Microbiology. 97(5): 950- 963. 49. Thomas, V., K. Herrera-Rimann, D. S. Blanc and G. Greub. 2006. Biodiversity of amoebae and amoeba-resisting bacteria in a hospital water network. Applied and Environmental Microbiology. 72(4): 2428-2438. 50. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 51. Thomas, V., G. McDonnell, S. P. Denyer and J. Y. Maillard. 2010. Free-living amoebae and their intracellular pathogenic microorganisms: risks for water quality. FEMS Microbiology Reviews. 34(3): 231-259.

11 Chapter 1. Introduction 52. Toze, S. 2006. Water reuse and health risks - real vs. perceived. Desalination. 187: 41- 51. 53. U.S. Federal Government. 2006. Safe Drinking Water Act. United States of America, U.S. EPA. 54. Vairavamoorthy, K. 2008. Innovation in water management for the city of the future. International Urban Water Conference, Heverlee, Belgium, CRC Press-Taylor & Francis Group. 55. Wilson, P. A. and J. Ferguson. 2005. Severe community-acquired pneumonia: an Australian perspective. Internal Medicine Journal. 35(12): 699-705. 56. Woodhead, M. 2002. Community-acquired pneumonia in Europe: causative pathogens and resistance patterns. European Respiratory Journal. 20(S36): 20S-27S.

12 CHAPTER 2

LITERATURE REVIEW AND AIMS 2. TABLE OF CONTENTS

2.1 INTRODUCTION______15 2.2 FREE LIVING AMOEBAE (FLA)______15 2.2.1 Ecology and biology ______15 2.2.2 FLA as human pathogens______17 2.3 AMOEBAE RESISTANT MICROORGANISMS (ARM) ______18 2.3.1 FLA interaction with ARM ______18 2.3.2 FLA as transport hosts ______22 2.4 FLA INFECTED WITH ARM ______22 2.4.1 FLA naturally infected with ARM in drinking water systems and applications ______22 2.4.2 Laboratory based research of FLA infections by ARM ______24 2.5 LEGIONELLA IN DRINKING WATER SYSTEMS ______27 2.6 FLA IN DRINKING WATER SYSTEMS ______30 2.6.1 Literature describing FLA in treated drinking water systems ______30 2.6.2 FLA in drinking water treatment plants ______33 2.6.3 FLA in drinking water distribution systems______37 2.6.4 FLA at in-premise point of use ______40 2.6.5 FLA in hospital hot water systems______43 2.7 LEGIONELLA AND FLA IN RECYCLED WATER SYSTEMS ______44 2.8 LEGIONELLA AND FLA IN APPLICATIONS OF DRINKING WATER ______45 2.8.1 Legionella spp. in applications of drinking water ______45 2.8.2 FLA in applications of drinking water ______47 2.9 CONCLUSIONS ______49 2.9.1 Theoretical model of FLA population in drinking water systems ______49 2.9.2 Research needs to identify the risks of FLA associated with ARM______52 2.10 REFERENCES ______53

Chapter 2. Literature review LIST OF TABLES

Table 2.1 Examples of ecosystems where FLA have been found______16 Table 2.2 FLA known to be pathogenic ______18 Table 2.3 The diversity of environmental FLA naturally infected with pathogenic ARM _____ 24 Table 2.4 FLA infected with water-based pathogenic ARM in laboratory research______26 Table 2.5 Detection of Legionella spp. in drinking water systems ______29 Table 2.6 Summary of aims, methods, samples sites and sample types ______32 Table 2.7 The detection of FLA in drinking water treatment plants______35 Table 2.8 The detection of FLA in treated water distribution systems ______39 Table 2.9 The detection of FLA in treated drinking water at point of use ______41 Table 2.10 The detection of FLA in hospital hot water systems______43 Table 2.11 The detection of Legionella spp. in pools, spas and cooling towers ______47 Table 2.12 The detection of FLA in pools, spas and cooling towers ______49

LIST OF FIGURES

Figure 2.1. A theoretical model for FLA density and diversity and ARM ______51

PUBLICATIONS

Part of this chapter was published as a critical review:

Thomas, J. M. and N. J. Ashbolt. 2011. Do free-living amoebae in treated drinking water systems present an emerging health risk? Environmental Science & Technology. 45(3): 860 - 869.

14 Chapter 2. Literature review

2.1 INTRODUCTION This literature review aimed to consolidate the current knowledge about FLA naturally infected with ARM in drinking water systems as well as those FLA that could be infected with ARM in laboratory based experiments. A further intent was to consolidate the literature describing Legionella spp. in drinking and recycled water systems and applications. Importantly the review aimed to bring together, for the first time, all the available knowledge about research on the diversity and density of environmental FLA in treated drinking and recycled water systems around the world. The scope of this review covers water treatment plants through to in- premise point of use and applications including cooling towers. To appreciate how these water- based pathogens are able to survive and grow in FLA it is important to understand the ecology and biology of FLA.

2.2 FREE LIVING AMOEBAE (FLA) 2.2.1 Ecology and biology FLA form a distinct group within the kingdom Protista based on their pseudopodia locomotion (Prescott et al., 2005). FLA are thought to have been first described in 1786 by O.F. Muller (Corliss, 2001) and today there are over 180 recognised species (Smirnov et al., 2005; Adl et al., 2007). New FLA species are frequently identified in the environment as it is estimated that the FLA group could consist of more than 600 species in total (Adl et al., 2007). Classification of FLA has traditionally been based on morphological features (Fenchel, 1987; Smirnov et al., 2005; Pawlowski and Burki, 2009) which required considerable skill, yet often resulted in misclassifications (Visvesvara, 1991; Rodriguez-Zaragoza, 1994; Patterson, 1996; Smirnov et al., 2005; Behets et al., 2007). In part, this inadequate classification system has hampered the study of amoebae (Corliss, 1973). However, the modern application of small sub- unit rDNA phylogeny has completely dismantled the old morphological classification system replacing it with two new supergroups and Rhizaria (Smirnov and Brown, 2004; Pawlowski and Burki, 2009). FLA now form the super-group Amoebozoa which includes which are more commonly known as slime moulds (Pawlowski and Burki, 2009). FLA occupy a diverse number of environments that range from frozen Antarctic soils to arid deserts and the nose of humans (Table 2.1). Freshwater and soil environments are a common habitat for FLA with over 53 genera described (Smirnov and Brown, 2004). Colonisation of an environment by FLA is dependent on the optimal physical and chemical conditions being present (Anderson, 1988); as FLA are only active when there is sufficient moisture present (Fenchel, 1987; Patterson, 1996; Day et al., 2007). In comparison, prokaryotes have a larger distribution

15 Chapter 2. Literature review than FLA as they are able to occupy environments where extremes of temperature or pH would not support FLA (Fenchel, 1987). The current ecological profile of FLA is acknowledged as being limited and further exploration is needed before a more complete understanding of the diversity and distribution can be formed (De Jonckheere, 1991; Rodriguez-Zaragoza, 1994; Adl et al., 2007; Pawlowski and Burki, 2009). This is especially true for treated drinking water systems where the current literature describing FLA is limited and disjointed. Table 2.1 Examples of ecosystems where FLA have been found Environment Country FLA presence Reference (# samples) Mexico City, Air 100 % (n= 8) (Rivera et al., 1991) Mexico Soil Antarctica 31 % (n=70) (Brown et al., 1982) South Australia, Arid lands 100 % (n=26) (Robinson et al., 2002) Australia France, Belgium, Water products - Turkey and 17 % (n=65) (Mergeryan, 1991) bottled water Germany Thermal water National Parks 48 % (n=23) (Sheehan et al., 2003) springs U.S.A. Sea water - Atlantic Ocean, 100 % (n=9) (Sawyer, 1980) sediments USA Plants – lettuce New York, 52 % (n=624) (Napolitano and leaves U.S.A. Collettieggolt, 1984) Reptiles – snakes Lausanne, (Ciurea-Van Saanen, 21 % (n=14) and lizards Switzerland 1980) Humans - nose, Mexico City, 40 % (n=90) (Rivera et al., 1986) mouth and parynx Mexico

FLA are able to live independently of other organisms (Corliss, 1973; Patterson, 1996) and display two distinct morphologies; an active feeding trophozoite and dormant cyst stage (Fenchel, 1987; Patterson, 1996). Trophozoites range in size from 3 μm to 3000 μm (Fenchel, 1987) and cysts are formed as protection against unfavourable environmental conditions including: starvation, desiccation, temperature, pH, light, and the ionic composition of water (Fenchel, 1987). Excystation occurs when favourable environmental conditions return (Fenchel, 1987; Prescott et al., 2005). Reproduction in FLA is asexual and predominantly by binary division or the fragmentation of the plasmodium (Fenchel, 1987; Smirnov and Brown, 2004; Prescott et al., 2005). Doubling times for FLA populations are linked to cell size with the smallest amoebae taking 2.5 h to double while the largest amoebae take several days (Fenchel, 1987). FLA meet their specific amino acid and other nutritional requirements by feeding on (algae and protozoa) and prokaryotes (bacteria) (Fenchel, 1987; Anderson, 1988; 16 Chapter 2. Literature review Rodriguez-Zaragoza, 1994; La Scola et al., 2003; Prescott et al., 2005; Day et al., 2007). Uptake of solid particles is via phagocytosis while attached to a solid surface, with saprozoic uptake of soluble organic nutrition also possible (Fenchel, 1987; Rodriguez-Zaragoza, 1994; Prescott et al., 2005). It is presently unclear if FLA feed on viruses or if they are up-taken unintentionally during the feeding process (La Scola et al. 2003). FLA play an important role in nutrient conversion between trophic levels and organic carbon cycling in a micro-ecosystem (Curds, 1973; Clarholm, 1981; Fenchel, 1987; Patterson, 1996). The variations in lifecycles and nutritional requirements of FLA make them a difficult group of organisms to isolate and identify from freshwater environments by culture (Day et al., 2007) which is a contributing factor to the limited understanding of FLA in drinking water systems.

2.2.2 FLA as human pathogens Some FLA are human pathogens in their own right (Rocha-Azevedo et al., 2009) and are known to cause human infections such as primary amoebic meningoencephalitis (PAM) and amoebic keratitis (Marciano-Cabral, 2006) (Table 2.2). PAM infections are acquired while bathing when water enters the nasal passages allowing a particular FLA, Naegleria fowleri to attach to the olfactory mucosa and migrate to the olfactory bulbs in the brain where they multiply rapidly (Rocha-Azevedo et al., 2009). N. fowleri, proliferates in warmer waters (25 °C and above) such as thermal springs, whirlpools, swimming pools, cooling systems and distribution systems and presents a significant public health risk (De Jonckheere, 2002). PAM infections are regularly fatal but infrequent with only five cases reported by US Centre for Disease Control (CDC) in 2005 - 2006; all of which were fatal (Yoder et al., 2008). Globally it is estimated that there were only 200 cases of PAM caused by N. fowleri in 2004 (Schuster and Visvesvara, 2004). Although under reporting of amoebic infections is expected as amoebic disease encephalitides are difficult to diagnose without an autopsy (Schuster and Visvesvara, 2004). Infections with amoebic keratitis are more common than PAM, particularly among contact lens wearers (Radford et al., 2002). The infection can take up to five months to treat due to the resistance of cysts to anti-microbial drugs and 30 % of those infected will loose some of their sight (Radford et al., 2002). Globally there were more than 3000 cases estimated for 2004 (Schuster and Visvesvara, 2004). While, in the United Kingdom it was estimated that an average of 1.2 people per million contracted amoebic keratitis annually between 1997 and 1999 (Radford et al., 2002). The use of tap water to wash or store contact lenses has been identified as a risk factor for microbial keratitis for contact lens wearers (Seal et al., 1992; Houang et al., 2001; Kilvington et al., 2004; Schuster and Visvesvara, 2004). Ophthalmology research to identify

17 Chapter 2. Literature review amoebic keratitis causing FLA in tap water forms a large part of the existing information on FLA in drinking water. Table 2.2 FLA known to be pathogenic (Adapted from the text of Rocha-Avezedo et al. (2009)) FLA Disease Pathology Etiology Acanthamoeba Granulomatous Fatal central Immune suppression and spp. amebic encephalitis nervous system contact with water or soil infection; lung infection Cutaneous lesions Skin infection Immune suppression and contact with water or soil Amoebic keratitis Sight threatening Contact lens wearing or corneal infection corneal trauma Balamuthia Balamuthia amebic Fatal central Contact with stagnant mandrillaris encephalitis nervous system recreational water infection Naegleria fowleri Primary amebic Fatal central Contact with recreational meningoencephalitis nervous system fresh water that is warm (PAM) infection Sappinia sp. Sappinia amebic Non fatal central Not yet known as only encephalitis nervous system one case reported. infection 2.3 AMOEBAE RESISTANT MICROORGANISMS (ARM)

2.3.1 FLA interaction with ARM

2.3.1.1 FLA facilitate growth and pathogenicity of ARM Once up-taken via active feeding ARM are contained within membrane-bound vacuoles in the FLA. To survive and grow within these vacuoles ARM must avoid digestion and replicate successfully (Snelling et al., 2006). The mechanisms used by Legionella pneumophila to survive within FLA cells are the most well researched. L. pneumophila has been shown to utilise 27 dot/icm genes to avoid digestion and replicate successfully in both FLA and human macrophage cells (Heidtman et al., 2009; Isberg et al., 2009). The dot/icm genes encode for proteins that prevent vacuole trafficking into the lysosomal network and that also recruit eukaryotic cell apparatus for intracellular multiplication of L. pneumophila (Bozue and Johnson, 1996; Isberg et al., 2009). Recent sequencing work has also revealed that L. pneumophila has a number of eukaryotic like proteins which are likely to be utilised to mimick host cell functions and facilitate intracellular replication (Habyarimana et al., 2009; Nora et al., 2009). A similar pathway appears to be used by Mycobaterium avium when the bacteria is up-taken by FLA into membrane bound vacuoles and here it significantly reduces digestion rates by preventing lysosome fusion with the vacuoles (Cirillo et al., 1997). Other ARM have been identified

18 Chapter 2. Literature review occupying membrane bound vacuoles within FLA; Burkholderia cepia (Marolda et al., 1999), Parachlamydia sp. (Heinz et al., 2007), Salmonella enterica (Tezcan-Merdol et al., 2004) and Mimivirus (La Scola et al., 2008). The mechanisms by which these ARM avoided digestion are yet to be fully explored but it is likely that similar pathways are utilised. The strategies used by ARM to survive and grow in FLA are complex and there is more to be understood. The ARM of FLA can be pathogenic (Rowbotham, 1980; Fields et al., 1989; Gao et al., 1999), symbiotic (Fritsche et al., 1993; Park et al., 2004; Gast et al., 2009) or capable of both modes of growth (Greub et al., 2003). Under the right conditions ARM may lyse their FLA host, releasing large numbers of ARM into the aquatic environment (Rowbotham, 1980; Marolda et al., 1999; Greub et al., 2003) or may be expelled from FLA in vesicles (Harb et al., 2000). Importantly, human pathogenic ARM use very similar mechanisms to infect and replicate within the FLA as they do in human lung macrophage cells (Isberg et al., 2009; Newton et al., 2010). Therefore, when ARM lyse FLA they are released with their virulence genes up-regulated (Nora et al., 2009; Schmitz-Esser et al., 2010) and are comparatively more virulent to humans than they were prior to FLA passage (Cirillo et al., 1994; Cirillo et al., 1997; Neumeister et al., 2000). It is known that after growth in A. castellanii that L. pneumophila is up to 100 times more invasive for human epithelial cells (Cirillo et al., 1994). Other Legionella spp. have also been shown to be significantly more infective for human monocyte cell lines after co-culture with A. castellanii (Neumeister et al., 2000). However, individual Legionella spp. behave differently as they do not all have the same cytopathogenicty, that is the ability to lyse their host cells (Gao et al., 1999). Mycobacterium avium has also been shown to be more virulent when grown in A. castellanii and subsequently used to infect human macrophage cells (Cirillo et al., 1997). For L. pneumophila there is some evidence that growth within the FLA is a requirement for future infection of multi-cellular organisms (Cirillo et al., 1994). Recent comparative genomic analysis of ARM has provided significant evidence to support the up-regulation of pathogenicity by FLA. The comparative genomic analysis of a number of ARM have shown that they many contain reduced metabolic capacities when compared to other non intracellular bacteria (Schmitz-Esser et al., 2010). Furthermore, some ARM have been shown to contain unusually high numbers of proteins with eukaryotic domains (Nora et al., 2009; Schmitz-Esser et al., 2010). These proteins which include; ankyrin repeats, TPR/SEL1 repeats, leucine-rich repeats and domains from the eukaryotic ubiquitin family are used for interaction and exploitation of FLA (Nora et al., 2009; Schmitz-Esser et al., 2010). There is evidence that some of the aforementioned proteins were acquired by horizontal gene transfer between ARM possibly while residing within FLA (Schmitz-Esser et al., 2010). Therefore, not only do FLA provide an environment in which there is up-regulation of proteins 19 Chapter 2. Literature review required for human infection but they also provide a location for gene exchange between microorganisms that will enhance ARM pathogenicity (Moliner et al., 2009; Schmitz-Esser et al., 2010). It is has even been proposed that the interaction of Legionella with aquatic protozoa such as FLA select virulence traits that result in some Legionella sp. evolving into human pathogens (Albert-Weissenberger et al., 2007). Logically protozoa like FLA could be the source of new pathogenic microorganisms in the environment (Harb et al., 2000). The ability to become more infective after growth within FLA may provide water-based pathogenic ARM with a significant biologic advantage and is even thought to be critical for their evolution to human pathogens, especially as there is evidence of horizontal gene transfer within FLA (Harb et al., 2000; Molmeret et al., 2005; Albert-Weissenberger et al., 2007; Moliner et al., 2010; Thomas and Greub, 2010). Association with FLA appears also to provide critical factors required for ARM growth, such as nutrients (Steinert et al., 1997; Steinert et al., 1998; Fields et al., 2002; Axelsson-Olsson et al., 2005). Specifically L. pneumophila has fastidious growth requirements (Fields et al., 2002) and are often present in drinking water systems in a viable but non culturable state (VBNC) (Diederen et al., 2007). The addition of A. castellanii trophozoites to VBNC L. pneumophila, produced after incubation for 125 days in tap water, results in the successful resuscitation of the pathogen (Steinert et al., 1997). Furthermore, in water system biofilm simulations L. pneumophila already present in the biofilm will significantly increase (by 2.9 log units) 48 h after the addition of A. castellanii (Declerck et al., 2009). Interestingly, is has also been shown that Mycobacterium avium can grow saprozoically on the by-products produced by A. polyphaga (Steinert et al., 1998). Hence, research has demonstrated that microorganisms do not need to be contained intra-cellularly within FLA to obtain physiological advantages from the association. In low nutrient environments such as drinking water the association with FLA is thought to be an adaptive survival mechanism of ARM that promotes their persistence and dispersal (King et al., 1988).

2.3.1.2 FLA protects ARM from physical and chemicals factors Ecologically FLA are very useful hosts for ARM in drinking water systems, as FLA form persistent cysts that can protect the ARM from physical extremes of temperatures (Winiecka- Krusnell and Linder, 2001; Aksozek et al., 2002; Storey et al., 2004), desiccation (Fenchel, 1987; Kahane et al., 2001) and exposure to ultraviolet radiation (Chang et al., 1985; Aksozek et al., 2002). With respect to temperature, Acanthamoeba spp. cysts are particularly resistant due to the presence of cellulose in their cell walls (Winiecka-Krusnell and Linder, 2001). Cysts of this genera can survive temperatures as high as 80 °C for up to 10 min (Storey et al., 2004).

20 Chapter 2. Literature review Furthermore, A. castellanii cysts can endure rapid changes in temperature and are resistant to repeated (n = 5) freeze thaw cycles (-160 °C to 45 °C) (Aksozek et al., 2002). Contained within the cysts of FLA, ARM are also protected from desiccation that would be lethal to individual cells in the environment (Winiecka-Krusnell and Linder, 2001). For example, Acanthamoeba polyphaga under desiccating conditions at 4 °C were able to maintain Simkania negevensis for 67 days longer than the bacteria maintained alone (Kahane et al., 2001). Cysts of FLA are also more resistant to ultraviolet (UV) radiation than bacteria or viruses and can require up to 15 times higher doses of UV for effective disinfection (Chang et al., 1985). After UV exposure times of 2 h (800 mJ.cm-2) cysts of A. castellanii were still viable (Aksozek et al., 2002). No literature on the survival of ARM contained with FLA cysts during UV radiation was identified. However, as UV disinfection is increasingly used to disinfect drinking and recycled water (Asano et al., 2007) more research is required to determine its efficacy against ARM that may be water-based pathogens. Free-living amoebae are able to protect ARM from a range of physical conditions, thus increasing their chances of survival in drinking and recycled water systems and subsequent water applications. FLA cysts also generally provide protection for ARM from a range of chemicals used in water disinfection including chlorine (De Jonckheere and van de Voorde, 1976; King et al., 1988; Kilvington and Price, 1990; Storey et al., 2004; Loret et al., 2008) and to a lesser degree monochloramine (Donlan et al., 2005). Specifically, cysts of A. polyphaga containing L. pneumophila were able to survive exposure to free chlorine at 50 mg.L-1 for 18 h (CT = 54 000) and upon subsequent excystation produced viable L. pneumophila (Kilvington and Price, 1990). These findings were confirmed in simulated water treatment plant conditions, where cysts of A. polyphaga were not inactivated by 2 h exposures to chlorine at 100 mg.L-1 (CT = 12 000)(Loret et al., 2008). However, the level of resistance varies considerably for different FLA species (De Jonckheere and van de Voorde, 1976). In one study Naegleria fowleri were shown to be the least resistant (1 h at 0.5 mg.L-1 ; CT = 30) while Acanthamoeba culbertsoni were the most resistant (24 h at 40 mg.L-1 ; CT = 57600) (De Jonckheere and van de Voorde, 1976). Monochloramine has been shown to be more effective than free chlorine at inactivating L. pneumophila associated with Hartmannella vermiformis after 3 h of contact time at 0.5 mg.L-1 (CT = 90) (Donlan et al., 2005). More research is needed to determine if monochloramine would be a better alternative to free chlorine to control FLA and ARM in water distribution systems. Other commonly used chemical disinfectants are also less effective against FLA cysts such as hydrogen peroxide and isopropyl alcohol (Aksozek et al., 2002), as well as bacteria-targeted antibiotics (Barker et al., 1992; Miltner and Bermudez, 2000; Rocha-Azevedo et al., 2009). The ability for FLA cysts to

21 Chapter 2. Literature review protect ARM from disinfection is of particular concern to the water industry and needs to be addressed (King et al., 1988; Thomas et al., 2004; Lau and Ashbolt, 2009).

2.3.2 FLA as transport hosts Another aspect of the ARM interaction with FLA is that the FLA may be a transport vehicle for water-based pathogenic ARM from drinking water via aerosols to the human respiratory tract (Rowbotham, 1980; Barker and Brown, 1994; Abu Kwaik et al., 1998; Berk et al., 1998; Winiecka-Krusnell and Linder, 1999). FLA have been shown to readily colonize the nasal tract and throat of healthy individuals (Rivera et al., 1986; Badenoch et al., 1988; Mergeryan, 1991). Furthermore, an American sero-prevalence study of healthy individuals indicated that 87 % (n = 55) of people tested positive for Acanthamoeba spp. infection (Chappell et al., 2001). Also, FLA were the only microorganism that consistently produce positive reactions from the sera of factory workers (MRC Symposium, 1977). These finding indicate that FLA commonly colonize the mucosal membrane and/or cause mild infections. It has even been proposed that Pontiac Fever, which is thought to be caused by infection with various Legionella spp., could actually be caused by hypersensitivity to Acanthamoeba spp. (Rowbotham, 1986). Hypothetically FLA infection of the human respiratory tract would easily facilitate the passage of any pathogenic ARM contained within the FLA, however research is needed to support this mode of transmission. FLA could also facilitate infection by an alternative pathway: the production of tiny aerosolizble vesicles (2-6 μm) containing infective ARM (Rowbotham, 1980; Berk et al., 1998) that can be inhaled. The sero-prevalence to FLA and frequency of FLA isolation in humans gives strong support to the hypothesis that FLA could act as transport hosts for water-based pathogenic ARM.

2.4 FLA INFECTED WITH ARM

2.4.1 FLA naturally infected with ARM in drinking water systems and applications FLA can facilitate the growth, survival, virulence and potentially transportation of a range of water-based pathogens. It is for these reasons highly plausible that FLA are a primary mechanism used by pathogenic water-based ARM to proliferate in treated drinking water systems. It is therefore surprising that there is very limited data on which FLA act as hosts in treated drinking water, what may control them or indeed the ecological conditions that promote ARM that may be human pathogens. This review aims to bring together, for the first time, all the available research on the diversity and density of environmental FLA in treated drinking water systems and applications around the world. The scope of this review covers drinking water

22 Chapter 2. Literature review treatment plants through to in-premise point of use including hospital hot water systems. Drinking water applications included are secondary uses for drinking water such as cooling towers and swimming pools and spas. Overall, culture-based biases have limited our understanding of FLA infected with ARM, and molecular methods required to screen for ARM have only been applied to drinking water samples in the last 10 - 20 years (Fritsche et al., 1993). Only eight studies were identified that had directly detected environmental FLA naturally infected with ARM in treated drinking water systems or applications (Table 2.3). Despite this scant research, six FLA genera have been shown to be naturally infected with a range of known and emerging pathogenic CAP causing ARM including Legionella spp., Mycobacterium sp., Flavobacterium spp., Pseudomonas spp. and Neochlamydia sp. Infected FLA were identified from within the water treatment plant right through to the point of use at the consumers’ taps and cooling towers. There is evidence that FLA infection rates might be consistently above zero, with one study reporting 16 % of FLA isolated from a drinking water system being infected with ARB (Thomas et al., 2008). It was reported that cooling towers are 16 times more likely to contain FLA infected with ARB than environmental water samples (Berk et al., 2006). This data supports the premise that FLA isolated from man-made systems may be more likely to contain ARB than FLA in natural sources waters (Thomas et al., 2008). Also ARM within FLA have been detected in hot tap water at temperature of 56 °C (Dykova et al., 2009) which emphasises the ability of these micro- organisms to occupy diverse niches within a water system. Though not all FLA isolates were identified with ARM, it is likely that there are many more symbiotic ARM that are yet to be discovered (Gast et al., 2009). Techniques utilising FLA co-culture and enrichment have enabled isolation of ARM that would not grow as pure cultures on traditional culture media (Rowbotham, 1980; Pagnier et al., 2008). Novel microorganisms such as Amoebosporidium minutum (microsporidian organism) (Hoffmann et al., 1998), Rhabdochlamydia spp. (Corsaro et al., 2009), Parachlamydia acanthamoebae (Birtles et al., 1997), and the giant mimivirus (La Scola et al., 2003) and its virophage (La Scola et al., 2008) have all been identified from aquatic environments with the aid of amoebae co-culture. As amoeba co-culture is utilised more extensively in environmental surveys more ARM will be discovered (Loret and Greub, 2010), especially as there are examples of FLA (Amoeba proteus) being reliant on their ARM for their own survival (Park et al., 2004). There is strong evidence that FLA are a natural host for known and emerging pathogens in drinking water systems (Hoffmann and Michel, 2001). As over 53 genera of FLA have been identified in freshwater/soil environments (Smirnov and Brown, 2004) it is likely that many

23 Chapter 2. Literature review more FLA infected with previously unidentified pathogenic ARM will continue to be found (Corsaro et al., 2010). Table 2.3 The diversity of environmental FLA naturally infected with pathogenic ARM isolated from treated drinking water systems and applications. Drinking FLA genera FLA species ARM* Reference water Flavobacterium Water indologines treatment Pseudomonas (Hoffmann and plant or stutzeri Michel, 2001) distribution Environmental Stenotrophomonas # system FLA maltophilia Legionella Cooling pneumophila (Berk et al., 2006) towers Legionella anisa sp. Pseudomonas In-premise (Michel et al., 1995) aeruginosa (tap) Distribution Acanthamoeba Legionella sp. (Corsaro et al., 2010) system T4 genotype Reservoir Neochlamydia sp. (Corsaro et al., 2009) sediment Mycobacterium mucogenicum Water Legionella Echinamoeba E. exundans treatment (Thomas et al., 2008) londiniensis plant Legionella donaldsonii Legionella Filamoeaba F. nolandi Hot tap water (Dykova et al., 2009) micdadei Water Neochlamydia sp. (Thomas et al., 2008) H. treatment Hartmannella vermiformis In-premise L. pneumophila (Thomas et al., 2006) (tap) Flavobacterium Water Naegleria N. andersoni johnsoniae treatment (Thomas et al., 2008) L. micdadei plant N. gruberi L. anisa     In-premise (Hoffmann et al., Vannella       (tap) 1998) * Only ARM naturally infecting environmental FLA are reported # FLA not identified in the publication cited

2.4.2 Laboratory based research of FLA infections by ARM The vast majority of research on FLA and ARM interactions has been conducted in the lab using axenic culture collection FLA from the genus Acanthamoeba spp. (Greub and Raoult, 2004). As Acanthamoeba spp. are human pathogens they are more readily available in culture collections and heavily utilised for experimentation, to the detriment of research on non- 24 Chapter 2. Literature review pathogenic amoebae genera (Rodriguez-Zaragoza, 1994). Already seven different genera of FLA have been identified as infected with ARM in treated drinking water (Table 2.3) thus highlighting that laboratory-based research that largely focuses on using Acanthamoeba spp. may be missing key ecological aspects that are related to other important drinking water genera, such as Echinamoeba sp. (Thomas et al., 2008). Furthermore, culture collection FLA when compared to non-axenic environmental FLA of the same species yield significantly different responses to environmental stresses (Rowbotham, 1980; Srikanth and Berk, 1993). To date in laboratory based research, seven other genera of FLA, besides Acanthamoeba, have been infected with pathogenic ARM (Table 2.4). The ARM include a CAP causing virus (Mimivirus) and CAP causing bacterial pathogens (L. pneumophila, Legionella sp., P. aeruginosa, Simkania negevensis and Mycobacterium avium) plus bacteria of faecal origin (Campylobacter jejuni, Vibrio cholerae and Salmonella enterica) (Table 2.4). Laboratory research has also given insight into the complexity of FLA interactions with ARM. A number of known and emerging water-based pathogens can live both independently and as ARM within freshwater FLA (Greub and Raoult, 2004). Facultative intracellular organisms is the term given to ARB that display an ability but not a requirement for intracellular growth (Prescott et al., 2005). Facultative intracellular ARB have very specific genetics that allow them to infect FLA, which is why an organism's species is not a determining factor in assessing the ability to be an ARM (Snelling et al., 2006; Isberg et al., 2009). Different species of the facultative intracellular ARB Mycobacterium display individual resistance and growth characteristics within the same strain of Acanthamoeba castellanii (Cirillo et al., 1997). Furthermore, the same species of Vibrio cholerae has different strains which are or are not resistant to a specific FLA host (Thom et al., 1992). Different serovars of Salmonella enterica display resistance and replication specificity for Acanthamoeba spp. (Tezcan-Merdol et al., 2004). Legionella is the only genera to date that contains a number of species that are able to infect a range of FLA consistently (Lau and Ashbolt, 2009), which is likely due the evolution of the bacteria (Albert-Weissenberger et al., 2007). Aside from the facultative intracellular ARM there are other micro-organisms that have different interactions with FLA. There is a second group of water-based pathogens which require FLA for growth; these ARM must reside within FLA (Greub and Raoult, 2004) and for ARB this is referred to as an obligate intracellular organism (Prescott et al., 2005). However, other ARM do not need to be intracellular but may require the specific extracellular products provided by FLA (Steinert et al., 1998; Greub and Raoult, 2004).

25 Chapter 2. Literature review

Table 2.4 FLA infected with known and emerging water-based pathogenic ARM in laboratory research. FLA genus Species Pathogen Source L. pneumophila (Rowbotham, 1980) A. castellani Methicillin resistant (Huws et al., 2006) Staphlycoccus aureus A. culbertsoni L. pneumophila (Fields et al., 1989) A. humidifier L. pneumophila (Rowbotham, 1980) Burkholderia cepacia (Marolda et al., 1999) Campylobacter jejuni (Axelsson-Olsson et al., 2005) L. pneumophila (Rowbotham, 1980) Acanthamoeba Mimivirus (La Scola et al., 2003) A. polyphaga Mycobacterium avium (Steinert et al., 1998) Parachlamydia (Greub et al., 2003) acanthamoebae Simkania negevensis (Kahane et al., 2004) A. palestinesis Legionella spp. (Rowbotham, 1980) A. rhysodes Salmonella enterica (Tezcan-Merdol et al., 2004) A. royreba L. pneumophila (Tyndall and Domingue, 1982) Amoeba A. proteus Legionella jeonii (Park et al., 2004) Dictyostelium D. discoideum L. pneumophila (Solomon et al., 2000) E. exudans L. pneumophila (Fields et al., 1989) Echinamoeba sp. P. aeruginosa (Michel et al., 1995) spp. L. pneumophila (Fields et al., 1989) Hartmannella H. vermiformis L. pneumophila (Fields et al., 1989) H. cantabrigiensis L. micdadei (Fallon and Rowbotham, 1990) N. fowleri L. pneumophila (Newsome et al., 1985) L. pneumophila (Rowbotham, 1980) N. gruberi Naegleria Vibrio cholerae (Thom et al., 1992) N. jadini L. pneumophila (Rowbotham, 1980) N. lovaniensis L. pneumophila (Tyndall and Domingue, 1982) Vahlkamphia V. jugosa Legionella spp. (Rowbotham, 1986)

Interactions between FLA and ARM are diverse, unique and determined by a host of factors including genetic and environmental factors, which we are only beginning to understand. The diversity of pathogenic ARM that are able to grow and replicate in FLA under laboratory

26 Chapter 2. Literature review conditions emphasises the range of FLA connections with ARM that are likely to exist in water systems but have not yet been identified. For the purposes of fully understanding the ecology of FLA and ARM in drinking water, research needs to place more emphasis on ARM found in FLA directly isolated from those systems. 2.5 LEGIONELLA IN DRINKING WATER SYSTEMS Legionella spp. are currently the third most common etiological agent for waterborne outbreaks in the USA (Craun et al., 2010). To understand how common Legionella spp are in drinking water systems around the world, including hospital hot water systems, information about the density and diversity of these bacteria was collated (Table 2.5). A selection of studies (n = 9) were identified that sampled from drinking water distribution systems, residence taps and showers and hospital hot water systems. Traditionally culture methods have been used to detect Legionella spp. and were used exclusively in two studies identified (Henke and Seidel, 1986; Sanden et al., 1992). Culture methods were compared to qPCR in six of the studies and all determined qPCR to be significantly more sensitive than detection by culture across a range of drinking water samples (Wellinghausen et al., 2001; Devos et al., 2005; Joly et al., 2006; Diederen et al., 2007; Chen and Chang, 2010; Wullings et al., 2011). One study reviewed used the technique of amoebae-co-culture to isolate Legionella spp. (Thomas et al., 2006). Amoebae- co-culture technique was first described by Rowbotham (1980) and involves incubating water samples with axenic cultures of FLA which allows for the growth and subsequent detection of a range of ARM (La Scola et al., 2003; Greub and Raoult, 2004; Pagnier et al., 2008). The frequency of Legionella spp. detection in drinking water distribution systems and at point of use ranged from 30 - 100 % with a mean of 75 % ( = 23). One study determined that the presence of FLA in samples significantly increased (p < 0.001) the likelihood of Legionella isolation (Sanden et al., 1992). The highest density of Legionella spp. reported was 400 cells.mL1 for shower samples using the sensitive qPCR detection technique (Chen and Chang, 2010). The greatest diversity of Legionella spp. identified was nine (Sanden et al., 1992) with three studies only reporting for a single species samples. The high mean frequency of isolation (75 %), diversity and density of Legionella spp. from a diverse range of sample and countries reveals how common the bacteria is in drinking water systems and point of use taps and showers. Biofilm is thought to be integral to the growth of Legionella spp. in drinking water systems (Declerck, 2010). Legionella are able to grow in two types of biofilms; solid-water interface biofilms and water-air interface (floating) biofilms (Declerck, 2010). In addition to detection in drinking water Legionella spp. were also detected in the accompanying biofilms

27 Chapter 2. Literature review distribution systems at densities as high as 390 cells.cm-2 (Wullings et al., 2011), supporting the role of biofilm in the facilitation of growth in water systems. However, it must be emphasised that Legionella spp. are only one small component of possible ARM bacteria present in water and biofilm samples. An extensive survey of showerhead biofilms in the U.S.A. revealed that Legionella spp. only account for 0.1 % (n = 3856) of the DNA sequences present (Feazel et al., 2009). In comparison, Mycobacterium spp. account for 28.1 % of the DNA sequences present (Feazel et al., 2009) indicating that this pathogenic ARM could present a greater health risk than Legionella spp. As Mycobacterium spp. have been identified in FLA in drinking water (Thomas et al., 2008) it is likely that controlling FLA will also reduce the risk presented. Due to ill-health, patients in hospitals are already more susceptible to infections; hence it is imperative to fully understand how pathogenic ARM multiply in hospital water systems and identify how to control them. In hospital hot water systems Legionella spp. were detected at a frequency that ranged from 15 - 99 % with a mean of 61 % ( = 31). As with the drinking water samples detection of Legionella spp. in hospital hot water systems was significantly (p < 0.001) correlated to the presence of FLA in the samples (Thomas et al., 2006). Furthermore, there was a strong correlation between the presence of Hartmannella spp. and particularly H. vermiformis and the detection of L. pneumophila in the samples (Breiman et al., 1990). Using qPCR the highest Legionella spp. density was reported as 400 cells.mL-1 (Joly et al., 2006). Only two studies reported a full range of Legionella spp. of which the greatest number of species identified was four (Kool et al., 1999). Of the water quality factors, temperature appears to be protective in the reduction of Legionella spp. as mean temperature of the samples was 56 °C accounting for reduced frequency of detection (15 %) (Thomas et al., 2006) compared to other systems which had considerably lower temperatures of 40 °C (Breiman et al., 1990) and 46 °C (Wellinghausen et al., 2001) and higher detection frequencies. Water disinfection residuals also impacted upon Legionella spp. detection, which were found to increase in a linear relationship as chlorine disinfection levels fell below 0.5 mg.L-1 (Kool et al., 1999). Sediment in the bottom of hot water tanks is an additional variable that can provide nutritional factors for the increased growth of environmental microorganisms which in turn significantly increased the growth of L. pneumophila via nutritional symbiosis (Stout et al., 1985). While the frequency of detection of pathogenic L. pneumophila in all of the studies emphasises that hospital hot water systems present an ideal environment for the growth of pathogenic ARM. Control strategies for Legionella spp. in hospitals involve super-heating, flushing and alternative treatment options such as copper - silver ionisation which have mixed success (Lin et al., 1998). It has already been recommended that FLA be controlled in hospitals hot water systems to assist in inhibiting the survival of the deadly methicillin resistant Staphlyococcus aureus (MRSA) which is causing 28 Chapter 2. Literature review increasing hospital fatalities (Huws et al., 2006). The significant association between FLA and Legionella spp. in both drinking water systems and hospital hot water systems gives strong support to the need to target FLA as part of the strategy to reduce the growth of these pathogenic ARM. Table 2.5 Detection of Legionella spp. in drinking water distributions systems, tap water and hospital hot water systems ) -1 Legionella spp. identified Location Reference (in order of frequency)

Density Diversity cells.mL (# species) % positive ( (# samples) DISTRIBUTION SYSTEMS L. bozemanae L. pneumophila 100 %ground ND - L. donaldsonii Netherlands 6 (Wullings et al.) n = 15 6 L. yabuuchiae L. anisa L. lytica TAP WATER 60 %tap ND - L. pneumophila (n = 5) 4 L. londiniensis (Devos et al., Belgium > 4 L. adelaidensis 2005) 83 % shower ND - L. fairfieldensis (n = 25) 80 Legionella-like amoebal pathogen 77 %hot ND - L. pneumophila (Joly et al., France 2* (n = 122) 100 Legionella spp. 2006) Legionella-like amoebal pathogen L. busanensis L. worsliensis L. pneumophila 87 %cold tap (Diederen et al., Netherlands - 8 L. macheachernii (n = 352) 2007) L. parisiensis L. adelaidensis L. erythra L. shakespeare (Henke and Israel 88 %cold tap >1* L. pneumophila Seidel, 1986) L. anisa L. bozemanii L. gormanii L. micdadei U.S.A 30 % (Sanden et al., - 9 L. pneumophila (Atlanta) (n = 207) 1992) L. rubrilucens L. sainthelensi L. steigerwaltii Unidentified - Legionella sp.

29 Chapter 2. Literature review 100 % L. pneumophila (Chen and Taiwan shower 400 2* Legionella spp. Chang, 2010) (n = 8) HOSPITAL HOT WATER SYSTEMS 63 % ND - L. pneumophila (Joly et al., France >2* (n = 91) 100 Legionella spp. 2006)

99 % 0.03 - Legionella spp. (Wellinghausen Germany >2* (n = 77) 10 L. pneumophila et al., 2001)

15 % L. anisa (Thomas et al., Switzerland - 2 (n = 26) L. pneumophila 2006) L. anisa L. dumoffii 73 % (Kool et al., U.S.A - 4 L. feeleii (n = 215) 1999) L. pneumophila (sero gp 1, 3, 6, 8 and 10) 57 % O (Breiman et al., U.S.A. - >1* L. pneumophila (n =14) 1990) * Detection of a single Legionella spp. reported O Outbreak of Legionnaires' disease or Pontiac Fever associated with the sample hospital ND - no detects of Legionella spp. in the samples 2.6 FLA IN DRINKING WATER SYSTEMS

2.6.1 Literature describing FLA in treated drinking water systems Given the ubiquitous occurrence of FLA in the environment, they are frequently transported into drinking water by water, soil, air, animals and plants (Rodriguez-Zaragoza, 1994). However, global research on the ecology of FLA within drinking water systems is scant and disjointed. In total 29 studies were identified where FLA were described for treated drinking water systems in 19 different countries (Table 2.6). Only six of the studies reviewed aimed to conduct comprehensive quantitative analysis for both FLA and ARM within drinking water systems. Slightly more studies (n = 10) looked broadly at FLA ecology while the majority (n = 13) only focused on two pathogenic FLA genera; Acanthamoeba and Naegleria spp. Hence there is a poor understanding of FLA diversity but there is more knowledge available on the frequency of FLA detection. In the majority of studies (n = 24) FLA were detected by culture based methods, which are known to underestimate both FLA diversity and density (Smirnov and Brown, 2004) (Table 2.6). Furthermore, some studies may have misclassified the FLA isolated because they identified by morphology alone (n = 13), which is frequently inaccurate (Smirnov et al., 2005). Fewer studies (n = 8) identified isolates using more exact molecular techniques of partial 18S rRNA

30 Chapter 2. Literature review sequencing. Only a small number of studies detected FLA directly using molecular methods; PCR (Boost et al., 2008) or qPCR (Puzon et al., 2009; Valster et al., 2009). Water samples from in-premise taps were the most frequently sampled location (n = 18) (Table 2.6). Five studies looked at FLA in both drinking water treatment plants and the distribution systems thus providing very useful data on FLA throughout water systems. Only four studies looked for FLA in hospital hot water systems which presented a unique environment due to the nature of the systems and presence of susceptible patients. Biofilm (n = 11) and sediments (n = 2) were sampled less frequently than water (n = 27) from all the sampling locations and this may have resulted in further underestimations of FLA, as they are thought to be predominantly water/solid surface associated (Rodriguez-Zaragoza, 1994)

31 Chapter 2. Literature review Table 2.6 Summary of aims, methods, samples sites and sample types for literature reviewed describing FLA in drinking water systems. Research aims and methods Sampling location and type

e y

. g p

stem Reference pp

Country lant ta y s Region FLA FLA p water ARM Source stora densit Treated eba spp. Hospital diversity n s reservoir hot water Naegleria Treatment Distributio In-premise In-premise Acanthamo Bulgaria C, M W W (Tsvetkova et al., 2004) France C C, D P, D W W W (Thomas et al., 2008) France/Spain C C, D P, D W (Loret et al., 2008) France V W W (Amblard et al., 1996) Germany C C, M V W W W W, S W (Hoffmann and Michel, 2001) Germany C, M W, B (Rohr et al., 1998) Netherlands qP qP, O W, B (Valster et al., 2009) Poland C, D C, M W (Lanocha et al., 2009) Spain C C, D P, D W W, B W B, S (Corsaro et al., 2010) Spain C C, D P, D W W W B, S (Corsaro et al., 2009) Canary Islands C, P W (Lorenzo-Morales et al., 2005) C, D P, D W, B

Europe/ Middle East Switzerland (Thomas et al., 2006) United Kingdom C, M C, O W, B (Kilvington et al., 2004) United Kingdom C, M C, M W (Seal et al., 1992) Israel O W (Kahane et al., 2004) Canada C C, M W (Barbeau and Buhler, 2001) Mexico C, M W W W (Bonilla-Lemus et al., 2010) United States C, D W (Marciano-Cabral et al., 2010) United States C, M B (Shoff et al., 2008) United States C, M W W (Sanden et al., 1992) United States C, M W, B (Breiman et al., 1990) C, M W

Americas United States (Fields et al., 1989) Nicaragua C, O C, O W (Leiva et al., 2008) West Indies C, P W (Lorenzo-Morales et al., 2005) Brazil C, D B (Carlesso et al., 2010) Australia qP W, B (Puzon et al., 2009) Hong Kong P W (Boost et al., 2008)

Asia Hong Kong C, M B (Houang et al., 2001) Korea C, M W W W (Jeong and Yu, 2005) Detection: C - detected by culture, P - detected by PCR, qP - detected by qPCR, V - visual detection (microscopy). Identification: M - identified by morphology, D - identified by DNA sequencing, and O - other. Samples: W - water, B - biofilm and S -sediment

32 Chapter 2. Literature review

2.6.2 FLA in drinking water treatment plants The effectiveness of water treatment processes in removing FLA from source waters was evaluated in six studies sampling from 16 individual drinking water treatment plants (Table 2.7). Although different water treatment plants were sampled the FLA data can be compared across the three main stages of water treatment; source water quality, clarification/filtration and ultra- filtration/disinfection. FLA were isolated from surface source waters at high frequency (75 - 100 %) while no FLA were detected in one ground water source (Corsaro et al., 2009). Compared to the other treatment plants the highest FLA densities (90 amoebae.mL-1) and diversities (17 genera) were reported in raw river water (Hoffmann and Michel, 2001) and after the first stage of water treatment (clarification/filtration) with FLA densities of 4 amoebae.mL-1 and diversity of 8 genera (Hoffmann and Michel, 2001). To gain a better picture of the removal rates of FLA the efficiencies of the treatment stage can be evaluated by calculating the respective log removals (log removal = -log(concout/concin) (Asano et al., 2007). For the first stage in a water treatment plant (clarification/filtration) there was a large range in log removals (0 - 4.6 log) with a modal log removal of 1-2 log across all the treatment plants (Table 2.7). The low log removals recorded for some treatment plants (1.3 log) may be due to being challenged by high FLA in the source waters (Hoffmann and Michel, 2001). Although some plants with low FLA source water densities (0.01 amoebae.mL-1) still recorded zero log removal (Thomas et al., 2008) in which case the filtration medium itself could be colonized by FLA (Corsaro et al., 2010). This is especially likely as protozoa (including FLA) preferentially colonize the top layer of sand filters where they can easily detach (Bomo et al., 2004). This was evident in one treatment plant where the source water diversity (seven genera) was dominated by Acanthamoeba spp. but in the sand biofilm of the filtration process Naegleria was the leading FLA of the five genera detected (Thomas et al., 2008). This changed again in the bulk water after the treatment stage where Hartmannella was the dominant FLA genera detected (Thomas et al., 2008). These clear changes in FLA community structure indicate that FLA genera are affected by water treatment processes differently and that some FLA genera are able to colonize at different stages. Clarification and filtration are important stages for the elimination of FLA during water treatment (Loret et al., 2008) and the variations seen in the density, diversity and removal of FLA across different water treatment plants needs to be further explored. The final stages of water treatment may involve filtration (often using carbon) and/or disinfection usually by chlorination. However, only five studies reported FLA data after this treatment stage. The percentage of samples found positive for FLA dropped compared to after the first treatment stage with less than 50 % containing detectable FLA (Table 2.7). The density 33 Chapter 2. Literature review of FLA present in the studies ranged from no detects to 0.11 amoebae.mL-1. Interestingly, the highest density of FLA (0.11 amoebae.mL-1) was recorded in a single monthly sample and was significantly higher than the other months (Thomas et al., 2008). Similarly, one-off high densities of FLA (0.1 amoebae.mL-1) were recorded in another study (Hoffmann and Michel, 2001) and it was identified that breakthrough events (low log removals) had been caused by the release of FLA that had colonized the filtration stage (Hoffmann and Michel, 2001; Thomas et al., 2008) and clarifying sludge (Corsaro et al., 2010). These observations are supported by other research reporting that FLA colonize new water treatment plant filters within weeks and that backwashing practices are not effective at removing them (Baudin et al., 2008). It appears that some genera of FLA (Hartmannella, Echinamoeba and Vannella) that are only detected in low densities in the source waters are able to effectively colonize and re-grow in the water treatment plants and be isolated in the final stages of water treatment (Thomas et al., 2008). Furthermore, FLA from the genus Hartmannella have been reported to resist disinfection entirely and enter the distribution in a number of different water treatment plants, indicating more disinfection efficacy studies should be undertaken on this FLA (Corsaro et al., 2010). Overall, FLA are removed during drinking water treatment with a modal efficiency of 1- 2 log but breakthrough events do occur and release potentially high numbers of FLA (~0.11 amoebae.mL-1) into the distribution system. It has been reported that FLA densities show positive correlation with dissolved organic matter levels in water treatment plants (Loret et al., 2008) and hence this water quality variable may be a potentially useful factor for gauging potential FLA populations. Work on the feasibility of various disinfection options for FLA has been explored (Loret and Greub, 2010; Thomas et al., 2010) but more research is required to determine effective water treatment methods to minimize the number of FLA entering a distribution system. Water treatment plants are an important stage in reducing the number of FLA entering a distribution system but it may be more important to take measures to control the growth of FLA once in the distribution system in order to minimize growth of pathogenic ARM.

34 Chapter 2. Literature review Table 2.7 The detection of FLA in drinking water treatment plants

a -1

Treatment steps FLA genera identified - sampled in order of frequency Density # genera Diversity % positive (# samples) amoebae.mL Log removal FRANCE -1 treatment plant (Thomas et al., 2008) Acanthamoeba, Naegleria, River source water 100 % Hartmannella, Vannella, 0.01-0.1 - 7 (Seine River) (n = 4) Echinamoeba, Glaeseria, Platyamoeba Naegleria, Echinamoeba Sand biofilm - - - 5 Acanthamoeba, Hartmannella Vannella Clarification; 75 % <0.005- 0-2 1 Hartmannella Rapid sand filtration (n = 4) 0.046 25 % Ozonation <0.005 - 2 Echinamoeba, Hartmannella (n = 4) Echinamoeba, Naegleria, Carbon biofilm - - - 4 Vannella 25 % GAC* - filtered <0.005 0-2 2 Echinamoeba, Hartmannella (n = 4) Chlorinated finished 50 % <0.005 – - - water (n = 4) 0.11 FRANCE AND SPAIN - 4 treatment plants (data listed per treatment plant) (Loret et al., 2008) 100 % 0.015- Surface source water - - - (n = 4) 0.15 Sludge blanket, sand 100 % <0.001- 0->2.2 - - filtration (n = 4) 0.046 100 % Surface water source 0.093 - - - (n = 1) Sludge blanket GAC* 100 % 0.004 1.4 - - filtration (n = 1) 100 % 0.048- Surface source water - - - (n = 2) 0.092 Sludge recirculation, 100 % >2.2- <0.0003 - - biolite filtration (n = 2) >2.5 100 % 0.21 – Surface source water - - - (n = 8) 4.6 Static settling, sand 100 % < 0.0003 2.1 – - - filtration (n = 8) – 0.042 3.4 SPAIN - 3 treatment plants (data combined) (Corsaro et al., 2010) Surface and ground 0.015- Naegleria, Acanthamoeba, (29) - 3 source water 4.6 Hartmannella Filtration/clarification (5) - - 1 Naegleria biofilm > Clarification sludge (7) - > 2 Naegleria, Hartmannella 1.1

35 Chapter 2. Literature review Sand/GAC filtered 0.0006- 0.9 - (33) - - water 0.0017 3.9 Finished water (12) - - 1 Hartmannella SPAIN – 1 treatment plant (Corsaro et al., 2009) (data included in (Corsaro et al., 2010) 75 % 0.21 - Acanthamoeba, Echinamoeba, River source water - 4 (8) 4.6 Naegleria, Vannella 0 % Ground source water - - - (8) 25 % >0.001 - 1.6 – Post sand filtration 2 Echinamoeba, Hartmannella (8) 0.005 3.7 0 % Ozonation < 0.001 - - (1) 13 % <0.01 - -1 – GAC filtration 1 Vannella (8) 0.03 -1.7 Chlorinated finished 0 % < 0.001 1 – 1.4 - water (8) GERMANY – 6 treatment plants (two treatment plant types) (Hoffmann and Michel, 2001) Reservoir source 100 % 0.002- - 11 Acanthamoeba, Naegleria water (15) 3 Coagulation, 40 % 0.3 – 0-0.3 6 - filtration (15) 3.5 100 % 0.2- River source water - 17 Naegleria, Acanthamoeba (11) 90 Coagulation, 100 % 0.002 – 1.3 – 8 Naegleria, Acanthamoeba sedimentation, (6) 30 4.6 Groundwater 82 % ND- - 5 Naegleria, Acanthamoeba infiltrate (11) 3 80 % Filtration ND -0.4 0–3.5 6 Naegleria, Acanthamoeba (10) Chlorinated finished 33 % ND -0.1 0–2.6 4 Hartmannella water (21) KOREA – 3 treatment plants (data combined) (97) (Jeong and Yu, 2005) 100 % Source water - - >2 Acanthamoeba (3) 100 % Precipitation - - >2 Acanthamoeba (3) 66 % Sand filtration - - >2 Acanthamoeba (3) 0 % Carbon filtration - - - - (3) 0 % Finished water - - - - (3) a Calculation used; log removal = -log(concout/concin) (Asano et al., 2007). *Granular activated carbon (GAC) ^ Chlorination data is aggregated for all water treatment plants in the study. ND - no detects of FLA in the samples

36 Chapter 2. Literature review 2.6.3 FLA in drinking water distribution systems Drinking water distribution systems consist of the distribution system (pipes) and storage reservoirs for the treated water. From the review of drinking water treatment plants FLA are known to enter treated drinking water distribution systems (Thomas et al., 2008) yet only eight of the 29 studies identified examined FLA in distribution systems and/or treated reservoirs (Table 2.8). Overall, the frequency of FLA isolation ranged from 10 % to 100 % for all distribution systems indicating that FLA are a pervasive resident. However, the highest detection frequencies are likely due to small sample sizes (n = 2) (Hoffmann and Michel, 2001) and sensitive molecular detection method (qPCR) (Puzon et al., 2009), while the lowest detection (10 %) could be due to the use of culture based methods (Corsaro et al., 2009). Three studies were identified where FLA were reported to enter the distribution system from the water treatment plant then persist at low densities < 0.005 amoebae.mL-1 (Thomas et al., 2008) and < 0.001 amoebae.mL-1 (Corsaro et al., 2009) and even grow (Hoffmann and Michel, 2001). Interestingly, these same three studies reported slightly higher diversities of FLA in the distribution system than detected in the finished water entering the distribution systems (Thomas et al., 2008). These small increases in diversity may indicate that cumulatively a greater number of FLA enter a distribution system than identified in the finished water, or alternatively FLA may be intruding post treatment through storages and mains breaks or leaks (Besner et al., 2008). More research utilising molecular typing techniques is needed to determine what portion of the FLA population present is due to intrusions. Dead-end legs in the distribution system appear to also allow greater FLA replication (Corsaro et al., 2009). Temperature could also play a role with the highest FLA densities (0.8 amoebae.mL-1) being recorded in summer which were significantly higher than the autumn sampling from the same point (Valster et al., 2009). How the FLA colonize the water pipes is not well understood but pipe biofilm is thought to be an integral aspect (Storey et al., 2004). Biofilms form on the internal surfaces of water systems and represent the sessile microbial community (Costerton et al., 1994) where FLA are thought to attach and feed (Rodriguez-Zaragoza, 1994). Only three studies were identified that sampled biofilms (Puzon et al., 2009; Valster et al., 2009; Corsaro et al., 2010) and one for sediment (Corsaro et al., 2010). Biofilm densities using qPCR for N. fowleri were 1 - 432 amoebae.10-7 bacteria (Puzon et al., 2009) while in the other study only one biofilm sample was positive for H. vermiformis (0.43 amoebae.cm-2) (Valster et al., 2009). Additionally, distribution system sediment was identified to contain FLA at greater density and diversity than biofilm in the same system (Corsaro et al., 2010). From this preliminary work it can be seen that FLA are readily isolated from pipe biofilm in potentially high densities and that sampling solely from the bulk water may result in underestimation of FLA density and diversity. Nonetheless, more work 37 Chapter 2. Literature review is needed to understand the full FLA community structure in biofilm; particularly as it has been proposed that FLA may release from the biofilm in large numbers (Storey et al., 2004) and colonize downstream regions of the water system (Costerton, 2004). More research is also needed to determine what factors influence FLA colonization, density and diversity in water distribution systems especially with respect to the role of biofilm microbiota. Treated drinking water reservoirs were the other part of the distribution system examined in the reported studies (Amblard et al., 1996; Hoffmann and Michel, 2001; Corsaro et al., 2009). Overall, FLA were isolated at higher frequencies (> 79 %) compared to the distribution systems samples (> 10 %) (Table 2.8). Also comparative densities were also higher with one study noting that the maximum number of FLA detected (4.5 amoebae.mL-1) in the reservoir was five times that measured in the distribution systems supplying it, despite the reservoir being emptied and cleaned prior to the start of the sampling period (Amblard et al., 1996). Similarly within another reservoir sediment samples had higher FLA densities (1 - 10 amoebae.mL-1) and diversity (Acanthamoeba, Hartmannella and Naegleria) than biofilm (0.01 - 0.1 amoebae.mL-1, Acanthamoeba and Naegleria) which in turn was greater than the distribution system supplying it (<0.001 amoebae.mL-1) (Corsaro et al., 2009). The accumulation of sediment and biofilm in drinking water reservoirs could partly explain the high frequency of FLA detection at these points in the distribution system. There appears to exist a complex microbial community within drinking water reservoirs with one research group finding that FLA densities did not show the same fluctuations as bacteria and flagellated protozoa but rather decreased in the warmer months due to predatory rotifer and crustacea populations (Amblard et al., 1996). This is consistent with a tap water study that reported no significant association between heterotrophic water bacteria presence and FLA (Seal et al., 1992). FLA populations were thought to be linked to bacteria as bacteria are their primary food source (Rodriguez-Zaragoza, 1994) but prey selectivity, and sessile (biofilm) versus planktonic (bulk water) habit make this relationship more complicated for treated drinking water systems. An additional factor that appears to have an effect on FLA densities and diversities is the length of the distribution system. One study reported higher densities and diversities of FLA with increasing water age in reservoirs that were further from the water treatment plant (Hoffmann and Michel, 2001). This indicates that changes in the water quality parameters, such as decrease in disinfectant residual and associated biofilm growth were facilitating re-growth and that reservoirs could be a site for further ingress of FLA. This is consistent with other research on total protozoa numbers in water distribution system biofilms that reported an increase in FLA and micro-flagellates as the length of the distribution system increased (70 - 2.9  103 cm-2) (Långmark et al., 2007). This increase was shown to correlate to 38 Chapter 2. Literature review the corresponding increase in residence time (<1 to >110 h) and decrease in chlorine residual (~0.25 to ~ 0.04 mg.L-1) (Långmark et al., 2007). It is possible that the distribution system biofilms are providing protection to the FLA as it has been reported that FLA are not removed from biofilm in a pilot drinking water system using common disinfection agents including; chlorine, monochloramine and copper/silver (Thomas et al., 2004). Overall, the literature reviewed highlights that distribution system reservoirs appear to be a point for FLA re-growth while distribution system pipes accommodated both persistence and growth. More research is needed to determine what factors contribute to these FLA populations and the extent to which they are infected with pathogenic ARM. Table 2.8 The detection of FLA in treated water distribution systems.

-1

FLA genera identified Location Reference in order of frequency

Density Diversity % positive (# samples) # genera amoebae.mL DISTRIBUTION SYSTEMS 50 % (Thomas et al., France  0.005 2 Acanthamoeba (n = 12) 2008) 50 % (Amblard et al., France ND - 0.5 - - (n = 14) 1996) ~ 100 % > ND - (Hoffmann and Germany > 4 Acanthamoeba (n = 2) 0.1 Michel, 2001) Hartmannella vermiformis* 50 %* ND - (Valster et al., Netherlands 4 Acanthamoeba, (n = 16) 0.8* 2009) Neoparamoeba, Echinamoeba (Corsaro et al., Spain (n = 33) < 0.003 >1 Acanthamoeba 2010) 10 % < 0.001 1 - (n = 10) (Corsaro et al., Spain 40 % end 0.01 - 2009) 2 Naegleria, Stenamoeba (n = 5) 0.1 13 %* (Bonilla-Lemus et Mexico - 1* Acanthamoeba* (n = 27) al., 2010) 100 %* <0.004 - Australia 1* Naegleria fowleri* (Puzon et al., 2009) (n = 6) 0.3* RESERVOIRS 79 % ND - (Amblard et al., France - - (14) 4.5 1996) ~ 100 % > ND - (Hoffmann and Germany > 4 Acanthamoeba (7) 0.1 Michel, 2001) * Detection of a single FLA genera or species ND - no detects of FLA in the samples

39 Chapter 2. Literature review 2.6.4 FLA at in-premise point of use In the literature reviewed, 18 studies across 13 different countries were identified that sampled from in-premise plumbing at the point of use (Table 2.9). For all studies the point of use was a tap within a dwelling except two studies where a toilet cistern (Shoff et al., 2008) and storage tanks (Bonilla-Lemus et al., 2010) were sampled. Overall, FLA were always detected in tap water (Table 2.9), with only one group of samples supplied by desalinated water in Israel not positive for the targeted FLA; Acanthamoeba polyphaga (Kahane et al., 2004). This same study detected A. polyphaga at high frequencies (92 %) from taps fed with treated surface water in the same city (Kahane et al., 2001), which indicates that the desalinization treatment process and/or the water produced are somewhat inhibitory to A. polyphaga growth and may present a yet unexplored control measure for FLA as it would effectively remove cysts from the source water. The mean frequency of FLA isolation across all the identified studies was 46 % (n = 17) with a very large standard deviation ( = 28); or 53 % (n = 9,  = 29) when studies that only focused exclusively on Acanthamoeba and Naegleria spp. were excluded. The data spread may be partly due to different methods used with the highest detection frequencies and densities reported for novel culture methods (100 % and 0.005 amoebae.mL-1) (Barbeau and Buhler, 2001) and membrane enzyme immunoassay (MEIA) (92 %) (Kahane et al., 2001). Similarly, the lowest frequencies of isolation of 10 % (Boost et al., 2008) and 8 % (Houang et al., 2001) which occurred when only Acanthamoeba spp. were targeted. The studies that reported on FLA diversity identified five different FLA genera on average (n = 8,  = 3). The most frequently identified genera in tap water was Acanthamoeba, although it should be noted that these studies mostly used morphology to identify FLA. As Acanthamoeba spp. have a distinctive polyhedral double walled cyst (Visvesvara, 1991) they are more readily identified which possibly skewed the data reported. The next three most frequently identified genera in order were Hartmannella, Vahlkampfia and Vannella (Table 2.9). FLA density and diversity at point of use was influenced by seasonal temperature variations (Hoffmann and Michel, 2001; Marciano-Cabral et al., 2010) and also increased during the summer months (Carlesso et al., 2010). Similarly, the diversity of FLA differed between the seasons. Specifically in one study Naegleria spp. were only isolated during the summer months while Acanthamoeba spp. were isolated throughout the year, including temperatures as low as 0.5 °C (Hoffmann and Michel, 2001). Meanwhile, Hartmannella vermiformis was detected in hospital hot water tanks that ranged in temperature between 42 - 52 °C (Fields et al., 1989). Hence, this analysis reveals that these FLA genera are an integral part of the water distribution system, but it is likely that less readily enriched/identified genera are under reported, and all require more attention with respect to their ability to host pathogenic ARM. 40 Chapter 2. Literature review The large variation in both the detection frequency and diversity of FLA at in-premise taps (Table 2.9), indicates that other factors aside from methods are responsible for the variation. One such variable is the practice of storing treated drinking water in tanks prior to distribution at the tap, this practice has been shown to increase the frequency of FLA detection in tap water (Bonilla-Lemus et al., 2010) as well as FLA diversity (Seal et al., 1992; Shoff et al., 2008). Specifically in the United Kingdom (UK) tap water sampling indicated a higher frequency of FLA isolated (48 %, n = 50) from bathroom taps supplied with mains water that had been stored in a tank in the roof prior to use than the direct mains-fed taps in the kitchen (26 %, n = 50) (Seal et al., 1992). Seal et al. (1992) proposed that the FLA were proliferating in the tanks in the roof, as the tanks were rarely cleaned and were not covered or well maintained (Seal et al., 1992). The UK has rates of amoebic keratitis that are 15 times higher than in the U.S. and the use of household roof tanks for storages of mains water is thought to be a contributing factor (Kilvington et al., 2004). Findings reported from Mexico support this mode of increase as Acanthamoeba spp. were detected at higher frequencies in water cisterns (49 %, n = 27) and roof tanks (22 %, n = 27) than in the mains water supplying them (13 %, n = 27). Similarly, rates of FLA detection were higher for apartment blocks with drinking water storage tanks in both Hong Kong (Boost et al., 2008) and Korea (Jeong and Yu, 2005) compared to blocks without tanks. Storage tanks have been identified as a contributing factor to FLA growth but there are likely many other factors that have yet to be identified. Globally FLA are consistently detected in treated drinking water at the point of use. The variation in the frequency of detection indicates that there are a number of individual water system variables that influence FLA persistence and growth. However the recording and correlation of water system variables with FLA has been very poorly reported and severely limits any deductions that can be drawn. FLA appear to be ubiquitous in treated drinking water but significantly more research is needed in order to determine the interactions with pathogenic ARM in situ before any risk to human health can be quantified. Table 2.9 The detection of FLA in treated drinking water at point of use

-1

FLA genera identified Location Reference in order of frequency

Density Diversity % positive (# samples) # genera amoebae.mL TAP WATER 18 % Hartmannella (Tsvetkova et al., Bulgaria - >2 (n = 60) Acanthamoeba 2004) 66 % > ND - Acanthamoeba, (Hoffmann and Germany > 4 (n = 3) 0.1 Hartmannella Michel, 2001)

41 Chapter 2. Literature review 58 %* (Lanocha et al., Poland - 1* Acanthamoeba* (n = 31) 2009) 60 %* (Lorenzo-Morales et Spain - 1* Acanthamoeba* (n = 148) al., 2005) Acanthamoeba, United 89 % Hartmannella, (Kilvington et al., - 5 Kingdom (n = 27)^ Naegleria, Vahlkampfia, 2004) Vannella Acanthamoeba, 48 % Tank United Hartmannella, (n = 50) Kingdom - 7 Vahlkamfia, Vannella, (Seal et al., 1992) 26 %Mains (London) Platyamoeba, Filamoeba (n = 50) Nuclearia 92 %* Israel ( n= 26) - >1* Acanthamoeba polyphaga* (Kahane et al., 2004) 0 %Desal Vannella, Hartmanella, 100 % 0.94 – (Barbeau and Buhler, Canada 5 Vahlkampfia, (n = 18) 5.3 2001) Acanthamoeba, Naegleria 49%Cistern* (Bonilla-Lemus et Mexico 22 %Tank* - 1* Acanthamoeba* al., 2010) (n = 27) Vahlkampfia, (Marciano-Cabral et U.S.A. - - 3 Acanthamoeba, Naegleria al., 2010) Vexillifera, Hartmannella, Acanthamoeba, Vahlkampfia, Vannella, U.S.A. 19 %Toilet - 12 Cochliopodium, (Shoff et al., 2008) (Florida) (n = 283) Limax, Platyamoeba, Mayorella, Echinamoeba, Parvamoeba, Saccamoeba Hartmannella, U.S.A. 70 % - 4 Acanthamoeba, (Sanden et al., 1992) (Atlanta) (n = 207)# Vahlkampfia, Rosculus 23 %* Acanthamoeba,* Nicaragua - 2* (Leiva et al., 2008) (n = 74) Naegleria* 36 %* (Lorenzo-Morales et West Indies - 1* Acanthamoeba* (n = 180) al., 2005) 35 % (Carlesso et al., Brazil - >1* Acanthamoeba* (n = 135)# 2010) Hong Kong, 10 %* - 1* Acanthamoeba* (Boost et al., 2008) China (n = 100) Hong Kong, 8 %* - >1* Acanthamoeba* (Houang et al., 2001) China (n = 90) 47 % (Jeong and Yu, Korea - >1* Acanthamoeba* (n = 207) 2005) *data only reported for a single FLA genera or species ^known sampling bias for certain FLA # data aggregated and some non-drinking water samples ND - no detects of FLA in the samples

42 Chapter 2. Literature review 2.6.5 FLA in hospital hot water systems There is evidence of significant association between FLA and pathogenic Legionella spp. in hospital water systems (Breiman et al., 1990; Thomas et al., 2006). For this reason it is critical to examine the density and diversity of FLA in these systems to determine if control measures are required. Only four studies from three countries were identified that reported on FLA in hospital hot water systems (Table 2.10). Part of one of these studies was undertaken after a Legionella outbreak in the hospital (Breiman et al., 1990) while no outbreak association was recorded in the other studies. All the studies used traditional culture methods to detect and identify the FLA which is likely to result in an underestimation of both the frequency of detection and diversity. The frequency of detection of FLA ranged from 12 - 52 % with an average of 35 % ( = 17). The reduced detection of FLA in Swiss samples (12 %) was linked to the higher mean temperature of the water system (56 °C) (Thomas et al., 2006). None of the studies identified reported on the density of FLA, which presents a noteworthy gap in the current understanding of FLA populations in these water systems. The highest diversity of FLA recorded was six genera (Breiman et al., 1990), with Hartmannella spp. being the most frequently isolated. The dominance of Hartmannella spp. in the hot water samples may be explained by its higher thermo-tolerance, with growth observed at 53 °C (Rohr et al., 1998). Hartmannella spp. have already been shown to host pathogenic ARM from the genera Neochlamydia (Thomas et al., 2008) and Legionella (Thomas et al., 2006) and consistent presence in hospital hot water systems needs to be explored further in order to calculate the risk that they present to the susceptible patient populations. Table 2.10 The detection of FLA in hospital hot water systems

-1

FLA genera identified - Location Reference in order of frequency

Density # genera Diversity % Positive (# samples) amoebae.mL HOSPITAL HOT WATER SYSTEMS Hartmannella, 52 % Echinamoebae, (Rohr et al., Germany - 4 (n = 56) Saccamoebae, 1998) Vahlkampfia

12 % Hartmannella, (Thomas et al., Switzerland - 3 (n = 26) Acanthamoeba, 2006) Unknown eukaryotic Hartmannella 43 % (Fields et al., U.S.A. - 3 Echinamoeba (n = 14) 1989) Acanthamoeba

43 Chapter 2. Literature review

45 % O Hartmannella (n = 75) Filamoeba Comandonia (Breiman et al., U.S.A. - 6 24 % Acanthamoeba 1990) (n = 33) Vahlkampfia Paratetramitus O Outbreak of Legionnaires' disease or Pontiac Fever associated with the sample hospital

2.7 LEGIONELLA AND FLA IN RECYCLED WATER SYSTEMS In the literature only one study was identified that sampled from treated recycled water systems. Recycled distribution systems were sampled at various locations across Australia for water and biofilm (Storey and Kaucner, 2009). The results reported were for recycled water that was treated to a standard required for non-drinking purposes such as irrigation (Storey and Kaucner, 2009) and this should be kept in mind when comparing the results to treated drinking water. Legionella spp. other than L. pneumophila (sero gps 1-4), were detected by culture in recycled water distribution system pipes at densities of 3.6 cells.mL-1 in water and 16 cells.cm-2 in biofilms (Storey and Kaucner, 2009). The frequency of FLA detection was not reported in this study but densities were. FLA were detected in the recycled water distribution system bulk water at a maximum density of 6 amoebae.mL-1 and in the biofilm at 60 amoeba.cm2 (Storey and Kaucner, 2009). Relative to drinking water samples (1.5 amoebae.mL-1) taken in the same study the recycled water densities are four times higher. Also the maximum recycled FLA density is also higher than any of the other treated drinking water studies reviewed (max 5.3 amoeba.mL-1) (Barbeau and Buhler, 2001). Equally, the recycled biofilm densities were higher than those reported for drinking water biofilm (28 amoebae.cm-2) in the same study. It must be emphasised that higher microbial loads are permitted in the recycled water sampled as it was only used for irrigation purposes (Asano et al., 2007). Despite this the maximum recycled FLA densities were only slightly higher than the treated drinking water densities. As the differences were not in the scale of orders of magnitude it is proposed that with greater treatment the recycled water FLA densities could easily be reduced to comparable levels to those observed in treated drinking water. More research is urgently needed to determine FLA presence in recycled water treated to a drinking water standard. The scant information on the diversity and density of FLA in treated recycled water is a significant gap that needs to be filled. In particular information about the effectiveness of different disinfection types on FLA in recycled water. It is critical that this data gap be addressed in order to determine the relative safety of recycled water for both drinking and non-drinking

44 Chapter 2. Literature review purposes. This is especially important, as due to water scarcity recycled water is being used more extensively on a global scale for both drinking and non-drinking purposes (Asano et al., 2007). 2.8 LEGIONELLA AND FLA IN APPLICATIONS OF DRINKING WATER Applications of drinking water relate to equipment and facilities that are frequently supplied by treated drinking water. These applications most commonly include pools, spas, humidifiers, air-conditioning units and cooling towers. Water is recirculated or retained in these applications and sometimes subject to additional treatment (Hill et al., 1990) which makes them distinct from drinking water supply systems. Spas and cooling towers produce aerosols which present a common exposure pathway for Legionella infection (Armstrong, 2005), which is why number of fatal Legionella outbreaks have been traced to infected cooling towers (Barbaree et al., 1986; Anonymous, 2000) and spas (Fallon and Rowbotham, 1990; Okada et al., 2005).

2.8.1 Legionella spp. in applications of drinking water A selection of the literature is presented here which reports on Legionella spp. in two pools and spas and eight cooling towers across eight different countries (Table 2.11). Three of the studies are likely to have positive bias for Legionella spp. detection as sampling was undertaken after a known Legionella outbreak (Barbaree et al., 1986; Fallon and Rowbotham, 1990; Okada et al., 2005). Traditional culture methods, which are known to underestimate both Legionella density and diversity were used exclusively in three studies (Barbaree et al., 1986; Bentham, 1993; Bentham, 2000). Amoebae co-culture, which is increasing in its popularity for the detection of ARM, was identified as a reliable means of isolating Legionella spp. by culture where traditional culture methods fail (Fallon and Rowbotham, 1990). One study used culture and PCR (Miyamoto et al., 1997) while five of the most recent studies solely used highly sensitive qPCR (Okada et al., 2005; Joly et al., 2006; Declerck et al., 2007; Wéry et al., 2008; Chen and Chang, 2010). The frequency of Legionella spp. had high positive detects of 100 % for pool and spa samples, although there is positive bias here due to the recent outbreaks linked to the locations sampled (Fallon and Rowbotham, 1990; Okada et al., 2005). The highest density of FLA reported was 1.5  104 cells.mL-1 which correlated with the highest diversity of FLA at greater than four genera (Okada et al., 2005). There was also a considerable diversity of FLA isolated simultaneously from the same samples indicating an association between the two (Fallon and Rowbotham, 1990; Okada et al., 2005). These studies highlight how under the right conditions

45 Chapter 2. Literature review pathogenic Legionella spp. can replicate to high densities with dire health consequences; 109 people hospitalised and seven deaths (Okada et al., 2005). In the literature selected Legionella spp. were detected in cooling towers at a frequency that ranged from 30 - 100 % with an average of 88 % (n = 8,  = 23). Temperature was a factor in the lowest detection frequency which occurred over winter, where below 16.5 °C there were no Legionella spp. detected by culture in any of the 31 Adelaide cooling towers sampled (Bentham, 1993). The density of Legionella spp. detected ranged from no detects to 2.5  103 cells.mL-1 with the highest densities being reported by studies using qPCR methods (Wéry et al., 2008; Chen and Chang, 2010). In one study Legionella numbers were correlated with the higher temperature of the ambient air and basin as well as the high frequency of cooling tower use (Bentham, 1993). The highest diversity of Legionella spp. reported was only three species (Wéry et al., 2008) but L. pneumophila was detected in all studies revealing the pervasiveness of this pathogen. Another interesting finding was the presence of a number of bacteria that were most closely related to the broad group of Legionella-like amoebal pathogens (LLAP) (Wéry et al., 2008). These LLAPs are likely to represent yet unidentified Legionella spp. (Pagnier et al., 2009) and have already been associated with human disease (Marrie et al., 2001). There appears to be a dynamic population of Legionella spp. in cooling towers with one study reporting that during 16 weeks of Legionella sampling by culture over summer it was found that samples taken two weeks apart were not statistically related (Bentham, 2000). Also an examination of the population dynamics of Legionella spp. in cooling towers revealed that L. pneumophila populations fluctuated over time and when this species increases the dominant species of other Legionella spp. reduce in number (Wéry et al., 2008). It has been strongly recommended that single or a limited number of Legionella tests may underestimate or overestimate the risk of Legionella and should not be relied upon for inquiry (Bentham, 2000). In addition to water samples biofilm samples were identified as an integral part of the location of Legionella spp. in cooling towers. A survey of Legionella spp. in ten cooling towers found that in attached and floating biofilm Legionella spp. were detected in the same orders of magnitude (102 cells.g-1 or cells.cm-1) as in the associated water (Declerck et al., 2007). An additional study described that surface biofilm was found to have a higher densities of Legionella spp. than water in cooling towers using qPCR (Chen and Chang, 2010). Biofilms are an important component of water systems especially due to the higher association with FLA. More research is needed to determine the association of FLA and Legionella spp. in reacreational pools, spas and cooling towers.

46 Chapter 2. Literature review Table 2.11 The detection of Legionella spp. in pools, spas and cooling towers

-1 Legionella spp. identified - Location Reference in order of frequency

Density Diversity cells.mL # species % positive (# samples) POOLS AND SPAS (Fallon and 100 % O U.K. 4 - 80 1* L. micdadei Rowbotham, (n = 6) 1990) L. londiniesis 100 % O  L. dumoffii (Okada et al., Japan > 4 (n = 55) 1.5  104 L. pneumophila 2005) Legionella spp. COOLING TOWERS 100 %# L. pneumophila (Declerck et Belgium 0.1 - 0.5 2* (n = 10) Legionella spp. al., 2007) 94 % ND - L. pneumophila (Joly et al., France 2* (n = 36) 30 Legionella spp. 2006) L. pneumophila 100 % <2 - L. fallonii (Wéry et al., France >3 (n = 40) 2.5  103 L. lytica 2008) Legionella-like amoebal 100 %O (Barbaree et U.S.A - 1* L. pneumophila (sero gp 1, 3, 5) (n = 4) al., 1986) L. pneumophila (sero gp 1) 75 %summer L. anisa (Bentham, Australia 30 %winter - 3 L. rubrilucens 1993) (n = 31) L. pneumophila (sero gp 2 -14) 2 - L. pneumophila (Bentham, Australia 100 %^  2* 2  103 Legionella spp. 2000) 92 %Hospital ND - L. pneumophila (Miyamoto et Japan  2* (n = 49) 100 Legionella spp. al., 1997) 100 % L. pneumophila (Chen and Taiwan 2.5  103 2* (n = 10) Legionella spp. Chang, 2010) O Outbreak of Legionnaires' disease or Pontiac Fever associated with sampled site * Data only reported for a single Legionella species ^ Cooling towers targeted known to contain Legionella spp. # Cooling towers were non-operational at time of sampling. 2.8.2 FLA in applications of drinking water A review of the literature identified only six studies that reported FLA in pools, spas and cooling towers in four different countries (Table 2.12). All of the studies used traditional culture methods to isolate FLA. However two of the studies also used molecular techniques to detect and identify FLA such as qPCR (Behets et al., 2007) or a comprehensive three tier approach consisting of enzyme-linked immunosorbent assay, PCR and isoenzyme electrophoresis

47 Chapter 2. Literature review (Declerck et al., 2007). The limited number of studies utilising molecular detection techniques has likely resulted in an underestimation of the density and diversity of FLA in applications of drinking water. For pools and spas the frequency of detection ranged from 43 - 100 %, with the highest detection frequency associated with a Legionella outbreak (Fallon and Rowbotham, 1990). The diversity of FLA was greater than three genera (Esterman et al., 1987). Using a model based on FLA data from spas it was found that free-chlorine residual needed to be at least 3.5 mg.L-1 in order to be 95 % sure that no FLA would be detected (Esterman et al., 1987). There was also found to be a correlation between increased spa temperature and reduced detection of FLA (Esterman et al., 1987). More studies are needed to determine how common FLA colonisation of pools and spas really is and any associations with pathogenic ARM. From the four studies that reported FLA in cooling towers the frequency of isolation ranged from 50 - 100 % with an average of 85 % ( = 23). Only one study reported FLA density with a maximum detection of 16 amoebae.mL-1 using sensitive qPCR techniques (Behets et al., 2006). The highest FLA diversity was five genera (Declerck et al., 2007) while Naegleria spp. were consistently isolated from the cooling towers probably due to their thermo-tolerance (Sykora et al., 1983). Furthermore, there was a significant positive correlation found in cooling towers between Naegleria fowleri and temperature but not Acanthamoeba spp. or Hartmannella spp. (Behets et al., 2007). FLA are not the only protozoa present in cooling towers with ciliates (Tetrahymena sp. and Cyclidium sp.) also isolated in some samples (Barbaree et al., 1986). As with other water samples FLA were also found in floating and attached biofilms along with Legionella spp. in non - operational cooling towers (Declerck et al., 2007). There exists a complex microbial ecosystem in cooling towers and other applications of drinking water where recirculation, elevated temperatures, insufficient disinfection and periods of inactivity can all facilitate the growth of FLA and their pathogenic ARM. More research is needed to determine what risks FLA and ARM present in drinking water applications and how to control them. This is especially important as water scarcity leads people to utilise alternative water sources, such as recycled water, for non-drinking purposes such as cooling towers.

48 Chapter 2. Literature review Table 2.12 The detection of FLA in pools, spas and cooling towers

-1

FLA genera identified Location Reference in order of frequency

Density # genera Diversity % positive (# samples) amoebae.mL POOLS AND SPAS UK Acanthameoba 100 %O (Fallon and culture - co- - 3 Hartmannella (n = 12) Rowbotham, 1990) culture Vannella Australia 43 % Acanthamoeba (Esterman et al., - > 3 culture (n = 98) Willertia 1987) Naegleria COOLING TOWERS Acanthamoeba, Naegleria, Belgium 90 %# (Declerck et al., - 5 Willaertia, Vahlkampfia culture - PCR (n = 10) 2007) Hartmanella Belguim 98 % Naegleria, Hartmannella ND - 16 3 (Behets et al., 2007) (culture,qPC (n = 123) Acanthamoeba U.S.A 50 %O (Barbaree et al., - 1 Naegleria culture (n = 4) 1986) U.S.A. 100 % - - - (Berk et al., 2006) culture (n = 40) O Outbreak of Legionnaires' disease associated with the sampled cooling towers # Cooling towers were non-operational at time of sampling ND - no detects of FLA in the samples 2.9 CONCLUSIONS 2.9.1 Theoretical model of FLA population in drinking water systems Based on the literature reviewed a theoretical model for FLA and ARM infection in treated drinking water systems was developed (Figure 2.1). Three key stages in the system have been identified where FLA and ARM interactions are likely to occur. The first stage is at the start of the distribution system where low FLA densities and diversities are anticipated due to higher disinfection residuals and low microbial diversity. Further along the distribution system pipeline is the second stage where storage sites and increasing microbial diversity due accumulation and growth as well as breaks in the distribution system integrity could also allow an increase in the density and diversity of FLA and their respective ARM. The final stage of the model is within the in-premise plumbing or applications of drinking water where increases in temperature, low disinfection residuals, large biofilm surface area and stagnation facilitate the increased FLA growth and interactions with ARM. People are exposed to these infective CAP causing ARM when they inhale aerosols during showering, spas or the breathing of air cooled by cooling towers. However, more research is critically needed to better understand actual ARM

49 Chapter 2. Literature review infected FLA along the length of a drinking water distribution systems in-order for this preliminary model to be further developed and applied to risk predictions and control strategies.

50 Figure 2.1 A theoretical model for FLA density and diversity including possible ARM infection from the end of the drinking water treatment plant to the in-premise tap. The schematic represents a cross section of drinking water pipe where the direction of water flow is from left to right with the dotted vertical lines representing transition to a distal section. Green deposits on the pipe surface are biofilm. Grey shapes and circles represent FLA in trophozoite and cyst form, respectively. Pathogenic ARM are represented as coloured lines and different colours indicate a different types of ARM.

2.9.2 Research needs to identify the risks of FLA associated with ARM This review has highlighted that FLA are isolated at high frequencies, density and diversity from treated drinking water systems globally but at present it is very difficult to quantify the health risk that they may present. A number of researchers in the field have already recommended that FLA need to be considered in any risk assessment for drinking water pathogens (Breiman et al., 1990; Szewzyk et al., 2000; Storey et al., 2004; Bichai et al., 2008; Loret et al., 2008; Thomas et al., 2008; Loret and Greub, 2010; Thomas et al., 2010) and proactively targeted for FLA reduction in order to lower drinking water pathogen numbers (Srikanth and Berk, 1993; Critchley and Bentham, 2009; Loret and Greub, 2010). The current regulatory focus on waterborne (fecal-oral route of transmission) pathogens along with a lack of understanding of the relative merits of FLA control in distribution systems (water utility responsibility) versus in-premise (currently not water utility responsibility) has left ARM pathogens and their diseases poorly controlled. This literature review has identified large gaps in knowledge about the density and diversity of FLA populations in treated drinking water. In particular it is not well understood what effects drinking water quality characteristics or system variables have on the density and diversity of FLA populations. There is also an absence of peer-reviewed data looking at FLA in recycled water systems. More importantly the significant lack of information about FLA and water-based pathogenic ARM interactions within the complex environments of treated drinking and recycled water systems needs to be addressed, given the recently recognized high health burden caused by just one ARM, Legionella pneumophila (Craun et al., 2010). Research needs to be directed towards determining the engineering/ecologic factors resulting in higher densities or diversities of FLA that are likely to lead to an interaction with water-based ARM. Additionally, quantification is needed of the water-based pathogenic ARM densities, growth characteristics and virulence that may be FLA facilitated. Once these information gaps are filled then extensive quantitative risk assessments can be undertaken and control strategies can be developed (Nwachuku and Gerba, 2004; Storey et al., 2004; Thomas et al., 2010). Yet even simplistic quantitative microbial risk assessments (QMRA) have demonstrated that FLA increase the risk of Legionella infection to unsafe levels (Storey et al., 2004).

Chapter 2. Literature review

2.10 REFERENCES 1. Abu Kwaik, Y., L. Y. Gao, B. J. Stone, C. Venkataraman and O. S. Harb. 1998. Invasion of protozoa by Legionella pneumophila and its role in bacterial ecology and pathogenesis. Applied and Environmental Microbiology. 64(9): 3127-3133. 2. Adl, S., B. Leander, A. Simpson, J. Archibald, R. Anderson, D. Bass, S. Bowser, G. Brugerolle, M. Farmer, S. Karpov, M. Kolisko, C. Lane, D. Lodge, D. Mann, R. Meisterfeld, L. Mendoza, O. Moestrup, S. Mozley-Standridge, A. Smirnov and F. Spiegel. 2007. Diversity, nomenclature, and of protists. Systematic Biology. 56(4): 684-9. 3. Aksozek, A., K. McClellan, K. Howard, J. Y. Niederkorn and H. Alizadeh. 2002. Resistance of Acanthamoeba castellanii cysts to physical, chemical, and radiological conditions. The Journal of Parasitology. 88(3): 621-623. 4. Albert-Weissenberger, C., C. Cazalet and C. Buchrieser. 2007. Legionella pneumophila - a human pathogen that co-evolved with fresh water protozoa. Cellular and Molecular Life Sciences. 64(4): 432-448. 5. Amblard, C., G. Bourdier, J. Carrias, N. Maurin and C. Quiblier. 1996. Seasonal evolution of microbial community structure in drinking water reservoir. Water Research. 30: 613-624. 6. Anderson, R. 1988. Comparative protozoology. New York, Springer-Verlag. 7. Anonymous. 2000. Outbreak of Legionnaires' disease associated with an aquarium in Australia. Communicable Disease Report Weekly. 10: 161. 8. Armstrong, T. 2005. A quantitative microbial risk assessment model for human inhalation exposure to Legionella. Philadelphia, Drexel University Doctor of Philosophy. 9. Asano, T., F. Burton, H. Leverenz, R. Tsuchihashi and G. Tchobanoglous. 2007. Water reuse issues, technologies and applications. New York, McGraw Hill 10. Axelsson-Olsson, D., J. Waldenstrom, T. Broman, B. Olsen and M. Holmberg. 2005. Protozoan Acanthamoeba polyphaga as a potential reservoir for Campylobacter jejuni. Applied and Environmental Microbiology. 71(2): 987-992. 11. Badenoch, P. R., T. R. Grimmond, J. Cadwgan, S. E. Deayton, M. S. L. Essery, Hill and B. D. 1988. Nasal carriage of free-living amoebae. Microbial Ecology in Health and Disease. 1(3): 209-211. 12. Barbaree, J. M., B. S. Fields, J. C. Feeley, G. W. Gorman and W. T. Martin. 1986. Isolation of protozoa from water associated with a legionellosis outbreak and demonstration of intracellular multiplication of Legionella pneumophila. Applied and Environmental Microbiology. 51(2): 422-424. 13. Barbeau, J. and T. Buhler. 2001. Biofilms augment the number of free-living amoebae in dental unit waterlines. Research in Microbiology. 152(8): 753-760. 14. Barker, J., M. R. Brown, P. J. Collier, I. Farrell and P. Gilbert. 1992. Relationship between Legionella pneumophila and Acanthamoeba polyphaga: physiological status and susceptibility to chemical inactivation. Applied and Environmental Microbiology. 58(8): 2420-2425. 15. Barker, J. and M. R. W. Brown. 1994. Trojan Horses of the microbial world: protozoa and the survival of bacterial pathogens in the environment. Microbiology. 140(6): 1253-1259. 16. Baudin, I., M. Jousset, C. Debeaupuis, C. Schlumberger, J. Dorko and L. Debreczeny. 2008. Decontamination of filtration media for controlling release of organisms. Water Quality Techonology Conference. Cincinnati, Ohio, American Water Works Association. 17. Behets, J., P. Declerck, Y. Delaedt, L. Verelst and F. Ollevier. 2006. Quantitative Detection and Differentiation of Free-Living Amoeba Species Using SYBR Green– Based Real-Time PCR Melting Curve Analysis. Current Microbiology. 53(6): 506-509.

53 Chapter 2. Literature review 18. Behets, J., P. Declerck, Y. Delaedt, L. Verelst and F. Ollevier. 2007. Survey for the presence of specific free-living amoebae in cooling waters from Belgian power plants. Parasitology Research. 100(6): 1249-1256. 19. Bentham, R. H. 1993. Environmental factors affecting the colonization of cooling towers by Legionella spp. in South Australia. International Biodeterioration & Biodegradation. 31(1): 55-63. 20. Bentham, R. H. 2000. Routine sampling and the control of Legionella spp. in cooling tower water systems. Current Microbiology. 41(4): 271-275. 21. Berk, S., R. Ting, G. Turner and R. Ashburn. 1998. Production of respirable vesicles containing live Legionella pneumophila cells by two Acanthamoeba spp. Applied and Environmental Microbiology. 64(1): 279-286. 22. Berk, S. G., J. H. Gunderson, A. L. Newsome, A. L. Farone, B. J. Hayes, K. S. Redding, N. Uddin, E. L. Williams, R. A. Johnson, M. Farsian, A. Reid, J. Skimmyhorn and M. B. Farone. 2006. Occurrence of infected amoebae in cooling towers compared with natural aquatic environments: implications for emerging pathogens. Environmental Science & Technology. 40(23): 7440-7444. 23. Besner, M.-C., P. Servais and M. Prevost. 2008. Efficacy of disinfectant residual on microbial intrusion: a review of experiments. Journal American Water Works Association. 100(10): 116-130. 24. Bichai, F., P. Payment and B. Barbeau. 2008. Protection of waterborne pathogens by higher organisms in drinking water: a review. Canadian Journal of Microbiology. 54(7): 509-524. 25. Birtles, R. J., T. J. Rowbotham, C. Storey, T. J. Marrie and D. Raoult. 1997. Chlamydia-like obligate parasite of free-living amoebae. The Lancet. 349(9056): 925- 926. 26. Bomo, A.-M., T. K. Stevik and I. Hovi. 2004. Bacterial removal and protozoan grazing in biological sand filters. Journal of Environmental Quality. 33(3): 1041-7. 27. Bonilla-Lemus, P., G. A. Ramírez-Bautista, C. Zamora-Muñoz, M. d. R. Ibarra- Montes, E. Ramírez-Flores and M. D. Hernández-Martínez. 2010. Acanthamoeba spp. in domestic tap water in houses of contact lens wearers in the metropolitan area of Mexico City. Experimental Parasitology. 126(1): 54-58. 28. Boost, M., P. Cho, S. Lai and W.-M. Sun. 2008. Detection of Acanthamoeba in tap water and contact lens cases using polymerase chain reaction. Optometry and Vision Science. 85(7): 526-530. 29. Bozue, J. and W. Johnson. 1996. Interaction of Legionella pneumophila with Acanthamoeba castellanii: uptake by coiling phagocytosis and inhibition of phagosome- lysosome fusion. Infection and Immunity. 64(2): 668-673. 30. Breiman, R. F., B. Fields, G. N. Sanden, L. Volmer, A. Meier and J. Spika. 1990. Association of shower use with Legionnaires' disease. Possible role of amoebae. Journal of the American Medical Association. 263(21): 2924 - 2926. 31. Brown, T. J., R. T. Cursons and E. A. Keys. 1982. Amoebae from antarctic soil and water. Applied and Environmental Microbiology. 44(2): 491-493. 32. Carlesso, A., G. Artuso, K. Caumo and M. Rott. 2010. Potentially pathogenic Acanthamoeba isolated from a hospital in Brazil. Current Microbiology. 60(3): 185-190. 33. Chang, J. C., S. F. Ossoff, D. C. Lobe, M. H. Dorfman, C. M. Dumais, R. G. Qualls and J. D. Johnson. 1985. UV inactivation of pathogenic and indicator microorganisms. Applied and Environmental Microbiology. 49(6): 1361-1365. 34. Chappell, C. L., J. A. Wright, M. Coletta and A. L. Newsome. 2001. Standardized method of measuring Acanthamoeba antibodies in sera from healthy human subjects. Clinical and Diagnostic Laboratory Immunology. 8(4): 724-730.

54 Chapter 2. Literature review 35. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634. 36. Cirillo, J., S. Falkow, L. Tompkins and L. Bermudez. 1997. Interaction of Mycobacterium avium with environmental amoebae enhances virulence. Infection and Immunity. 65(9): 3759-3767. 37. Cirillo, J. D., S. Falkow and L. S. Tompkins. 1994. Growth of Legionella pneumophila in Acanthamoeba castellanii enhances invasion. Infection and Immunity. 62(8): 3254-3261. 38. Ciurea-Van Saanen, M. 1980. Isolation of free-living amoebae from cold blooded animals Schweizer Archiv für Tierheilkunde 122(11): 653-658. 39. Clarholm, M. 1981. Protozoan grazing of bacteria in soil-impact and importance. Microbial Ecology. 7(4): 343-350. 40. Corliss, J. O. 1973. Protozoan ecology: a note on its current status. American Zoologist. 13(1): 145-148. 41. Corliss, J. O. 2001. Two most remarkable amoeba men: Joseph Leidy (1823-1891) of Philadelphia and EugËne Penard (1855-1954) of Geneva. Protist. 152(1): 69-85. 42. Corsaro, D., V. Feroldi, G. Saucedo, F. Ribas, J.-F. Loret and G. Greub. 2009. Novel Chlamydiales strains isolated from a water treatment plant. Environmental Microbiology. 11(1): 188-200. 43. Corsaro, D., R. Michel, J. Walochnik, K.-D. M¸ller and G. Greub. 2010. Saccamoeba lacustris, sp. nov. (Amoebozoa: Lobosea: ), a new lobose amoeba, parasitized by the novel chlamydia `Candidatus Metachlamydia lacustris' (Chlamydiae: Parachlamydiaceae). European Journal of Protistology. 46(2): 86-95. 44. Corsaro, D., G. S. Pages, V. Catalan, J.-F. Loret and G. Greub. 2010. Biodiversity of amoebae and amoeba-associated bacteria in water treatment plants. International Journal of Hygiene and Environmental Health. 213(3): 158-166. 45. Costerton, J. W. 2004. A short history of the development of the biofilm concept. Microbial Biofilms. M. Ghannoum and G. O'Toole. Washington, D.C., ASM Press: 4-18. 46. Costerton, J. W., Z. Lewandowski, D. DeBeer, D. Caldwell, D. Korber and G. James. 1994. Biofilms, the customized microniche. Journal of Bacteriology. 176(8): 2137-2142. 47. Craun, G. F., J. M. Brunkard, J. S. Yoder, V. A. Roberts, J. Carpenter, T. Wade, R. L. Calderon, J. M. Roberts, M. J. Beach and S. L. Roy. 2010. Causes of outbreaks associated with drinking water in the United States from 1971 to 2006. Clinical Microbiology Reviews. 23(3): 507-528. 48. Critchley, M. and R. Bentham. 2009. The efficacy of biocides and other chemical additives in cooling water systems in the control of amoebae. Journal of Applied Microbiology. 106(3): 784-789. 49. Curds, C. R. 1973. The role of protozoa in the activated-sludge process. American Zoologist. 13(1): 161-169. 50. Day, J., U. Achilles-Day, S. Brown and A. Warran. 2007. Cultivation of algae and protozoa Manual of Environmental Microbiology. C. Hurst, R. Crawford, J. Garlandet al. Washington, DC, ASM Press. 51. De Jonckheere, J. 2002. A century of research on the amoeboflagellate genus Naegleria. Acta Protozoologica 41: 309-342. 52. De Jonckheere, J. and H. van de Voorde. 1976. Differences in destruction of cysts of pathogenic and nonpathogenic Naegleria and Acanthamoeba by chlorine. Applied and Environmental Microbiology. 31(2): 294-297. 53. De Jonckheere, J. F. D. 1991. Ecology of Acanthamoeba. Reviews of Infectious Diseases. 13: S385-S387. 54. Declerck, P. 2010. Biofilms: the environmental playground of Legionella pneumophila. Environmental Microbiology. 12(3): 557-566.

55 Chapter 2. Literature review 55. Declerck, P., J. Behets, A. Margineanu, V. van Hoef, B. De Keersmaecker and F. Ollevier. 2009. Replication of Legionella pneumophila in biofilms of water distribution pipes. Microbiological Research. 164(6): 593-603. 56. Declerck, P., J. Behets, V. van Hoef and F. Ollevier. 2007. Detection of Legionella spp. and some of their amoeba hosts in floating biofilms from anthropogenic and natural aquatic environments. Water Research. 41(14): 3159-3167. 57. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925. 58. Diederen, B., C. d. Jong, I. Aarts, M. F. Peeters and A. v. d. Zee. 2007. Molecular evidence for the ubiquitous presence of Legionella species in Dutch tap water installations. Journal of Water and Health. 5(3): 375 - 383. 59. Donlan, R., T. Forster, R. Murga, E. Brown, C. Lucas, J. Carpenter and B. Fields. 2005. Legionella pneumophila associated with the protozoan Hartmannella vermiformis in a model multi-species biofilm has reduced susceptibility to disinfectants. Biofouling. 21(1): 1-7. 60. Dykova, I., L. Hruba, M. Kostka and H. Peckova. 2009. Fliamoeba sp. isolated from hot- piping system, a host of Legionella closely related to Legionella micdadei. Acta Protozoologica. 48: 321-327. 61. Esterman, A., I. Calder, S. Cameron, D. Roder, R. Walters, P. Christy and B. Robinson. 1987. Determinants of the microbiological characteristics of spa pools in South Australia. Water Research. 21(10): 1231-1235. 62. Fallon, R. J. and T. J. Rowbotham. 1990. Microbiological investigations into an outbreak of Pontiac fever due to Legionella micdadei associated with use of a whirlpool. Journal of Clinical Pathology. 43(6): 479-483. 63. Feazel, L. M., L. K. Baumgartner, K. L. Peterson, D. N. Frank, J. K. Harris and N. R. Pace. 2009. Opportunistic pathogens enriched in showerhead biofilms. Proceedings of the National Academy of Sciences of the United States of America. 106: 16393- 16399. 64. Fenchel, T. 1987. Ecology of protozoa. The biology of free-living phagotrophic protists. Madison, Wisconsin, Science Tech. 65. Fields, B., R. Benson and E. Besser. 2002. Legionella and Legionnaires' disease: 25 years of investigation. Clinical Microbiology Reviews. 15(3): 506-526. 66. Fields, B. S., G. N. Sanden, J. M. Barbaree, W. E. Morrill, R. M. Wadowsky, E. H. White and J. C. Feeley. 1989. Intracellular multiplication of Legionella pneumophila in amoebae isolated from hospital hot water tanks. Current Microbiology. 18(2): 131-137. 67. Fritsche, T. R., R. K. Gautom, S. Seyedirashti, D. L. Bergeron and T. D. Lindquist. 1993. Occurrence of bacterial endosymbionts in Acanthamoeba spp. isolated from corneal and environmental specimens and contact lenses. Journal of Clinical Microbiology. 31(5): 1122-1126. 68. Gao, L. Y., M. Susa, B. Ticac and Y. A. Kwaik. 1999. Heterogeneity in intracellular replication and cytopathogenicity of Legionella pneumophila and Legionella micdadei in mammalian and protozoan cells. Microbial Pathogenesis. 27(5): 273-287. 69. Gast, R. J., R. W. Sanders and D. A. Caron. 2009. Ecological strategies of protists and their symbiotic relationships with prokaryotic microbes. Trends in Microbiology. 17(12): 563-569. 70. Greub, G., B. La Scola and D. Raoult. 2003. Parachlamydia acanthamoeba is endosymbiotic or lytic for Acanthamoeba polyphaga depending on the incubation temperature. Annals of the New York Academy of Sciences. 990: 628-634. 71. Greub, G. and D. Raoult. 2004. Microorganisms resistant to free-living amoebae. Clinical Microbiology Reviews. 17(2): 413-433.

56 Chapter 2. Literature review 72. Habyarimana, F., C. T. Price, M. Santic, S. Al-Khodor and Y. A. Kwaik. 2009. Molecular characterization of the Dot/Icm-translocated AnkH and AnkJ eukaryotic-like effectors of Legionella pneumophila. Infection and Immunity. 78(3): 1123-1134. 73. Harb, O. S., L. Y. Gao and Y. A. Kwaik. 2000. From protozoa to mammalian cells: a new paradigm in the life cycle of intracellular bacterial pathogens. Environmental Microbiology. 2(3): 251-265. 74. Heidtman, M., E. J. Chen, M.-Y. Moy and R. R. Isberg. 2009. Large-scale identification of Legionella pneumophila Dot/Icm substrates that modulate host cell vesicle trafficking pathways. Cellular Microbiology. 11(2): 230-248. 75. Heinz, E., Irina Kolarov, Christian K‰stner, Elena R. Toenshoff, Michael Wagner and Matthias Horn. 2007. An Acanthamoeba sp. containing two phylogenetically different bacterial endosymbionts. Environmental Microbiology. 9(6): 1604-1609. 76. Henke, M. and K. M. Seidel. 1986. Association between Legionella pneumophila and amoebae in water. Israel Journal of Medical Sciences. 22(9): 690-695. 77. Hill, G., E. Pring and P. Osborn. 1990. Cooling towers, principles and practice. London, Butterworth-Heinemann. 78. Hoffmann, R. and R. Michel. 2001. Distribution of free-living amoebae (FLA) during preparation and supply of drinking water. International Journal of Hygiene and Environmental Health. 203(3): 215-219. 79. Hoffmann, R., R. Michel, E. N. Schmid and K.-D. Müller. 1998. Natural infection with microsporidian organisms (KW19) in Vannella spp. (Gymnamoebia) isolated from a domestic tap-water supply. Parasitology Research. 84(2): 164-166. 80. Houang, E., D. Lam, D. Fan and D. Seal. 2001. Microbial keratitis in Hong Kong: relationship to climate, environment and contact-lens disinfection. Transactions of the Royal Society of Tropical Medicine and Hygiene. 95(4): 361-367. 81. Huws, S., A. Smith, M. Enright, P. Wood and M. Brown. 2006. Amoebae promote persistence of epidemic strains of MRSA. Environmental Microbiology. 8(6): 1130-1133. 82. Isberg, R. R., T. J. O'Connor and M. Heidtman. 2009. The Legionella pneumophila replication vacuole: making a cosy niche inside host cells. Nature Reviews Microbiology. 7(1): 13-24. 83. Jeong, H. J. and H. S. Yu. 2005. The role of domestic tap water in Acanthamoeba contamination in contact lens storage cases in Korea. Korean Journal of Parasitology. 43(2): 47-50. 84. Joly, P., P.-A. Falconnet, J. Andre, N. Weill, M. Reyrolle, F. Vandenesch, M. Maurin, J. Etienne and S. Jarraud. 2006. Quantitative real-time Legionella PCR for environmental water samples: data interpretation. Applied and Environmental Microbiology. 72(4): 2801-2808. 85. Kahane, S., B. Dvoskin, M. Mathias and M. G. Friedman. 2001. Infection of Acanthamoeba polyphaga with Simkania negevensis and S. negevensis survival within amoebal cysts. Applied and Environmental Microbiology. 67(10): 4789-4795. 86. Kahane, S., N. Platzner, B. Dvoskin, A. Itzhaki and M. G. Friedman. 2004. Evidence for the Presence of Simkania negevensis in drinking water and in reclaimed wastewater in Israel. Applied and Environmental Microbiology. 70(6): 3346-3351. 87. Kilvington, S., T. Gray, J. Dart, N. Morlet, J. R. Beeching, D. G. Frazer and M. Matheson. 2004. Acanthamoeba keratitis: the role of domestic tap water contamination in the United Kingdom. Investigative Ophthalmology & Visual Science. 45(1): 165-169. 88. Kilvington, S. and J. Price. 1990. Survival of Legionella pneumophila within cysts of Acanthamoeba polyphaga following chlorine exposure. Journal of Applied Microbiology. 68(5): 519-525. 89. King, C. H., E. B. Shotts, Jr, R. E. Wooley and K. G. Porter. 1988. Survival of coliforms and bacterial pathogens within protozoa during chlorination. Applied and Environmental Microbiology. 54(12): 3023-3033. 57 Chapter 2. Literature review 90. Kool, J. L. M. D., D. M. P. H. Bergmire-Sweat, J. C. M. D. Butler, E. W. Brown, D. J. M. D. Peabody, D. S. M. D. Massi, J. C. P. E. Carpenter, J. M. B. A. Pruckler, M. S. Robert F. Benson and B. S. P. Fields. 1999. Hospital characteristics associated with colonization of water systems by Legionella and risk of nosocomial Legionnaires' Disease: a cohort study of 15 hospitals. Infection Control and Hospital Epidemiology. 20(12): 798-805. 91. La Scola, B., S. Audic, C. Robert, L. Jungang, X. de Lamballerie, M. Drancourt, R. Birtles, J.-M. Claverie and D. Raoult. 2003. A giant virus in amoebae. Science. 299(5615): 2033. 92. La Scola, B., C. Desnues, I. Pagnier, C. Robert, L. Barrassi, G. Fournous, M. Merchat, M. Suzan-Monti, P. Forterre, E. V. Koonin and D. Raoult. 2008. The virophage as a unique parasite of the giant mimivirus. Nature. 455(7209): 4. 93. Långmark, J., M. V. Storey, N. J. Ashbolt and T.-A. Stenström. 2007. The effects of UV disinfection on distribution pipe biofilm growth and pathogen incidence within the greater Stockholm area, Sweden. Water Research. 41(15): 3327-3336. 94. Lanocha, N., D. Kosik-Bogacka, A. Maciejewska, M. Sawczuk, A. Wilk and W. Kuzna- Grygiel. 2009. The occurrence Acanthamoeba (free living amoeba) in environmental and respiratory samples in Poland. Acta Protozoologica. 48: 271-279. 95. Lau, H. Y. and N. J. Ashbolt. 2009. The role of biofilms and protozoa in Legionella pathogenesis: implications for drinking water. Journal of Applied Microbiology. 207(2): 368-378. 96. Leiva, B., E. Clasdotter, E. Linder and J. Winiecka-Krusnell. 2008. Free-living Acanthamoeba and Naegleria spp. amoebae in water sources of Leon, Nicaragua. Revista De Biologia Tropical. 56(2): 439-446. 97. Lin, Y.-s. E., R. D. Vidic, J. E. Stout and V. L. Yu. 1998. Legionella in water distribution systems. Journal American Water Works Association. 90(9): 11. 98. Lorenzo-Morales, J., J. F. Lindo, E. Martinez, D. Calder, E. Figueruelo, B. Valladares and A. Ortega-Rivas. 2005. Pathogenic Acanthamoeba strains from water sources in Jamaica, West Indies. Annals of Tropical Medicine & Parasitology. 99(8): 751-758. 99. Lorenzo-Morales, J., A. Ortega-Rivas, P. Foronda, E. Martínez and B. Valladares. 2005. Isolation and identification of pathogenic Acanthamoeba strains in Tenerife, Canary Islands, Spain from water sources. Parasitology Research. 95(4): 273-277. 100. Loret, J.-F. and G. Greub. 2010. Free-living amoebae: Biological by-passes in water treatment. International Journal of Hygiene and Environmental Health. 213(3): 167-175. 101. Loret, J.-F., M. Jousset, S. Robert, C. Anselme, G. Saucedo, F. Ribas, L. Martinez and V. Catalan. 2008. Elimination of free-living amoebae by drinking water treatment processes. European journal of Water Quality 39(1): 37-50. 102. Loret, J. F., M. Jousset, S. Robert, G. Saucedo, F. Ribas, V. Thomas and G. Greub. 2008. Amoebae-resisting bacteria in drinking water: risk assessment and management. Water Science and Technology. 58(3): 571-577. 103. Marciano-Cabral, F. 2006. Advances in free-living amebae research 2006: summary from the IX international workshop on opportunistic protists (IWOP-9). Journal of Eukaryotic Microbiology. 53(s1): S8-S9. 104. Marciano-Cabral, F., M. Jamerson and E. S. Kaneshiro. 2010. Free-living amoebae, Legionella and Mycobacterium in tap water supplied by a municipal drinking water utility in the USA. Journal of Water and Health. 8(1): 71-82. 105. Marolda, C., B. Hauroder, M. John, R. Michel and M. Valvano. 1999. Intracellular survival and saprophytic growth of isolates from the Burkholderia cepacia complex in free-living amoebae. Microbiology 45(7): 1509-1517. 106. Marrie, T., D. Raoult, B. La Scola, R. Birtles and E. Carolis. 2001. Legionella-like and other amoebal pathogens as agents of community-acquired pneumonia. Emerging Infectious Diseases. 7(6): 1026. 58 Chapter 2. Literature review 107. Mergeryan, H. 1991. The prevalence of Acanthamoeba in the human environment. Reviews of Infectious Diseases. 13: S390-S391. 108. Michel, R., H. Burghardt and H. Bergmann. 1995. Acanthamoebae isolated from a highly contaminated drinking water system of a hospital exhibited natural infections with Pseudomonas aeruginosa. Zentralblatt Fur Hygiene Und Umweltmedizin. 196(6): 532- 544. 109. Miltner, E. C. and L. E. Bermudez. 2000. Mycobacterium avium grown in Acanthamoeba castellanii is protected from the effects of antimicrobials. Antimicrobial Agents and Chemotherapy. 44(7): 1990-1994. 110. Miyamoto, H., H. Yamamoto, K. Arima, J. Fujii, K. Maruta, K. Izu, T. Shiomori and S. Yoshida. 1997. Development of a new seminested PCR method for detection of Legionella species and its application to surveillance of Legionellae in hospital cooling tower water. Applied and Environmental Microbiology. 63(7): 2489-2494. 111. Moliner, C., P. E. Fournier and D. Raoult. 2010. Genome analysis of microorganisms living in amoebae reveals a melting pot of evolution. FEMS Microbiology Reviews. 34(3): 281-294. 112. Moliner, C., D. Raoult and P. E. Fournier. 2009. Evidence of horizontal gene transfer between amoeba and bacteria. Clinical Microbiology and Infection. 15(s2): 178-180. 113. Molmeret, M., M. Horn, M. Wagner, M. Santic and Y. Abu Kwaik. 2005. Amoebae as training grounds for intracellular bacterial pathogens. Applied and Environmental Microbiology. 71(1): 20-28. 114. MRC Symposium. 1977. Humidifier fever. Thorax. 32(6): 653-663. 115. Napolitano, J. J. and C. Collettieggolt. 1984. Occurrence of on oak leaf lettuce (Lactuca-Sativa Var Crispa) and Boston lettuce (Lactuca-Sativa Var Capitata). Journal of Protozoology. 31(3): 454-455. 116. Neumeister, B., G. Reiff, M. Faigle, K. Dietz, H. Northoff and F. Lang. 2000. Influence of Acanthamoeba castellanii on intracellular growth of different Legionella species in human monocytes. Applied and Environmental Microbiology. 66(3): 914-919. 117. Newsome, A. L., R. L. Baker, R. D. Miller and R. R. Arnold. 1985. Interactions between Naegleria fowleri and Legionella pneumophila. Infection and Immunity. 50(2): 449-452. 118. Newton, H. J., D. K. Y. Ang, I. R. van Driel and E. L. Hartland. 2010. Molecular pathogenesis of infections caused by Legionella pneumophila. Clinical Microbiology Reviews. 23(2): 274-298. 119. Nora, T., M. Lomma, L. Gomez-Valero and C. Buchrieser. 2009. Molecular mimicry: an important virulence strategy employed by Legionella pneumophila to subvert host functions. Future Microbiology. 4: 691-701. 120. Nwachuku, N. and C. Gerba. 2004. Health effects of Acanthamoeba spp. and its potential for waterborne transmission. Reviews of Environmental Contamination and Toxicology: 93-131. 121. Okada, M., K. Kawano, K. Fumiaki, J. Amemura-Maekawa, H. Watanabe, K. Yagita, T. Endo and S. Suzuki. 2005. The largest outbreak of legionellosis in Japan associated with spa baths : Epidemic curve and environmental investigation. Kansenshogaku zasshi 79(6): 365-374. 122. Pagnier, I., M. l. Merchat and B. La Scola. 2009. Potentially pathogenic amoeba- associated microorganisms in cooling towers and their control. Future Microbiology. 4(5): 615-629. 123. Pagnier, I., D. Raoult and B. La Scola. 2008. Isolation and identification of amoeba- resisting bacteria from water in human environment by using an Acanthamoeba polyphaga co-culture procedure. Environmental Microbiology. 10(5): 1135-1144. 124. Park, M., S. T. Yun, M. S. Kim, J. Chun and T. I. Ahn. 2004. Phylogenetic characterization of Legionella like endosymbiotic X-bacteria in Amoeba proteus: a 59 Chapter 2. Literature review proposal for 'Candidatus Legionella jeonii' sp. nov. Environmental Microbiology. 6(12): 1252-1263. 125. Patterson, D. J. 1996. Free living freshwater protozoa. Sydney, University of New South Wales. 126. Pawlowski, J. and F. Burki. 2009. Untangling the phylogeny of amoeboid protists. Journal of Eukaryotic Microbiology. 56(1): 16-25. 127. Prescott, L., J. Harley and D. Klein. 2005. Microbiology. Boston, McGraw-Hill. 128. Puzon, G. J., J. A. Lancaster, J. T. Wylie and J. J. Plumb. 2009. Rapid detection of Naegleria Fowleri in water distribution pipeline biofilms and drinking water samples. Environmental Science & Technology. 43(17): 6691-6696. 129. Radford, C., D. Minassian and J. Kart. 2002. Acanthamoeba keratitis in England and Wales: incidence, outcome, and risk factors. British Journal of Ophthalmology. 86: 6. 130. Rivera, F., F. Lares, E. Ramirez, P. Bonilla, S. Rodriguez, A. Labastida, R. Ortiz and D. Hernandez. 1991. Pathogenic Acanthamoeba Isolated during an Atmospheric Survey in Mexico City. Reviews of Infectious Diseases. 13: S388-S389. 131. Rivera, F., I. Rosas, M. Castillo, M. Chávez, R. Gómez, R. E. Chío and J. Islas. 1986. Pathogenic and free-living protozoa cultured from the nasopharyngeal and oral regions of dental patients: II. Environmental Research. 39(2): 364-371. 132. Robinson, B. S., S. S. Bamforth and P. J. Dobson. 2002. Density and diversity of protozoa in some arid Australian soils. Journal of Eukaryotic Microbiology. 49(6): 449- 453. 133. Rocha-Azevedo, B., H. B. Tanowitz and F. Marciano-Cabral. 2009. Diagnosis of infections caused by pathogenic free-living amoebae. Interdisciplinary Perspectives on Infectious Diseases. 2009(251406): 14. 134. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241. 135. Rohr, U., S. Weber, R. Michel, F. Selenka and M. Wilhelm. 1998. Comparison of free- living amoebae in hot water systems of hospitals with isolates from moist sanitary areas by identifying genera and determining temperature tolerance. Applied and Environmental Microbiology. 64(5): 1822-1824. 136. Rowbotham, T. 1980. Preliminary report on the pathogenicity of Legionella pneumophila for freshwater and soil amoebae. Journal of Clinical Pathology. 33: 1179-1183. 137. Rowbotham, T. 1986. Current views on the relationships between amoebae, legionellae and man. Israel Journal of Medical Sciences. 22: 678-689. 138. Sanden, G. N., W. E. Morrill, B. Fields, R. F. Breiman and J. M. Barbaree. 1992. Incubation of water samples containing amoebae improves detection of Legionellae by the culture method. Applied and Environmental Microbiology. 58(6): 2001-2004. 139. Sawyer, T. 1980. Marine amebae from clean and stressed bottom sediments of the Atlantic Ocean and Gulf of Mexico. Journal of Eukaryotic Microbiology. 27(1): 13-32. 140. Schmitz-Esser, S., P. Tischler, R. Arnold, J. Montanaro, M. Wagner, T. Rattei and M. Horn. 2010. The genome of the amoeba symbiont "Candidatus Amoebophilus asiaticus" reveals common mechanisms for host cell interaction among amoeba-associated bacteria. Journal of Bacteriology. 192(4): 1045-1057. 141. Schuster, F. L. and G. S. Visvesvara. 2004. Free-living amoebae as opportunistic and non-opportunistic pathogens of humans and animals. International Journal for Parasitology. 34(9): 1001-1027. 142. Seal, D., F. Stapleton and J. Dart. 1992. Possible environmental sources of Acanthamoeba spp. in contact lens wearers. British Journal of Ophthalmology. 76(7): 424-427. 143. Sheehan, K. B., J. A. Fagg, M. J. Ferris and J. M. Henson. 2003. PCR detection and analysis of the free-living amoeba Naegleria in hot springs in Yellowstone and Grand Teton National Parks. Applied and Environmental Microbiology. 69(10): 5914-5918. 60 Chapter 2. Literature review 144. Shoff, M., A. Rogerson, S. Kessler, S. Schatz and D. Seal. 2008. Prevalence of Acanthamoeba and naked amoebae in South Florida domestic water. Journal of Water and Health. 6(1): 5. 145. Smirnov, A. and S. Brown. 2004. Guide to the methods of study and identification of soil gymnamoebae. Protistology. 3(3): 148-190. 146. Smirnov, A., E. Nassonova, C. Berney, J. Fahrni, I. Bolivar and J. Pawlowski. 2005. Molecular phylogeny and classification of the lobose amoebae. Protist. 156(2): 129-142. 147. Snelling, W., J. Moore, J. McKenna, D. Lecky and J. Dooley. 2006. Bacterial-protozoa interactions; an update on the role these phenomena play towards human illness. Microbes and Infection. 8: 578-587. 148. Solomon, J. M., A. Rupper, J. A. Cardelli and R. R. Isberg. 2000. Intracellular growth of Legionella pneumophila in Dictyostelium discoideum, a system for genetic analysis of host-pathogen interactions. Infection and Immunity. 68(5): 2939-2947. 149. Srikanth, S. and S. G. Berk. 1993. Stimulatory effect of cooling tower biocides on amoebae. Applied and Environmental Microbiology. 59(10): 3245-3249. 150. Steinert, M., K. Birkness, E. White, B. Fields and F. Quinn. 1998. Mycobacterium avium bacili grow saprozoically in coculture with Acanthamoeba ployphaga and survive within cyst walls. Applied and Environmental Microbiology. 64(6): 2256-2261. 151. Steinert, M., L. Emody, R. Amann and J. Hacker. 1997. Resuscitation of viable but nonculturable Legionalla pneumophila Philadelphia JR32 by Acanthamoeba castellanii. Applied and Environmental Microbiology. 64(5): 2047-2053. 152. Storey, M., N. Ashbolt and T. A. Stenström. 2004. Biofilms, thermophilic amoebae and Legionella pneumophila - a quantitative risk assessment for distributed water. Water Science & Technology. 50(1): 77-82. 153. Storey, M. and C. Kaucner. 2009. Understanding the growth of opportunistic pathogens in distribution systems. Cooperative Research Centre for Water Quality and Treatment. Adelaide. Report 79: 120. 154. Storey, M., J. Winiecka-Krusnell, N. Ashbolt and T. A. Stenström. 2004. The efficacy of heat and chlorine treatment against thermotolerant Acanthamoebae and Legionellae. Scandinavian Journal of Infectious Diseases. 36(9): 656-622. 155. Stout, J. E., V. L. Yu and M. G. Best. 1985. Ecology of Legionella pneumophila within water distribution systems. Applied and Environmental Microbiology. 49(1): 221-228. 156. Sykora, J. L., G. Keleti and A. J. Martinez. 1983. Occurrence and pathogenicity of Naegleria fowleri in artificially heated waters. Applied and Environmental Microbiology. 45(3): 974-979. 157. Szewzyk, U., R. Szewzyk, W. Manz and K. H. Schleifer. 2000. Microbiological safety of drinking water. Annual Review of Microbiology. 54: 81-127. 158. Tezcan-Merdol, D., M. Ljungstrom, J. Winiecka-Krusnell, E. Linder, L. Engstrand and M. Rhen. 2004. Uptake and replication of Salmonella enterica in Acanthamoeba rhysodes. Applied and Environmental Microbiology. 70(6): 3706-3714. 159. Thom, S., D. Warhurst and B. S. Drasar. 1992. Association of Vibrio cholerae with fresh water amoebae. Journal of Medical Microbiology. 36(5): 303-306. 160. Thomas, V., T. Bouchez, V. Nicolas, S. Rober, J. F. Loret and Y. Lèvi. 2004. Amoebae in domestic water systems: resistance to disinfection treatments and implication in Legionella persistence. Journal of Applied Microbiology. 97(5): 950-963. 161. Thomas, V. and G. Greub. 2010. Amoeba/amoebal symbiont genetic transfers: lessons from giant virus neighbours. Intervirology. 53(5): 254-267. 162. Thomas, V., K. Herrera-Rimann, D. S. Blanc and G. Greub. 2006. Biodiversity of amoebae and amoeba-resisting bacteria in a hospital water network. Applied and Environmental Microbiology. 72(4): 2428-2438.

61 Chapter 2. Literature review 163. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 164. Thomas, V., G. McDonnell, S. P. Denyer and J. Y. Maillard. 2010. Free-living amoebae and their intracellular pathogenic microorganisms: risks for water quality. FEMS Microbiology Reviews. 34(3): 231-259. 165. Tsvetkova, N., M. Schild, S. Panaiotov, R. Kurdova-Mintcheva, B. Gottstein, J. Walochnik, H. Aspöck, M. Lucas and N. Müller. 2004. The identification of free- living environmental isolates of amoebae from Bulgaria. Parasitology Research. 92(5): 405-413. 166. Tyndall, R. L. and E. L. Domingue. 1982. Cocultivation of Legionella pneumophila and free-living amoebae. Applied and Environmental Microbiology. 44(4): 954-959. 167. Valster, R. M., B. A. Wullings, G. Bakker, H. Smidt and D. van der Kooij. 2009. Free- living protozoa in two unchlorinated drinking water supplies, identified by phylogenic analysis of 18S rRNA gene sequences. Applied and Environmental Microbiology. 75(14): 4736-4746. 168. Visvesvara, G. S. 1991. Classification of Acanthamoeba. Reviews of Infectious Diseases. 13: S369-S372. 169. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 170. Wéry, N., V. Bru-Adan, C. Minervini, J.-P. Delgénes, L. Garrelly and J.-J. Godon. 2008. Dynamics of Legionella spp. and bacterial populations during the proliferation of L. pneumophila in a cooling tower facility. Applied and Environmental Microbiology. 74(10): 3030-3037. 171. Winiecka-Krusnell, J. and E. Linder. 1999. Free- living amoebae protecting Legionella in water: The tip of an iceberg? Scandinavian Journal of Infectious Diseases. 31(4): 383- 385. 172. Winiecka-Krusnell, J. and E. Linder. 2001. Bacterial infections of free-living amoebae. Research in Microbiology. 152(7): 613-619. 173. Wullings, B. A., G. Bakker and D. van der Kooij. 2011. Concentration and diversity of uncultured Legionella spp. in two unchlorinated drinking water supplies with different concentrations of natural organic matter. Applied and Environmental Microbiology. 77(2): 634-641. 174. Yoder, J., M. Hlavsa, G. F. Craun, V. Hill, V. Roberts, P. A. Yu, L. A. Hicks, N. T. Alexander, R. L. Calderon, S. L. Roy and M. J. Beach. 2008. Surveillance for waterborne disease and outbreaks associated with recreational water use and other aquatic facility-associated health events. United States, 2005-2006. Morbidity and Mortality Weekly Report. 57(SS09): 1-29.

62 CHAPTER 3

METHODOLOGY 3. TABLE OF CONTENTS

3.1 INTRODUCTION ______65 3.2 CONTROL MICROORGANISMS ______67 3.2.1 FLA control cultures ______67 3.2.2 Legionella control cultures ______69 3.3 SAMPLING______70 3.3.1 Sample collection ______70 3.3.2 Sample sizes ______70 3.3.3 Sample concentration ______70 3.4 WATER QUALITY MEASUREMENTS ______72 3.4.1 Physical characteristics ______72 3.4.2 Chemical characteristics ______73 3.4.3 Biological characteristics______74 3.5 CULTURE METHODS ______76 3.5.1 FLA detection ______76 3.5.2 Legionella detection ______77 3.5.3 FLA uptake of Legionella ______77 3.6 MOLECULAR METHODS ______77 3.6.1 DNA extraction ______77 3.6.2 Identification of isolated microorganisms ______81 3.6.3 Detection of microorganisms by qPCR ______84 3.6.4 Gel electrophoresis ______89 3.6.5 Cloning ______90 3.6.6 DNA sequencing ______90 3.6.7 DNA sequence analysis ______91 3.6.8 Phylogenetic tree creation ______91 3.7 MICROSCOPIC METHODS ______92 3.7.1 Light microscopy of microorganisms ______92 Chapter 3. Methodology 3.7.2 Fluorescent microscopy of environmental samples______92 3.7.3 Fluorescent microscopy of FLA infection with stained Legionella______94 3.7.4 Confocal scanning laser microscopy ______96 3.8 STATISTICAL METHODS ______96 3.8.1 Raw data analysis ______96 3.8.2 Graphing and tests of significance______97 3.9 REFERENCES ______98

LIST OF TABLES

Table 3.1. Control FLA sourced from the AWQC. ______67 Table 3.2. Control FLA sourced from the ATCC. ______67 Table 3.3 Control Legionella cultures sourced from BABS and ATCC. ______69 Table 3.4 Preliminary experiment to scope density of FLA in samples. ______70 Table 3.5 PCR primers specificity using BLAST. ______81

LIST OF FIGURES

Figure 3.1 Experimental flow diagram______66 Figure 3.2 Biofilm removal from sample surfaces by scraping ______71 Figure 3.3 Quantification of bioifilm removal efficiency by scrapping ______72 Figure 3.4 Heterotrophic plate count ______74 Figure 3.5 Examples of plaques formed on non-nutrient agar plates with E. coli ______76 Figure 3.6 DNA extraction efficiencies ______79 Figure 3.7 Inhibition of qPCR due to extracted DNA ______80 Figure 3.8 Standard curves for FLA qPCR ______87 Figure 3.9 Legionella pneumophila standard curve for qPCR ______89 Figure 3.10 Staining of Acanthamoeba sp. cysts ______94 Figure 3.11 Vybrant stained Legionella pneumophila (ATCC 33152) ______95

64 Chapter 3. Methodology

3.1 INTRODUCTION A range of culture, molecular and microscopic methods were used to accurately estimate the density and diversity of FLA and Legionella in samples. As FLA are frequently associated with biofilm, both biofilm and water samples were taken from water systems and experimental set-ups. Additionally water quality characteristics were measured to explore the factors that may facilitate the growth of FLA and Legionella in drinking and recycled water systems and applications. Culture based methods have been the traditional techniques to detect microorganisms although it is well accepted that culture based methods underestimate both the density and diversity of the microorganisms targeted (Besner et al., 2008). However, culture based methods were still employed in this study as it was essential to isolate living FLA from the samples for imaging and to determine if the FLA could host Legionella as ARB. Molecular techniques are an important tool in determining the density and diversity of microorganisms and were therefore used extensively in the methods. Developments in molecular methods have resulted in techniques such as quantitative polymerase chain reaction (qPCR) being very accessible (White, 1996). Also cloning and sequencing were utilised for the identification of the detected microorganisms (FLA and Legionella). Light and fluorescent microscopy is a very useful technique to assist in the identification of FLA isolated by culture. In particular, fluorescent microscopy combined with target stains allow for microorganisms to be located in situ. Staining of living microorganisms allowed in vivo tracking and imaging over a number of days of Legionella cells infecting FLA. Additionally with the aid of confocal scanning laser microscopes (CSLM) the location of those Legionella cells within a FLA could be viewed in three dimensions giving greater evidence to the hosting function of the FLA. Particular culture, molecular and microscopic methods were selected for use based on their validity, accuracy and availability. The methods were applied in parallel across samples and experiments (Figure 3.1) and will be described in detail in this chapter.

65 Chapter 3. Methodology

Water and biofilm samples Water qqualityuality 

TempTemp pH TurbTurb Cl TOC N P Concentrationcentra  FixationFixati 

DNA extractionn

Molecularolecu Cultureultur Microscopyrosc mmethodsethods methodsmethods mmethodsethods BioBiofillmm quantifyquantify qPCRqPqPCR HPCHPC + FLAFLA  Legionellagione  Direct fluouorescentre E.cE.colioli microscopymicroscopy 

FLA  LegionellaLegigiononellae 

PCRR Isolatessolate  LightL mmicroscopyic 

CloningClonin  Selectlect FFLAL  StainedStained ConfocalConfocal Sequencingquencq g isolatesisolates LegionellaLegionella scanningscanning laserlaser microscopymicroscopy

Phylogeneticylogen aanalysisnalysis

Figure 3.1 Experimental flow diagram; samples were analysed by culture, molecular and microscopic methods

66 Chapter 3. Methodology

3.2 CONTROL MICROORGANISMS Control cultures were supplied and purchased from different culture collections in Australian and the USA. A range of FLA and Legionella species were sourced to act as positive controls for culture and molecular detection techniques.

3.2.1 FLA control cultures FLA were supplied by Dr Bret Robinson at the Australian Water Quality Centre (AWQC) (Adelaide, Australia) and purchased from the American Type Culture Collection (ATCC). The cultures supplied by the AWQC had only been classified to genus level based on morphology. Therefore all the cultures were sequenced using partial 18S rRNA FLA targeted PCR (Section 3.6.2.1) or FLA genera specific qPCR (Section 3.6.3.1). Products from the PCR were then visualised, cloned, sequenced and analysed as described in this chapter (Sections 3.6.4 to 3.6.7) (Table 3.1). The ATCC cultures purchased had already been identified to species level so were not further analysed (Table 3.2). Table 3.1 Control FLA sourced from the AWQC % similarity Culture # BLAST best match Source location (accession #) 100 % Recycled water biofilm (ACT, AC362 Acanthamoeba sp. (HQ833439) Australia) 99 % HM060 Hartmannella vermiformis Drinking water (WA, Australia) (AY238942) 99 % HM061 Harmannella vermiformis Bore water (NSW, Australia) (DQ407573) 99 % Water treatment plant (NSW, NG1020 Naegleria lovaniensis (X96569) Australia) Fresh surface water (NSW, TM036 Tetramitus/Vahlkampfia^ ^ Australia) 100 % WT053 Willaertia magna Drinking water (WA, Australia) (X96579) ^ named to genus level as not successfully sequenced Table 3.2 Control FLA sourced from the ATCC. ATCC® # Species Source location 30234 Acanthamoeba castellanii Yeast culture (London, UK) 30461 Acanthamoeba polyphaga Eye infection (Houston, USA) 50237 Hartmannella vermiformis Cooling tower (South Dakota, USA) 30894 Naegleria fowleri Cerebrospinal fluid (Richmond, USA)

67 Chapter 3. Methodology

3.2.1.1 Maintenance of FLA cultures The cultures supplied by AWQC were xenic and grown on non-nutrient agar (NNA) plates with overlays of E. coli as a food source (Esterman et al., 1987). E. coli (TransforMax™ EPI300) stored at -80 °C in 15 % glycerol stock were grown in soya tryptone broth (Oxoid or BD Bacto™, Franklin Lakes, USA) at 37 °C for 72 hr then concentrated by centrifugation (3270  g) for 10 min. Concentrated E. coli were spread onto non-nutrient agar plates (1 % w/v) (Oxoid) to create a lawn. Using a sterile loop the FLA from the stock cultures were innoculated onto the E. coli lawn at one edge of the plate and then allowed to grow across the plate. Plates were sealed with a sealing film (Parafilm M®, Pechiney Plastic Packaging, Chicago, USA) and incubated at room temperature out of direct sunlight and re-cultured at least once a month onto fresh plates. The cultures supplied by the ATCC were axenic and grown in nutrient rich mediums in 25 cm2 vented cap tissue culture flasks. A. castellanii ATCC 30234 and A. polyphaga ATCC 30461 were grown in ATCC medium 712 (protease yeast glucose) at 30 °C and sub-cultured weekly. ATCC medium 712 is a protease yeast glucose medium and made to a volume of 1 L using three components. The first component was made with 20 g of proteose peptone (Oxoid), 1 g of yeast extract (Invitrogen), 950 mL of filtered water (Milli-Q) and autoclaved. The second component were individual buffer elements which were added directly to the first component to avoid precipitation; 10 mL of 0.4 M magnesium sulphate (Sigma-Aldrich, St Louis, USA), 8 mL of 0.05 M calcium chloride (Univar, Redmond, USA), 34 mL of 0.1 M trisodium citrate (Univar), 10 mL of 0.005 M ammonium iron (II) sulphate (Univar), 10 mL of 0.25 M disodium hydrogen phosphate (Univar) and 10 mL of 0.25 M postassium dihydrogen phosphate (Univar). The third component was 50 mL of 2 M glucose (Sigma-Aldrich). All the components were sterilised individually then added together, mixed well and stored at 4 °C until used. Hartmannella vermiformis ATCC 50237 and Naelgeria fowleri ATCC 30894 were grown in ATCC medium 1034 (modified PYNFH medium) and sub-cultured fortnightly. However, H. vermiformis ATCC 50237 was grown at 30 °C while N. fowleri ATCC 30894 was grown at 35 °C. ATCC medium 1034 is a modified peptone yeast nucleic acid folic acid and hemin medium (PYNFH). The medium is made to a volume of 1 L using three components. The first component was made with 10 g of peptone (BD Bacto), 10 g of yeast extract (Invitrogen), 1 g of yeast nucleic acid (Sigma-Aldrich), 15 mg of folic acid (Sigma-Aldrich), 1 mg of hemin (Sigma-Aldrich) and 880 mL of filtered water (Milli-Q) and was autoclaved at 121 °C for 15 min. The second component was 20 mL of buffer solution from; 18.1 g of

68 Chapter 3. Methodology phosphate dihydrogen phosphate (Univar) and 25 g of disodium hydrogen phosphate (Univar) in 1 L of sterile water (Milli-Q), autoclaved and adjusted to pH 6.5. The third component was 100 mL of heat inactivated fetal bovine serum (Sigma-Aldrich). All three components were mixed well and stored at 4 °C until used.

3.2.2 Legionella control cultures Legionella cultures were sourced from the UNSW School of Biological and Medical Sciences (BABS) culture collection and purchased from ATCC (Table 3.3). As all the cultures were already identified to species level no further identification was required. Table 3.3 Control Legionella cultures sourced from BABS and ATCC Culture # Species Source location BABS culture collection 12500 Legionella pneumophila Unknown ATCC 33152 Legionella pneumophila Human lung (Philadelphia, USA) 33155 Legionella pneumophila Creek water (Bloomington, USA) 33215 Legionella pneumophila Human lung (Chicago, USA) 33216 Legionella pneumophila Cooling tower (Dallas, USA) 33279 Legionella dumoffii Cooling tower (New York, USA) 33462 Legionella longbeacheae Human lung (Long Beach, USA) All Legionella cultures were stored at -80 °C in 15 % glycerol stocks. Cultures were initially plated onto buffer charcoal yeast extract (BCYE) agar plates and incubated at 37 °C for 72 hr. BCYE agar plates were either purchased (BD Bacto) or made. BCYE agar was made to a volume of 1 L with 12.5 g of BCYE agar (Oxoid) added to 900 mL of filtered water (Milli-Q) then autoclaved. To the cooled but not solidified BCYE agar Legionella BCYE growth supplement (Oxoid) made up to volume with 100 mL of sterile filtered water (Milli-Q) was added then the plates poured. Final concentration of Legionella BCYE growth supplement (Oxoid) components were 1 g of potassium hydroxide, 0.25 g of ferric pyrophosphate, 0.4 g of L- hydrochloride and 1.0 g of -Ketoglutarate. Legionella colonies from plates were picked and inoculated into buffered yeast extract (BYE) broth in sterile tubes at 37 °C for 48 - 72 hr. BYE broth was made to a volume of 1 L with 10 g of yeast extract (Oxoid) and 900 mL of filtered water (Milli-Q) then autoclaved. Once the media had cooled, 100 mL of Legionella BCYE growth supplement (Oxoid) was added.

69 Chapter 3. Methodology

3.3 SAMPLING

3.3.1 Sample collection Water samples were collected in accordance with the international standard (ISO 19458:2006, Water quality - sampling for microbial analysis). All samples were collected in sterilised sampling containers containing thiosulphate (1 % w/v) (Univar, Redmond, USA) to neutralise the chlorine disinfectant present (up to 5 mg.L-1). Samples were cold transported and processed within 24 hr.

3.3.2 Sample sizes Treated water volumes collected to target FLA in the studies reviewed ranged from 0.1 mL (Hoffmann and Michel, 2001) to 1 L (Thomas et al., 2008). As the density of FLA was not known in the water system sampled a preliminary study was conducted using the Annular Reactors set-up as described in Chapter 6. The reactors were supplied with drinking water and had been in operation for two months as the optimisation stage prior to commencement of the main experiment. Water (10 - 100 mL) and biofilm (1 - 3 coupons) were sampled and processed as described for sample concentration (Section 3.3.3) and detection of FLA by culture (Section 3.5.1). It was found that in drinking water FLA were detected at mean densities of 1.7 amoebae.mL-1 in water and 0.7 amoebae.cm-2 in biofilm (Table 3.4). Based on these preliminary results it was determined that triplicate sampling of 10 mL for water samples and one side of a biofilm coupon (7.5 cm2) would be the minimum requirement to detect FLA by culture and the more sensitive molecular methods. Table 3.4 Preliminary experiment to scope density of FLA in samples Source n FLA range FLA mean Drinking water 12 0 - 4 amoebae.mL-1 1.7 amoebae.mL-1 Drinking biofilm 12 0 - 2.7 amoebae.cm-2 0.7 amoebae.cm-2

3.3.3 Sample concentration

3.3.3.1 Water samples Concentration by centrifugation has been shown to be more effective for FLA from water samples than concentration by filtration (Pernin et al., 1998). Concentration by centrifugation has also been used to remove FLA from environmental water sources (Tyndall et al., 1989). Therefore samples volumes (100 - 250 mL) were placed into 250 mL sterile tubes (Corning®, Lowell, USA) and concentrated by centrifugation. Smaller sample volumes (10 - 50 mL) were

70 Chapter 3. Methodology concentrated in 50 mL sterile tubes (Sarstedt). Centrifugation of samples was at 3270  g for 15 min at 22 °C using a bench-top centrifuge (Allegra-X-12R, Beckman Coulter, Brea, USA).

3.3.3.2 Biofilm samples Scrapping of biofilm from solid surfaces has been shown to be effective for the removal of FLA from environmental samples (Långmark et al., 2007). Additionally, the method is practical and efficient and therefore used in this research. Biofilm was removed from solid surfaces by scraping with sterile straight edge forceps and rinsing repeatedly (Zelver et al., 2001) into individual 50 mL tubes (Sarstedt) with 10 mL of 1/4 strength Ringers solution (Oxoid, Cambridge, UK). To confirm that the biofilm had been effectively removed stainless steel coupons taken from the preliminary set-up of the annular reactor (Chapter 6) were fixed, stained with a chitin cell wall stain and visualised as described (Section 3.7.2). Biofilm was quantified by measuring the pixels present using an image analysis software (ImageJ) as described (Section 3.7.2.3) (Figure 3.2). A B

 

Figure 3.2 Biofilm removal from sample surfaces by scraping. A: stainless steel coupon with drinking water biofilm present. B: the same sample type after biofilm removal by scraping. Both coupons were stained with a cell wall stain (Calcofluor™ White) and imaged with a UV filter by a fluorescent microscope (DMB400, Lieca) at 400  magnification. Scale bar is 10 μm. Mean biofilm quantity was 34 pixels.μm-2 ( = 9.9) before scraping (Figure 3.3). After scrapping the quantity of biomass was significantly reduced (student t-test, p = 0.048). The mean quantity of biofilm remaining after scraping was only 17.6 pixels.μm-2 ( = 2.2). However, as the fluorescent stain is likely to stain other biological materials on the coupon surface no adjustment was made for any estimated remaining biofilm after scraping in the subsequent experiments.

71 Chapter 3. Methodology

Biofilm removal by scrapping

60

50

40

30

20

Quantity (pixels) of DNA 10

0 Before After

Figure 3.3 Quantification of bioifilm removal efficiency by scraping. Fluorescently stained biofilm samples from ambient reactors were quantified before and after scraping. Each bar graph represents the mean and one standard deviation for three replicate samples. 3.4 WATER QUALITY MEASUREMENTS Physical, chemical and biological characteristics of water are measured as indicators of water quality. Some characteristics were measured regularly by water utilities while others were included for the specific foci of this research.

3.4.1 Physical characteristics Temperature, pH and turbidity are all physical characteristics that water quality guidelines recommend sampling weekly to monthly (Australian Government, 2004). Temperature was analysed at time of sampling using a handheld digital thermometer (LCD Multi-Thermometer). For pH analysis samples were transported back to the laboratory and pH recorded using a pH meter (Aqua, TPS, Springfield, Australia). The pH meter was calibrated weekly by laboratory staff. For turbidity analysis samples were transported back to the lab and turbidity was measured using a portable HI98703 Turbidimeter (Hanna Instruments, Woonsocket, US) or a bench top 2100N Turbidimeter (Hach, Loveland, US). Before sampling, the turbidity meter was checked against a series of standards and if it was found to be inaccurate the machine was re-calibrated.

72 Chapter 3. Methodology

3.4.2 Chemical characteristics When chlorine is used as a disinfectant regular monitoring is required as recommended by the Australian Drinking Water Guidelines (Australian Government, 2004). The other chemical characteristics measured in this research (total organic carbon, nitrogen and phosphate) were selected because they have a positive influence on the growth of microorganisms and biofilm in water (Volk and LeChevallier, 1999; Chandy and Angles, 2004; Simoes et al., 2006). Additionally, they are frequently found at higher concentrations in recycled water compared to drinking water (Asano et al., 2007).

3.4.2.1 Chlorine disinfectant Free and total chlorine were measured using photometric tests (Spectroquant, Merck, Frankfurt, Germany) in a photometer (Spectroquant Nova 60, Merck) as per manufacturer's instructions. The method involves acidification of the sample and addition of dipropyl-p- phenylenediamine (DPD) which reacts with free-chlorine to form a red-violet dye that is measured photometrically. To measure bound chlorine, potassium iodide is also added to the sample (Merck).

3.4.2.2 Total organic carbon Total organic carbon (TOC) was measured as non-purgeable organic carbon using a combustion oxidation TOC analyser (TOC- 5000A, Shimadzu, Kyoto, Japan) according to the manufacturer's instructions (Shimadzu, 1993). Briefly all samples were filtered using 0.45 μm filters (Millex- HA, Millipore™, Billerica, USA) and acidified with 1 % v/v hydrochloric acid (Univar). A potassium phthalate (Sigma-Aldrich) standard (1, 2.5, 5 and 10 ppm) was used for quantification and blanks were filtered water (Milli-Q, Millipore™).

3.4.2.3 Total nitrogen Photometric tests (Spectroquant, Merck) were used to measure total nitrogen in a photometer (Spectroquant Nova 60) as per manufacturer’s instructions. The method involves treatment with an oxidizing agent in a thermo-reactor to transform the nitrogen compounds in the sample into nitrate. Addition of concentrated sulphuric acid and benzoic acid derivative causes a reaction which produces a red nitro compound that is measured photometrically (Merck).

3.4.2.4 Total phosphate Photometric tests (Spectroquant) were used to measure phosphate in a photometer (Spectroquant Nova 60) as per manufacturer's instructions. The basis of the method is that

73 Chapter 3. Methodology orthophosphate ions react in a sulphuric acid solution with molybdate ions to form molybdophosphoric acid. The addition of ascorbic acid reduces this to phosphomolybdenum blue which is measured photometrically (Merck).

3.4.3 Biological characteristics The monitoring of heterotrophic bacteria and total coliforms is recommended by a number of water guidelines (Australian Government, 2004) as an indicator of water quality. Biofilm quantification was undertaken in this research to determine how much biofilm was present and if there was any relationships between biofilm quantity and the microorganisms detected.

3.4.3.1 Heterotrophic plate counts Heterotrophic plate counts were performed as per the standard except spread plating rather than pour plating was used (AS/NZS 4276.3.1:2007- Water quality microbiology. Method 3.1: Heterotrophic colony count methods - pour plate method using yeast extract agar). Briefly an aliquot of concentrated sample was spread plated onto yeast exact agar plates (Oxoid, Cambridge, UK) in duplicate then one set of plates was incubated at 22 °C and the other at 37 °C both for 48 h (Figure 3.4). Plates were counted at this time and the number of colony forming units (cfu) enumerated. Negative controls for the plates were spread with sterile filtered water (Milli-Q), while positive controls were spread plated with dilutions of fresh E. coli (TransforMax™ EPI300, Epicentre®, Madison, USA) culture grown in soya tryptone broth (Oxoid) at 37 °C for 72 h.

Figure 3.4 Heterotrophic plate count. Heterotrophic bacterial growth from a recycled water sample.

74 Chapter 3. Methodology

3.4.3.2 Total coliforms Escherichia coli and other coliform bacteria were detected using a selective agar (Brilliance™, Oxoid). The medium contained two chromogenic agents which selectively results in pink coliform colonies and purple E. coli colonies while other Gram-negative bacterial colonies appear white or blue. Plates were inoculated with 0.5 mL of concentrated sample which was spread-plated in duplicate and incubated at 37 °C for 48 h, after which any colonies of the correct colour were enumerated. Negative and positive controls were the same as those used for the HPC method (Section 3.4.3.1)

3.4.3.3 Biofilm quantification Quantification of biofilm is possible through a range of techniques including staining and counter staining, depth analysis with confocal microscopy, culture based methods and measurement of the physical components such as protein and DNA (Peeters et al., 2008; Taff et al., 2011). As a percentage of biofilm DNA is present in equal proportions to proteins and is therefore a good measure of biofilm (Sutherland, 2001). Furthermore for some microorganisms DNA is not just present as an intracellular component but produced as an extracellular product essential to the formation of biofilm (Whitchurch et al., 2002). DNA is easier to quantify at low concentrations than proteins or carbohydrates (Daniels et al., 2007) and hence is a good surrogate for biofilm biomass in fresh water samples.

DNA was quantified based on its specific absorbance at 260 nm (A260) (Daniels et al., 2007). For this research DNA was extracted from biofilm as well as water samples because FLA have been associated with floating biofilms (Declerck et al., 2007). DNA was extracted from samples using a DNA extraction kit (Section 3.6.1.2). DNA was quantified by absorbance at A260 on a microspectrophotometer (NanoDrop 2000, Thermo Scientific, Wilmington, USA). A small aliquot (1 μL) of the DNA extract was placed on the optical surface of the microspectrophotometer for reading and the machine was blanked against nuclease free water (Gibco®, Invitrogen, Carlsbad, USA) (Gallagher and Desjardins, 2006). Concentration of DNA was determined by a function of A260 and the ratio of absorbance at A260 and A280 was also examined to determine levels of contaminating material in the samples. To complement the DNA quantification in some samples the fluorescently stained biofilm was also quantified using an imaging software (ImageJ) as previously described (Section 3.3.3.2).

75 Chapter 3. Methodology

3.5 CULTURE METHODS

3.5.1 FLA detection FLA detection by culture is the traditional method of quantification (Rodriguez- Zaragoza, 1994). Different bacteria can be used as the food source for the FLA including Enterobacter aerogenes (Tsvetkova et al., 2004) but E. coli (Hoffmann and Michel, 2001; Loret et al., 2008; Thomas et al., 2008; Lanocha et al., 2009) is the most frequently used. Hence, for comparability of results E. coli were used as the food source. FLA were detected by culture using non-nutrient agar (NNA) plates with E. coli overlays as described for FLA control cultures (Section 3.2.1.1). Aliquots of concentrated water and biofilm samples were spread plated in duplicate. One plate from each replicate sample were incubated at room temperature (22 °C) and the other at 36 °C. The plates were examined daily for the first seven days for plaque formation then once every seven days until total incubation time was 30 days. A long incubation time is recommend to allow for any FLA in cyst form to ex-cyst and grow (Houang et al., 2001). Plaques were formed by both FLA and other microorganisms or viruses such as bacteriophage (Figure 3.5). Therefore microorganisms from the plaques were examined under a light microscope (DM4000B, Leica, Wetzlar, Germany) at 400  magnification to confirm the presence of FLA. Re-culturing of all FLA identified was attempted by picking the plaque onto fresh NNA with E. coli overlay. FLA with distinct morphologies were imaged and identified by partial 18S rRNA sequencing (Section 3.6.2.1).

Figure 3.5 Examples of plaques formed on non-nutrient agar plates with E. coli overlays. A: plaques formed due to non-FLA microorganisms from WRP samples. B: plaques formed by FLA from MRD samples.

76 Chapter 3. Methodology

3.5.2 Legionella detection Legionella bacteria were detected in concentrated water and biofilm samples using the initial stages of the standard method for the detection of Legionella bacteria by culture in water (AS/NZS 3896:2008, Waters - Examination for Legionella spp. including Legionella pneumophila). Aliquots of concentrated samples from drinking and recycled water and biofilm were spread plated in duplicate onto glycine vancomycin hydrochloride polymyxin B sulphate cycloheximide (GVPC) (Sigma-Aldrich) plates and incubated at 37 °C for seven days with plates examined at four and seven days. Colonies from the GVPC plates were picked and plated onto BCYE (Oxoid) and incubated for three days at 37 °C. Growth of these colonies indicate presumptive Legionella spp. which needed to be confirmed by further tests. Here the methods departed from the standard and the colonies identity was determined by applying targeted PCR for Legionella 16S rRNA gene (Section 3.6.2.3).

3.5.3 FLA uptake of Legionella To determine the potential interactions of L. pneumophila with the isolated FLA species a range of co-innoculation growth experiments were conducted. Two Legionella pneumophila cultures (ATCC® 33152 and ATCC® 33155) were stained with a fluorescent cell tracing dye (Vybrant® CFDA SE Cell Tracer Kit, Intitrogen™) (Section 3.7.3). Legionella and FLA cells were grown as described (Section 3.2) and then enumerated using an improved Neubauer hemocytometer with a depth of 0.1 mm (Cole-Parmer, Vernon Hills, USA). L. pneumophila (ATCC® 33152 and ATCC® 33155) were added independently to the FLA cultures in a ratio of 5:1 in sterile tap water in 24 well cell tissue culture plates (Sarstedt). The experiments were conducted over seven days at a range of temperatures (22 and 37 °C) and Legionella cells were enumerated by fluorescent microscopy. The specific combinations of experiments conducted are detailed in the relevant chapter (Chapter 7). 3.6 MOLECULAR METHODS 3.6.1 DNA extraction

3.6.1.1 From cultures of microorganisms For the extraction of DNA from cultures of microorganisms a method based on cell lysis and a resin DNA binding (InstaGene Matrix™, Bio-Rad) was used that has proved successful for FLA cysts (Robinson et al., 2006; Puzon et al., 2009). The cell lysis and resin DNA binding system (InstaGene Matrix™) was used as per manufacturer’s instructions. Briefly the method involves picking the microorganism from colonies or plaques on plates using a sterile loop. Then

77 Chapter 3. Methodology the microorganisms were re-suspended in 1 mL of sterile filtered water (Milli-Q) and centrifuged into a pellet. The binding matrix was then added and cells were lysed by vortexing and heating. A final centrifugation step left the extracted DNA in the supernatant ready for use. Extracted DNA was stored at -20 °C until further processed.

3.6.1.2 From environmental and experimental samples Successful DNA extraction from environmental samples requires efficient DNA extraction, removal of potentially PCR inhibitory compounds and reproducibility hence the use of commercially available kits is often recommended (Strap, 2007). DNA from FLA and other microorganisms has been successfully extracted from fresh water and biofilm using the FastDNA® SPIN Kit for Soil (Bio101®, Q-biogene, MP Biomedicals, Solon, USA) and applied to qPCR methods (Kuiper et al., 2004; Kuipier et al., 2006; Larsen et al., 2007; Huws et al., 2007 ; Chang et al., 2010). As this kit is widely used in the literature it was selected for DNA extraction. The FastDNA® SPIN Kit for Soil kit was used as per manufacturer’s instructions. The first stage in the DNA extraction process was cell lysis using glass beads and a detergent in a bead beating machine (FastPrep FP120, Bio101®). Cell debris was removed by centrifugation and protein was removed by protein precipitation. DNA was then bound to a DNA binding matrix and washed with ethanol. DNA was eluted in 100 μL of nuclease free water (Gibco®) and heated for 5 min at 55 °C to increase yield. Extracted DNA was stored at -20 °C until used.

3.6.1.3 DNA extraction efficiency from environmental and experimental samples DNA extraction efficiency was conducted in triplicate in recycled, drinking and filtered water (Milli-Q) matrices spiked with 400 ng of control human genomic DNA containing a brain- derived neurotrophic factor (BDNF) (K562 cells, PCR Reagent System, Invitrogen). DNA was extracted from the spiked samples as described for other samples and eluted in 100 μL of nuclease free water (Gibco®). Quantification of the human genomic DNA present in the DNA extracts was achieved using qPCR with primers targeting the BDNF gene to produce a 764 bp product. Reactions for qPCR were performed in duplicate using 10 μL of extracted DNA, 0.5 μM of BDNF-forward primer (5'-AUGGAGAUCUCUGGAT CCATGACCATCCTTTTCCTT-3') and 0.5 μM of BDNF-reverse primer (5'- ACGCGUACUAGUGGATCCCTATCTTCCCCTTTTAAT-3') (PCR Reagent System, Invitrogen), 12.5 μL of qPCR reagent mix (iQ™ SYBR® Green Supermix, Bio-Rad) and nuclease free water (Gibco®) up to 25 μL. Positive controls for the reaction was 40 ng of control human genomic DNA added to duplicate reactions. Negative controls contained no template and

78 Chapter 3. Methodology were made up to volume with nuclease free water (Gibco®). Reactions for qPCR were conducted in a real-time PCR machine (Chromo™, Bio-Rad) under the following thermo-cycling conditions: activation at 94 °C for 3 min then 35 cycles of denaturation at 94 °C for 45 s, annealing at 55 °C for 30 s and extension at 72 °C for 1.5 min, the final extension was at 72 °C for 10 min. Extraction efficiencies were estimated using a standard curve of the cyclic thresholds across known concentrations of control human genomic DNA. This standard curve was used to determine the quantity of DNA in the treatment reactions measured during qPCR compared to the positive controls. Average reductions in the detection of the control human genomic DNA was used to estimate the percentage loss of DNA during DNA extraction from different sample types. DNA lost during extraction averaged 9 % ( = 1.3) for filtered water (MilliQ), 19 % ( = 3) for tap drinking water and 12 % ( = 0.4) for recycled water (Figure 3.6). The differences were found to be significant using a parametric test (1-way ANOVA, p = 0.0017) and individual sample adjustments were made when quantifying subsequent qPCR assays. However, one limitation of this method was that it did not allow for the DNA loses resulting from failure to lyse cells to be accurately estimated.

DNA extraction efficiencies 100

80

60

40 DNA lost (%) DNA

20

0 Control Purified Drinking Recycled

Figure 3.6 DNA extraction efficiencies. Extraction efficiencies for purified, drinking and recycled water. Each bar graph represents the mean and one standard deviation for three replicate samples. 3.6.1.4 Quantification of DNA inhibitors present in environmental sample DNA extracts To determine if any inhibitors were present in the final DNA extracted from drinking water, recycled biofilm and filtered water an internal human genomic DNA (BDNF, Invitrogen)

79 Chapter 3. Methodology control was used. Reactions for qPCR inhibition quantification were run in duplicate with each reaction containing 60 ng of control human genomic DNA, 0.5 μM of each BDNF-forward primer and BDNF-reverse primer (PCR Reagent System), 12.5 μL of qPCR reagent mix (iQ™ SYBR® Green Supermix), 5 μL of sample DNA extract and nuclease free water (Gibco®) up to 25 μL. Positive controls for the reaction was the use of 60 ng of control human genomic DNA in duplicate with no sample DNA extract. Negative controls contained no template and were made up to volume with nuclease free water (Gibco®). Reactions for qPCR were conducted in a real- time PCR machine (Chromo™, Bio-Rad) as previously described (Section 3.6.1.3). Inhibition present in the samples during qPCR was quantified using the standard curve of control human genomic DNA which compared the treatment reactions to the positive controls. Average reductions in the detection of the control human genomic DNA was used to estimate the percentage inhibition in DNA extracts from different sample types. Sample inhibition was observed with DNA detection in qPCR found to be 6 % ( = 4.8) less in the presence of drinking water DNA extracts and 11 % ( = 3.5) less in the presence of recycled biofilm DNA extracts. Using parametric tests these averages were found not be significantly different (1-way ANOVA, p = 0.1) based on sample type, qPCR quantification were adjusted to account for the effects of inhibition within the samples.

qPCR inhibition of extracted DNA 100

80

60

40 Inhibition (%) Inhibition

20

0 Control Drinking Recycled

Figure 3.7 Inhibition of qPCR due to extracted DNA. Inhibition calculated for drinking water and recycled water biofilm samples. Each bar graph represents the mean and one standard deviation for three replicate qPCR reactions.

80 Chapter 3. Methodology

3.6.2 Identification of isolated microorganisms

3.6.2.1 Identification of FLA Due to the phylogenetic diversity of FLA it is not possible to design a primer specifically for FLA based on the small subunit (SSU) gene (18S rDNA) (Nassonova et al., 2010). In the literature a PCR primer set was identified that claimed to be specific for FLA (Tsvetkova et al., 2004). The primer had been used in other studies to identify FLA from environmental samples (Behets et al., 2007; Declerck et al., 2007). An in silico PCR reaction was undertaken with the Tsvetkova et al. (2004) FLA targeted primers using the basic local alignment search tool (BLAST) on the National Centre for Biotechnology Information (NCBI) (http://blast.ncbi.nlm.nih.gov). The search result returned a vast range of taxonomic hits for the primer set which included fungi/metozoa organisms as the third most frequent hit over Amoebozoa which were only the sixth most frequent (Table 3.5). This emphasises the difficulty in designing FLA genus specific primers. Table 3.5 PCR primers specificity using BLAST

Search Taxonomy Search Taxonomy rank rank 1 Eukaryota 6 Amoebozoa 2 Fungi/Metozoa 7 Haptophyaceae 3 Alveolata 8 Rhodophyta 4 Viridiplantae 9 Rhizaria 5 Stramenopiles 10 Cryptophyta

The Tsvetkova et al. (2004) primers set was still utilised with the knowledge that they would also amplify fungi and other eukaryotic microorganisms isolated with the FLA. The PCR reaction produced a range of product sizes (500 - 1500 bp) (Tsvetkova et al., 2004). PCR reactions were performed in duplicate 5 μL of template DNA, 100 nM each for forward primer (P-FLA-F, 5'-CGCGGTAATTCCAGCTCCAATAGC-3') and reverse primer (P-FLA-R, 5'- CAGGTTAAGGTCTCGTTCGTTAAC-3'), 2.5 U of Taq polymerase, 0.2 mM each of dNTPs,

3 mM MgCl2 , final PCR buffer concentration of 100 mM KCl and 40 mM Tris-HCl at pH 8.4 (Invitrogen) then made up to 50 L with nuclease free water (Gibco®)(Tsvetkova et al., 2004). Positive controls were DNA extracts from culture collections of an Acanthamoeba sp., H. vermiformis and N. lovaniensis supplied by AWQC (Table 3.1). No template negative controls were made up with nuclease free water (Gibco®). PCR reactions were performed in dual block PCR machine (Bio-Rad) under the following thermo-cycling conditions; activation at

81 Chapter 3. Methodology

94 °C for 3 min, then 40 cycles of denaturation at 94 °C for 1 min, annealing at 63 °C for 1 min and extension at 74 °C for 3.5 min, the final extension was at 72 °C for 10 min. PCR products were visualised using the Water Research Centre gel electrophoresis method and products of the correct size were sequenced as per the Water Research Centre lab DNA sequencing method.

3.6.2.2 Identification of other eukaryotic microorganisms A number of universal 18S rDNA primers published in the literature were evaluated (Sogin and Gunderson, 1987; Medlin et al., 1988; Amann et al., 1990; Lopez- Garcia et al., 2001; lapeta et al., 2005). The 18S rDNA universal forward (Sogin and Gunderson, 1987) and reverse primer (Amann et al., 1990) were selected because they had been successfully used to amplify FLA DNA from drinking water samples (Valster et al., 2009). These primers were used to amplify the DNA from other eukaryotic microorganisms and FLA that were not successfully identified by the FLA PCR described (Section 3.6.2.1). Specifically, a portion (532 bp) of the 18S rRNA gene was amplified using the forward (Sogin and Gunderson, 1987) and reverse primer (Amann et al., 1990) with the PCR conditions outlined (Valster et al., 2009). PCR reactions were prepared in duplicate using 5 μL of target DNA, 2.5 U of Taq polymerase (Takara, Madison, USA), 0.2 mM dNTP (Takara), PCR buffer

(10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl2 and pH 8.3, Takara), 500 nM each of forward (Euk1a-F, 5’-CTGGTTGATCCTGCCAG-3’) and reverse primers (Euk516-R, 5’ACCAGACTTGCCCTCC -3’) and nuclease free water (Fisher Scientific, Pittsburgh, USA) to total volume of 25 μL. DNA extracted from A. castellanii (ATCC® 30234) was used as a positive control and a no-template negative control was included. Thermo-cycling was performed in a four block thermo-cycler (DNA Engine Tetrad®, Biorad) under the following conditions: initial activation for 2 min 10 s at 94 °C then 35 cycles of denaturation for 30 s at 94 °C, annealing for 45 s at 56 °C then extension for 2 min 10 s at 72 °C, followed by a final extension of 7 min at 72 °C. PCR products were visualised using the US EPA gel electrophoresis method (Section 3.6.4.2) and products of the correct size were cloned and sequenced (Section 3.6.5 -3.6.7)

3.6.2.3 Identification of Legionella Two published primers sets targeting Legionella spp. were evaluated (Jonas et al., 1995; Miyamoto et al., 1997). The Miyamoto et al. (1997) primers were shown to not be as effective in water samples as other primers (Devos et al., 2005). However, as the primers were only being applied to isolated cultures this was not a concern plus they produced a larger product size (654 bp) which was advantageous (Miyamoto et al., 1997).

82 Chapter 3. Methodology

Presumptive Legionella colonies identified by the initial stages of the standard method were identified by applying targeted PCR for Legionella 16S rRNA gene (Miyamoto et al., 1997). DNA was extracted from single colonies as described (Section 3.6.1.1). PCR reactions were preformed in duplicate with 5 μL of template DNA, 500 nM each for forward primer (LEG225, 5'-AAGATTAGCCTGCGTCCGAT-3') and reverse primer (LEG858, 5'- GTCAACTTATCGCGTTTGCT-3'), 2.5 U of Taq polymerase , 0.2 mM each of dNTPs,

1.5 mM MgCl2 , PCR buffer concentration of 50 mM KCl and 10 mM Tris-HCl at pH 8.3 (Takara) then made up to 50 L with nuclease free water (Fisher-Scientific). DNA extracted from a L. pneumophila (ATCC® 33155) culture was used a positive control while a no template negative control of nuclease free water (Fisher-Scientific) was used. PCR reactions were performed in four block thermo-cycling machine (DNA Engine Tetrad®, Bio-Rad) under the following thermo-cycling conditions; activation at 95 °C for 1.5 min, then 30 cycles of denaturation at 94 °C for 10 s, annealing at 64 °C for 1 min and extension at 74 °C for 1 min, then a final extension was at 72 °C for 10 min. PCR products were visualised using the US EPA gel electrophoresis method (Section 3.6.4.2) and products of the correct size were cloned and sequenced (Section 3.6.5 -3.6.7).

3.6.2.4 Identification of other prokaryotic microorganisms To identify other prokaryotic microorganisms a number of universal 16S rRNA targeted primers were evaluated for their suitability (Amann et al., 1995 ; Marchesi et al., 1998; Nakatsu and Marsh, 2007). The Amann et al. (2005) primers were selected because they amplified a large portion of the 16S rDNA gene (795 bp)(Amann et al., 1995 ). Specifically, the PCR reactions were undertaken using the forward primer (8F, 5’-TGAGCCAGGATCAAACTCT-3’) and reverse primer (785R, 5’-CTACCAGGGTATCTAATCC-3’) (Amann et al., 1995 ). PCR reactions were prepared in duplicate with 5 μL of target DNA, 1.2U of Taq polymerase

(Takara), 0.2 mM dNTP (Takara), PCR buffer (10 mM Tris-HCl, 50 mM KCl, 1.5 mM MgCl2 and pH 8.3, Takara), 200 nM of forward and reverse primers and nuclease free water (Fisher- Scientific) to make a total volume of 50 μL (Barkovskii and Fukui, 2004). DNA from E. coli (EPI-300) was used as a positive control and a no-template negative control was included. Thermo-cycling was performed in a thermal cycler (DNA Engine Tetrad®, Biorad) under the following conditions: initial activation for 2 min at 94 °C, 40 cycles of de-naturation for 30 s at 92 °C, annealing for 1 min at 56 °C then extension for 45 s at 68 °C, followed by a final extension of 7 min at 72 °C (Barkovskii and Fukui, 2004). PCR products were visualised using

83 Chapter 3. Methodology the USEPA gel electrophoresis method (Section 3.6.4.2) and products of the correct size were cloned and sequenced (Section 3.6.5 -3.6.7).

3.6.3 Detection of microorganisms by qPCR

3.6.3.1 FLA detection by qPCR For detection of FLA by qPCR three different genera of FLA were targeted with individual primer sets in individual reactions.

3.6.3.1.1. Acanthamoeba spp. Two qPCR primers were identified that targeted the Acanthamoeba genus (Qvarnstrom et al., 2006; Rivière et al., 2006). The Rivière et al. (2006) primers were selected because they could be used without the designed probe and still be specific for Acanthamoeba spp. This option was selected because the aim was to clone and sequence all the qPCR products and bound probes make this more problematic compared to qPCR using primers and fluorescent dye only (Grunenwald and Kramer, 2004). Using this method the specificity of the qPCR reactions were verified by sequencing rather than the use of specific probes. Comparing the qPCR reactions when applied to environmental samples using their specific probes the Qvarnstrom et al. (2006) qPCR out performed the Rivière et al. (2006) qPCR (Chang et al., 2010) and this should be kept in mind if those reactions are to be used with the probes. Acanthamoeba spp. were detected by amplification of a 100 bp portion of the 18S rRNA gene (Rivière et al., 2006). Each reaction was run in duplicate and consisted of 5 μL of sample DNA extract, 0.2 μM each of forward primer (TaqAcF1, 5'- CGACCAGCGATTAGGAGACG-

3') and reverse primer (TaqAcR1, 5'- CCGACGCCAAGGACGAC-3'), 12.5 μL of qPCR reagent mix (iQ™ SYBR® Green Supermix) and volume made up to 25 μL with nuclease free water (Gibco®). Reactions were run in a real-time PCR machine (Chromo™, Bio-Rad) under the following cycling conditions: activation at 95 °C for 2 min, then 40 cycles of denaturation at 95 °C for 15 s, annealing at 60 °C for 30 s and extension at 72 °C for 40 s, then the fluorescence profile of the plate was read, a final extension at 72 °C for 10 min. The standard curve was created using supplied control cultures of Acanthamoeba sp. (WQRC) trophozoite and cysts culture.

3.6.3.1.2. Hartmannella vermiformis Only one qPCR method was identified that targeted Hartmannella and it was species specific for H. vermiformis (Kuipier et al., 2006). Although the reaction was species specific it

84 Chapter 3. Methodology was still utilised as H. vermiformis is often associated with drinking water (Thomas et al., 2006; Thomas et al., 2008; Valster et al., 2009). H. vermiformis was detected using a qPCR that targeted a 502 bp section of the 18S rRNA gene that did not use a probe (Kuipier et al., 2006). Each reaction was run in duplicate and consisted of 5 μL of sample DNA extract, 0.2 μM each of forward primer (Hv1227F, 5'-

TTACGAGGTCAGGACACTGT-3') and reverse primer (Hv1728R, 5'-

® GACCATCCGGAGTTCTCG -3'), 12.5 μL of qPCR reagent mix (iQ™ SYBR Green Supermix) and volume made up to 25 μL with nuclease free water (Gibco®). Reactions were run on a real-time PCR machine (Chromo™, Bio-Rad) under the following cycling conditions: activation at 95 °C for 3 min, then 40 cycles of denaturation at 95 °C for 20 s, annealing at 56 °C for 30 s and extension at 72 °C for 40 s, then the fluorescence profile of the plate was read, with a final extension at 72 °C for 10 min. The standard curve was created using supplied control cultures of H. vermiformis (Australian Water Quality Centre, Adelaide) trophozoite and cysts culture.

3.6.3.1.3. Naegleria spp. Only one qPCR method was identified which targeted the Naegleria genus (De Jonckheere, 1998) and another which targeted N. fowleri specifically (Qvarnstrom et al., 2006). The Naegleria genus qPCR (Qvarnstrom et al., 2006) method was selected as the aim was to identity as broad a range of FLA as possible. Naegleria spp. were detected using qPCR that targeted the internal transcribed spacer (ITS) region of the rRNA gene and produced products that ranged in size from 304 to 729 bp depending on the species detected (De Jonckheere, 1998). Each reaction was run in duplicate and consisted of 5 μL of sample DNA extract, 0.2 μM each of forward primer (NgITSFw, 5'- AACCTGCGTAGGGATCATTT-3') and reverse primer (NgITSRv, 5'- TTTCCTCCCCTTATTAATAT-3'), 12.5 μL of qPCR reagent mix (iQ™ SYBR® Green Supermix) and volume made up to 25 μL with nuclease free water (Gibco®). Reactions were run in real-time PCR machine (Chromo™, Bio-Rad) under the following cycling conditions: activation at 95 °C for 3 min, then 40 cycles of denaturation at 94 °C for 20 s, annealing at 50 °C for 20 s and extension at 72 °C for 20 s, then the fluorescence profile of the plate was read, a final extension at 72 °C for 10 min. It was also identified that the reaction was able to identify Willertia magna as well. The standard curve was created using supplied control cultures of N. lovaniensis (WQRC) trophozoite and cysts culture.

85 Chapter 3. Methodology

3.6.3.2 Melt curves and standard curves At the conclusion of each of the FLA qPCR reactions a melt curve was performed from 75 - 95 °C with 20 s holds and a plate read at every 0.5 °C increment. The profile of the melt curve was visually analysed to estimate the similarity of the DNA products present. Additionally all qPCR products were run on agarose gels as described (Section 3.6.4.1) to determine the presence of a product of the correct size. Duplicate qPCR reactions with products of the correct size were pooled and purified using a PCR purification kit (Purelink, Invitrogen) as per manufacturer’s instructions. A selection of qPCR products from the different reactions were then cloned (Section 3.6.5) and sequenced (Section 3.6.6.2). The 18S rRNA gene copy number varies for different FLA species; Acanthamoeba sp. has approximately 660 copies while H. vermiformis has twice as many with 1330 copies (Valster et al., 2009). For these reasons it is essential to complete individual calibration curves for each FLA species. Additionally, it is possible to detect FLA at less than single cell levels. The limit of detection for all of the FLA was determined to be 0.1 amoebae per qPCR reaction based on the standard curves (Figure 3.8) For all FLA qPCR reactions standard curves were generated from eight fold serial dilutions in sterile water (Gibco®) of DNA extracted as described (Section 3.6.1.2) from trophozoites and cysts. Trophozoites were detected at lower cyclic thresholds than equivalent cysts numbers over all three amoeba species (Figure 3.8). Standard curves were generated using linear regression and the goodness-of-fit (r2) was greater than 0.98 for all except one standard (Figure 3.8 - F, r2 = 0.95). The amplification efficiency of each standard curve was calculated using Equation 1.1 (Pfaffl, 2004). Knowing the amplification efficiency of a qPCR reaction is important to determine if the reaction conditions are optimal and to determine if standard and experimental results are comparable. (1/slope) E =10 (Equation 1.1) Where: E = the amplification efficiency of the qPCR reaction slope = the slope of the linear regression function for the standard curve. The amplification efficiency for Acanthamoeba sp. trophzoite standard curve was calculated at 2.36 which was higher than the 1.83 for H. vermiformis trophozoites and 2.05 for Naegleria sp. trophozoites. However, all these values are close to the optimal efficiency of 2.0 which was recommend (Pfaffl, 2004).

86 Chapter 3. Methodology

ABAcanthamoeba sp. trophozoites standard curve Acanthamoeba sp. cysts standard curve 5 5

y = -0.37x + 11.0 y = -0.50x + 13.6 (r2 = 1.00) 2 4 4 (r = 0.98)

3 3

2 2

1 1 Log of amoeba cell numbers numbers cell amoeba Log of Log of amoeba cell numbers numbers cell amoeba Log of 0 0

-1 -1 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Cyclic threshold (Ct) Cyclic threshold (Ct)

C H.vermiformis trophozoites standard curve D H. vermiformis cysts standard curve 4 4 y = -0.26x + 8.4 y = -0.36x + 12.2 2 (r = 0.98) (r2 = 0.99)

3 3

2 2

1 1

0 0 Log of amoeba cell numbers numbers cell amoeba Log of Log of amoeba cell numbers numbers cell amoeba Log of

-1 -1 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Cyclic threshold (Ct) Cyclic threshold (Ct) E F Naegleria lovaniensis trophozoites standard curve Naegleria lovaniensis cysts standard curve 4 4 y = -0.32x + 12.9 y = -0.30x + 9.7 (r2 = 0.95) (r2 = 0.99) 3 3

2 2

1 1

0 0 Log of amoeba cell numbers numbers cell amoeba Log of numbers cell amoeba Log of

-1 -1 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 Cyclic threshold (Ct) Cyclic threshold (Ct)

Figure 3.8 Standard curves for FLA qPCR. Graphs A - F were calculated using DNA extracted from three species of FLA (trophozoites and cysts). Straight lines were fitted using linear regression with the formula and regression squared value in the top right of each figure.

87 Chapter 3. Methodology

3.6.3.3 Legionella detection by qPCR Two qPCR methods for Legionella spp. were evaluated for their suitability (Wellinghausen et al., 2001; Chang et al., 2009). The Chang et al. (2009) qPCR method was selected because it produced a slightly larger product (434 bp) than the other method (Wellinghausen et al., 2001). Legionella was detected using a qPCR that targeted the 16S rDNA gene of Legionella spp. (Chang et al., 2009). Quantitative PCR reactions were preformed in duplicate with 2.5 μl of template DNA, 12.5 μl of qCPR master mix (SYBR® green PCR master mix, Applied Biosystems) and 0.2 μM each of forward primer (LEG427F, 5'- GTAAAGCACTTTCAGTGGGGAG-3') and reverse primer (LEG880R, 5'- GGTCAACTTATCGCGTTTGCT-3') and made up to a final volume of 25 μl with nuclease free water (Fisher Scientific) (Chang et al., 2009). Positive controls were DNA extracts from L. pneumophila cultures (ATCC® 33215 or BABS 12500) and negative controls contained no template DNA and were made up the required volume using nuclease free water (Fisher Scientific). Using 96 well plates qPCR was preformed in an real-time PCR detection system (I- Cycler iQ Multi-Color, Bio-Rad). Thermal cycling conditions used were activation at 95 °C for 3 min; 40 cycles of denaturisation at 95 °C for 20 s, annealing at 58 °C for 30 s and extension at 72 °C for 40 s and a final extension step of at 72 °C for 10 min. The fluorescent intensity of the dye was record during the annealing steps. A melt curve analysis was conducted at the end of each run from 75 - 95 °C with 20 s holds and a plate ready at every 0.5 °C increment. For quantification a cell based calibration curve was determined under the same conditions using a eight fold serial dilution in nuclease free water (Fisher-Scientific) of DNA extracted from fresh growth cultures of two different L. pneumophila (ATCC® 33155 and ATCC® 33215) (Figure 3.9). The detection limit was determined to be five Legionella cells per reaction which correlated to the maximum cyclic threshold value of 40. The amplification efficiency of the reaction was 1.91 based on slope of the curve and equation 1.

88 Chapter 3. Methodology

Standard curve for Legionella pneumophila 6

y = -0.28x + 11.9 (r2 = 0.90) 5

4 cell numbers cell 3

2 Legionella

1 Log of Log of

0 20 22 24 26 28 30 32 34 36 38 40 Cyclic threshold (Ct)

Figure 3.9 Legionella pneumophila standard curve for qPCR. L. pneumophila ATCC® 33155 and L. pneumophila ATCC® 33215 cultures were used to form a joint standard curve calibrating cells numbers to cyclic thresholds (Ct) for qPCR. The box in the top right corner is the linear regression function and r2 value for the plotted line. 3.6.3.4 Legionella detection in isolated FLA FLA that had been isolated from drinking and recycled water and identified and sub- cultured for number of months were screened for the presence of Legionella sp. residing as amoebae-resistant bacteria (ARB). DNA was extracted from the FLA cultures as described (Section 3.6.1.1.) Legionella spp. specific qPCR was then applied as previously described using 5 μL of DNA extract (Section 3.6.3.3).

3.6.4 Gel electrophoresis

3.6.4.1 Water Research Centre lab gel electrophoresis method PCR products were visualised on a 1.5 % agarose gel (UltraPure™1000, Invitrogen) made and run in tris-base boric acid disodium -ethylenediaminetetraacetic (EDTA) acid buffer (TBE). The buffer was made to a working concentration of 50 mM Tris base (Univar), 42 mM boric acid (Sigma-Aldrich) and 10 mM EDTA (Gibco®) in sterile water (Milli-Q) and adjusted to pH 8.3 (Hendrickson and Walthers, 2007). PCR products were loaded with 2 μL of loading dye (BlueJuice™, Invitrogen) and 5 μL of the 100 bp DNA molecular ladder (EZ Load, Bio- Rad) was diluted 5  in TBE before loading. Gels were run at 100 V for 1 h. Gels were visualised using in a nucleic acid binding gel stain (SYBR Gold, Invitrogen) which was diluted

89 Chapter 3. Methodology

10 000  in TBE and incubated with the gel for 30 min. Gels were imaged using a UV illumination and imaging system (Gel-Doc, Bio-Rad).

3.6.4.2 US EPA lab gel electrophoresis method PCR products were visualised on a 1 % (w/v) agarose gel (Amresco, Solon, USA) gel containing nucleic acid stain (DNA Gel Star, Cambrex, East Rutherford, USA) and run in a lithium borate buffer. The buffer was made to a working solution of 10 mM lithium hydroxide monohydrate (Fisher-Scientific), 30 mM boric acid (Fisher-Scientific) and 100 mL filtered water (Milli-Q) and adjusted to pH 8.2. PCR products were loaded with 1 μL of loading dye (Orange G, Sigma-Aldrich) and run with 10 μL of a 100 - 2000 bp molecular ladder (E-Gel, Invitrogen) Gels were run at 220 V for 30 min. Gels were imaged using UV illumination and imaging system (Alpha Innotech, Santa Clara, USA).

3.6.5 Cloning All PCR and qPCR products selected for cloning were purified using the PCR purification kits (Micro PureLinkTM or PurelinkTM, Invitrogen) as per manufactures instructions. Briefly, purification was achieved by binding of the double stranded DNA to silica-based membrane and washing with ethanol. The DNA was then eluted in nuclease free water and the quantity of the purified qPCR products was then checked on agarose gels as previously described (Section 3.6.4.2). Select qPCR products were then cloned into vectors (pCR®4-TOPO) and transformed into E. coli (One Shot®TOP10) using a cloning kit (TOPO TA Cloning®Kits for Sequencing, Invitrogen) as per manufacturer’s instructions with ampicillin selection media. PCR products containing the correct size DNA insert were submitted to the US EPA DNA sequencing facility (Section 3.6.6.2).

3.6.6 DNA sequencing

3.6.6.1 Water Research Centre lab DNA sequencing method PCR products of correct size range were pooled from duplicate reactions and purified using a PCR purification kit (Purelink, Invitrogen) as per the manufactures instructions for high cut off products (< 300 bp). DNA sequence labeling PCR was performed with 10 μL of purified PCR product, with 50 nM of forward primer (P-FLA-F), 3.5 μL of 5  sequencing buffer (Applied Biosystems, Life Technologies™, Carlsbad, USA), 2 μL of sequencing reagent mix (BigDye® Terminator v3.1, Applied Biosystems) and nuclease free water (Gibco®) up to 20 μL. Labeling reactions were run in PCR machine (Bio-Rad) under the following thermo-cycling conditions: 96 °C for 1 min then 100 cycles of de-naturation at 96 °C for 10 s, annealing at 50 °C

90 Chapter 3. Methodology for 5 s and extension at 60 °C for 4 min. Labeled PCR products were then purified up using an ethanol/EDTA precipitation method as per the protocol supplied by the DNA sequencing facility (The Ramaciotti Centre, 2010). Purified labeled PCR products were sequenced as per manufactures instructions using a capillary electrophoresis DNA sequencer (3730, Applied Biosystems) by The Ramaciotti Centre for Gene Function Analysis at The University of New South Wales (UNSW).

3.6.6.2 US EPA lab DNA sequencing method Cloned DNA fragments were cleaned and sequenced in the forward direction only using the provided forward primer (M13, 5´-GTAAAACGACGGCCAG-3´) using DNA labeling reactions (BigDye chemistry, Applied Biosystems) and a capillary electrophoresis DNA sequencer (ABI 377, Applied Biosystems) performed by Dynamac Contractors at the USEPA.

3.6.7 DNA sequence analysis DNA sequences produced were checked for quality and trimmed using DNA editing software (FinchTV, Geospiza, Seattle, USA). Trimmed sequences were then imported into DNA analysis and management software (Vector NTI®, Invitrogen) and searched against the NCBI (www.ncbi.nlm.nih.gov) using the basic local alignment search tool (BLAST). Sequences in the database with greater than 95 % similarity to the sequences identified from the samples were noted. Additionally, likely Legionella sequences were also searched against the Ribosomal Database Project (RPD) (http://rdp.cme.msu.edu/).

3.6.8 Phylogenetic tree creation A selection of sample DNA sequences that were clearly similar to target microorganism (FLA or Legionella) were selected for the formation of a phylogenetic tree. First the selected sequences were aligned using ARB SILVA (Pruesse et al., 2007) and then a phylogenetic tree was constructed using ARB software (Ludwig et al., 2004) using the joining neighbour algorithm. For the formation of FLA phylogenetic trees 40 DNA sequences in total were selected from the database from 13 FLA genera known to be associated with fresh water; Acanthamoeba, Echinamoeba, Filamoeba, Hartmannella, Heterolobosea, Naegleria, Neoparamoeba, Platyamoebae, Sappinia, Tetramitus, Thecamoeba, Vahlkampfia, and Vannella (Smirnov and Brown, 2004; Shoff et al., 2008). The root for the FLA tree was the common yeast Saccharomyces cerevisiae (accession number EF153844).

91 Chapter 3. Methodology

For the formation of Legionella phylogenetic trees 50 DNA sequences in total were selected from the database including 38 different Legionella sp. and Legionella-like amoebal pathogens. The root for the tree was the bacterium Chlamydophila pneumoniae (accession number Z49873).

3.7 MICROSCOPIC METHODS

3.7.1 Light microscopy of microorganisms When using the light microscope Köhler illumination was first established for the objective that was to be used (Salmon et al., 2005). Isolated microorganisms were re-suspended in a drop of sterile water (Milli-Q) on a microscope slide with a cover slip. Microorganism were observed and enumerated using phase-contrast microscopy under a light microscope (DM4000B, Leica, Wetzlar, Germany) at 400  magnification. While microorganisms were enumerated using a counting chamber; improved Neubauer hemocytometer with a depth of 0.1 mm (Cole- Parmer) under the same conditions. Microorganisms were imaged using phase-contrast microscopy at 1000  magnification using an oil immersion lens on a light microscope (DM4000B, Leica). Images were taken using the camera and software (Lieca Application Suite) attached to the microscope. Post image analysis such as addition of scale bars and over-lays was preformed on the imaging software (Lieca Application Suite).

3.7.2 Fluorescent microscopy of environmental samples

3.7.2.1 Fixation of environmental samples Fixation of water and biofilm samples involved the cross-linking of proteins using common fixative agents (Surman et al., 1996). Water samples were concentrated to leave 1 mL of concentrated sample which was transferred to a 2 mL tube (Sarstedt). Biofilm coupons were simply removed from the storage 50 mL tubes (Sarstedt) prior to fixation. The fixative agent consisted of 2 % (weight/volume) of formaldehyde (Invitrogen) , 0.1 % glutaraldehyde (Invitrogen), 50 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES) buffer (Sigma-Aldrich) and made up to volume with filtered water (Milli-Q) (Surman et al., 1996). Biofilm coupons were immersed in the fixative agent and 1 mL of the fixative agent was added to the concentrated water samples. Incubation for both sample types was 1 h at room temperature. At completion of fixation water samples in 2 mL tubes were centrifuged at 2000  g for 10 min then the supernatant was removed leaving a cell pellet which was re-

92 Chapter 3. Methodology suspended in 50 mM HEPES buffer. Biofilm coupons were also rinsed in HEPES buffer. Both fixed sample types were stored at -20 °C prior to staining and imaging.

3.7.2.2 Staining and imaging The density of microorganisms with eukaryotic morphology is a useful supplement to the data collected in this study but is not a direct quantification of FLA. A selection of stains had been reviewed for their suitability for application to FLA in the literature (Packroff et al., 2002). Diagnosis of amoebic keratitis infections reported the use of a stain specific for the chitin in the FLA cells walls; Calcofluor™ White (Wilhelmus et al., 1986). While the quantification of biofilm using a live/dead stain is often utilised (Storey and Ashbolt, 2002; Thomas et al., 2004). Hence propidium iodide (component of LIVE/DEAD® BacLight™ Bacterial Viability Kit, Invitrogen) and Calcofluor™ White M2R (component of LIVE/DEAD® Yeast Viability Kit, Invitrogen) were used in this research. Propidium iodide diffuses only through damaged cell membranes and stains nucleic acids. It has an excitation maxima of 480 nm, an emission maxima of 500 nm and is viewed with a Texas Red filter setting. Calcofluor™ White stains chitin, a component of cells walls. It has an excitation maxima of 365 nm, an emission maxima of 435 nm and is viewed with a UV filter setting such as 4',6-diamidino-2-phenylindole (DAPI). Both fluorescent stains were made up to working concentrations as per manufacturer’s instructions. Fixed samples were thawed and stained with final concentrations of 4 μM of propidium iodide and 2.5 μM of Calcofluor™ White. For biofilm samples an aliquot (100 μL) of fixed sample was removed and stained. While for biofilm coupons the stain was added directly to the surface and allowed to sit. Samples were incubated with the stains for 30 min in the dark. Imaging occurred immediately after staining using a fluorescent filters UV filter (filter cube A) for Calcofluor™ White and the red filter (filter cube N2.1) for propidium iodide on a fluorescent microscope (DM4000B, Leica) at 400  magnification. Per sample replicate 12 fields of view were taken which were enumerated for microorganisms. Images were over-laid and analysed using the software (Lieca Imaging Suite) accompanying the microscope (Figure 3.10).

3.7.2.3 Quantification of fluorescently stained samples To quantify the number of fluorescently stained particles present in a particular field of view from a sample the images were imported into a software that quantified pixels (ImageJ) (Abramoff et al., 2004). The ImageJ is a free software available on line (http://rsbweb.nih.gov/ij/docs/faqs.html) and allows the user to define the area and pixel dimensions to be counted. The number of pixels can be used as a surrogate for the quantity of stain biomass present (Surman et al., 1996).

93 Chapter 3. Methodology

A B

 

Figure 3.10 Staining of Acanthamoeba sp. cysts. Cysts were stained with both Calcofluor™ White and propidium iodide. A: cysts viewed with UV filter showing the blue fluorescence of cell wall chitin stained by the Calcofluor™ White in the live cysts. B: the same cysts viewed with the red-filter showing the red fluorescence of nucleic acids stained by the propidium iodide which permeated the only cyst that was dead. Image taken at 400  magnification using a fluorescence microscope (DM400B, Lieca). Images labeled with a 10 μm scale bar. 3.7.3 Fluorescent microscopy of FLA infection with stained Legionella There are a number of stains available to track living cells. Mycobacterium avium was successfully tracked being up-taken by Acanthamoeba sp. after being stained with a fluorescent cell tracing dye (Vybrant® CFDA SE Cell Tracer Kit, Invitrogen™) (Berry et al., 2010). Although other ARM experiments using Cryptosporidium oocysts found that the stain could not be visualised after uptake although the concentration of stain used was not reported making it difficult to determine if low stain concentration may have been a contributing factor as well as the oocysts morphology (Scheid and Schwarzenberger, 2011). The stain consisted of a carboxyfluorescein diacetate, succinimidyl ester CFDA SE whichpassively diffuses into living cells where it reacts with intracellular esterases to yield fluorescence (Figure 3.11). The stain is retained and passed from parent to daughter cells but not to other cells in the population. The stain has maxima for excitation at 492 nm and emission at 517 nm and is viewed by fluorescent microscopy with a filter setting for the common fluorescein isothiocyanate stain (FITC) (Invitrogen). The cell tracing dye was applied to Legionella sp. being uptake by FLA species in culture based experiments (Figure 3.11).

94 Chapter 3. Methodology

Figure 3.11 Vybrant® stained Legionella pneumophila (ATCC® 33152) within Acanthamoeba castellanii (ATCC® 30234). White arrow indicates the stained Legionella up- taken by the FLA. Image taken using CSLM (FV1000) with a FITC filter and DIC at 1000  magnification with 2 μm scale bar. Fresh cultures of Legionella were grown as described (Section 3.2.2) and then 1 mL was concentrated by centrifugation 2000  g for 15 min and the supernatant replaced with 1 mL of warmed (37 °C) phosphate buffer saline (PBS) with 10 μm of stain (Vybrant® CFDA) and incubated at 37 °C for 15 min. PBS was made up to 1 L with 137 mM sodium chloride (Univar), 2.7 mM postassium chloride (Univar), 8.1 mM of disodium hydrogen phosphate (Univar), 1.76 mM of postassium dihydrogen phosphate (Univar) and 1000 mL of filtered water (Milli-Q) adjusted to pH 7.4. After incubation the stained Legionella was centrifuged again (2000  g for 15 min) and the supernatant replaced with 1 mL of warmed (37 °C) BYE broth and incubated at 37 °C for 30 min. After a final centrifugation under the same conditions the supernatant was removed and re-suspended in sterile filtered water (Milli-Q) and enumerated ready for immediate addition. Fresh FLA cultures were grown as described for culture collections (Section 3.2.1.1) and environmental isolates (Section 3.5.1). For the axenic ATCC FLA grown in high nutrient medium 1 mL of culture was centrifuged at 400  g for 6 min and the supernatant was replaced with filtered sterile water (Milli-Q) and enumerated for addition. FLA grown on NNA with an E. coli overlay were scraped and re-suspended in 1 mL of filtered sterile water (Milli-Q) and then enumerated. The experimental set-up is detailed in the relevant chapter (Chapter 7).

3.7.3.1 Water Research Centre lab fluorescent microscopy FLA and Legionella up-take experiments were enumerated using an improved Neubauer hemocytometer (Cole-Parmer) and a fluorescent microscope (DM400B, Lieca) at 400  magnification and a green filter (filter cube I3). All images were taken and modified with the software attached to the microscope (Lieca application suite).

95 Chapter 3. Methodology

3.7.3.2 US EPA lab fluorescent microscopy FLA and Legionella up-take experiments were enumerated using an improved Neubauer hemocytometer (Cole-Parmer) and a fluorescent microscope (Standard 2S, Zeiss, Oberkochen, Germany) at 400  magnification and a green filter (filter cube FT580). All images taken were modified using the software connected to the microscope (Axio-vision, Zeiss).

3.7.4 Confocal scanning laser microscopy A confocal scanning laser microscope (CSLM) (Fluoview FV1000, Olympus, Tokyo, Japan) at the Biomedical Medical Imaging Facility (UNSW) was used to image the fluorescent Legionella within the FLA cells. A FITC filter was used for the fluorescent Legionella which was over-laid with a differential interference contrast (DIC) image of the FLA (Figure 3.11). Images were taken in three dimensions at 1000  magnification oil immersion lens. Post image modification was undertaken using the software attached to the microscope (FV1000 viewer software, Olympus).

3.8 STATISTICAL METHODS 3.8.1 Raw data analysis

3.8.1.1 Software All raw data was entered into a spreadsheets (Excel, Microsoft, Redmond, USA) or a statistical software program (Prism®, GraphPad Software, La Jolla, USA). All statistical analysis and graphs were produced using a statistical software program (Prism®).

3.8.1.2 Normality tests Initially all raw data collect was first displayed in a scatter plot to gauge if it followed Gaussian distribution or not. Any outlying values were also checked and excluded only if a clear error in observing or recording the result could be determined. Tests for normalacy were conducted if greater than seven observations were present using the D'Agostino - Pearson test (D'Agostino, 1986). If the raw data was found not to be distributed according to Gaussian distribution it was transformed to logarithms in base 10. If after transformation the data followed Guassian distribution it was analysed in this form but all means were back transformed for reporting in the results text. However, standard deviations were not back transformed and reported in the text as this would give nonsense values. Data that still followed non-Guassian distribution after transformation was back transformed and analysed as non-Guassian distributed data.

96 Chapter 3. Methodology

3.8.2 Graphing and tests of significance

3.8.2.1 Gaussian distribution Data that displayed Gaussian distribution was displayed as bar graphs with arithmetic means, standard deviations above and below the mean. For each graph the total number of sample observations listed per treatment or sample type is given. Parametric statistical tests were applied to determine if any significant differences were present between data sets. The level of confidence was set at 95 % and hence to satisfy significant difference the p-value had to be equal to or less than 0.05. For the comparison of two un-matched data sets student t-tests were applied. For the comparison of three or more un-paired data sets 1 way and 2 way ANOVA tests were used with the Tukey-Kramer post test analysis which compares all pairs of results.

3.8.2.2 Non-Gaussian distribution For data that did not follow Gaussian distribution the data was presented as box plots with median, means (+), inter-quartile range (25 - 75 percentiles), whiskers from the 5 to 95 percentiles and outliers (•). For each graph of box plots the total number of sample observations was listed. Non-parametric statistical tests were applied to determine if any significant differences were present between data sets. For the comparison of two un-matched data sets the Mann-Whitney test was applied but only to samples that had greater than three observations. For comparison of three or more un-paired data sets the Kruskal-Wallis test was applied with Dunn's post test analysis to compare all pairs of results.

3.8.2.3 Linear regression Linear regressions were preformed for data sets and the slope and the y-intercept of the line were used to form the linear equation for the line. To test the goodness-of-fit of the linear regression the r2 value was calculated. The closer the r2 value was to a value of 1 the better the goodness-of-fit of the linear regression.

97 Chapter 3. Methodology

3.9 REFERENCES 1. Abramoff, M. D., P. J. Magalhaes and S. J. Ram. 2004. Image Processing with ImageJ. Biophotonics International. 11(7): 36-42. 2. Amann, R., W. Ludqwig and K. Schleifer. 1995 Phylogentic identification and in situ detection of individual microbial cells without cultivation. Microbiological Reviews. 59: 143-69. 3. Amann, R. I., B. J. Binder, R. J. Olson, S. W. Chisholm, R. Devereux and D. A. Stahl. 1990. Combination of 16S rRNA-targeted oligonucleotide probes with flow cytometry for analyzing mixed microbial populations. Applied and Environmental Microbiology. 56(6): 1919-1925. 4. Asano, T., F. Burton, H. Leverenz, R. Tsuchihashi and G. Tchobanoglous. 2007. Water reuse issues, technologies and applications. New York, McGraw Hill 5. Australian Government. 2004. Australian Drinking Water Quality Guidelines. National Health And Medical Research Council. 6. Barkovskii, A. L. and H. Fukui. 2004. A simple method for differential isolation of freely dispersed and particle-associated peat microorganisms. Journal of Microbiological Methods. 56(1): 93-105. 7. Behets, J., P. Declerck, Y. Delaedt, L. Verelst and F. Ollevier. 2007. Survey for the presence of specific free-living amoebae in cooling waters from Belgian power plants. Parasitology Research. 100(6): 1249-1256. 8. Berry, D., M. Horn, C. Xi and L. Raskin. 2010. Mycobacterium avium infections of Acanthamoeba strains: host strain variability, grazing acquired infections, and altered dynamics of inactivation with monochloramine. Applied and Environmental Microbiology. 76(19): 6685-6688. 9. Besner, M.-C., P. Servais and M. Prevost. 2008. Efficacy of disinfectant residual on microbial intrusion: a review of experiments. Journal American Water Works Association. 100(10): 116-130. 10. Chandy, J. and M. Angles. 2004. Factors influencing the development of biofilms under controlled conditions. Adelaide Cooperative Research Centre for Water Quality and Treatment 11. Chang, B., K. Sugiyama, T. Taguri, J. Amemura-Maekawa, F. Kura and H. Watanabe. 2009. Specific detection of viable Legionella cells by combined use of photoactivated ethidium monoazide and PCR/Real-Time PCR. Applied and Environmental Microbiology. 75(1): 147-153. 12. Chang, C.-W., Y.-C. Wu and K.-W. Ming. 2010. Evaluation of real-time PCR methods for quantification of Acanthamoeba in anthropogenic water and biofilms. Journal of Applied Microbiology. 9999(9999). 13. D'Agostino, R. B. 1986. Tests for normal distribution. Goodness-of-fit techniques. R. B. D'Agostino and M. A. Stepenes. New York, Macel Decker. 14. Daniels, L., R. Hanson and J. Phillips. 2007. Chemical analysis. Methods for General and Molecular Microbiology. C. A. Reddy, J. A. Breznak, G. Marzluf, T. M. Schmidt and L. R. Snyder. Washington, D.C., ASM Press. 15. De Jonckheere, J. F. 1998. Sequence Variation in the Ribosomal Internal Transcribed Spacers, Including the 5.8S rDNA, of Naegleria spp. Protist. 149(3): 221-228. 16. Declerck, P., J. Behets, V. van Hoef and F. Ollevier. 2007. Detection of Legionella spp. and some of their amoeba hosts in floating biofilms from anthropogenic and natural aquatic environments. Water Research. 41(14): 3159-3167.

98 Chapter 3. Methodology

17. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925. 18. Esterman, A., I. Calder, S. Cameron, D. Roder, R. Walters, P. Christy and B. Robinson. 1987. Determinants of the microbiological characteristics of spa pools in South Australia. Water Research. 21(10): 1231-1235. 19. Gallagher, S. and P. Desjardins. 2006. Quantitation of DNA and RNA with Absorption and Fluorescence Spectroscopy. Current Protocols in Molecular Biology. Appendix 3D. 20. Grunenwald, H. and K. Kramer. 2004. "Cloning a Real-Time PCR product: Does SYBR Green I dye interfere? ." Epicentre Forum 11, 17- 19. 21. Hendrickson, W. and D. Walthers. 2007. Nucleic acid analysis. Methods for General and Molecular Microbiology. C. A. Reddy. Washinton, D.C., ASM Press. 22. Hoffmann, R. and R. Michel. 2001. Distribution of free-living amoebae (FLA) during preparation and supply of drinking water. International Journal of Hygiene and Environmental Health. 203(3): 215-219. 23. Houang, E., D. Lam, D. Fan and D. Seal. 2001. Microbial keratitis in Hong Kong: relationship to climate, environment and contact-lens disinfection. Transactions of the Royal Society of Tropical Medicine and Hygiene. 95(4): 361-367. 24. Huws, S., J. Edwards, E. Kim and N. Scollan. 2007 Specificity and sensitivity of eubacterial primers utilized for molecular profiling of bacteria within complex microbial ecosystems. Journal of Microbiological Methods. 70 565-569. 25. Jonas, D., A. Rosenbaum, S. Weyrich and S. Bhakdi. 1995. Enzyme-linked immunoassay for detection of PCR-amplified DNA of legionellae in bronchoalveolar fluid. Journal of Clinical Microbiology. 33(5): 1247-1252. 26. Kuiper, M. W., B. A. Wullings, A. D. L. Akkermans, R. R. Beumer and D. van der Kooij. 2004. Intracellular proliferation of Legionella pneumophila in Hartmannella vermiformis in aquatic biofilms grown on plasticized polyvinyl chloride. Applied and Environmental Microbiology. 70(11): 6826-6833. 27. Kuipier, M., R. Valster, B. Wullings, H. Boonstra, H. Smidt and D. van der Kooji. 2006. Quantitative detection of the free-living amoeba Hartmannella vermiformis in surface water using real-time PCR. Applied and Environmental Microbiology. 72(9): 5750-5756. 28. Långmark, J., M. V. Storey, N. J. Ashbolt and T.-A. Stenström. 2007. The effects of UV disinfection on distribution pipe biofilm growth and pathogen incidence within the greater Stockholm area, Sweden. Water Research. 41(15): 3327-3336. 29. Lanocha, N., D. Kosik-Bogacka, A. Maciejewska, M. Sawczuk, A. Wilk and W. Kuzna- Grygiel. 2009. The occurrence Acanthamoeba (free living amoeba) in environmental and respiratory samples in Poland. Acta Protozoologica. 48: 271-279. 30. Larsen, P., J. L. Nielsen, M. S. Dueholm, R. Wetzel, D. Otzen and P. H. Nielsen. 2007. Amyloid adhesins are abundant in natural biofilms. Environmental Microbiology. 9(12): 3077-3090. 31. Lopez- Garcia, P., F. Rodriguez-Valera, C. Pedros-Allo and D. Moreira. 2001. Unexpected diversity of small eukaryotes in deep-sea Antarctic plankton. Nature. 409: 603-607. 32. Loret, J. F., S. Robert, G. Saucedo, V. Catalan, D. Corsaro and G. Grueb. 2008. Characterization of amoebae and intra-amoebal bacteria in drinking water and indentification of control strategies. American Water Works Association Water Quality and Technology Conference, Cincinnati, Ohio, American Water Works Association. 33. Ludwig, W., O. Strunk, R. Westram, L. Richter, H. Meier, Yadhukumar, A. Buchner, T. Lai, S. Steppi, G. Jobb, W. Frster, I. Brettske, S. Gerber, A. W. Ginhart, O.

99 Chapter 3. Methodology

Gross, S. Grumann, S. Hermann, R. Jost, A. Knig, T. Liss, R. Lºümann, M. May, B. r. Nonhoff, B. Reichel, R. Strehlow, A. Stamatakis, N. Stuckmann, A. Vilbig, M. Lenke, T. Ludwig, A. Bode and K. Ä. Schleifer. 2004. ARB: a software environment for sequence data. Nucleic Acids Research. 32(4): 1363-1371. 34. Marchesi, J. R., T. Sato, A. J. Weightman, T. A. Martin, J. C. Fry, S. J. Hiom and W. G. Wade. 1998. Design and Evaluation of Useful Bacterium-Specific PCR Primers That Amplify Genes Coding for Bacterial 16S rRNA. Applied and Environmental Microbiology. 64(2): 795-799. 35. Medlin, L., H. Elwood, S. Stickel and M. Sogin. 1988. The characterization of enzymatically amplified eukaryotic 16S -like rRNA coding regions. Gene. 71: 491-499. 36. Miyamoto, H., H. Yamamoto, K. Arima, J. Fujii, K. Maruta, K. Izu, T. Shiomori and S. Yoshida. 1997. Development of a new seminested PCR method for detection of Legionella species and its application to surveillance of Legionellae in hospital cooling tower water. Applied and Environmental Microbiology. 63(7): 2489-2494. 37. Nakatsu, C. and T. Marsh. 2007. Analysis of microbial communities and denaturing gradient gel electrophoresis and terminal restriction fragment length polymorphism. Methods for General and Molecular Microbiology. Washington D.C., ASM Press. 38. Nassonova, E., A. Smirnov, J. Fahrni and J. Pawlowski. 2010. Barcoding amoebae: comparison of SSU, ITS and COI genes as tools for molecular identification of naked lobose amoebae. Protist. 161(1): 102-115. 39. Packroff, G., J. Lawrence and T. Neu. 2002. In situ confocal laser scanning microscopy of protozoans in cultures and complex biofilm communities. Acta Protozoologica. 41: 245-253. 40. Peeters, E., H. J. Nelis and T. Coenye. 2008. Comparison of multiple methods for quantification of microbial biofilms grown in microtiter plates. Journal of Microbiological Methods. 72(2): 157-165. 41. Pernin, P., M. Pelandakis, Y. Rouby, A. Faure and F. Siclet. 1998. Comparative Recoveries of Naegleria fowleri Amoebae from Seeded River Water by Filtration and Centrifugation. Applied and Environmental Microbiology. 64(3): 955-959. 42. Pfaffl, M. 2004. Quantification strategies in real-time PCR. A-Z of quantitative PCR. S. A. Bustin. La Jolla, International University Line: 87 - 111. 43. Pruesse, E., C. Quast, K. Knittel, B. M. Fuchs, W. Ludwig, J. Peplies and F. O. Glockner. 2007. SILVA: a comprehensive online resource for quality checked and aligned ribosomal RNA sequence data compatible with ARB. Nucleic Acids Research. 35(21): 7188-7196. 44. Puzon, G. J., J. A. Lancaster, J. T. Wylie and J. J. Plumb. 2009. Rapid detection of Naegleria Fowleri in water distribution pipeline biofilms and drinking water samples. Environmental Science & Technology. 43(17): 6691-6696. 45. Qvarnstrom, Y., G. Visvesvara, R. Sriram and A. Da Silva. 2006. Multiplex real-time PCR assay for simultaneous detection of Acanthamoeba spp., and Naegleria fowleri. Journal of Clinical Microbiology. 44(10): 3589-3595. 46. Rivière, D., F. Ménard Szczebara, J.-M. Berjeaud, J. Frère and Y. Héchard. 2006. Development of a real-time PCR assay for quantification of Acanthamoeba trophozoites and cysts. Journal of Microbiological Methods. 64: 78-83. 47. Robinson, B., P. Monis and P. Dobson. 2006. Rapid, sensitive and discriminating identification of Naegleria spp. by real time PCR and melting curve analysis. Applied and Environmental Microbiology. 72(9): 5857 - 5863. 48. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241.

100 Chapter 3. Methodology

49. Salmon, E., K. von Lackum and J. Cabnabm. 2005. Proper alignment and adjusment of the light microscope. Current Protocols in Microbiology. 2A.1.1. 50. Scheid, P. and R. Schwarzenberger. 2011. Free-living amoebae as vectors of cryptosporidia. Parasitology Research: 1-6. 51. Shimadzu. 1993. Total Organic Carbon Analyzer model TOC-5000A, Shimadzu. 52. Shoff, M., A. Rogerson, S. Kessler, S. Schatz and D. Seal. 2008. Prevalence of Acanthamoeba and naked amoebae in South Florida domestic water. Journal of Water and Health. 6(1): 5. 53. Simoes, L., N. Azevedo, A. Pacheco, C. W. Keevil and M. Vieira. 2006. Drinking water biofilm assessment of total and culturable bacteria under different operating conditions. Biofouling 22(2): 91-99. 54. lapeta, J., D. Moreira and P. López-García. 2005. The extent of protist diversity: insights from molecular ecology of freshwater eukaryotes. Proceedings of the Royal Society B: Biological Sciences. 272(1576): 2073-2081. 55. Smirnov, A. and S. Brown. 2004. Guide to the methods of study and identification of soil gymnamoebae. Protistology. 3(3): 148-190. 56. Sogin, M. L. and J. H. Gunderson. 1987. Structural diversity of eukaryotic small subunit ribosomal RNAsa. Annals of the New York Academy of Sciences. 503(Endocytobiology): 125-139. 57. Storey, M. V. and N. J. Ashbolt. 2002. A comparison of methods and models for the analysis of water distribution pipe biofilms. Water Science and Technology: Water Supply. 2(4): 73-80. 58. Strap, J. 2007. Characterization of microeukaryota in natural environments. Manual of Environmental Microbiology. C. Hurst, R. Crawford, J. Garlandet al. Washington, DC, ASM Press. 59. Surman, S. B., J. T. Walker, D. T. Goddard, L. H. G. Morton, C. W. Keevil, W. Weaver, A. Skinner, K. Hanson, D. Caldwell and J. Kurtz. 1996. Comparison of microscope techniques for the examination of biofilms. Journal of Microbiological Methods. 25(1): 57-70. 60. Sutherland, I. W. 2001. The biofilm matrix - an immobilized but dynamic microbial environment. Trends in Microbiology. 9(5): 222-227. 61. Taff, H., J. Nett and D. Andes. 2011. Comparative analysis of Candida biofilm quantitation assays. Medical Mycology. Online: 1-5 62. The Ramaciotti Centre. 2010. Sequencing Protocol for ABI 3730 Capiliary Sequencer. T. R. Centre. 63. Thomas, V., T. Bouchez, V. Nicolas, S. Rober, J. F. Loret and Y. Lèvi. 2004. Amoebae in domestic water systems: resistance to disinfection treatments and implication in Legionella persistence. Journal of Applied Microbiology. 97(5): 950-963. 64. Thomas, V., K. Herrera-Rimann, D. S. Blanc and G. Greub. 2006. Biodiversity of amoebae and amoeba-resisting bacteria in a hospital water network. Applied and Environmental Microbiology. 72(4): 2428-2438. 65. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 66. Tsvetkova, N., M. Schild, S. Panaiotov, R. Kurdova-Mintcheva, B. Gottstein, J. Walochnik, H. Aspöck, M. Lucas and N. Müller. 2004. The identification of free- living environmental isolates of amoebae from Bulgaria. Parasitology Research. 92(5): 405-413. 67. Tyndall, R. L., K. S. Ironside, P. L. Metler, E. L. Tan, T. C. Hazen and C. B. Fliermans. 1989. Effect of thermal additions on the density and distribution of

101 Chapter 3. Methodology

thermophilic amoebae and pathogenic Naegleria fowleri in a newly created cooling lake. Applied and Environmental Microbiology. 55(3): 722-732. 68. Valster, R. M., B. A. Wullings, G. Bakker, H. Smidt and D. van der Kooij. 2009. Free- living protozoa in two unchlorinated drinking water supplies, identified by phylogenic analysis of 18S rRNA gene sequences. Applied and Environmental Microbiology. 75(14): 4736-4746. 69. Volk, C. J. and M. W. LeChevallier. 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Applied and Environmental Microbiology. 65(11): 4957-4966. 70. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 71. Whitchurch, C. B., T. Tolker-Nielsen, P. C. Ragas and J. S. Mattick. 2002. Extracellular DNA Required for Bacterial Biofilm Formation. Science. 295(5559): 1487. 72. White, T. 1996. The future of PCR technology: diversification of technologies and applications. Trends Biotechnology. 14(12): 478-83. 73. Wilhelmus, K. R., M. S. Osato, R. L. Font, N. M. Robinson and D. B. Jones. 1986. Rapid Diagnosis of Acanthamoeba Keratitis Using Calcofluor White. Arch Ophthalmol. 104(9): 1309-1312. 74. Zelver, N., M. Hamilton, D. Goeres and J. Heersink. 2001. Development of a standardized antibiofilm test. Methods in Enzymology, Academic Press. Volume 337: 363-376.

102

CHAPTER 4

FLA AND LEGIONELLA IN A RECYCLED WATER SCHEME 4 TABLE OF CONTENTS

4.1 INTRODUCTION ______105 4.1.1 Recycling water ______105 4.1.2 Water recycling plants (WRP)______106 4.1.3 Dual - distribution systems ______107 4.2 AIMS ______108 4.3 METHODS ______108 4.3.1 Water recycling plant sampling______108 4.3.2 Modified Robbins Device sample collection______110 4.3.3 Water quality ______111 4.3.4 Sample processing ______111 4.3.5 Culture methods______112 4.3.6 Molecular detection ______112 4.3.7 Microscopic detection ______113 4.4 RESULTS ______113 4.4.1 Water quality ______113 4.4.2 Biofilm quantification ______116 4.4.3 FLA detected by culture ______117 4.4.4 FLA detected by qPCR ______120 4.4.5 Legionella detection ______125 4.4.6 Protozoan detected by microscopy ______129 4.5 DISCUSSION ______131 4.5.1 WRP ______131 4.5.2 MRD ______132 Chapter 4. Recycled water scheme 4.5.3 Conclusions ______135 4.6 REFERENCES ______136

LIST OF TABLES

Table 4.1 Water recycling plant treatment processes and sampling______108 Table 4.2 FLA isolated by culture and identified by partial 18S rRNA sequencing __ 120 Table 4.3 FLA detected by qPCR and identified using cloning and sequencing _____ 124 Table 4.4 Identification of Legionella spp. detected by qPCR______128

LIST OF FIGURES

Figure 4.1 Water recycling plant UV disinfection unit ______109 Figure 4.2 Accessing the modified Robbins Device______109 Figure 4.3 Modified Robbins Device ______110 Figure 4.4 WRP water quality ______114 Figure 4.5 MRD water quality ______115 Figure 4.6 MRD heterotrophic plate counts ______116 Figure 4.7 DNA quantification from MRD ______117 Figure 4.8 FLA detected by culture ______118 Figure 4.9 Selected FLA identified by morphological and 18S rRNA______119 Figure 4.10 Gel Electrophoresis of H. vermiformis qPCR products______120 Figure 4.11 WRP FLA detected by qPCR ______122 Figure 4.12 MRD FLA detected by qPCR ______123 Figure 4.13 Phylogenetic tree of FLA detected in WRP and MRD ______125 Figure 4.14. Example of Legionella qPCR results for WRP samples ______126 Figure 4. 15. Detection of Legionella spp. in WRP using qPCR ______126 Figure 4.16. Detection of Legionella spp. in MRD using qPCR ______127 Figure 4. 17. Isolated FLA cultures positive for Legionella spp. ______128 Figure 4. 18. Phylogenetic tree of Legionella detected by qPCR in MRD______129 Figure 4. 19. Micro-organisms detected by fluorescent microscopy in MRD______130 Figure 4. 20. MRD protozoa direct counts by microscopy______131

104 Chapter 4. Recycled water scheme

4.1 INTRODUCTION

4.1.1 Recycling water Over two billion people live in areas of the world where there is not sufficient freshwater to meet basic needs (Oki and Kanae, 2006). Climate change and population growth will place increasing pressures on available freshwater resources especially in urban areas (Vairavamoorthy, 2008). Recycling wastewater for both drinking and non-drinking purposes is a viable solution with increasing up-take worldwide (Asano, 2004). As a water source recycled water offers a number of advantages including predictable volumes of wastewater to recycle and conservation of natural water resources via reduced freshwater withdrawal and decreased pollution by wastewater (Asano et al., 2007). Recycled water in some places is used to supplement traditional drinking water and hence used for drinking, cooking and washing (Asano et al., 2007). While more commonly recycled water is used for non-drinking purposes including cooling towers (Levine, 2003), toilet flushing (Lazarova et al., 2003), irrigation and pools (Asano et al., 2007). Highly problematic to the use of recycled water is the public's perception of consuming treated wastewater and fears about the safety of the recycled water (Toze, 2006). There are documented occurrences of infection associated with recycled water. In Australia there was a recent Legionella outbreak at a self-service carwash which utilised recycled water which resulted in seven people being hospitalised (McArthur, 2008). Furthermore both FLA and Legionella spp. have been isolated from recycled water distribution systems (Storey and Kaucner, 2009) indicating that there is likely to be an interaction present that may be facilitating the growth of CAP causing pathogen ARB. An additional reason for concern is that there are documented cases where recycled water has been used in applications for which it was not intended such as filling swimming pools (Catchpole, 2004) because it is cheaper than the non-recycled water. Also there have been documented cases of cross connections within water recycling schemes where recycled water is used as drinking water (Storey et al., 2007). For these reasons it is essential that microbial pathogens be controlled in recycled water as exposure to recycled water aerosols (O'Toole et al., 2009) containing pathogenic CAP can occur. To examine if CAP causing Legionella bacteria were present in a recycled water system and if FLA were facilitating that growth, water was examined from a recycled water scheme. The recycled water scheme consists of a water recycling plant (WRP) and drinking and recycled water distribution system. The scheme has been in operation for nearly a decade and treats 4.7 billion litres

105 Chapter 4. Recycled water scheme of wastewater to supply over 19000 residences with recycled water for non-drinking purposes including toilet flushing, garden watering, washing cars and other outdoor uses (Fairbairn, 2006). Overall the provision of recycled water reduced demand for drinking water by 40 % in the area (Fairbairn, 2006). The recycled water is supplied to residences via a second distributions system that runs in parallel to the traditional water pipes supplying treated surface water of drinking water quality. This type of distribution is commonly referred to as a dual-distribution system and is present in a number of other urban developments both in Australia and internationally (Jimenez and Asano, 2008).

4.1.2 Water recycling plants (WRP) The quality of the recycled water produced is dependant on both the type of wastewater to be recycled and the treatment processes used (Toze, 2006). In the case of the WRP sampled the recycled wastewater being treated was sewage and the treatment processes were primary, secondary and then tertiary treatment with final disinfection via ultraviolet radiation and chlorination (Table 4.1). No literature could be identified where FLA and pathogenic ARB were examined in recycled water treatment plants. However, some of the primary and secondary processes used in recycled water treatment are similar to drinking water treatment plants that have been surveyed for FLA and pathogenic ARM. In particular it was found that FLA were on occasion able to avoid disinfection and breakthrough the treatment process at densities up to 1 amoebae.mL-1 (Hoffmann and Michel, 2001; Thomas et al., 2008). The break-through events were linked to the fact that FLA had colonised the filtration stage (Hoffmann and Michel, 2001; Thomas et al., 2008) and clarifying sludge (Corsaro et al., 2010) of the treatment process. Furthermore, FLA from the genus Hartmannella were more likely than other FLA genera to be reported breaking through into the water distribution system (Corsaro et al., 2010). FLA isolated from drinking water treatment plants are known to host a range of pathogenic ARB (Hoffmann and Michel, 2001; Thomas et al., 2008) and the ability of FLA to by-pass the treatment process could carry those ARB into the distribution system. Once in the distribution system it has been shown that some pathogenic ARB are able to regrow and cause infections namely Legionella spp. (Stout et al., 1992) and Mycobacterium spp. (Marshall et al., 2011). The absence of data on FLA removal, and any association with Legionella spp., in a water recycling plant was addressed in this study by sampling from the final treatment stages of the WRP.

106 Chapter 4. Recycled water scheme 4.1.3 Dual - distribution systems The presence of FLA and pathogenic ARB within recycled water drinking systems has only been briefly addressed. A survey of six recycled water distribution systems across Australia found that by culture Legionella spp. (other than L. pneumophila) found with the highest densities detected to be 36 cells.mL-1 in water and 16 cells.cm-2 in biofilm (Storey and Kaucner, 2009). While FLA were detected at a maximum density of 60 amoebae.mL-1 in water and 60 amoebae.cm-2 in the biofilm (Storey and Kaucner, 2009). Specifically, in the distribution system of the recycled water scheme sampled no Legionella spp. were detected by culture but FLA were detected at low densities (Storey and Kaucner, 2009). Detection of both FLA and Legionella by culture is known to underestimate the occurrence of these micro-organisms compared to PCR based molecular techniques (Wellinghausen et al., 2001; Smirnov and Brown, 2004; Devos et al., 2005; Joly et al., 2006; Diederen et al., 2007; Chen and Chang, 2010; Wullings et al., 2011). Hence, the studies of recycled water systems reviewed provide only very limited information about the density and diversity of FLA. More research is critically needed to determine the potential of FLA to host pathogenic ARM in recycled water schemes. Recycled water distribution systems differ from drinking water distribution systems only with respect to the type of water being distributed. Hence, the review findings that FLA are detected at a mean frequency of 52 % (n = 8) in drinking water distribution systems (Table 2.6) highlights that FLA can grow in distribution systems and these sites should be targeted in any investigation of a water system. Distribution systems are likely to facilitate the growth of FLA in part due to the formation of biofilm on the internal surface of pipes. There have been comparatively few studies looking at FLA in the biofilm of distribution systems. Using qPCR targeting N. fowleri the densities of this species of FLA were estimated at between 1 - 432 amoebae.10-7 bacteria (Puzon et al., 2009) while in the other study only one biofilm sample was positive (0.43 amoebae. cm-2) for H. vermiformis (Valster et al., 2009). Additionally, distribution system sediment has been identified to contain FLA at greater density and diversity than biofilm in the same system (Corsaro et al., 2010). Our limited understanding of FLA in the biofilm of drinking water distribution systems and the absence of any detailed data about FLA in recycled water systems needs to be addressed. This study aims to fill these identified knowledge gaps by examining both the water and biofilm of the drinking and recycled water systems utilising a pipe sampling device termed a modified Robbins Device (MRD).

107 Chapter 4. Recycled water scheme 4.2 AIMS The first aim of this research was to determine the density and diversity of FLA and Legionella spp. in the final stages of the water recycling plant (WRP) and the finished recycled water. The second aim was to determine the density and diversity of FLA and Legionella spp. in the water and biofilm of the dual-distribution system. FLA and Legionella spp. were investigated using both culture and qPCR in all samples except the WRP samples where culture methods were not used for Legionella spp. due to sampling limitations. During both sampling stages water quality characteristics were measured and supplied by the water utility to determine any relationship with the microorganisms of interest. Finally select FLA that had been isolated and sub-cultured were screened for the presence of any Legionella spp. residing as stable ARB.

4.3 METHODS

4.3.1 Water recycling plant sampling The water recycling plant (WRP) was sampled over a six week period during the Autumn months of April and May in 2008. Grab samples of treated water were collected once a week from each of the final four stages of the WRP (Table 4.1, Figure 4.1) by water utilities monitoring teams using 500 mL bottles (polyethylene terephthalate) and cold transported back to the lab. Samples were processed within 24 hr of collection.

Table 4.1 Water recycling plant treatment processes and sampling Flow Sampling Treatment stages Treatment process details rate points (L.s-1) Primary settlement Screening and solids separation 180 - Biological reactor - FeCl addition - Secondary clarification 3 180 Clarification - NaOH addition Rapid mixing - alum addition Post Tertiary clarification 180 Flocculation - Cationic polymer clarification Post deep bed Deep bed filtration Slow sand filtration 180 filtration UV photolysis with medium Post UV UV disinfection 180 pressure 10 ML.d-1 reactor disinfection NaOCl addition to achieve 6 - 7 Post super- Super-chlorination 153 mg.L-1 for 80 - 140 min. chlorination Stored prior to pumping into - Finished recycled water 153 distribution system

108 Chapter 4. Recycled water scheme

Figure 4.1 Water recycling plant UV disinfection unit. The recycled water flows from left to right through the unit and is disinfected by medium pressure UV lamps.

Figure 4.2 Accessing the modified Robbins device. The device is located underground in a confined space and is accessed by a man hole at street level.

109 Chapter 4. Recycled water scheme 4.3.2 Modified Robbins Device sample collection The modified Robbins Device (MRD) is a biofilm collecting device installed in the dual- distribution system and requires confined space training to access it (Figure 4.2). The device is located approximately 3 km from the nearest serving potable and recycled reservoirs and 10 km from supplying WRP. The dual-distribution system at this point services 160 residential dwellings. The original Robbins Device design (McCoy et al., 1981) was modified to consist of two un- plasticised polyvinyl chloride (uPVC) pipes 2 m in length with an internal diameter of 150 mm. The pipes each contain 60 biofilm sampling ports each filled with a stainless steel screw plug fitted with a removable biofilm coupon (15 mm x 50 mm) (Storey, 2002) (Figure 4.3 - A). Drinking water MRD coupons were made of stainless steel to replicate the inert hydrophilic surface of the cement- lined ductile iron pipes of the drinking water distribution system (Storey, 2002) (Figure 4.3-B). While recycled water MRD coupons were cut from the same uPVC piping material of the recycled water distribution system (Storey, 2002) (Figure 4.3 - C). The coupons had been in situ in the MRD collecting biofilm for nearly five years.

Figure 4.3 Modified Robbins device. A: the MRD underground with the recycled water mains pipe (purple) in the fore ground and the drinking water pipe (blue) behind. The screws in the pipes are the points at which the coupons are inserted. B: drinking water biofilm coupon made from stainless steel. C: recycled water biofilm coupon made from uPVC un-plasticised polyvinyl chloride.

110 Chapter 4. Recycled water scheme Due to restriction of access sampling from the MRD occurred once in February 2009 with the assistance of the water utilities maintenance team (Figure 4.2). Twelve biofilm coupons and 2 L of water were collected from both the drinking and recycled water MRD by either draining from the MRD or using the sampling taps present. Glass bottles were used for water sample collection while biofilm coupons were removed from screw points with sterile forceps and then each coupon was placed in a 50 mL sterile tube (Sarstedt, Nümbrecht, Germany) containing water from the corresponding distribution pipe, as described (Section 3.3).

4.3.3 Water quality Water quality data were supplied by the water utility for the sampled stages of the WRP over the six weeks of sampling. For the MRD water quality data for four months prior to sampling was supplied from a routine sampling site approximately 200 m from the MRD. Water quality variables supplied and analysed for both sampling sites were: temperature, pH, turbidity, total organic carbon, free and combined chlorine (both mono-chlorine and di-chloramine). Additionally for the treated water samples from the WRP and the MRD the water utility supplied microbiological water quality data including heterotrophic plates counts (HPC), total coliform and Escherichia coli counts. Also, at time of sampling from the MRD, grab samples were taken for water quality testing to determine if data supplied from the routine sampling sites were representative data for the MRD site. Water quality characteristics analysed were temperature, pH, turbidity, total organic carbon, total phosphate, and free and combined chlorine as described (Section 3.4.1 and 3.4.2). Using 0.5 mL of concentrated sample in duplicate heterotrophic plate counts and total coliform counts were performed as described (Section 3.4.3). Additionally the biomass in water and biofilm samples was estimated by quantification of the extracted DNA as described (Section 3.4.3).

4.3.4 Sample processing Water and biofilm samples from both water types were analysed by culture, molecular and microscopic methods. Hence, the 12 samples taken for each sample and water type were split into three groups giving four samples for each type of analysis (culture, molecular, and microscopic). Specifically each drinking and recycled water sample (2 L) was split into 12 replicates each of 100 mL which were concentrated by centrifugation as described (Section 3.3.3). Supernatant was removed to leave 10 mL of concentrate in four replicates used for culture analysis and only 1 mL in eight replicates used for molecular and microscopic analysis. Of the 12 MRD biofilm coupons for each water type, eight of the coupons to be analysed by culture and molecular methods had the

111 Chapter 4. Recycled water scheme biofilm removed by scraping. The remaining four biofilm coupons and four water concentrates replicates were fixed directly for microscopic analysis as described (Section 3.7.2).

4.3.5 Culture methods

4.3.5.1 FLA detection by culture FLA were detected and isolated from samples by culture using non-nutrient agar (NNA) plates with Escherichia coli overlays. Aliquots (0.5 mL) of the four concentrated water and four biofilm samples were spread plated in replicates of four for each of the water types (64 plates in total) (Section 3.5.1). Three isolated FLA were identified by partial 18S rRNA amplification using primers with a higher specificity for FLA (Section 3.6.2.1) and direct sequencing (Section 3.6.6.1). While the remaining isolates (n = 14)were identified by partial 18S rRNA amplification using a universal primer (Section 3.6.2.2) and cloning (Section 3.6.5) and sequencing (Section 3.6.6.2).

4.3.5.2 Detection of Legionella by culture Legionella bacteria were detected using 0.5 mL aliquots of concentrated water and biofilm samples in duplicate and the initial stages of the standard method (Section 3.5.2). Presumptive Legionella colonies were then confirmed using targeted PCR for Legionella 16S rRNA gene (Section 3.6.3). If products of the correct size were identified they were cloned (Section 3.6.5) and sequenced (Section 3.6.6.2).

4.3.6 Molecular detection

4.3.6.1 DNA extraction For molecular analysis DNA was extracted individually from the scrapings of four biofilm coupons and four concentrated water samples for both drinking and recycled water (16 samples in total). DNA was extracted using a cell lysis and DNA purification kit for environmental samples (FastDNA® SPIN Kit for Soil) as previously described (Section 3.6.1.2). DNA extraction losses and qPCR inhibition previously calculated were used to adjust qPCR results (Section 3.6.1.3).

4.3.6.2 Detection of FLA by qPCR For the detection of FLA by qPCR three different genera of FLA were targeted with individual PCR reactions in duplicate with 5 μL of DNA extract. Cell based standard curves for trophozoites were used to quantify each FLA qPCR reaction as described (Section 3.6.3.1). A selection of positive qPCR samples were selected for cloning and sequencing. Selected from the Naegleria spp. targeted qPCR were all three positive samples from the WRP. From the Acanthamoeba spp. targeted

112 Chapter 4. Recycled water scheme qPCR three out of 14 positive samples were selected. Finally from the H. vermiformis targeted qPCR three out of the 13 positive samples were selected.

4.3.6.3 Legionella detection by qPCR Legionella were detected by qPCR using duplicate reactions with 2.5 μL in replicates for four. Legionella cell based calibration curves were used for quantification (Section 3.6.3.3). From this qPCR all of the three positive samples were submitted for cloning and sequencing.

4.3.6.4 Legionella spp. detection in isolated FLA A selection of FLA (n = 9) from drinking and recycled water that had been isolated, identified and subcultured for 18 months were screened for the presence of Legionella sp. residing as ARB as described (Section 3.6.3.4). Both positive samples were submitted for cloning and sequencing.

4.3.6.5 Cloning, sequencing and identification of positive PCR/qPCR products For each cloning reaction two to six cloned DNA fragments (Section 3.6.5) were sequenced in the forward direction only using the provided forward primer (M13) (Section 3.6.6.2) and analysed (Section 3.6.7).

4.3.6.6 Fluorescent microscopy of samples Replicate (n = 4) drinking and recycled water (100 mL) and biofilm samples were concentrated, fixed, stained and imaged as described (Section 3.7.2). For each sample 12 fields of view were imaged and enumerated for large microorganisms.

4.4 RESULTS

4.4.1 Water quality

4.4.1.1 Water recycling plant water quality From the water recycling plant (WRP) water quality data was reported for the four treatment stages sampled (Figure 4.4). There was minimal change in water temperature (range: 18.8 - 22.6 °C) and total organic carbon (range: 6.4 - 9.0 mg.L-1) over the sampling period. While mean pH increased gradually from clarification (pH 6.8,  = 0.10) to chlorination (pH 7.4,  = 0.12). Chlorine was not added in the first two stages of sampling but after UV disinfection the disinfectant was added and was highest after super chlorination with free chlorine mean of 4.9 mg.L-1 ( = 0.7) and combined chlorine mean of 1.1 mg.L-1 ( = 0.06) (Figure 4.4 - D). The recycled water was

113 Chapter 4. Recycled water scheme maintained at this super-chlorination level for 80 to 140 min prior to release into the distribution system (CT = 480 - 840).

Figure 4.4 WRP water quality. Graphs A, B, C and D present the mean and standards deviation of six weeks of water quality data collected at four WRP stages. Each bar graph represents six samples. 4.4.1.2 MRD water quality Water quality data for the modified Robbins device (MRD) was collected at the time of sampling and also supplied by the water utility for four months period prior to collection. The mean recycled water temperature (26.6 °C,  = 1.42) was significantly higher (t-test, p = 0.008) than drinking water (24.3 °C,  = 0.49) (Figure 4.5). While the pH of recycled water (pH 7.6,  = 0.14) was significantly lower (t-test, p = 0.038) than drinking water (pH 8.0,  = 0.04). Total organic carbon (TOC) mean for recycled water (9.5 mg.L-1) was significantly higher (t-test, p = 0.0003) than drinking water (4.1 mg.L-1). Mean total nitrogen (N) and total phosphorus (P) were also higher for

114 Chapter 4. Recycled water scheme recycled water (N = 3.4 mg.L-1 and P = 0.03 mg.L-1) than drinking water (N = 1.8 mg.L-1 and P = 0.01 mg.L-1). Free and combined chlorine concentrations for both drinking and recycled water varied considerably. Recycled water had an mean free (0.49 mg.L-1) and combined (0.28 mg.L-1) chlorine concentration that was higher than drinking water free (0.28 mg.L-1) and combined (0.10 mg.L-1) chlorine but this was not significant (t-test, p = 0.3). The water quality of recycled water was consistent with the recycled water leaving the WRP except that the mean chlorine residual had dropped one fold from 4.9 to 0.49 mg.L-1.

Figure 4.5 MRD water quality. Graphs A, B and D present water quality data collected at the MRD and water utility testing sight. While graph C represent total organic carbon value water taken at time of sampling from the MRD. Each bar shows the mean and standard deviation for six replicate samples except for graph C where two replicate samples were used.

115 Chapter 4. Recycled water scheme As part of the water quality data heterotrophic plate counts (HPC) were taken at the time of sampling from the MRD (Figure 4.6). Consistent with the lower disinfectant residuals (Figure 4.5 - D), drinking water had the highest mean HPC count (1.7  103 cfu.mL-1). Recycled water and biofilm had comparable HPC counts (140 cfu.mL-1and 141 cfu.cm-2), while drinking water biofilm had the lowest HPC counts (62 cfu.mL-1). The differences in HPC counts between the sample types were significant (1-way ANOVA , p < 0.0001) and may represent a useful parameter for comparison to FLA isolation frequencies.

Figure 4.6 MRD heterotrophic plate counts. Data was log10 transformed for drinking and recycled water and biofilm samples. Each bar represents the mean and one standard deviation for four replicate samples.

4.4.1.3 Biofilm quantification DNA quantities were determined for each sample type, recycled water biofilm had the highest mean quantities of microbial biomass present as DNA extract (502 ng.cm-2,  = 122) but this was not higher than drinking water biofilm (448 ng.cm-2,  = 67) (Figure 4.7). The DNA quantities for the water samples were significantly lower (1-way ANOVA, p < 0.0001) but followed a similar pattern to biofilm samples. Recycled water samples had the highest mean DNA concentrations but with high variability (117 ng.mL-1,  = 116) followed by drinking water (88 ng.mL-1,  = 5.9).

116 Chapter 4. Recycled water scheme

Figure 4.7 DNA quantification from MRD. Each bar represents the mean with one standard deviation for four replicate samples. 4.4.2 FLA detected

4.4.2.1 FLA detection by culture FLA were detected using non-nutrient agar plates with E. coli overlays. The data from both the WRP and MRD were found not to follow Gaussian distribution hence non parametric tests were used (Figure 4.8). In the WRP FLA were detected after clarification (1.8 amoebae.mL-1) and then increased after the filtration stage (4.8 amoebae.mL-1) but not significantly (Mann Whitney test, p = 0.5). However, the highest FLA densities (13 amoebae.mL-1) were also detected after filtration. After UV and chlorine disinfection no FLA were detected by culture which was significantly lower compared to the initial treatment stages (Kruskal-Wallis test, p = 0.01). FLA were detected in the MRD at the highest numbers in the recycled water biofilm (0.56 amoebae.cm-2, max = 4.5 amoebae.cm-2) which was slightly higher than drinking water (0.49 amoebae.mL-1, max = 4.0 amoebae.mL-1). While fewer FLA were detected in recycled water samples (0.11 amoebae.cm-2) and drinking water biofilm (0.19 amoebae.mL-1) these differences were not significant (Kruskal-Wallis test, p = 0.36).

117 Chapter 4. Recycled water scheme

Figure 4.8 FLA detected by culture. A: WRP data for water over six weeks sampling from four different stages with 12 samples represented per box plot. B: MRD data for drinking and recycled water and biofilm samples with 16 samples per box plot. Box plot of the non-Guassian data, with means (+) and whiskers at 5-95 percentiles. FLA isolated with distinct morphologies were identified by partial 18S rRNA sequencing (Table 4.2) and were imaged (Figure 4.9). The FLA successfully sequenced from the WRP was identified as Naegleria gruberi which was not identified by culture in the recycled water of the MRD (Table 4.2). The two MRD drinking water isolates successfully identified were H. vermiformis while the recycled water isolates (n = 5) were identified as an Acanthamoeba sp. with no assigned speciation. However, none of the FLA isolated by culture from biofilm were successfully re-cultured and hence none were sequenced. Also a number of other FLA were not able to be sub-cultured and identified by the chosen methods. Fungi were also co-isolated with the FLA identified and include Auriculariaceace sp., Aspergillus sp., Fusarium spp., and Penicillum sp.

118 Chapter 4. Recycled water scheme

Figure 4.9. Selected FLA identified by morphological and 18S rRNA based PCR Naegleria gruberi in cyst (A) and trophozoite (B) form isolated from the clarification stage of WRP. Acanthamoeba sp. cyst (C) and trophozoite (D) from MRD recycled water. Hartmannella vermiformis cyst (E) and trophzoite (F) from MRD drinking water. Images taken at 1000  magnification using light microscopy and a 10 μm scale bar.

119 Chapter 4. Recycled water scheme

Table 4.2 FLA isolated by culture and identified by partial 18S rRNA sequencing

Source Sampling point Isolate # Seq. BLAST best match and BLAST % length ARB alignment (accession #) 99 % WRP Clarification 1-3-50ai 653 Naegleria gruberi (AB298288) 99 % MRD Drinking water pfcii 575 Hartmannella vermiformis (DQ084364) 99 % pfcih 575 Hartmannella vermiformis (DQ084364) 99 % Recycled water rfaih 629 Acanthamoeba sp. (AY148956) 99 % rfbi 629 Acanthamoeba sp. (AY148956) 99 % rfbii 629 Acanthamoeba sp. (AY148956) 99 % rfbih 629 Acanthamoeba sp. (GQ397476) 99 % rfbiih 629 Acanthamoeba sp. (GQ397476) 4.4.2.2 FLA detected by qPCR All qPCR reaction products were imaged using gel electrophoresis to confirm the correct products sizes and to rule out non-specific amplification. H. vermiformis qPCR gel with an expected product length of 502 bp is provided as an example (Figure 4.10).

Figure 4.10 Gel Electrophoresis of H. vermiformis qPCR products for WRP and MRD. L: ladder 3 kb - 100 bp; P: positive control; Samples 1, 4: clarification stage; 2, 5: filtration stage. 6: UV disinfection; 3, 7: chlorination; 8-9: drinking water biofilm; 10-11: recycled water biofilm; 12- 13: drinking water; 14-16: recycled water; N: negative control. Run on 1.5 % agarose gel in TBE buffer, with Sybr Gold visualisation.

120 Chapter 4. Recycled water scheme Based on direct qPCR, FLA in the recycled water treatment plant were primarily detected after the clarification and filtrations stages (Figure 4.11). In the WRP samples Acanthamoeba spp. were detected at mean trophozoite densities of 3.4 amoebae.mL-1 after clarification but at the highest mean and maximum after the filtration stage (mean = 3.8 amoebae.mL-,1, max = 14.6 amoebae.mL-,1 ) (Figure 4.11 - A). While after UV and chlorine disinfection Acanthamoeba spp. were detected at less than a single organism equivalent (< 0.003 amoebae.mL-1). Similarly, detection of H. vermiformis appeared highest after the clarification (1.7 amoebae.mL-1) but it was after the filtration stages that had the highest maximum detections (3.5 amoebae.mL-1) (Figure 4.11 - B). H. vermiformis was also detected in one sample after super chlorination at low densities (2.7 amoebae.mL-1). Naegleria were detected in a similar pattern to Acanthamoeba although were present at lower mean densities after clarification (0.36 amoebae.mL-1) and filtration stages (1.5 amoebae.mL-1) (Figure 4.11 - C). Within each set of the non-Gaussian distributed qPCR results there was found to be no significant difference between the treatment stages and the FLA detected (Kruskal-Wallis test, p = 0.39).

121 Chapter 4. Recycled water scheme

Figure 4.11. WRP FLA detected by qPCR. A: Acanthamoeba spp. targeted qPCR data. B: Hartmannella vermiformis targeted qPCR. C: Naegleria spp. targeted qPCR. Data is presented a box plot with means (+), whiskers from 5-95 percentiles and for four replicate samples. Acanthamoeba was detected at the highest mean densities (4.6 amoebae.cm-2) and maximum detection (22 amoebae.cm-2) in drinking water biofilm from the MRD (Figure 4.12). The next highest mean concentration was detected in recycled water biofilm (2.2 amoebae.cm-2), followed by drinking water (0.9 amoebae.mL-1) and the lowest mean concentration was found in recycled water (0.2 amoebae.mL-1). The detection of H. vermiformis by qPCR did not follow the same pattern as Acanthamoeba. The highest H. vermiformis were detected in drinking water (2.9 amoebae.mL-1). The remaining samples all had very similar detection frequencies for H. vermiformis with detection decreasing slightly from recycled water (0.9 amoebae.mL-1), drinking water biofilm (0.7 amoebae.cm-2) to recycled water biofilm (0.5 amoebae.cm-2). There was no detection of Naegleria 122 Chapter 4. Recycled water scheme spp. in any of the MRD samples. For both FLA qPCR results there was found to be no significant difference between FLA densities and the water types (Kruskal-Wallis test, p = 0.39).

Figure 4.12 MRD FLA detected by qPCR. A: Acanthamoeba spp. targeted qPCR data. B: Hartmannella vermiformis targeted qPCR. Data presented a box plot with means (+) , whiskers from 5-95 percentiles for four replicate samples.

4.4.2.3 Identification of FLA species detected by qPCR The Naegleria species detected by qPCR in the WRP was confirmed as N. spitzbergeniensis (Table 4.3). While, from the MRD the Acanthamoeba detected was identified only as Acanthamoeba sp. for both drinking and recycled water samples. The specificity of the H. vermiformis primers was confirmed with the sequencing results returning all H. vermiformis positive. BLAST best matches were compared to the alignment within the ARB software generated phylogenetic tree (Figure 4.13).

123 Chapter 4. Recycled water scheme Table 4.3 FLA detected by qPCR and identified using cloning and sequencing Sampling sample Seq. BLAST best match and BLAST % Source point # length ARB alignment (accession #) 99% RHS 1-3 415 Naegleria spitzbergeniensis (AM157660) Clarification 99% WRP RHS 2-3 418 Naegleria spitzbergeniensis (AM157660) 99 % Filtration RHS 1-4 414 Naegleria spitzbergeniensis (AM157660) 100 % Acpfa 132 Acanthamoeba sp. Drinking (HM055890) water 99 % Hvpfb 502 Hartmannella vermiformis (DQ407567) Drinking 99 % MRD Hvpa 501 Hartmannella vermiformis water biofilm (DQ407567) 99 % Hvrb10 502 Hartmannella vermiformis Recycled (DQ407567) water biofilm 100 % Acrb10 63 Acanthamoeba sp. (EU683884)

124 Chapter 4. Recycled water scheme

Figure 4.13 Phylogenetic tree of FLA detected in WRP and MRD. Sequences of FLA detected by qPCR were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the drinking water pipe are blue and are red for the recycled water samples. The scale bar for the branch lengths is 0.10 changes per nucleotide. 4.4.3 Legionella detection

4.4.3.1 Legionella detection by culture

Of the 11 positive samples that were identified by the initial stages of the standard method for Legionella detection none were found to be Legionella spp. after the application of the Legionella spp. PCR.

125 Chapter 4. Recycled water scheme 4.4.3.2 Quantification of Legionella spp. qPCR results All qPCR reaction products were imaged using gel electrophoresis to confirm the correct products sizes and to rule out non-specific amplification. Legionella targeted qPCR gel with an expected product length of 452 bp is provided as an example (Figure 4.14).

Figure 4.14 Example of Legionella qPCR results for WRP samples. Lanes L: E-Gel DNA ladder, P: Legionella pneumophila positive control, 1: clarification sample, 2: filtration sample, 3: chlorination sample, N: negative control. WRP samples were positive for Legionella spp. by qPCR after clarification (401 cells.mL-1,  = 256) and filtration (150 cells.mL-1,  = 150). No positive Legionella spp. qPCR results were recorded after UV or chlorination disinfection stages over the three weeks of sampling (Figure 4.15). The difference between the treatment stages were found not to be significant (1-way ANOVA, p = 0.097).

Figure 4.15 Detection of Legionella spp. in WRP using qPCR. Legionella cells numbers calculated for water (mL-1) using Legionella pnuemophila standard curves. Each bar graph represents the mean and one standard deviations for four replicate samples.

126 Chapter 4. Recycled water scheme For the MRD samples only one recycled water and one recycled biofilm sample were positive out of all eight replicates. Recycled water had one positive Legionella spp. sample had 40 cells.mL-1. While, the only Legionella spp. positive recycled water biofilm sample was 14 cells.cm-2. There were no positive results for any of the drinking water or biofilm samples taken (Figure 4.16). One positive qPCR product from the MRD samples was cloned, sequenced and identified as Legionella anisa (Table 4.4) and aligned in a phylogenetic tree using ARB (Figure 4.18).

Figure 4.16 Detection of Legionella spp. in MRD using qPCR. Legionella cells numbers calculated for water (mL-1) and biofilm (cm-2) using Legionella pnuemophila standard curves. Each box plot with means (+) and whiskers from 5-95 percentiles for four replicate samples. 4.4.3.3 Detection of Legionella spp. in FLA isolates. All FLA isolates (n = 16) were screened for the presence of Legionella using PCR (Figure 4.17).

127 Chapter 4. Recycled water scheme

Figure 4.17 Isolated FLA cultures positive for Legionella spp. Lanes L: E-Gel DNA ladder, P1: Legionella pneumophila positive control, P2: Environmental Willertia magna, 1: drinking water isolate (pfcii), 2: recycled water isolate (rfbii), N: negative control. Due to low transformation success during cloning none of the Legionella PCR positives from these FLA isolates were sequenced. However, other PCR products from FLA isolates samples from the annular reactors were successfully sequenced and identified as Legionella micadedi (Chapter 6). From the PCR it can be deduced that ARB were present in the FLA cultures and it is likely that the drinking water isolate contained a species of Legionella while the recycled water isolate ARB may not be Legionella spp. due to the larger size of the PCR product (Figure 4.17).

4.4.3.4 Identification of positive Legionella spp. qPCR products There were only two positive qPCR products from the recycled MRD biofilm and water samples (Figure 4.16).These samples were cloned, sequenced, BLAST searched (Table 4.4) and aligned using nearest neighbour algorithms in ARB (Figure 4.18). Only the recycled water sample was sequenced successfully and positively identified as Legionella anisa.

Table 4.4 Identification of Legionella spp. detected by qPCR. BLAST best match, # Seq. % ID Source Sampling point ARB alignment and PCR length (accession #) RDP 99 % MRD Recycled water 1 454 Legionella anisa (FN667912)

128 Chapter 4. Recycled water scheme

Figure 4.18 Phylogenetic tree of Legionella detected by qPCR in MRD. Sequences of Legionella spp. detected by qPCR were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the drinking water are blue. The scale bar for the branch lengths is 0.10 changes per nucleotide. 4.4.4 Protozoa detected by microscopy

4.4.4.1 Fluorescent microscopy of fixed samples Fluorescent microscopy of fixed samples revealed a number of micro-organisms with a length ranging from 3 - 10 μm (Figure 4.19).

129 Chapter 4. Recycled water scheme

Figure 4.19 Microorganisms detected by fluorescent microscopy in MRD samples. A: drinking water concentrate with the presences of three micro-organisms with similar morphology to FLA trophozoites. B: recycled water concentrate with four likely FLA trophozoites. C: drinking water biofilm sample with a single likely FLA cyst. D: Recycled water biofilm with a likely FLA trophozoite and four FLA cysts. The background colour for this image appears green due to the purple plastic of the slide. Images taken using fluorescence microscopy with UV filter, 400  magnification and 10 μm scale bar.

Microorganisms of the correct size range (3 - 10 μm) were counted over 12 random fields of view then adjusted (Figure 4.20). Total mean direct counts were highest in recycled water biofilm (8  104 cells.cm-2) and lowest in the drinking water biofilm (3  104 cells.cm-2). However the data was highly variable and the differences between sample types was not significant (Kruskal-Wallis test, p = 0.37).

130 Chapter 4. Recycled water scheme

Figure 4.20 MRD protozoa direct counts by microscopy. Likely protozoa (3 - 10 μm in length) were enumerated by direct fluorescent microscopy. Data displayed as box plots with means (+) and whiskers from 5-95 percentiles for 12 replicate samples. 4.5 DISCUSSION

4.5.1 Water Recycling Plant The water quality in the WRP varied little for temperature, pH or TOC concentration between the four treatment stages sampled (Figure 4.4). Over six weeks of sampling FLA and Legionella were detected in the filtration and clarification stages of the WRP. From the FLA detected by culture (filtration mean 4.8 amoebae.mL-1) only Naegleria spp. was positively identified by DNA sequencing using the FLA targeted primers (Tsvetkova et al., 2004) as fungi isolated with the FLA were dominating the PCR reactions. However, based on morphology other FLA species were also present. Of the three FLA targeted by qPCR, Acanthamoeba were found to most readily colonise the filtration stage of the WRP (3.8 amoebae.mL-1, max = 14.6 amoebae.mL-1). While, H. vermiformis and Naegleria spp. also colonised but at lower densities (Figure 4.11). Legionella spp. were also found at high densities in clarification (401 cells.mL-1) and filtration stages (150 cells.mL- 1). The filtration stage in water treatment plants has been identified as site for FLA and pathogenic ARB interactions due to the presence of biofilm (Thomas et al., 2008). 131 Chapter 4. Recycled water scheme After UV and super-chlorination stages no FLA were detected by culture and no Legionella spp. were detected by qPCR. However FLA were detected by qPCR in two samples which were positive for H. vermiformis. In one water sample taken after super chlorination (free chlorine 4.9 mg.L-1) H. vermiformis was detected at a concentration of 2.7 amoebae.mL-1. This may indicate that the FLA, possibly in the resistant cyst form, can effectively withstand disinfection in the WRP. However, from the qPCR method used the viability of the H. vermiformis detected could not be established but these results are consistent with other studies of FLA in wastewater treatment plants. Of all the FLA genera detected in wastewater treatment plants Hartmannella in particular are reported to colonise the filtration stage of the treatment plant then resist disinfection and pass into finished water (Hoffmann and Michel, 2001; Thomas et al., 2008; Corsaro et al., 2010). The densities of Hartmannella reported were only low (0.1 amoebae.mL-1)(Hoffmann and Michel, 2001) but this is likely due to the utilisation of less sensitive culture methods. The detection of Hartmannella in some WRP samples after super-chlorination indicates that the FLA could be transporting ARB into the recycled water system. To determine the full extent of FLA by-passing treatment it is recommend that greater volumes of water (> 1 L) be sampled from the final stages of treatment over a number of months rather than the six weeks used in this study. Furthermore, the qPCR reactions for Acanthamoebae and H. vermiformis need to be further optimised for the filtration and clarification samples as non- specific amplification in some samples (Figure 4.10) may have led to an underestimation of FLA densities. Additionally, if more cloning and sequencing resources (for bi-directional sequencing) were available then FLA isolated by culture should be rapidly screened for both identification by 18S rRNA sequencing and by targeted PCR for ARB such as Legionella and Mycobacterium. From the six weeks of sampling conducted at the WRP it can be determined that a range of FLA and Legionella spp. effectively colonise both the clarification and filtration stages leading to possible residing of Legionella spp. in FLA as ARB. Further that some FLA of the genera Hartmannella may be able resist disinfection and pass into the recycled water system harbouring pathogenic ARB such as Legionella spp.

4.5.2 Modified Robins Device The drinking and recycled water from the MRD had distinct water quality profiles. Recycled water had a significantly higher temperature, TOC, nitrogen and phosphorous than drinking water . The mean free chlorine residual for recycled water (0.49 mg.L-1) was also higher than drinking water (0.28 mg.L-1). Higher concentrations of nutrients (TOC, N and P) in water has been shown to

132 Chapter 4. Recycled water scheme contribute to greater microbial growth (Volk and LeChevallier, 1999) and biofilm formation (Chandy and Angles, 2004). Mean biofilm concentrations for recycled water were higher than drinking water but not significantly. However, the higher free chlorine concentration in the recycled water at the WRP (4.9 mg.L-1) and in the distribution system appear to be effectively controlling the growth of microorganisms. This is evident by the fact that the heterotrophic plate counts were 10 times higher for drinking water than recycled water (Figure 4.6). Also the mean biofilm concentrations for drinking and recycled water were nearly equivalent (Figure 4.7). The consistently higher free chlorine concentration in the recycled water also appears to have influenced FLA and Legionella growth. By culture FLA were detected in the drinking water (0.49 amoebae.mL-1) and recycled water biofilm (0.56 amoebae.cm-2) at similar mean densities (Figure 4.8). Only seven isolates were successfully identified by RNA sequencing and were found to be most closely related to H. vermiformis and Acanthamoeba sp. (Table 4.2). From the small sample number it is not possible to draw any conclusions from the fact the H. vermiformis isolates were from drinking water while Acanthamoeba sp. isolates were from recycled water. Using the standard method for detection of Legionella spp. a number of bacteria were isolated on the selective plates (GVPC and BCYE) but by PCR none of these bacteria were confirmed to be Legionella spp. Using qPCR it was shown that Acanthamoeba and H. vermiformis are present in both the drinking and recycled distribution systems although at higher densities in the drinking water system. The highest densities by qPCR for Acanthamoebae were in drinking water biofilm (4.6 amoebae. cm-2) and for H. vermiformis in drinking water (2.9 amoebae. mL-1) (Figure 4.12). Sequencing of nine of the qPCR products confirmed the specificity of the qPCRs and that the Acanthamoeba isolates could not be given clear speciation based on 18S rRNA sequence alignment (Figure 4.13). The classification of FLA is in a period of change with old morphological classification systems being replaced with new phylogenetic based systems (Pawlowski and Burki, 2009). Hence, from the 18S rRNA sequence alignments undertaken in ARB software (Ludwig et al., 2004) Acanthamoeba sp. previously assigned to the species A. castellanii and A. polyphaga based on morphology are in fact not different based on phylogentic analysis. Therefore the Acanthamoeba sp. detected by both culture and qPCR will not be assigned species names as they do not belong to any distinct species. Interestingly no Naegleria spp. were detected by qPCR in any of the MRD samples indicating that these FLA was effectively removed during the treatment process of the WRP (Figure 4.11). This is consistent with other studies that detected Naegleria spp. in the water treatment plants but not in the finished water (Thomas et al., 2008). Naegleria spp. have been isolated from the

133 Chapter 4. Recycled water scheme drinking water distribution system biofilms in Australia previously by qPCR (Puzon et al., 2009). However, Naegleria spp. do not appear to be a member of FLA community in the dual drinking and recycled water system surveyed. This could be related to temperature as Naegleria spp. are more frequently found in water systems from warmer environments and water sources (Blair et al., 2008). The overall diversity of FLA identified in this research is likely to be an underestimate. This is because the E. coli overlay NNA methods are known to bias certain FLA and exclude the growth of others (Smirnov and Brown, 2004). Additionally there is only a limited number of qPCR primers available to target FLA and the 18S rRNA gene is not optimal for identification of FLA as a group (Nassonova et al., 2010). The diversity of FLA is likely to be greater than identified within this study and new methods need to be developed in order to identify that diversity. In particular strain identification is needed in order to track the sources of FLA within the system. Legionella spp. were detected in the recycled water distribution system at low densities but not in the drinking water system by qPCR. Only two samples were positive at densities of 40 cells.mL-1 in recycled water and 15 cells.cm-2 in recycled biofilm (Figure 4.16). From these low detection frequencies it is not possible to determine if the recycled water facilitated the growth of Legionella spp. to a greater degree than the drinking water. Another consideration is that Legionella pneumophila has been shown to grow more extensively on surfaces made of polyvinyl chloride (PVC) compared to stainless steel (Armon et al., 1997). This may have influenced the detection rates on the drinking water system with cement lined pipes and stainless steel biofilm coupons compared to the recycled water pipes and coupons made from PVC. Also as molecular techniques are not able to determine the viability of the microorganisms whose DNA is detected it is possible that the Legionella DNA may not have been from viable cells. The detected Legionella spp. was most closely related to Legionella anisa (Table 4.4 and Figure 4.18) which has been associated with disease (Fields et al., 2002) but is not reported as a contributor to legionellosis infections worldwide (Bartram et al., 2007). Detection of Legionella sp. in chlorinated distribution systems could not be identified in the literature (Table 2.5) so it is difficult to determine if these results are consistent with other distribution systems. From the MRD nine of the FLA isolates were screened for ARB Legionella spp. and two isolates were found to be positive, one from drinking water and the other from recycled water (Figure 4.17). This is despite the fact that the FLA had been isolated over 12 months prior and sub- cultured repeatedly. This indicates that Legionella spp. may reside as a stable ARB within FLA and use this mechanism to survive and grow in the water distribution system. At present the Australian

134 Chapter 4. Recycled water scheme Drinking Water Guidelines have not set a guideline value for Legionella spp. in drinking water and disinfection action is only advised when the use of the water is likely to present an exposure risk such as in hospitals and cooling towers. However, considering the ability for FLA to increase the growth and virulence of Legionella species (Cirillo et al., 1994; Neumeister et al., 2000)) more research is needed to determine if the densities and diversity of FLA and Legionella spp. isolated from the dual-drinking and recycled water system present a health risk. It must also be kept in mind that FLA are not the only eukaryotic microorganism capable of hosting pathogenic ARB. Cilliates such as Tetrahymena sp. have also been shown to host Legionella spp. as ARB (Fields et al., 1984) and it is plausible that other eukaryotic microorganisms can act as hosts and/or transport vectors. Using fluorescent microscopy the concentrated water and biofilm samples were examined for larger microorganisms between 3 - 10 μm in length that would be likely eukaryotes. The densities based on microscopy were a lot higher than the culture or qPCR results for FLA. Overall there was no significant difference between the sample types recycled water biofilm has the highest number of larger microorganisms by direct count (8 x 104 cells.cm-2) and the lowest was in the drinking water biofilm (3 x 104 cells.cm-2) (Figure 4.20). The high densities of large microorganisms in the samples of unknown identity presents an avenue that should be pursued further to determine the role that these microorganisms are having in the overall health related quality of the drinking and recycled water.

4.5.3 Conclusions This study revealed that FLA are consistently isolated from drinking and recycled water and biofilm and that some of those FLA contain pathogenic Legionella as ARB. To increase our understanding of the FLA populations and the other ARB present be future research should sample from the MRD are more regular intervals. Additionally, as with the WRP samples if cloning and sequencing resources are available then all FLA isolated from the system should be promptly identified and screened for a range of pathogenic ARB. The MRD set-up presents a unique piece of equipment to sample distribution system biofilm for drinking and recycled water. To make full utilisation of this resource a study that included drinking and recycled water samples from the residence supplied by the distribution system pipes within the MRD would give a complete picture of the how the density and diversity of the FLA populations and also to what extent end water uses may be exposed to FLA containing pathogenic ARB.

135 Chapter 4. Recycled water scheme

4.6 REFERENCES 1. Armon, R., J. Starosvetzky, T. Arbel and M. Green. 1997. Survival of Legionella pneumophila and Salmonella typhimurium in biofilm systems. Water Science and Technology 35(11-12): 293-300. 2. Asano, T. 2004. Water recycling - A relevant solution? Workshop on Water Crisis - Myth or Reality, Santander, SPAIN, Taylor & Francis Ltd. 3. Asano, T., F. Burton, H. Leverenz, R. Tsuchihashi and G. Tchobanoglous. 2007. Water reuse issues, technologies and applications. New York, McGraw Hill 4. Bartram, J., Y. Chartier, J. Lee, K. Pond and S. Surman-Lee. 2007. Legionella and the prevention of Legionellosis. Geneva, World Health Organisation 5. Blair, B., P. Sarkar, K. Bright, F. Marciano-Cabral and C. Gerba. 2008. Naegleria fowleri in well water. Emerging Infectious Diseases. 14(9): 1499-1501. 6. Catchpole, H. 2004. Recycled water could risk health. A. N. i. Science. 7. Chandy, J. and M. Angles. 2004. Factors influencing the development of biofilms under controlled conditions. Adelaide Cooperative Research Centre for Water Quality and Treatment 8. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634. 9. Cirillo, J. D., S. Falkow and L. S. Tompkins. 1994. Growth of Legionella pneumophila in Acanthamoeba castellanii enhances invasion. Infection and Immunity. 62(8): 3254-3261. 10. Corsaro, D., G. S. Pages, V. Catalan, J.-F. Loret and G. Greub. 2010. Biodiversity of amoebae and amoeba-associated bacteria in water treatment plants. International Journal of Hygiene and Environmental Health. 213(3): 158-166. 11. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925. 12. Diederen, B., C. d. Jong, I. Aarts, M. F. Peeters and A. v. d. Zee. 2007. Molecular evidence for the ubiquitous presence of Legionella species in Dutch tap water installations. Journal of Water and Health. 5(3): 375 - 383. 13. Fairbairn, I. 2006. Operation of an STP for recycled water production plant. 69th Annual Water Industry Engineers and Operators' Conference. Bendigo, Australia. 14. Fields, B., R. Benson and E. Besser. 2002. Legionella and Legionnaires' disease: 25 years of investigation. Clinical Microbiology Reviews. 15(3): 506-526. 15. Fields, B. S., E. B. Shotts, Jr, J. C. Feeley, G. W. Gorman and W. T. Martin. 1984. Proliferation of Legionella pneumophila as an of the ciliated protozoan Tetrahymena pyriformis. Applied and Environmental Microbiology. 47(3): 467-471. 16. Hoffmann, R. and R. Michel. 2001. Distribution of free-living amoebae (FLA) during preparation and supply of drinking water. International Journal of Hygiene and Environmental Health. 203(3): 215-219. 17. Jimenez, B. and T. Asano. 2008. Water reuse - an international survey of current practice, issues and needs. London, International Water Association Publishing. 18. Joly, P., P.-A. Falconnet, J. Andre, N. Weill, M. Reyrolle, F. Vandenesch, M. Maurin, J. Etienne and S. Jarraud. 2006. Quantitative real-time Legionella PCR for environmental water samples: data interpretation. Applied and Environmental Microbiology. 72(4): 2801- 2808.

136 Chapter 4. Recycled water scheme 19. Lazarova, V., S. Hills and R. Birks. 2003. Using recycled water for non-potable, urban uses: a review with particular reference to toilet flushing. Water Science & Technology: Water Supply. 3(4): 69-77. 20. Levine, A. D. 2003. Use of Reclaimed Wastewater for Cooling Tower Applications. Membranes for Industrial Wastewater Recovery and Reuse, International Water Association, International Water Association. 21. Ludwig, W., O. Strunk, R. Westram, L. Richter, H. Meier, Yadhukumar, A. Buchner, T. Lai, S. Steppi, G. Jobb, W. Frster, I. Brettske, S. Gerber, A. W. Ginhart, O. Gross, S. Grumann, S. Hermann, R. Jost, A. Knig, T. Liss, R. Lºümann, M. May, B. r. Nonhoff, B. Reichel, R. Strehlow, A. Stamatakis, N. Stuckmann, A. Vilbig, M. Lenke, T. Ludwig, A. Bode and K. Ä. Schleifer. 2004. ARB: a software environment for sequence data. Nucleic Acids Research. 32(4): 1363-1371. 22. Marshall, H. M., R. Carter, M. J. Torbey, S. Minion, C. Tolson, H. E. Sidjabat, F. Huygens, M. Hargreaves and R. Thomson. 2011. Mycobacterium lentiflavum in drinking water supplies, Australia. Emerging Infectious Diseases. 17(3): 7. 23. McArthur, G. 2008. Legionella outbreak puts seven in hospital Herald Sun. Melbourne. 24. McCoy, W. F., J. D. Bryers, J. Robbins and J. W. Costerton. 1981. Observations of fouling biofilm formation. Canadian Journal of Microbiology. 27: 910-917. 25. Nassonova, E., A. Smirnov, J. Fahrni and J. Pawlowski. 2010. Barcoding Amoebae: Comparison of SSU, ITS and COI Genes as Tools for Molecular Identification of Naked Lobose Amoebae. Protist. 161(1): 102-115. 26. Neumeister, B., G. Reiff, M. Faigle, K. Dietz, H. Northoff and F. Lang. 2000. Influence of Acanthamoeba castellanii on intracellular growth of different Legionella species in human monocytes. Applied and Environmental Microbiology. 66(3): 914-919. 27. O'Toole, J., M. Keywood, M. Sinclair and K. Leder. 2009. Risk in the mist? Deriving data to quantify microbial health risks associated with aerosol generation by water-efficient devices during typical domestic water-using activities. Water Science and Technology. 60(11): 2913- 2920. 28. Oki, T. and S. Kanae. 2006. Global hydrological cycles and world water resources. Science. 313(5790): 1068 - 1072. 29. Pawlowski, J. and F. Burki. 2009. Untangling the phylogeny of amoeboid protists. Journal of Eukaryotic Microbiology. 56(1): 16-25. 30. Puzon, G. J., J. A. Lancaster, J. T. Wylie and J. J. Plumb. 2009. Rapid detection of Naegleria Fowleri in water distribution pipeline biofilms and drinking water samples. Environmental Science & Technology. 43(17): 6691-6696. 31. Smirnov, A. and S. Brown. 2004. Guide to the methods of study and identification of soil gymnamoebae. Protistology. 3(3): 148-190. 32. Storey, M. 2002. The ecology of enteric viruses within distribution pipe biofilm. School of Civil and Environmental Engineering. Sydney, The University of New South Wales. PhD. 33. Storey, M., D. Deere, A. Davision, T. Tam and A. Lovell. 2007. Risk management and cross- connection detection of a dual reticulation system. Australian Water Association Water Reuse and Recycling conference, Sydney, Australia. 34. Storey, M. and C. Kaucner. 2009. Understanding the growth of opportunistic pathogens in distribution systems. Cooperative Research Centre for Water Quality and Treatment. Adelaide. Report 79: 120. 35. Stout, J. E., V. L. Yu, P. Muraca, J. Joly, N. Troup and L. S. Tompkins. 1992. Potable water as a cause of sporadic cases of community-acquired Legionnaires' Disease. New England Journal of Medicine. 326(3): 151-155.

137 Chapter 4. Recycled water scheme 36. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 37. Toze, S. 2006. Water reuse and health risks - real vs. perceived. Desalination. 187: 41-51. 38. Tsvetkova, N., M. Schild, S. Panaiotov, R. Kurdova-Mintcheva, B. Gottstein, J. Walochnik, H. Aspöck, M. Lucas and N. Müller. 2004. The identification of free-living environmental isolates of amoebae from Bulgaria. Parasitology Research. 92(5): 405-413. 39. Vairavamoorthy, K. 2008. Innovation in water management for the city of the future. International Urban Water Conference, Heverlee, Belgium, CRC Press-Taylor & Francis Group. 40. Valster, R. M., B. A. Wullings, G. Bakker, H. Smidt and D. van der Kooij. 2009. Free- living protozoa in two unchlorinated drinking water supplies, identified by phylogenic analysis of 18S rRNA gene sequences. Applied and Environmental Microbiology. 75(14): 4736-4746. 41. Volk, C. J. and M. W. LeChevallier. 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Applied and Environmental Microbiology. 65(11): 4957-4966. 42. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 43. Wullings, B. A., G. Bakker and D. van der Kooij. 2011. Concentration and diversity of uncultured Legionella spp. in two unchlorinated drinking water supplies with different concentrations of natural organic matter. Applied and Environmental Microbiology. 77(2): 634-641.

138 CHAPTER 5 FLA AND LEGIONELLA IN GARDEN HOSES 5. TABLE OF CONTENTS

5.1 INTRODUCTION ______141 5.1.1 Garden hoses ______141 5.1.2 Legionella in garden hoses ______141 5.1.3 FLA in garden hoses ______141 5.2 AIMS ______142 5.3 METHODOLOGY______142 5.3.1 Experimental set up______142 5.3.2 Water quality______143 5.3.3 Sample processing______143 5.3.4 Culture methods ______144 5.3.5 Molecular detection______145 5.4 RESULTS ______145 5.4.1 Water quality______145 5.4.2 FLA detection ______149 5.4.3 Legionella detection ______154 5.5 DISCUSSION ______158 5.5.1 Hose water quality______158 5.5.2 FLA in hoses ______158 5.5.3 Legionella in hoses ______159 5.5.4 Conclusions ______160 5.6 REFERENCES______161

Chapter 5. Garden hoses LIST OF TABLES

Table 5.1 FLA isolated by culture from hoses and identified by sequencing______143 Table 5.2 FLA detected by qPCR from hoses and identified by sequencing ______152 Table 5.3 Identity of Legionella qPCR products confirmed by cloning and sequencing _____ 156

LIST OF FIGURES

Figure 5.1. Drinking water garden hose set-up______143 Figure 5.2 Cross-sections of garden hoses ______144 Figure 5.3 Water quality for the supplying drinking water and the two hoses______147 Figure 5.4 Hose heterotrophic plate counts ______148 Figure 5.5 Quantification of hose biofilm using DNA ______149 Figure 5.6 Hose FLA detected by culture______150 Figure 5.7 H. vermiformis qPCR gel electrophoresis image ______151 Figure 5.8 Hose FLA detected by qPCR over 18 months ______151 Figure 5.9 Phylogenetic tree of FLA detected in hoses ______153 Figure 5.10 Legionella spp. qPCR gel electrophoresis image ______154 Figure 5.11 Detection of Legionella sp. in hoses using qPCR ______155 Figure 5.12 Phylogenetic tree of Legionella spp. detected in hoses by qPCR ______157

140 Chapter 5. Garden hoses

5.1 INTRODUCTION

5.1.1 Garden hoses Garden hoses are ubiquitous in household gardens throughout the developed world. In Australia there are over 6.4 million private dwellings with yards or gardens (Australian Bureau of Statistics, 2006) and based on conservative estimates of at least one garden hose per private dwelling there are six million garden hoses in use across Australia. Garden hoses have relatively large internal surface areas which present an excellent environment for the growth of microorganisms especially as water often sits in hoses for extended periods of time between uses and becomes stagnant (Schofield, 1985; Rollins and Colwell, 1986; Lee et al., 2002). A garden hose with a length of 20 m and an internal diameter of 12.5 mm has approximately 7.9  104 cm2 of internal surface area for biofilm formation. Biofilm is known to facilitate the growth and persistence of a range of potentially pathogenic microorganisms in a range of environments (LeChevallier et al., 1996; Storey and Ashbolt, 2001; Costerton, 2004; Declerck, 2010). For example, hoses with rubber linings were found to facilitate the growth of heterotrophic bacteria biofilms in hospital water distribution systems and the biofilms could not be removed by purging or repeated chlorine disinfection (Buffet-Bataillon et al., 2010).

5.1.2 Legionella in garden hoses There is limited data available on the legionellosis risk from exposure to garden hose aerosols. Nonetheless, garden hoses have been linked to an outbreak of Legionnaires disease and Pontiac Fever after four people became infected after using a private garden spa (Euser et al., 2010). Identical Legionella pneumophila serogroup 1 was detected in two garden hose samples, the garden spa and the garden shower of the private dwelling (Euser et al., 2010). While it was likely that the garden spa was the source of infection, the affected people also had contact with the other contaminated sources of water so transmission from the garden hoses could not be excluded (Euser et al., 2010). The growth of pathogenic Legionella within other hose types has been reported also. Pathogenic L. pneumophila has been shown to grow in the biofilm lining extension hoses of hospital showers at densities greater than 3  103 CFU.mL-1 and were only removed after disinfection with chlorine dioxide (Walker et al., 1995). Hoses, and garden hoses in particular, present an ideal environment for the growth and replication of pathogenic Legionella which have been over looked by previous researchers.

5.1.3 FLA in garden hoses No literature was identified that reported on FLA in any types of garden hoses. However, FLA have been reported in hoses of dental units (Barbeau and Buhler, 2001) that have 141 Chapter 5. Garden hoses similar characteristics to garden hoses in that they are used intermittently, left moist and have a large internal surface area for the growth of biofilm. FLA were detected in dental unit lines at densities of up to 330 amoebae.mL-1 with five different genera identified (Barbeau and Buhler, 2001). Compared to dental unit hoses garden hoses have an additional factor which is likely to facilitate the growth of FLA in that they are often left in the sun and the water inside the hoses warms to higher temperatures than the surrounding air temperature. It is highly likely that garden hoses present an ideal environment for both FLA growth and interaction with pathogenic ARM. Furthermore, the use of garden hoses presents a common human exposure pathway to aerosols potentially containing up-regulated pathogenic ARM.

5.2 AIMS The main aim of this research was to characterise the density and diversity of FLA and Legionella in the water and biofilm of two garden hose types supplied with drinking water over 18 months. Furthermore, isolated FLA were screened for any infection by Legionella and used in up-take and release experiments with L. pneumophila. Water quality variables were also collected to examine any relationship with FLA and Legionella occurrence.

5.3 METHODOLOGY

5.3.1 Experimental set up

Brand new garden hoses 20 m in length were attached to outdoor drinking water taps and set in place for a total of 18 months. Two separate hose types were used in the experiments: hose A and hose B. Hose A set-up utilised a standard 12.5 mm internal diameter green and black nylon braided plastic garden hose (Pope, Beverly, Australia). While hose B set-up used a 12.5 mm internal diameter lilac plastic garden hose recommend for use with recycled water (Dual Irrigation, Dural, Australia). The hoses were connected to the tap using a brass two way adapter with valves and brass click-on adaptors. The hoses were filled with water and sealed at the ends with poly pipe adaptors and a brass valve (Figure 5.1).

142 Chapter 5. Garden hoses A B

Figure 5.1. Drinking water garden hose set-up. A: hose A set-up with a standard green garden hose. B: hose B set -up with a lilac hose recommended for use with recycled water. 5.3.2 Water quality Samples were taken from the taps supplying the garden hoses and from the hose set-ups at least once every two months with a higher frequency (fortnightly) during the summer months over the course the experiment. Water quality components analysed were temperature, pH, turbidity, total organic carbon, total phosphate, and free and combined chlorine as described (Section 3.4.1 and 3.4.2). Using 0.5 mL of water and re-suspended biofilm in duplicate heterotrophic plate counts and total coliform counts were also performed as described (Section 3.4.3) for each of the five samplings. Additionally the biomass in water and biofilm samples was estimated by quantification of the extracted DNA as described (Section 3.4.3).

5.3.3 Sample processing The hoses were sampled at six, seven, eight, nine and 18 months during the experiment. However, only samples from six, nine and 18 months were analysed by molecular methods. To sample the hoses 50 mL of water was collected from the tap supplying the hoses and also from the end of each hose. The water samples were split into three samples of 10 mL. One 10 mL sample was used directly for culture analysis and the remaining two 10 mL volumes were concentrated by centrifugation to leave 1 mL for molecular and microscopic analysis as described (Section 3.3.3). Hose biofilm was collected by cutting a 30 cm length of hose from the end closest to the supplying tap with a sterile Stanly knife and then the hose was reconnected. The outside of the hose was wiped down with disinfectant (0.1 % sodium hypochlorite (Barkley 143 Chapter 5. Garden hoses and Michelson, 2003)), the ends sealed with film (Parafilm) and transported in sterile plastic bags to prevent contamination of the samples. Biofilm was clearly visible on the inside of the garden hoses (Figure 5.2). The removed hose length was cut into three 10 cm lengths and biofilm was removed from two sections by scraping as described (Section 3.3.3.2). The cells scrapings were used directly for culture methods and concentrated to 1 mL for molecular methods. The remaining 10 cm hose length and concentrated water samples were fixed directly for microscopic analysis as described (Section 3.7.2). A B

1 cm

Figure 5.2 Cross-sections of garden hoses. A: Hose A cross-section. B: Hose B cross-section with white biofilm clearly visible in contrast to the cleared area scraped in the top left corner. Scale bar represents 1 cm. 5.3.4 Culture methods

5.3.4.1 FLA detection by culture FLA were detected and isolated from samples by culture using non-nutrient agar (NNA) plates with E. coli overlays. Aliquots (0.5 mL) of water and scraped biofilm from the six samples (two taps, two hose water and two hose biofilm) were spread plated in duplicate for each of the water types (12 plates in total) for each sampling (Section 3.5.1). Twelve isolated FLA were identified by partial 18S rDNA amplification using primers with a higher specificity for FLA (Section 3.6.2.1) and direct sequencing (Section 3.6.6.1).

5.3.4.2 Detection of Legionella by culture Legionella bacteria were detected using 0.5 mL aliquots of water and biofilm samples in duplicate and the initial stages of the standard method (Section 3.5.2).

144 Chapter 5. Garden hoses 5.3.5 Molecular detection

5.3.5.1 DNA extraction For molecular analysis DNA was extracted individually from the four concentrated water samples and two concentrated biofilm samples for each of the three sampling periods (18 samples in total). DNA was extracted using a cell lysis and DNA purification kit for environmental samples (FastDNA® SPIN Kit for Soil) as previously described (Section 3.6.1.2). DNA extraction losses and qPCR inhibition previously calculated for drinking water were used to adjust qPCR results (Section 3.6.1.3).

5.3.5.2 Detection of FLA by qPCR For the detection of FLA by qPCR three different genera of FLA were targeted with individual PCR reactions in duplicate with 5 μL of DNA extract. Cell based standard curves for trophozoites were used to quantify each FLA qPCR reaction as described (Section 3.6.3.1). A selection of positive qPCR samples were selected for cloning and sequencing. From the Acanthamoeba spp. target qPCR two out of 18 positive samples were selected and four out of the 17 H. vermiformis target qPCR positive samples.

5.3.5.3 Legionella detection by qPCR Legionella were detected by qPCR using duplicate reactions with 2.5 μL in replicates of four. Legionella cell based calibration curves were used for quantification (Section 3.6.3.3). From this qPCR six positive samples were submitted for cloning and sequencing.

5.3.5.4 Legionella detection in isolated FLA A selection of FLA (n = 12) had been isolated, identified and sub-cultured for up to 21 months were screened for the presence of Legionella sp. residing as ARB as described (Section 3.6.3.4). Positive samples were submitted for cloning and sequencing.

5.3.5.5 Cloning, sequencing and identification of positive PCR/qPCR products For each cloning reaction two to six cloned DNA fragments (Section 3.6.5) were sequenced in the forward direction only using the provided forward primer (M13) (Section 3.6.6.2) and analysed (Section 3.6.7).

5.4 RESULTS

5.4.1 Water quality The water temperature within the hoses were consistently higher than the drinking water supplying the hoses over the 18 months of sampling (Figure 5.3 - A). At day 173 in particular during summer the water from hose A (standard garden hose) was 23.6 °C higher and hose B 145 Chapter 5. Garden hoses (lilac hose) was 7.3 °C higher than the supplying drinking water (29.7 °C). There was no observable difference between the supplying water pH and the that of the water from the hoses (Figure 5.3 - B). The turbidity of hose B water was also consistently higher than the supplying drinking water and hose A (Figure 5.3 - C). At day 242 the turbidity of hose B was more than 19 NTU higher than both the supplying drinking water (0.34 NTU) and hose A (0.98 NTU). The available free chlorine in both the supplying drinking water and the garden hoses was quite low and ranged from only 0.023 to 0.24 mg.L-1(Figure 5.3 - D). The concentrations of total organic carbon (TOC) was consistently higher in the hose water compared to the supplying drinking water (Figure 5.3 - E). The maximum TOC concentration was 161 mg.L-1 for hose A and 94 mg.L-1 for hose B compared to just 13 mg.L-1 for the supplying drinking water. Total nitrogen and phosphorous data was only recorded for the supplying drinking water and it was found that total nitrogen increased to 2.95 mg.L-1 while total phosphorous decreased to 0.06 mg.L-1 over the course for the sampling (Figure 5.3 - F).

146 Chapter 5. Garden hoses

ABHose water temperature Hose water pH 50 10 Drinking water 45 Hose A Hose B 9 40 ° C) ( 35 8 30 pH

25 7

Temperature 20 6 15

10 5 0 50 100 150 200 250 300 350 400 450 500 550 0 50 100 150 200 250 300 350 400 450 500 550 Days Days

CDHose water turbidity Hose water free chlorine 25 0.30

0.25 20 ) -1 0.20 15 0.15 10 0.10 Turbidity (NTU) Turbidity Chlorine (mg.L 5 0.05

0 0.00 0 50 100 150 200 250 300 350 400 450 500 550 0 50 100 150 200 250 300 350 400 450 500 550 Days Days

Hose water total organic carbon EFHose water nitrogen and phosphorus 200

) 4 0.4 -1 Phosphorus (mg.L

150 ) 3 0.3 -1 Nitrogen

100 2 0.2 -1 Nitrogen (mg.L Nitrogen 1 0.1 50 ) Phosphorus Total organic carbon (mg.L carbon organic Total 0 0 0.0 0 50 100 150 200 250 300 350 400 450 500 550 0 50 100 150 200 250 300 350 400 450 500 550 Days Days

Figure 5.3 Water quality for the supplying drinking water and the two hoses. A: water temperature. B: water pH. C: water turbidity. D: water free chlorine. E: water total organic carbon. F: total nitrogen and phosphorus for the supplying drinking water only. Each data point on the graphs represent a single value or a mean with error bars showing the range of data. For the supplying drinking water there were two replicates samples but for hose water only a single sample was taken for each data point.

147 Chapter 5. Garden hoses 5.4.1.1 Heterotrophic plate counts HPC had the highest mean concentrations in hose water compared to biofilm over the 18 months of sampling (Figure 5.4). The highest mean detection rate was for hose A water was 3.0  105 cfu.mL-1 which was not quite significantly higher (Mann-Whitney test, p = 0.06) than the mean hose A biofilm counts (1.7  104 cfu.cm-2). Similarly the HPC counts for hose B water (1.9  105 cfu.mL-1) were higher than hose B biofilm (1.0  104 cfu.cm-2) but not significantly (Mann-Whitney test, p = 0.10). The HPC counts for hose A water were significantly higher (Mann-Whitney test, p = 0.008) than for the supplying drinking water (979 cfu.mL-1).

Hose heterotrophic plate counts

800000 Water (mL-1) Biofilm (cm-2) 700000 )

-2 600000

500000 or cm or -1 400000

300000

HPC (cfu.mLHPC 200000

100000

0 Water Water Biofilm Water Biofilm Supply Hose A Hose B

Figure 5.4 Hose heterotrophic plate counts. Combined HPC data for five sampling events during the 18 month experiment. Each box plot represents the mean (+) and whiskers at 5-95 percentiles for > 10 replicate samples. 5.4.1.2 Biofilm quantification The quantification of biofilm in hoses using DNA found that hose B biofilm had the highest mean concentration of DNA (127 ng.cm-2,  = 29) (Figure 5.5). In comparison the supplying drinking water had the lowest DNA concentrations (66 ng.mL-1,  = 33). However, there was found no significant differences between any of the samples types (1-way ANOVA, p = 0.42).

148 Chapter 5. Garden hoses

Hose DNA quantification

200 Water (mL-1) Biofilm (cm-2) 175 ) -2 150

125 or ng.cm or

-1 100

75

50 DNA (ng.mL

25

0 Water Water Biofilm Water Biofilm Supply Hose A Hose B

Figure 5.5 Quantification of hose biofilm using DNA. Combined DNA quantity data for five sampling events during the 18 month experiment. Each bar graph represents the mean and one standard deviation for > 5 samples. 5.4.2 FLA detection

5.4.2.1 FLA detected by culture FLA were detected at the highest mean (111 amoebae.mL-1) and maximum concentrations (800 amoebae.mL-1) in water from hose B (Figure 5.6). The second highest mean concentration was for hose B biofilm (37 amoebae.cm-2). In contrast the lowest mean concentration of FLA was detected in hose A water (0.4 amoebae.mL-1) which was significantly lower than hose B biofilm (Mann-Whitney test, p = 0.02). There appeared to be only a small quantity of FLA entering the hoses via the water supply with a mean detection of 1.4 amoebae.mL-1.

149 Chapter 5. Garden hoses

Hose FLA detected by culture

800 Water (mL-1) Biofilm (cm-2) 600 ) -2 400 or cm or

-1 200

120

100

80

60 FLA (amoebae.mL

40

20

0 Water Water Biofilm Water Biofilm Supply Hose A Hose B

Figure 5.6 Hose FLA detected by culture. Combined FLA data for five sampling events during 18 month experiment. For each sample type there were greater than 18 samples taken. Box plots with means (+), whiskers at 5-95 percentiles and outlying scores (•). In total 12 FLA isolates were successfully identified by partial 18S rDNA sequencing (Table 5.1). The majority of the isolates (10 out of 12) were identified as H. vermiformis in both hose types and the supplying drinking water. Using phylogenetic tree alignment it was confirmed that a diverse range of H. vermiformis were present in the samples (Figure 5.9). In the supply water Echinamoeba exudans was also identified which was in contrast to the hose samples were this particular FLA was not identified by culture. Interestingly a FLA (SDB2ii) isolated from hose A biofilm had a best BLAST match with an uncultured freshwater eukaryotic clone (AY919786). However, using phylogenetic alignment the sequence was identified as being most closely related to other Acanthamoeba spp. (Figure 5.9).

5.4.2.2 FLA detected by qPCR All qPCR reaction products were imaged using gel electrophoresis to confirm the correct products sizes and check for non-specific amplification. H. vermiformis qPCR gel with an expected product length of 502 bp is provided as an example (Figure 5.7). The gel shows amplification for the hose A water (lane 2), hose A biofilm (lane 3) and hose B biofilm (lane 6) from DNA extracts taken after 6 months of intermittent use.

150 Chapter 5. Garden hoses

L P 1 2 3 4 5 6 N L 1000 bp

600 bp 400 bp

Figure 5.7 H. vermiformis qPCR gel electrophoresis image for duplicate hose systems sampled at 6 months (December 2008) of continuous operation. Lanes, L: 100 bp EZ Load DNA ladder, P: H. vermiformis positive control, 1: supply tap, 2: hose A water, 3: hose A biofilm, 4: supply tap, 5: hose B water, 6: hose B biofilm, N: DNA free water negative control.

ABHose Acanthamoeba spp. qPCR Hose H. vermiformis qPCR 2000 2000 Water (mL-1) 1800 1800 Biofilm (cm-2) ) ) -2 1600 -2 1600

1400 1400 or cm or or cm or -1 1200 -1 1200

1000 1000

800 800

600 600

400 400 FLA (amoebae.mL FLA (amoebae.mL

200 200

0 0 Water Water Biofilm Water Biofilm Water Water Biofilm Water Biofilm Supply Hose A Hose B Supply Hose A Hose B

Figure 5.8 Hose FLA detected by qPCR over 18 months. A: Acanthamoeba spp. targeted qPCR. B: H. vermiformis targeted qPCR. Box plots with means (+) and whiskers from 5-95 percentiles for > 6 replicates. FLA were detected by qPCR using the Acanthamoeba spp. and H. vermiformis targeted primers but no FLA were detected using the Naegleria spp. primers (Figure 5.8). The highest mean detection of Acanthamoeba spp. using qPCR was for hose B water at 324 amoebae.mL-1 which also had the highest single detection of 1.6  103 amoebae.mL-1 (Figure 5.8- A). Mean Acanthamoeba spp. densities detected in biofilm samples where also high for both hoses with 230 amoebae.cm-2 for hose A and 81 amoebae.cm-2 for hose B. The differences between the sample types were found not to be significant (Kruskal-Wallis test, p = 0.92). The results for H. vermiformis qPCR show a similar pattern with the highest mean detection for hose B water (300 amoebae.mL-1) and similar detection densities in hose B biofilm (71 amoebae.cm-2) (Figure 5.8 - B). However, in contrast to the Acanthamoeba spp. qPCR detection pattern there were very low detects of H. vermiformis for hose A biofilm (0.5 amoebae.cm-2) but higher detection mean

151 Chapter 5. Garden hoses for the supplying drinking water tap (109 amoebae.mL-1). The differences between the sample types were found not to be significant (Kruskal-Wallis test, p = 0.15).

Table 5.1 FLA isolated by culture from hoses and identified by partial 18S rRNA sequencing. Sampling Months Isolate Seq. BLAST best match and BLAST % point (date) # length ARB alignment (accession #) 100% SDF1i 24 Echinamoeba sp. (EU273832) Supply 18 100% drinking SDF1ih 670 Hartmannella vermiformis (Dec 09) (EU137741) water 100% SDF1ii 480 Hartmannella vermiformis (AB525839) 6 100% DB2-h 300 Hartmannella vermiformis (Dec 08) (AB525839) 7 100% Hose A JB2h 404 Hartmannella vermiformis biofilm (Jan 09) (AB525839) 18 Uncultured freshwater 99% SDB2ii 600 (Dec 09) eukaryotic clone (AY919786) 8 100% FB7b 490 Hartmannella vermiformis (Feb 09) (AB525839) 100% SDB7i 470 Hartmannella vermiformis Hose B (FJ628003) biofilm 18 100% SDB7ih 550 Hartmannella vermiformis (Dec 09) (M95168) 100% SDB7ii 215 Hartmannella vermiformis (AB525839) 100% SDF7i 255 Hartmannella vermiformis Hose B 18 (AB525839) water (Dec 09) 100% SDF7ii 550 Hartmannella vermiformis (AB525839)

Table 5.2 FLA detected by qPCR from hoses and identified by partial 18S rDNA sequencing. Sampling Months Isolate Seq. BLAST best match and BLAST % point (date) # length ARB alignment (accession #) 100 % DF1 47 Acanthamoeba castellanii 6 (JF437606) Supply (Dec 08) 100 % drinking DF6_F8 65 Acanthamoeba sp. (HQ833439) tap water 18 99 % SDF1 502 Hartmannella vermiformis (Dec 09) (EU137741) Hose A 6 99 % DF2 502 Hartmannella vermiformis water (Dec 08) (DQ407573) 99 % DB2_E7 502 Hartmannella vermiformis Hose A 6 (DQ407573) biofilm (Dec 08) 99 % DB2_F7 502 Hartmannella vermiformis (EU137741) 152 Chapter 5. Garden hoses From the Acanthamoeba spp. qPCR two sequences were confirmed as Acanthamoeba spp. (Table 5.2). Additionally four H. vermiformis qPCR sequences were confirmed as belonging to two distinct H. vermiformis. The phylogenetic alignment further confirmed the correct identification as well as a range of diversity in the H. vermiformis detected (Figure 5.9).

AF338419, Naegleria clarkinida M18732, Naegleria gruberi X93224, Naegleria minor AY576367, Naegleria sp. AF338423, Naegleria fowleri U80057, Naegleria andersoni EF378693, Tetramitus sp. DQ388520, Heterolobosea sp. DQ122381, Thecamoeba quadrilineata AF293896, Filamoeba nolandi AY929916, Platyamoeba sp. DQ913104, Vannella epipetala EU980613, Sappinia pedata AJ489262, Echinamoeba thermarum AF293895, Echinamoeba exundans AY121848, Neoparamoeba aestuarina culture_Hose_A_biofilm_JB2bh culture_Hose_B_biofilm_SDB7ii culture_Hose_B_biofilm_JB7bh culture_Hose_A_biofilm_FB2a culture_tap_supply_water_SDF1ii culture_Hose_B_biofilm_SDB7i culture_Hose_B_biofilm_FB7b culture_Hose_A_biofilm_JB2h culture_Hose_B_biofilm_SDB7ih culture_Hose_B_water_SDF7ii AF426157, Hartmannella vermiformis DQ084366, Hartmannella vermiformis culture_Hose_A_biofilm_DB2h culture_tap_supply_water_SDF1ih Hv_qPCR_Hose_A_water_DF2_G8 Hv_qPCR_Hose_A_biofilm_DB2_E7 culture_Hose_B_water_SDF7i Hv_qPCR_Hose_A_water_SDFl_C12 Hv_qPCR_Hose_A_biofilm_DB2_F7 culture_Hose_A_biofilm_SDB2ii Ac_qPCR_tap_supplywater_DF6_F8 Ac_qPCR_tap_supply_water_DF1 AF260721, Acanthamoeba castellanii AF019057, Acanthamoeba culbertsoni AF260720, Acanthamoeba rhysodes AF260725, Acanthamoeba polyphaga U07406, Acanthamoeba rhysodes M13435, Acanthamoeba castellanii AF019061, Acanthamoeba polyphaga AF019068, Acanthamoeba hatchetti S81337, Acanthamoeba griffini AY026244, Acanthamoeba polyphaga AF019067, Acanthamoeba culbertsoni AF019066, Acanthamoeba comandoni AF019065, Acanthamoeba tubiashi EF153844, Saccharomyces cerevisiae (baker’s yeast)

0.10 Figure 5.9 Phylogenetic tree of FLA detected in hoses. FLA sequences were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the supplying drinking water tap are blue, hose A are green and hose B are purple. The scale bar for the branch lengths is 0.10 changes per nucleotide.

153 Chapter 5. Garden hoses 5.4.3 Legionella detection

5.4.3.1 Legionella detection by culture Using the standard method for the detection of Legionella by culture no presumptive Legionella bacteria were detected from water or biofilm sampled from the garden hose system.

5.4.3.2 Legionella detection by qPCR Gel electorphoresis of the products from the Legionella genus directed 16S rDNA qPCR were used to confirmed that the products were of the correct size (Figure 5.10). Lanes 1 through to 6 show positive Legionella qPCR products for DNA extracted from samples taken after 6 months.

Figure 5.10. Legionella spp. qPCR gel electrophoresis image for duplicate hose systems sampled at 6 months (December 2008) of continuous operation. Lanes, L: E-Gel DNA ladder, P: Legionella pneumophila positive control, 1: tap A water, 2: hose A water, 3: hose A biofilm, 4: tap B water, 5: hose B water, 6: hose B biofilm, N: DNA free water negative control.

All samples for the two hose set ups were positive using the Legionella spp. qPCR at least once during the 18 months of sampling. The highest mean detection of Legionella spp. was for hose B water with 6.5  103 cells.mL-1 ( = 2.7  103). This result was found to be significantly higher than all other samples (1 way ANOVA, p < 0.0001). The second highest detection was for hose B biofilm with 295 cells.cm-2 ( = 304). The supplying drinking water taps were positive for Legionella spp. with a mean detection of 66 cells.mL-1 ( = 37). The lowest mean detection rates for Legionella spp. were for hose A water (39 cells.mL-1) and biofilm (22 cells.cm-2).

154 Chapter 5. Garden hoses

Hose qPCR detection of Legionella spp. 10000 Water (mL-1) -2 9000 Biofilm (cm ) ) -2 8000

or cm or 7000 -1 6000

5000 (cells.mL 4000

3000

2000 Legionella Legionella 1000

0 Water Water Biofilm Water Biofilm Supply Hose A Hose B

Figure 5.11. Detection of Legionella spp. in hoses using qPCR. Combined Legionella qPCR data for three sampling events during 18 month experiment. For each sample type there were greater than 6 samples taken. Bar graphs present the mean and one standard deviation. A selection of Legionella spp. qPCR positive samples (six out of 13) were cloned and sequenced. In total nine sequences were identified as belonging to the family Legionellaceae (Figure 5.12). Five sequences were classified as specific Legionella sp. while the remainder were unclassified (Figure 5.12 and Table 5.3). As there is only one genus within the family, this confirms that the qPCR primers were specific for Legionella spp. for the water and biofilm samples. Generally, the phylogenetic tree allocation agrees with the BLAST best matches but the speciation of some of the Legionella spp. are different. Specifically, the hose B water sample (DF7_1) aligns with a cluster of uncultured Legionella bacteria in the phylogenetic tree which is removed from L. taurinensis which was the BLAST best match. The phylogenetic tree reveals the diversity of Legionella present in the hose samples despite the small number of qPCR products that were sequenced (Figure 5.12).

155 Chapter 5. Garden hoses Table 5.3 Identity of Legionella qPCR products confirmed by cloning and sequencing.

Sampling Months Seq. % ID Isolate # BLAST best match point (date) length (accession #) 96 % 9 MF1_1 448 Legionella dreseniensis Supply (Mar 09) (AM747393) drinking water 99 % MF1_2 454 Legionella waltersii (NR024969) 95 % MF2_10 450 Uncultured Legionella sp. (GU797444) 97 % MF2_11 454 Uncultured Legionella sp. 9 (GU797444) (Mar 09) 96 % Hose A water MF2_14 454 Uncultured Legionellales (EF667907) 96 % MF2_13 454 Uncultured Legionellales (EF667907) 18 SDF2_2 96 % 454 Uncultured Legionellales (Dec 09) 6 (EF667907) Hose A 9 96 % MB2_21 454 Uncultured Legionellales biofilm (Mar 09) (EF667907) 6 96 % Hose B water DF7_1 450 Legionella taurinensis (Dec 08) (FN667604)

5.4.3.3 Detection of Legionella spp. in FLA isolates. Of the 6 FLA isolated and screened from the supply drinking water one H. vermiformis isolate (SDF1ih) was identified by PCR to contain Legionella bacteria nine months after isolation (Figure 6.13). The PCR product could not be successfully sequenced however Legionella PCR products from other FLA were identified by sequencing as Legionella micadedi (Section 6.4.3.3).

156 Chapter 5. Garden hoses

AJ969023, Legionella anisa U59697, Legionella parisiensis X73400, Legionella steigerwaltii Z32644, Legionella tucsonensis Z49720, Legionella cherriinellales Z49738, Legionella wadsworthii Z49725, Legionella gratiana X73407, Legionella cincinnatiensis Z49735, Legionella santicrucis AY444741, X73399, Legionella sainthelensi DQ408661, Legionella pneumophila CR628337, Legionella pneumophila str. CR628336, Legionella pneumophila str. X73402, Legionella pneumophila qPCR Drinking supply tap water_MF1_A10_2 AF122886, Legionella waltersii Z49729, Legionella moravica Z49732, Legionella quateirensis Z49739, Legionella worsleiensis Z49736, Legionella shakespearei U66104, Legionellalike amoebal X97366, Legionella drancourtii X97364, Legionella lytica qPCR Hose A water_MF2_A11_10 qPCR Hose A water_MF2_A11_11 X97361, Legionella sp. U64035, Legionellalike amoebal pathogen Z49724, Legionella donaldsonii X73406, Legionella feeleii U64034, Legionellalike amoebal pathogen AY957915, uncultured bacterium AY957913, uncultured bacterium qPCR Drinking supply tap water_MF1_A10_1 Z49722, Legionella fairfieldensis AM747393, Legionella dresdeniensis Z49717, Legionella birminghamensis Z49733, Legionella quinlivanii qPCR Hose B water_DF7_A5_1 EU409111, uncultured bacterium EU835422, uncultured bacterium Z49723, Legionella geestiana Z32638, Legionella erythra DQ667196, Legionella taurinensis X73409, Legionella jamestowniensis Z32667, Legionella jordanis X73403, Legionella brunensis Z32640, Legionella israelensis Z49730, Legionella londiniensis X73397, Legionella oakridgensis AB233211, Legionella yabuuchiae AF122884, Legionella beliardensis AF122883, Legionella gresilensis Z49727, Legionella lansingensis qCPR Hose A biofilm_MB2_A12_21 qPCR Hose A water SDF2_B8_26 qPCR Hose A water MF2_A11_14 qPCR Hose A water MF2_A11_13 EF667907, uncultured Legionellales bacterium EF612966, uncultured bacterium EF409862, uncultured bacterium AY050590, uncultured bacterium Z49873, Chlamydophila pneumoniae

0.10 Figure 5.12. Phylogenetic tree of Legionella spp. detected in garden hoses by qPCR. Sequences of Legionella spp. detected by qPCR were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the supplying drinking water tap are blue, hose A are green and hose B are purple. The scale bar for the branch lengths is 0.10 changes per nucleotide.

157 Chapter 5. Garden hoses 5.5 DISCUSSION

5.5.1 Hose water quality Garden hose water had higher water temperature, turbidity and total organic carbon compared to the supplying drinking water. The temperature of the water of both hoses warmed by the sun were consistently between 0.5 to 19.5 °C higher than the supplying water (Figure 5.3 - A). Higher temperatures are known to be the single largest factor for increasing the growth of microorganisms in water systems (LeChevallier et al., 1996). Furthermore, 1 cm2 of biofilm can contain 10 to 1  103 times more HPC bacteria than 1 mL of water from the same system (Volk and LeChevallier, 1999). Therefore, the 10 - 2  103 times higher HPC counts in the water and biofilm of both garden hoses compared to the supplying water (Figure 5.4), are likely due to a combination of higher temperatures, large available internal surface area (7.9  104 cm2) and increased quantities of biofilm (Figure 5.5). Also the higher turbidity and TOC counts for the hose water compared to the supplying water (Figure 5.3 - C and E) are likely linked to the increased microbial growth. The estimation of biofilm using DNA concentration in the extracts did not allow for differentiation between water and biofilm samples due to the high cell count in the water. For future work staining of the biofilm in situ and determining the physical depth with confocal microscopy (Beyenal et al., 2004) would give more relevant information about the composition of the biofilm present inside the hoses. Furthermore, the DNA extraction efficiencies estimations could be improved using water samples from the hoses rather than from the supplying water.

5.5.2 FLA in hoses Detection of FLA by culture revealed the highest mean detects for hose B water (111 amoebae.mL-1) followed by hose B biofilm (37 amoebae.cm-2) (Figure 5.6). FLA detects in hose B were significantly higher (Mann-Whitney test, p = 0.02) than for hose A with only 0.4 amoebae.mL-1. These significant differences between the hoses were not observed for the water quality HPC where hose A water was actually higher than hose B water (Figure 5.4). This indicates that the hose material used in hose B is the only differing variable between the two hoses and may have been a contributing factor to the growth of FLA. Common green garden hoses have been reported to contain the antibiotic triclosan (Smith and Lourie, 2009). It is possible that the lilac hose lacked some type of anti-microbial component that was present in the green garden hose and resulted in the reduced growth of FLA and Legionella. The majority (10/12) of FLA isolated by culture were identified as H. vermiformis which is consistent with the results from the MRD drinking water distribution system samples that supplied the water (Section 4.4.3). However, the qPCR results indicate that Acanthamoeba sp. were present in 158 Chapter 5. Garden hoses equivalent proportions to the H. vermiformis which raises questions about the viability of the FLA detected by qPCR. For the qPCR reactions hose water B was found to have the highest mean densities of Acanthamoeba spp. (324 amoebae.mL-1) and H. vermiformis (300 amoebae.mL-1) compared to the other sample types (Figure 5.8). Acanthamoeba spp. were found at higher levels in the hose A biofilm (231 amoebae.cm-2) compared to H. vermiformis (0.5 amoebae.cm-2) which may indicate a preference between the two FLA genera. The qPCR results gave an indication as to the source of the H. vermiformis in the hoses with a high detection average in the supplying drinking water (110 amoebae.mL-1). The identification of H. vermiformis predominantly by culture may be due to this genera dominating the NNA E.coli overlay plates and therefore biasing the identification. The diversity of H. vermiformis revealed by the phylogenetic tree (Figure 5.9) reveals that the species is more varied than previously thought. Further strain identification work could be used to determine the association between these FLA. A second explanation could be that there were base pair errors in the sequences reads which artificially increased the diversity. For future work this could be rectified by sequencing PCR products in both directions rather than a single forward read as used in these methods. Overall this initial finding of high densities of FLA in garden hoses is not surprising but are higher than any reported FLA densities in the literature (Chapter 2). The highest densities reported in the literature were 16 amoebae.mL-1 for cooling tower water (Behets et al., 2007) and 90 amoebae.mL-1 for river water (Hoffmann and Michel, 2001) and which are 3 - 20 times lower than the average FLA detected by qPCR in the hoses. The higher temperatures in the hoses are likely to have facilitated the growth of FLA as it is known that a positive correlation exists between the two (Carlesso et al., 2010). Also the detection of higher FLA densities in the hose water samples compared to biofilm may be due to stagnation of the water within the hose and the suspension of large quantities of sloughed biofilm (Storey and Ashbolt, 2003) from the hose internal surface resulting in elevated particles to supported the attachment and feeding of FLA in water (Rodriguez-Zaragoza, 1994). The increased turbidity readings for the hose water supports this explanation.

5.5.3 Legionella in hoses The detection of Legionella spp. by qPCR matched the pattern of FLA detection indicating a relationship between the two in the garden hoses. Furthermore, one H. vermiformis isolate (SDF1i) was positive for Legionella spp. by PCR nine months after isolation; revealing that the FLA could be infected in situ by pathogenic Legionella. For Legionella spp. the highest mean detection was from hose B water with 6.5  103 cells.mL-1 which was significantly higher

159 Chapter 5. Garden hoses than the other sample types (1 way ANOVA, p < 0.0001) (Figure 5.11). Comparably hose A water had a much lower average detection density of 39 cells.mL-1 despite also containing FLA. The high concentrations of Legionella detected in hose B water were in the same order of magnitude as the maximum Legionella detections in cooling towers (Bentham, 2000; Wéry et al., 2008; Chen and Chang, 2010) and only one order of magnitude lower than the maximum detection in a spa legionellosis outbreak (Okada et al., 2005). A selection of the Legionella spp. detected by qPCR were identified to be closely related to non-pathogenic species L. dreseniensis, L. waltersii, L. taurinensis as well as a number of unidentified Legionella spp. and Legionellales with unknown pathogenicity (Table 5.3). The diversity of Legionella sp. revealed by sequencing only a small number of positive qPCR products (n = 6) and the identification of species with unknown pathogenicity requires more attention in future research.

5.5.4 Conclusions Garden hoses are a common application of drinking water and their ability to facilitate the growth of FLA and Legionella spp. to high densities presents a health risk to hose users that is yet to be quantified. This study indicates that the lilac hose type recommended for use with recycled water may facilitate the growth of Legionella spp. significantly more than a standard green garden hose. For households using recycled water and this hose type there may be an elevated risk of exposure to pathogenic Legionella. More research on garden hoses are needed to determine if the high FLA and Legionella densities observed are consistent across a larger number of garden hoses which have been in use for years rather than months. Furthermore, data is needed on the diversity of other pathogenic ARM residing in garden hoses and the relationship with the FLA present.

160 Chapter 5. Garden hoses

5.6 REFERENCES 1. Australian Bureau of Statistics. 2006. 04 August 2011. "Census QuickStats: Australia." Retrieved 12 August, 2011, from http://www.censusdata.abs.gov.au. 2. Barbeau, J. and T. Buhler. 2001. Biofilms augment the number of free-living amoebae in dental unit waterlines. Research in Microbiology. 152(8): 753-760. 3. Barkley, E. and C. Michelson. 2003. Laboratory safety Methods for General and Molecular Microbiology. C. A. Reddy, T. J. Beveridge, J. A. Breznaket al. Washington D.C., American Society for Microbiology. 4. Behets, J., P. Declerck, Y. Delaedt, L. Verelst and F. Ollevier. 2007. Survey for the presence of specific free-living amoebae in cooling waters from Belgian power plants. Parasitology Research. 100(6): 1249-1256. 5. Bentham, R. H. 2000. Routine sampling and the control of Legionella spp. in cooling tower water systems. Current Microbiology. 41(4): 271-275. 6. Beyenal, H., C. Donovan, Z. Lewandowski and G. Harkin. 2004. Three-dimensional biofilm structure quantification. Journal of Microbiological Methods. 59(3): 395-413. 7. Buffet-Bataillon, S., M. Bonnaure-Mallet, A. d. l. Pintiere, G. Defawe, A.-L. Gautier- Lerestif, S. Fauveau and J. Minet. 2010. Heterotrophic bacterial growth on hoses in a neonatal water distribution system. Journal of Microbiology and Biotechnology. 20(4): 779-781. 8. Carlesso, A., G. Artuso, K. Caumo and M. Rott. 2010. Potentially pathogenic Acanthamoeba isolated from a hospital in Brazil. Current Microbiology. 60(3): 185-190. 9. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634. 10. Costerton, J. W. 2004. A short history of the development of the biofilm concept. Microbial Biofilms. M. Ghannoum and G. O'Toole. Washington, D.C., ASM Press: 4-18. 11. Declerck, P. 2010. Biofilms: the environmental playground of Legionella pneumophila. Environmental Microbiology. 12(3): 557-566. 12. Euser, S. M., M. Pelgrim and J. W. den Boer. 2010. Legionnaires' disease and Pontiac fever after using a private outdoor whirlpool spa. Scandinavian Journal of Infectious Diseases. 42(11-12): 910-916. 13. Hoffmann, R. and R. Michel. 2001. Distribution of free-living amoebae (FLA) during preparation and supply of drinking water. International Journal of Hygiene and Environmental Health. 203(3): 215-219. 14. LeChevallier, M. W., N. Welch and D. Smith. 1996. Full-scale studies of factors related to coliform regrowth in drinking water. Applied and Environmental Microbiology. 62(7): 2201-2211. 15. Lee, S. H., D. A. Levy, G. F. Craun, M. J. Beach and R. L. Calderon. 2002. Surveillance for waterborne-disease outbreaks: United States, 1999–2000. Morbidity and Mortality Weekly Report. 51(8): 1-47. 16. Okada, M., K. Kawano, K. Fumiaki, J. Amemura-Maekawa, H. Watanabe, K. Yagita, T. Endo and S. Suzuki. 2005. The largest outbreak of legionellosis in Japan associated with spa baths : Epidemic curve and environmental investigation. Kansenshogaku zasshi 79(6): 365-374. 17. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241. 18. Rollins, D. M. and R. R. Colwell. 1986. Viable but nonculturable stage of Campylobacter jejuni and its role in survival in the natural aquatic environment. Applied and Environmental Microbiology. 52(3): 531-538.

161 Chapter 5. Garden hoses 19. Schofield, G. M. 1985. A note on the survival of Legionella pneumophila in stagnant tap water. Journal of Applied Microbiology. 59(4): 333-335. 20. Smith, R. and B. Lourie. 2009. Slow death by rubber duck. How the toxic chemistry of everyday life affects our health. Ontario, Random House. 21. Storey, M. and N. J. Ashbolt. 2001. Persistence of two model enteric viruses (B40-8 and MS-2 bacteriophages) in water distribution pipe biofilms. Water Science and Technology. 43(12): 133-138. 22. Storey, M. and N. J. Ashbolt. 2003. A risk model for enteric virus accumulation and release from recycled water distribution pipe biofilms. Water Science and Technology: Water Supply. 3(3): 93-100. 23. Volk, C. J. and M. W. LeChevallier. 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Applied and Environmental Microbiology. 65(11): 4957-4966. 24. Walker, J. T., C. W. Mackerness, D. Mallon, T. Makin, T. Williets and C. W. Keevil. 1995. Control of Legionella pneumophila in a hospital water system by chlorine dioxide. Journal of Industrial Microbiology & Biotechnology. 15(4): 384-390. 25. Wéry, N., V. Bru-Adan, C. Minervini, J.-P. Delgénes, L. Garrelly and J.-J. Godon. 2008. Dynamics of Legionella spp. and bacterial populations during the proliferation of L. pneumophila in a cooling tower facility. Applied and Environmental Microbiology. 74(10): 3030-3037.

162 CHAPTER 6 FLA AND LEGIONELLA IN HEATED ANNULAR REACTORS 6. TABLE OF CONTENTS

6.1 INTRODUCTION ______165 6.1.1 Legionella outbreaks ______165 6.1.2 Legionella in heated applications of water ______165 6.1.3 FLA and Legionella______166 6.2 AIMS ______166 6.3 METHODS ______167 6.3.1 Experimental set-up______167 6.3.2 Water quality ______169 6.3.3 Sampling from annular reactors supplied with drinking water ______169 6.3.4 Culture methods______170 6.3.5 Molecular detection ______170 6.4 RESULTS ______171 6.4.1 Water quality ______171 6.4.2 FLA detection ______174 6.4.3 Legionella detection ______181 6.5 DISCUSSION ______185 6.5.1 FLA populations in annular reactors ______185 6.5.2 Legionella populations in annular reactors ______186 6.5.3 FLA infected with Legionella ______187 6.5.4 Conclusions and future research ______187 6.6 REFERENCES ______189

Chapter 6. Heated annular reactors LIST OF TABLES

Table 6.1 FLA isolated by culture and identified by sequencing ______176 Table 6.2 FLA detected by qPCR and identified by sequencing ______179 Table 6.3 Legionella qPCR products identified by sequencing ______183 Table 6.4 Identification of Legionella spp. detected by PCR in FLA isolates ____ 184

LIST OF FIGURES

Figure 6.1 Annular reactor set-up______167 Figure 6.2 Annular reactor set up components______168 Figure 6.3 Sampling from the annular reactors ______169 Figure 6.4 Annular reactors water quality ______172 Figure 6.5 Annular reactor heterotrophic plate counts ______173 Figure 6.6 Quantification of annular reactor biofilm ______174 Figure 6.7 Annular reactor FLA detected by culture ______175 Figure 6.8 Acanthamoeba spp. qPCR gel electrophoresis ______177 Figure 6.9 Annular FLA detected by qPCR ______178 Figure 6.10 Phylogenetic tree of FLA detected in annular reactors ______180 Figure 6.11 Legionella spp. qPCR gel electrophoresis______181 Figure 6.12 Detection of Legionella sp. in annular reactors using qPCR______182 Figure 6.13 Example of isolated FLA cultures positive for Legionella spp. _____ 183 Figure 6.14 Phylogenetic tree of Legionella spp. ______184

164 Chapter 6. Heated annular reactors

6.1 INTRODUCTION

6.1.1 Legionella outbreaks Legionella outbreaks are predominantly traced to hospital warm water networks (Breiman et al., 1990; Lin et al., 1998), cooling towers (Heath et al., 1998; García-Fulgueiras et al., 2003) and spas (Jernigan et al., 1996; Den Boer et al., 2002; Okada et al., 2005; Euser et al., 2010). Furthermore, residential hot water systems and showers have been identified as a potential source of Legionella infections for individuals (Breiman et al., 1990; Stout et al., 1992). Heated water applications of drinking water are known to increase the growth rates of pathogenic Legionella in part due to the bacteria's temperature tolerance and growth at 35 - 45 °C (Bartram et al., 2007). However, another contributing factor are FLA whose presence has been significantly correlated with the detection of Legionella in hospital warm water systems (Breiman et al., 1990; Thomas et al., 2006).

6.1.2 Legionella in heated applications of water Heated applications of drinking water include hot and warm water systems, cooling towers and spas. In these applications the water is frequently maintained within the heating vessel for a period of time before being replaced or re-circulated. The most common applications of this nature in residential homes are hot water systems which are reported to be maintained at mean temperatures at about 50 °C (Eyring et al., 2008) with maximums of 60 °C (Ménard- Szczebara et al., 2008). Shower heads are known to contain Legionella which people are exposed to during showering (Falkinham et al., 2008; Feazel et al., 2009). In residential and also recreational settings spas and heated pools facilitate the growth of Legionella spp. (Fallon and Rowbotham, 1990; Okada et al., 2005). Spas generally operate at 40 °C (Armstrong and Haas, 2007) while heated pools are generally cooler at 20 - 24 °C (erva, 1971). People directly using the spa as well as those some 15 m away have been reported to inhale aerosols containing pathogenic Legionella and fall ill (Den Boer et al., 2002). In facilities for the care of the sick and elderly and education of the young, warm water systems are often maintained at temperatures between 38 - 43.5 °C to prevent scalding but which is known to facilitate the growth of Legionella (NSW Department of Health, 2004). However, hospital warm water systems are frequently maintained at higher temperatures between 40 to 60°C (Fields et al., 1989; Rohr et al., 1998) in an attempt to limit the growth of Legionella and other pathogens (Lin et al., 1998). Nevertheless, pathogenic L. pneumophila is regularly detected in hospital hot water systems (Breiman et al., 1990; Kool et al., 1999; Wellinghausen et al., 2001; Joly et al., 2006; Thomas et al., 2006). 165 Chapter 6. Heated annular reactors Cooling towers are another industrial applications of drinking water where the water absorbs heat in order to cool air and then it is re-circulated. There are two main types of cooling towers; large industrial hyperbolic cooling towers which use natural drafts and smaller cross flow cooling towers often used in buildings and frequently located on rooftops (Hill et al., 1990). Cooling towers operate at temperatures between 10 - 40 °C (Berk et al., 2006; Behets et al., 2007) and consistently contain a diverse range of pathogenic and non-pathogenic Legionella (Barbaree et al., 1986; Bentham, 1993; Miyamoto et al., 1997; Bentham, 2000; Joly et al., 2006; Declerck et al., 2007; Wéry et al., 2008; Chen and Chang, 2010). Water aerosols are released in large quantities from cooling towers during usual operation through evaporation, drift and blow down (Lindahl, 2004).

6.1.3 FLA and Legionella Despite FLA being directly correlated with the detection of Legionella (Breiman et al., 1990; Thomas et al., 2006), only three other studies were identified where both FLA and Legionella populations were examined simultaneously in warm water applications; cooling towers (Barbaree et al., 1986; Declerck et al., 2007) and spas (Fallon and Rowbotham, 1990). Additional studies have identified FLA independently in hospital hot water systems (Fields et al., 1989; Rohr et al., 1998), cooling towers (Berk et al., 2006; Behets et al., 2007) and spas (Esterman et al., 1987). However, only one study reported on the density and diversity of FLA in a warm water application (cooling tower) (Behets et al., 2007). Therefore, colonisation rates and details of FLA density and diversity in water and biofilm and the influence this has on Legionella growth under elevated temperatures is still poorly understood. Understanding how FLA and Legionella colonise a water vessel under elevated temperatures and in which part of the system their interaction occurs (water or biofilm) is very important. This knowledge is important as it forms the basis upon which risk assessments and control strategy recommendations can be made for FLA and Legionella in warm water applications of a particular drinking water types.

6.2 AIMS The aim of this research was to examine the colonisation rates, density and diversity of FLA and Legionella in water and biofilm under elevated temperatures (42 °C) and water retention conditions. The experiment was conducted over 13.5 months to determine any population variations linked to changes in water quality or microbial community characteristics.

166 Chapter 6. Heated annular reactors 6.3 METHODS

6.3.1 Experimental set-up Annular reactors are devices that allow for the sampling of biofilm that forms on removable coupons under different operating conditions. Annular reactors have been widely used in water research to determine the factors associated with biofilm formation (Volk and LeChevallier, 1999; Ollos et al., 2003) and specific micro-organism retention within biofilms (Storey and Ashbolt, 2001; Packroff et al., 2002; Declerck et al., 2009). The annular reactors (Biofilm Reactors, CRCWQT, Adelaide, Australia) used in this research consisted of a polycarbonate drum with removable stainless steel rectangular coupons (15 mm  50 mm) contained within a glass outer casing. The coupon surfaces were 10 mm from the glass casing and the total water volume was 1.15 L. The reactors could be operated under different flow rates and drum rotation speeds (Figure 6.1 - B). Four annular reactors were set-up in the field and supplied with drinking water (Figure 6.1 - A). Two of the annular reactors were heated to approximately 42 °C using heating coils, while the remaining reactors operated at ambient temperature. The temperature setting of 42 °C was selected because it was common temperature across the range of heated water applications (Section 6.1.2) but was still within the growth range of Legionella (Bartram et al., 2007). The reactors were all set to rotate at the same speed (setting 10) equivalent to approximately 12 drum rotations per minute or shear expected from 30 cm per second flow rate.

 

Figure 6.1 Annular reactor set-up. A: Four annular reactors in place with the two reactors in the foreground heated to approximately 42 °C while the two reactors in the rear were maintained at ambient temperatures. The central plastic tub contained the thermostats and electrical cables. B: A single heated annular reactor in place with the rotational setting (10) visible in the digital display.

167 Chapter 6. Heated annular reactors Drinking water from the outside taps used in the hose experiments (Chapter 5) was piped to the reactors through inert 9.5 mm diameter blue polyurethane tubing (Surethane, Advanced Technology Pneumatic, Acacia Ridge, Australia). Blue polyurethane tubing 10.5 m in length connected one outdoor tap to a header bottle and from the second tap 21.5 m of the same tubing was connected to a second header bottle (Figure 6.2 - A). All tubing from the outdoor taps were insulated with a 10 mm edge width foam insulating tubing (Armaflex, Armacell, Sydney, Australia) to limit temperature fluctuations in the feed tube. To ensure that feed tubing were not colonised by bacteria they were detached, disinfected (3.75 % sodium hypochlorite, Univar) and flushed after three months of operation. Water was delivered into two header containers which consisted of two 2 L plastic bottles (Naeglene, Thermo Fisher Scientific) that had been modified (Figure 6.2 - A). Each header bottle included at the top an air vent fitted with a filter (F6 0.2 μm, Millipore), an overflow outlet with black tubing (Norprene, Masterflex, Cole- Parmer) at the base and clear tubing (Tygon LFL, Masterflex) water feed to the pumping unit. Water was pumped into reactors at rate of between 15 - 20 mL.min-1 using duplicate console drives (Masterflex) and two pump heads per drive (Model 7518-00, Masterflex). Black peristaltic tubing (Norprene) was used in the pump heads and then connected to 0.5 m of clear tubing (Tygon LFL) to attach to each annular reactors. The residence time for water in the annular reactor was approximately 57 - 77 min. Heating coils and thermostats were custom made (Cynebar, Brisbane, Australia). The heating coil consisted of 10 m of heating wire which was wrapped around the glass casing of the annular reactors with aluminum tape (Figure 6.2 - B). For each heated annular reactor an individual thermostat was used in order to control the temperature accurately. Thermostats were set to achieve a water temperature within the drum of approximately 42 °C.  

Figure 6.2 Annular reactor set up components. A: Header water bottle with drinking water feed via the blue tube and then drawn from the base via the clear tubing by the pump unit and pumped into the annular reactors. A black over flow tube was also was also fitted from the base of the bottle. B: A glass casing removed from the rest of the annular reactor and fitted with a 168 Chapter 6. Heated annular reactors heating coiling and aluminium tape. The thermostat unit and the wire for the thermometer is present in the foreground. 6.3.2 Water quality Samples were taken from the taps supplying the annular reactor and the supplying feed hose approximately once every month during the experiment. Water characteristics analysed were temperature, pH, turbidity, total organic carbon, total phosphate, and free and combined chlorine as described (Section 3.4.1 and 3.4.2). Using 0.5 mL of water and re-suspended biofilm in duplicate heterotrophic plate counts and total coliform counts were also performed as described (Section 3.4.3) for each of the four samplings (1.5, 4, 5 and 13.5 months). Additionally, the biofilm on the annular reactor coupons was fluorescently stained and quantified as described (Section 3.4.3.3).

6.3.3 Sampling from annular reactors supplied with drinking water

Sampling from the reactors occurred after 1.5, 4, 5 and 13.5 months of operation. Water (50 mL) was sampled from the hoses supplying the reactors and from within each reactor. Water and coupons were removed from each reactor via the sampling plug at the top of the reactor bowl (Figure 6.3). Water samples were taken with sterile serological pipettes (Figure 6.3 - A) and stainless steel coupons were removed with the use of a wire hook (Figure 6.3 - B). Three rows of coupons (nine in total) had biofilm removed by scraping as described (Section 3.3.3.2).  

Figure 6.3 Sampling from the annular reactors. A: Water sampling using serological pipette from the sampling point at the top of the annular reactor. B. Close-up of the sampling point and the stainless steel coupon removal using a sterile wire hook.

169 Chapter 6. Heated annular reactors 6.3.4 Culture methods

6.3.4.1 FLA detection by culture FLA were detected and isolated from samples by culture using non-nutrient agar (NNA) plates with E. coli overlays. Aliquots (0.5 mL) of water and scraped biofilm from the 12 types of samples (duplicate taps, feed hoses, ambient reactor and heated reactor water and biofilm) were spread plated in duplicate for each of the water types (24 plates in total) for each sampling period (Section 3.5.1). Twenty three isolated FLA were identified by partial 18S rRNA amplification using primers with a higher specificity for FLA (Section 3.6.2.1) and direct sequencing (Section 3.6.6.1).

6.3.4.2 Detection of Legionella by culture Legionella bacteria were detected using 0.5 mL aliquots of water and biofilm samples in duplicate and the initial stages of the standard method (Section 3.5.2).

6.3.5 Molecular detection

6.3.5.1 DNA extraction For molecular analysis DNA was extracted individually from the eight concentrated water samples and four concentrated biofilm samples for each of the three sampling periods (36 samples in total). DNA was extracted using a cell lysis and DNA purification kit for environmental samples (FastDNA® SPIN Kit for Soil) as previously described (Section 3.6.1.2). DNA extraction losses and qPCR inhibition previously calculated for drinking water were used to adjust qPCR results (Section 3.6.1.3).

6.3.5.2 Detection of FLA by qPCR For the detection of FLA by qPCR three different genera of FLA were targeted with individual PCR reactions in duplicate with 5 μL of DNA extract. Cell based standard curves for trophozoites were used to quantify each FLA qPCR reaction as described (Section 3.6.3.1). A selection of positive qPCR samples were selected for cloning and sequencing. From the Acanthamoeba spp. target qPCR eight out of 31 positive samples and nine out of the 18 H. vermiformis target qPCR positive samples were selected.

6.3.5.3 Legionella detection by qPCR Legionella were detected by qPCR using duplicate reactions with 2.5 μL in replicates of four. Legionella cell based calibration curves were used for quantification (Section 3.6.3.3). From this qPCR six positive samples were submitted for cloning and sequencing.

170 Chapter 6. Heated annular reactors 6.3.5.4 Legionella detection in isolated FLA A selection of FLA (n = 13) had been isolated, identified and sub-cultured for up to 21 months were screened for the presence of Legionella sp. residing as ARB as described (Section 3.6.3.4). Positive samples were submitted for cloning and sequencing.

6.3.5.5 Cloning, sequencing and identification of positive PCR/qPCR products For each cloning reaction two to six cloned DNA fragments (Section 3.6.5) were sequenced in the forward direction only using the provided forward primer (M13) (Section 3.6.6.2) and analysed (Section 3.6.7).

6.4 RESULTS

6.4.1 Water quality The flow rates for both reactors ranged from 10 to 27.5 mL.min-1 over the 404 days of operation (Figure 6.4 - A). The mean flow rates for the ambient reactors was 18.6 mL.min-1 and 19.7 mL.min-1 for heated reactors. The temperature for the water in the ambient reactors was consistent with that of the supplying water and ranged from 16.4 to 29.5 °C (Figure 6.4 - B). The heated annular reactors temperatures were maintained at a mean of 42.4 °C and ranged from 41.0 to 45.3 °C over the course of the experiment. The pH in both heated and ambient reactors ranged from 6.9 to 8.9 and followed the same fluctuations observed for the supplying water (Figure 6.4 - C). Similarly the turbidity, free chlorine and total organic carbon generally followed the same fluctuations as the supplying drinking water. Turbidity ranged from 0.19 to 1.6 NTU for the supplying drinking water and annular reactors (Figure 6.4 - D). Free chlorine was low for both the supply water and within the annular reactors and ranged from 0.02 to 0.19 mg.mL-1 (Figure 6.4 - E). Total organic carbon fluctuations in the supplying drinking water were larger (0.11 to 40.3 mg.mL-1) compared to the reactor fluctuations (0.15 - 7.5 mg.mL-1) (Figure 6.4 - F).

171 Chapter 6. Heated annular reactors

A Annular reactor flow rates B Annular reactor water temperatures 30 50

25 45 ) -1

° C) 40 20 ( 35 15 30 25 10 Flow (mL.min Flow

Temperature 20 5 15 0 10 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 Day Days

Supply water Ambient reactor Heated reactor

Annular reactor turbidity C Annular reactor water pH D 10.0 2.0 9.5 9.0 1.5 8.5 1.0

pH 8.0 7.5

7.0 (NTU) Turbidity 0.5 6.5 6.0 0.0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 Days Days

EFAnnular reactor free chlorine Annular reactor total organic carbon )

0.25 -1 50

) 0.20 40 -1

0.15 30

0.10 20

Chlorine (mg.L Chlorine 0.05 10

0.00 (mg.L carbon organic Total 0 0 50 100 150 200 250 300 350 400 0 50 100 150 200 250 300 350 400 Days Days

Figure 6.4 Annular reactors water quality. A: Flow rates for ambient and heated reactors. B: Water temperature. C: Water pH. D: Water turbidity. E: Water free chlorine F: Water total organic carbon. Each data point on the graphs represent the mean with error bars showing the range. For the supplying drinking water there were four samples but for each reactor there were two replicate samples (one from each reactor). 6.4.1.1 Heterotrophic plate counts Heated annular reactor water had highest combined mean (1.4  103 cfu.mL-1) and maximum (2.9  103 cfu.mL-1) HPC combined over the course of the experiment (Figure 6.5).

172 Chapter 6. Heated annular reactors The heated annular reactor biofilm had the lowest mean (300 cfu.cm2) and maximum (890 cfu.cm2) which was lower than the supplying drinking water mean (440 cfu.mL-1) and maximum (1.2  103 cfu.mL-1). However, none of the differences between between HPC results for the sample types were found to be significant (Kruskal - Wallis test, p = 0.42).

Annular reactor HPC

3500 Water (mL-1) Biofilm (cm-2) 3000 ) -2 2500 or cm or

-1 2000

1500

1000 HPC (cfu.mLHPC 500

0 Water Water Biofilm Water Biofilm Supply Ambient Heated reactor reactor

Figure 6.5 Annular reactor heterotrophic plate counts. Combined HPC data for four sampling events during the 13.5 month experiment. Box plots with means (+) and whiskers at 5- 95 percentiles for > 16 samples. 6.4.1.2 Biofilm quantification Annular reactor biofilm was quantified by fluorescent microscopy and average pixel counts of the cells on the biofilm coupons. Heated annular reactors had slightly higher mean biofilm counts at 32 pixels.μm-2 ( = 12) than ambient annular reactors (34 pixels.μm-2,  = 15) (Figure 6.6) However, the difference were found not to be significant (t-test, p = 0.7).

173 Chapter 6. Heated annular reactors

Annular reactor biofilm

60

50 ) 2

40

30

20 Biofilm (pixels.mmBiofilm 10

0 Ambient Heated reactor reactor

Figure 6.6 Quantification of annular reactor biofilm using fluorescent microscopy. Combined biofilm quantification for three sampling events during the 13.5 month experiment. Bar graphs represent the mean and one standard deviation for 18 samples. 6.4.2 FLA detection

6.4.2.1 FLA detection by culture FLA were detected at the highest concentrations in the ambient reactor water with a mean of 6.3 amoebae.mL-1 and a maximum single detect of 34 amoebae.mL-1 (Figure 6.7). From here the detection rates decline across the other samples to the lowest FLA mean concentration of 0.2 amoebae.cm-2 for the heated reactor biofilm. The supplying drinking water mean concentration (1.1 amoebae.mL-1) was lower than all the reactor samples means except for the heated reactor biofilm. However, these apparently declining values were not found to be significantly different (Kruskal-Wallis test, p = 0.31) from each other.

174 Chapter 6. Heated annular reactors

Annular reactor FLA detected by culture

20 Water (mL-1) -2

) Biofilm (cm ) -2

15 or cm or -1

10

5 FLA (amoebae.mL

0 Water Water Biofilm Water Biofilm Supply Ambient Heated reactor reactor

Figure 6.7 Annular reactor FLA detected by culture. Combined FLA data for four sampling events during 13.5 month experiment. Box plots with means (+) and whiskers at 5-95 percentiles for > 24 samples.

From all the FLA detected by culture 23 isolates were successfully identified by partial 18S rRNA sequencing (Table 6.1). Three of the isolates from the supplying drinking water were already reported in Chapter 5 (Table 5.1) but are included here as a reference for the annular reactor results. The majority (15/23) of isolates were identified as H. vermiformis. The next most frequently identified FLA was Echinamoeba exudans with five isolates while only three isolates of Acanthamoeba sp. were identified. There was no pattern of FLA species detected relative to the sample type except that no FLA were successfully identified from heated annular reactor biofilm samples. Nearly all the BLAST results were confirmed by aligning the sequences in a phylogentic tree (Figure 6.10). Except for one of the supply drinking water isolates (SDF1h) which was best aligned with Vahlkampfia inornata (AJ224887) over Echinamoeba exudans.

175 Chapter 6. Heated annular reactors Table 6.1 FLA isolated by culture from annular reactors and identified by partial 18S rRNA sequencing. Sampling Months Seq. BLAST best match and BLAST % Isolate # point (date) length ARB alignment (accession #) 100 % SDF1i 24 Echinamoeba exudans EU273832) Supply 13.5 100 % drinking (Dec SDF1ih 670 Hartmannella vermiformis water 09) (EU137741) 100 % SDF1ii 480 Hartmannella vermiformis (AB525839) 100 % 1.5 DF4 180 Hartmannella vermiformis (AB525839) (Dec 100 % 08) DF9 170 Hartmannella vermiformis (AB525839) Ambient 100 % reactor SDF4ii 418 Echinamoeba exudans (AB520727) water 13.5 100 % (Dec SDF4iihC2 125 Hartmannella vermiformis (AB525839) 09) 100 % SDF9i 440 Echinamoeba exudans (EU377601) 100 % 1.5 DB9 500 Hartmannella vermiformis (AB525839) (Dec 100 % 08) DB9ii 249 Hartmannella vermiformis (AB525839) 4 84 % FB4ai 90 Acanthamoeba sp. (Feb 09) (GQ397478) 5 100 % MB4ii 400 Acanthamoeba sp. (Mar09) (GQ397478) Ambient 100 % reactor SDB4i 500 Hartmannella vermiformis (AB525836) biofilm 100 % SDB4ih 510 Hartmannella vermiformis (AB525836) 13.5 100 % SDB4ii 470 Acanthamoeba sp. (Dec 09 (GQ397478) 100 % SDB4iih 460 Hartmannella vermiformis (AB525836) 100 % SDB9ii 249 Hartmannella vermiformis (AB525839) 100 % DF5h 390 Echinamoeba exundans (EU377601) 100 % 1.5 DF5b 560 Echinamoeba exundans (EU377601) (Dec 100 % Heated 08) DF5bh 96 Hartmannella vermiformis (AB525836) reactor 100 % water DF10h 440 Hartmannella vermiformis (AB525839) 98 % FF5a 90 Hartmannella vermiformis 4 (AB525839) (Feb 09) 100 % FF5bh 96 Hartmannella vermiformis (AB525836)

176 Chapter 6. Heated annular reactors 6.4.2.2 FLA detected by qPCR All qPCR reaction products were imaged using gel electrophoresis to confirm the correct products sizes and check for non-specific amplification. Acanthamoeba spp. qPCR gel with an expected product length of 100 bp is provided as an example from samples taken after 1.5 months (December 2008) of reactor operation (Figure 6.8). Positive qPCR products can be seen for all samples except one of the feed hose samples (Lane 2).

L P 1 2 3 4 5 6 7 8 9 10 11 12 N L 1000 bpp 500 bp

100 bp

Figure 6.8 Acanthamoeba spp. qPCR gel electrophoresis image for annular reactor samples taken at 1.5 months (December 2008). Lane L: 100 bp EZ Load DNA ladder, Lane P: Acanthamoeba sp. positive control. Lanes 1 and 7: drinking supply water. Lanes 2 and 8: feed hose water. Lanes 3 and 9: ambient reactor water. Lanes 4 and 10: ambient reactor biofilm. Lane 5 and 11: heated reactor water; Lanes 6 and 12: heated reactor biofilm. DNA free water negative control. FLA were detected by qPCR using the Acanthamoeba spp. and H. vermiformis targeted primers but no FLA were detected using the Naegleria spp. primers (Figure 6.9). Using the Acanthamoeba spp. qPCR the highest mean (46 amoebae.mL-1) was detected for supply drinking water although this was skewed due to a high maximum count (432 amoebae.mL-1) (Figure 6.9- A). Ambient reactor water had the second highest mean (15 amoebae.mL-1) and maximum (59 amoebae.mL-1) detection of Acanthamoeba spp. Detection mean and maxima were lower for ambient reactor biofilm and heated reactor water. The lowest detection mean (0.3 amoebae.cm2) was recorded for heated reactor biofilm samples. However, the differences observed between the samples types were not statistically significant (Kruskal-Wallis test, p = 0.14). Detection of H. vermiformis was highest in the supplying drinking water with a mean of 44 amoebae.mL-1 which was skewed by a maximum detection of 348 amoebae.cm-2 (Figure 6.9 - B). In contrast to the Acanthamoeba spp. qPCR results the second highest mean detection for H. vermiformis was for heated annular reactor water at 21 amoebae.mL-1. Followed by the ambient reactor biofilm with a mean detection of 17 amoebae.cm-2. Heated annular reactor biofilm again add the lowest qPCR detection rates with a mean of 1 amoebae.cm-2. Comparing the difference observed for the sample types it was found that they were not statistically significant (Kruskal- Wallis test, p = 0.51).

177 Chapter 6. Heated annular reactors

ABAnnular reactor Acanthamoeba spp. qPCR Annular reactor H. vermiformis qPCR 450 450 Water (mL-1) -2 ) 400 ) 400 Biofilm (cm ) -2 -2 350 350 or cm or 300 cm or 300 -1 -1

250 250 100 100

80 80

60 60

40 40 FLA (amoebae.mL FLA FLA (amoebae.mL 20 20

0 0 Water Water Biofilm Water Biofilm Water Water Biofilm Water Biofilm Supply Ambient Heated Supply Ambient Heated reactor reactor reactor reactor

Figure 6.9 Annular FLA detected by qPCR over 13.5 months. A: Acanthamoeba spp. targeted qPCR. B: H. vermiformis targeted qPCR. For each sample type there were greater than 12 samples taken. Box plots with means (+), whiskers from 5-95 percentiles and outlying scores (•). From the qPCR eight products from the Acanthamoeba spp. qPCR and nine products from H. vermiformis qPCR were sequenced (Table 6.2). All the Acanthamoeba spp. qPCR were BLAST best matched with Acanthamoeba sp. While all the H. vermiformis qPCR were BLAST best matched with H. vermiformis thus confirming the specificity of the qPCR when applied to the annular reactor samples. The BLAST best matches were confirmed by the use of a phylogenetic tree (Figure 6.10). The supplying drinking water samples were also presented previously (Table 5.2) but are included here to assist in comparison. The only sample type that did not have any qPCR products sequenced was heated reactor biofilm due to the very low detection frequencies (Figure 6.9).

178 Chapter 6. Heated annular reactors Table 6.2 FLA detected by qPCR from annular reactors and identified by sequencing Sampling Months Seq. BLAST best match and BLAST % Isolate # point (date) length ARB alignment (accession #) 100 % DF1 47 Acanthamoeba castellanii 1.5 (JF437606) Supply (Dec 08) 100 % drinking DF6_F8 65 Acanthamoeba sp. (HQ833439) tap water 13.5 99 % SDF1 502 Hartmannella vermiformis (Dec 09) (EU137741) 100 % SDF4_D1 502 Hartmannella vermiformis (EU137741) 99 % Ambient SDF4_E1 502 Hartmannella vermiformis 13.5 (EU137741) reactor (Dec 09) 99 % water SDF9_A2 502 Hartmannella vermiformis (EU808634) 99 % SDF9_H2 502 Hartmannella vermiformis (EU137741) 100 % DB4_H7 63 Acanthamoeba sp. (JF437606) Ambient 1.5 99 % reactor DB9_G11 502 Hartmannella vermiformis (Dec 08) (EU137741) biofilm 100 % DB9_H9 58 Acanthamoeba sp. (JF437606) 99 % DF5b_C3 502 Hartmannella vermiformis (DQ407573) 1.5 100 % DF10_G12 65 Acanthamoeba sp. (Dec 08) (HQ833439) 100 % DF10_H12 65 Acanthamoeba sp. (JF437606) Heated 100 % reactor SDF5_D11 74 Acanthamoba sp. (HQ833439) water 100 % SDF5_H11 65 Acanthamoba sp. 13.5 (JF437606) (Dec 09) 99 % SDF10_D3 502 Hartmannella vermiformis (DQ407573) 99 % SDF10_C3 362 Hartmannella vermiformis (DQ407573)

179 Chapter 6. Heated annular reactors

AF260721, Acanthamoeba castellanii AF019057, Acanthamoeba culbertsoni AF260725, Acanthamoeba polyphaga U07406, Acanthamoeba rhysodes AF260720, Acanthamoeba rhysodes culture_ambient_reactor_biofilm_MB4bii culture_ambient_reactor_biofilm_SDB4iiC1 M13435, Acanthamoeba castellanii AF019061, Acanthamoeba polyphaga AF019067, Acanthamoeba culbertsoni AY026244, Acanthamoeba polyphaga S81337, Acanthamoeba griffini AF019068, Acanthamoeba hatchetti Ac_qPCR_tap_supply_water_DF1 Ac_qPCR_ambient_reactor_biofilm_DB4_H7 Ac_qPCR_heated_reactor_water_SDF5_H11 Ac_qPCR_ambient_reactor_biofilm_DB9_A9 Ac_qPCR_tap_supplywater_DF6_F8 Ac_qPCR_heated_reactor_water_DF10_G12 Ac_qPCR_heated_reactor_water_DF10_H12 Ac_qPCR_ambient_reactor_biofilm_DB9_H9 culture_ambient_reactor_water_SDF4iihC2 culture_heated_reactor_water_FF5a culture_heated_reactor_water_FF5bh culture_ambient_reactor_biofilm_DB5 Ac_qPCR_heated_reactor_water_SDF5_D11 culture_ambient_reactor_water_DF9 culture_ambient_reactor_water_DF4 culture_ambient_reactor_biofilm_SDB9ii Hv_qPCR_ambient_reactor_water_SDF4_E1 Hv_qPCR_ambient_reactor_water_SDF4_D1 Hv_qPCR_ambient_reactor_water_SDF9_A2 culture_ambient_reactor_biofilm_SDB4i culture_ambient_reactor_biofilm_SDB4iih culture_tap_supply_water_SDF1ii Hv_qPCR_ambient_reactor_water_SDF9_H2 DQ084366, Hartmannella vermiformis culture_tap_supply_water_SDF1ih culture_ambient_reactor_biofilm_DB9ii AF426157, Hartmannella vermiformis culture_ambient_reactor_biofilm_SDB4ih culture_heated_reactor_water_DF5b Hv_qPCR_heated_reactor_water_SDF10_D3 culture_ambient_reactor_biofilm_DB9 culture_heated_reactor_water_DF10h Hv_qPCR_ambient_reactor_biofilm_DB9_G11 Hv_qPCR_heated_reactor_water_SDF10_C3 culture_heated_reactor_water_DF5h EF378693, Tetramitus sp. culture_ambient_reactor_biofilm_FB4ai Hv_qPCR_heated_reactor_water_DF5b_C3 AF293895, Echinamoeba exundans culture_ambient_reactor_water_SDF4ii culture_ambient_reactor_water_SDF9i AJ489262, Echinamoeba thermarum AF338419, Naegleria clarki M18732, Naegleria gruberi X93224, Naegleria minor AY576367, Naegleria sp. AF338423, Naegleria fowleri U80057, Naegleria andersoni DQ388520, Heterolobosea sp. AJ224887, Vahlkampfia inornata culture_tap_supply_water_SDF1i AJ224886, Vahlkampfia avara DQ122381, Thecamoeba quadrilinea EU980613, Sappinia pedata AF293896, Filamoeba nolandi AY929916, Platyamoeba sp. DQ913104, Vannella epipetala AY121848, Neoparamoeba aestuarina EF153844, Saccharomyces cerevisiae (baker’s yeast)

0.10 Figure 6.10 Phylogenetic tree of FLA detected in annular reactors. FLA sequences were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the supplying drinking water are blue, ambient annular reactor are black bolded and heated annular reactor are orange. The scale bar for the branch lengths is 0.10 changes per nucleotide.

180 Chapter 6. Heated annular reactors 6.4.3 Legionella detection

6.4.3.1 Legionella detection by culture Using the standard method for the detection of Legionella by culture no presumptive Legionella bacteria were detected from water or biofilm sampled from the annular reactors.

6.4.3.2 Legionella detection by qPCR Gel electrophoresis of the products from the Legionella genus directed 16S rRNA gene qPCR were used to confirmed that the products were of the correct size (Figure 6.11). Lanes 1 and 5 through to 12 show positive Legionella qPCR products for DNA extracted from samples taken after 1.5 months (December 2008).

Figure 6.11 Legionella spp. qPCR gel electrophoresis image for annular reactor samples taken at 1.5 months (December 2008) of continuous. Lanes L: E-Gel DNA ladder. P: Legionella pneumophila positive control. Lanes 1 and 7: drinking supply water. Lanes 2 and 8: feed hose water. Lanes 3 and 9: ambient reactor water. Lanes 4 and 10: ambient reactor biofilm. Lane 5 and 11: heated reactor water; Lanes 6 and 12: heated reactor biofilm. Lane N: DNA free water negative control. Legionella spp. were detected in all sample types at least once during the three sampling events (Figure 6.12). The highest mean detection (80 cells.mL-1) was for the heated annular reactor water but this was in part due to a single high maximum detection of 278 cells.mL-1. The second highest detection mean was for ambient reactor water with 63 cells.mL-1 followed by the supplying drinking water with 55 cells.mL-1. Heated annular reactor biofilm had the lowest detection mean of 13 cells.mL-1. The differences between the Legionella spp. qPCR sample types was found not to be statistically significant (Kruskal-Wallis test, p = 0.67).

181 Chapter 6. Heated annular reactors

Annular reactor Legionella qPCR

300 Water (mL-1)

) Biofilm (cm-2) -2 250 or cm or

-1 200

150 (cells.mL 100

50 Legionella Legionella

0 Water Water Biofilm Water Biofilm Supply Ambient Heated reactor reactor

Figure 6.12 Detection of Legionella sp. in annular reactors using qPCR. Combined Legionella qPCR data for three sampling events during 13.5 month experiment. Each box plot has means (+) and whiskers from 5-95 percentiles for > 12 samples. Of the Legionella spp. qPCR positives samples six out of 20 were successfully cloned and sequenced. All unique cloned qPCR products that were sequenced (n=9) were identified as belonging to the family Legionellaceae (Table 6.3). Three sequences were positively classified as specific Legionella spp. while the remainder were unclassified - Legionellales bacteria (Table 6.3). The BLAST results were mostly complemented by the ARB software phylogenetic tree alignment (Figure 6.14) except that supplying water sample MF1_1 aligned with L. fairfieldiensis over L. dreseniensis. This confirms the specificity of the qPCR for environmental Legionella spp. in the water and biofilm samples.

182 Chapter 6. Heated annular reactors Table 6.3 Legionella qPCR products from annular reactors identified by sequencing

Sampling Months Seq. % ID Isolate # BLAST best match point (date) length (accession #) 96 % Supply 5 MF1_1 448 Legionella dreseniensis (AM747393) drinking (Mar 09) 99 % water MF1_2 454 Legionella waltersii (NR024969) 5 95 % MF4_1 414 Legionellales bacterium (Mar 09) (EF667907) Ambient 97 % reactor SDF9_6 376 Legionellales bacterium 13.5 (EF667907) water (Dec 09) 96 % SDF9_10 414 Legionellales bacterium (EF667907) 96 % DB9_1 410 Legionellales bacterium 1.5 (EU808634) Ambient (Dec 08) 96 % reactor DB9_15 413 Legionellales bacterium (EU808634) biofilm 13.5 96 % SDB4_1 414 Legionella yabuuchiae (Dec 09) (FJ542894) Heated 5 95 % reactor MF5_1 203 Legionellales bacterium (Mar 09) (EU808634) water

6.4.3.3 Detection of Legionella spp. in FLA isolates In total 13 FLA isolates from the annular reactors were screened for the presence of Legionella using PCR more than nine months after they were isolated. One H. vermiformis isolate (SDF1ih) from the supplying drinking water and an Acanthamoeba sp. isolate (SDB4ii) from the ambient reactor biofilm were found to contain Legionella (Figure 6.13). Additionally an Echinamoeba exudans isolate (SDF4ii) from the ambient reactor water was found to contain Legionella also but was not included in the gel electrophoresis image.

Figure 6.13 Example of isolated FLA cultures positive for Legionella spp by PCR. Lanes L: E-Gel DNA ladder, P1: Legionella pneumophila positive control, P2: Environmental Willertia magna, Set up A lane 1: Tap water H. vermiformis isolate (SDFlih), 2: Ambient reactor water Acanthamoeba sp. isolate (SDB4ii), N: negative control. Only the PCR product from the Acanthamoeba sp. isolate (SDB4ii) was successfully cloned and sequenced and identified as Legionella micdadei (Table 6.4). 183 Chapter 6. Heated annular reactors Table 6.4 Identification of Legionella spp. detected by PCR in FLA isolates. Sampling Months FLA Seq. BLAST % BLAST best match point (date) (isolate #) length (accession #) Ambient 13.5 Acanthamoeba sp. 100 % reactor 455 Legionella micdadei (Dec 09) (SDB4ii) (AF227162) water

Z49724, Legionella donaldsonii X73406, Legionella feeleii X97355, Legionella sp. AY957915, uncultured bacterium Z49717, Legionella birminghamensis Z49733, Legionella quinlivanii supply drinking tap water_MF1_A10_1 Z49722, Legionella fairfieldensis AM747393, Legionella dresdeniensis DQ667196, Legionella taurinensis Z32638, Legionella erythra X73409, Legionella jamestowniensis EF036512, Legionella jordanis X73403, Legionella brunensis Z49723, Legionella geestiana; EU835422, uncultured bacterium Z49727, Legionella lansingensis AB233209, Legionella impletisoli AB233211, Legionella yabuuchiae X73397, Legionella oakridgensis Z49730, Legionella londiniensis AJ919271, Legionella nautarum Z32640, Legionella israelensis AF424887, Legionella busanensis AF122883, Legionella gresilensis AF122884, Legionella beliardensis Ambient reactor biofilm_DB9_A8_1 Ambient reactor biofilm_DB9_A8_15 Ambient reactor water_MF4_B1_1 Ambient reactor water_SDF9_B11_10 ) Ambient reactor water_SDF9_B11_6 EF520576, uncultured gamma proteobacterium EF667907, uncultured Legionellales bacterium AY050590, uncultured bacterium Ambient reactor biofilm_SDB4_B9_1 Z49716, Legionella adelaidensis AY744776, Legionella anisa AF122882, Legionella sp. ATCC U59697, Legionella parisiensis X73400, Legionella steigerwaltii Z32644, Legionella tucsonensis X73401, Legionella wadsworthii Z49720, Legionella cherrii X73407, Legionella cincinnatiensis AY444741, Legionella longbeachae Z49725, Legionella gratiana CR628336, Legionella pneumophila X73402, Legionella pneumophila DQ408661, Legionella pneumophila CP000675, Legionella pneumophila Drinking tap water_MF1_A10_2 AF122886, Legionella waltersii X97366, Legionella drancourtii X97357, Legionella sp. X97364, Legionella lytica X97363, Legionella sp. U64035, Legionellalike amoebal pathogen Z49729, Legionella moravica Z49739, Legionella worsleiensis Z49736, Legionella shakespearei U66104, Legionellalike amoebal pathogen Z49873, Chlamydophila pneumoniae

0.10 Figure 6.14 Phylogenetic tree of Legionella spp. detected in annular reactors by qPCR. Sequences of Legionella spp. detected by qPCR were aligned using the neighbour-joining distance matrix method in ARB software. Sequences isolated from the supplying drinking water tap are blue, ambient annular rectors are bolded black and heated annular reactors are orange. The scale bar for the branch lengths is 0.10 changes per nucleotide.

184 Chapter 6. Heated annular reactors 6.5 DISCUSSION

6.5.1 FLA populations in annular reactors FLA were detected by culture at the highest mean concentration in ambient reactor water (6.3 amoebae.mL-1) which was higher than the supplying water (1 amoebae.mL-1) and the heated reactor water (1 amoebae.mL-1) (Figure 6.7). This indicated that the FLA were introduced via the supplying drinking water but effectively colonised the ambient reactor in the initial months of operation and found the elevated temperatures of the heated reactor not conducive to growth. H. vermiformis was the most frequent FLA isolated by culture (Table 6.1) which was consistent with the findings for the distribution system supplying the reactor (Section 4.4.2) and other reported in-premise FLA detections (Sanden et al., 1992; Barbeau and Buhler, 2001; Ménard- Szczebara et al., 2008; Shoff et al., 2008). In contrast to the culture data the qPCR results indicate that supplying drinking water had the highest detection of Acanthamoeba sp. and H. vermiformis but this was likely skewed by high single maximum detections (Figure 6.8). The variability in the data points toward a heterogeneous distribution of FLA where high single detects may be associated with biofilm containing FLA sloughing off into the water (Storey et al., 2004). In contrast to Acanthamoeba sp. qPCR and culture results, H. vermiformis were detected at the highest mean by qPCR in the heated reactor water (21 amoebae.mL-1). H. vermiformis appears to be more thermo-tolerant than Acanthamoeba sp., which is consistent with other research on heated water applications which report H. vermiformis as the most frequently isolated FLA under the higher temperatures (Rohr et al., 1998; Thomas et al., 2006). The densities of FLA detected in the annular reactor were higher than the maximum reported in the only study identified describing FLA densities in a warmed water application where water from cooling towers were reported to have 16 amoebae.mL-1 (Behets et al., 2007). Biofilm was readily colonised by FLA in the ambient reactor and detected by both culture and qPCR methods (Figure 6.7 and 6.8). Conversely, the biofilm in the heated ambient reactor recorded the lowest detection density of FLA by both culture and qPCR. Furthermore, no FLA isolates nor qPCR products were successfully sequenced in these samples due to the low detection densities (Table 6.2 and 6.3). Quantification of the biofilm present on the coupons by fluorescent microscopy confirmed that biofilm was present on both coupon types and was not significantly different in density (Figure 6.6). Also the HPC for the biofilm samples were similar with mean ambient reactor counts of 430 cfu.cm2 compared to heated reactor counts of 300 cfu.cm2. It appears that under the elevated temperature the FLA were preferentially residing in the water and not colonising the biofilm. These results appear to be the first clue that under

185 Chapter 6. Heated annular reactors elevated temperatures in heated water applications that biofilm may not be an important factor for the colonisation or replication of FLA, although further strain .

6.5.2 Legionella populations in annular reactors Legionella were not detected by culture but were detected by qPCR reinforcing the limitations of culture methods for water samples (Devos et al., 2005). Using qPCR Legionella were detected at the highest mean concentration (80 cells.mL-1) in heated reactor water (Figure 6.12). The growth in the heated annular reactor is not unexpected as it has been previously reported that Legionella bacteria in the presence of other bacteria grow in water at 42 °C (Yee and Wadowsky, 1982). Also that the elevated Legionella detection in the heated annular reactor water was matched with H. vermiformis detection by qPCR but was not statistically correlated (Spearman rank test, r = 0.87, p = 0.33). Analogous to the FLA results the lowest detection of Legionella (13 cells.cm-2) was in the heated annular reactor biofilm. Others studies have reported a significant correlation between Legionella detection and the presence of FLA (Breiman et al., 1990; Thomas et al., 2006). As with the FLA the Legionella appear to have been introduced into the reactors by the supplying drinking water and then colonising and possibly increasing in concentration slightly over the course of the experiment. The density of Legionella detected is in the same order of magnitude as other tests from heated water applications such as cooling towers (Miyamoto et al., 1997; Bentham, 2000; Joly et al., 2006; Wéry et al., 2008) and spas (Fallon and Rowbotham, 1990). However, the maximum density detected in heated annular reactor water (280 cells.mL1) was less than the maximums detected in some cooling towers (2.5  103 cells.mL-1) (Chen and Chang, 2010) and spas (1.5  104 cells.mL-1) (Okada et al., 2005). The role of biofilm in proliferation of the Legionella in the annular reactors does not appear as important as the association with FLA. Other studies have reported Legionella to be detected in biofilm of cooling towers at similar (Declerck et al., 2007) or increased concentrations (Chen and Chang, 2010) compared to corresponding water. However, cooling towers tend to operate at ambient temperatures and hence the conditions in the ambient reactor more closely resemble the water present in a cooling tower except for the absence of recirculation. In the ambient reactor Legionella was detected in the biofilm at mean concentrations of 39 cells.cm-2 but at much lower densities (13 cells.cm-2) in the heated reactor biofilm. However, there was no literature identified that reported on the density of Legionella in biofilm from a heated water system for comparison purposes. In these experiments the presence of FLA appears to be a more important factor than biofilm quantity in the density of Legionella bacteria. This findings are complemented by other research using annular reactors where

186 Chapter 6. Heated annular reactors Legionella numbers increased by 2.9 log units after the addition of A. castellannii despite the presence of an established biofilm before the FLA addition (Declerck et al., 2009). There was a large diversity of Legionella bacteria detected in the small number of qPCR samples (n = 6) that were sequenced with the majority (6 out of 9) belonging to unclassified Legionella or Legionalles bacteria with unknown pathogenicity (Table 6.3 and Figure 6.14). Other research looking at the diversity of Legionella in drinking water also reported a large number of yet unclassified Legionella sp. (Wéry et al., 2008; Wullings et al., 2011). The diversity of Legionella in drinking water applications needs to be explored further particularly as the pathogenicity of the detected but unclassified Legionella bacteria is presently unknown making any risk assessments difficult.

6.5.3 FLA infected with Legionella Of the 13 FLA isolates screened for Legionella by PCR three were positive from the supplying water and ambient reactor (Section 6.4.3.3). Only two heated reactor water isolates Echinamoeba exudans (DF5b) and H. vermiformis (FF5bh) were screened for Legionella so it is possible that Legionella infection was present in the heated reactor but not captured in the very small number of isolates screened, as other research has reported replication of Legionella in FLA isolated from a hospital hot water system at 42 °C (Fields et al., 1989). The diversity of FLA found to be naturally infected with Legionella sp. in drinking water is comparable to other research which identified Legionella sp. in Acanthamoeba sp. (Corsaro et al., 2010), E. exudans (Thomas et al., 2008) and H. vermiformis (Thomas et al., 2006). Some of the FLA population in the supplying drinking water and annular reactors appear to be permanently infected with the Legionella sp. present which highlights that FLA are facilitating the growth of Legionella in the treated drinking water and their applications. The identity of the Legionella sp. present in the annular reactor biofilm isolate (Acanthamoeba sp. - SDB4ii) was identified as a possibly pathogenic Legionella micdadei (Table 6.4). L. micdadei is a known pathogen (Fields et al., 2002) and as an example was responsible for a mild outbreak of Legionella infections (Pontiac Fever) from a whirlpool in Scotland (Fallon and Rowbotham, 1990). The detection of pathogenic Legionella in a FLA isolates from the annular reactors emphasises the role that FLA play in facilitating pathogenic Legionella growth in applications of water where water is retained and biofilm can grow.

6.5.4 Conclusions and future research This research has contributed valuable knowledge on the density and diversity of FLA and Legionella in drinking water and heated water applications. The finding that the biofilm in the heated reactor was not as readily colonised by FLA requires further exploration using 187 Chapter 6. Heated annular reactors different biofilm materials, such as PVC and copper (van der Kooij et al., 2005). If biofilm is not a large contributing factor to the proliferation of Legionella in heated water applications then it will have considerable influence on the health risk assessment models and control strategies. Another aspect that requires further attention is the density of FLA present in applications of drinking water. More sensitive experiments with a greater number of replicates need to be designed in order to quantify the correlation between FLA and Legionella density and diversity. Additionally it must not be forgotten that other pathogenic ARM are likely to be infecting the FLA population and the pathogenicity of the Legionella sp. present need to be determined in order to accurately estimate the overall health risk presented by inhaling aerosols containing these pathogens.

188 Chapter 6. Heated annular reactors

6.6 REFERENCES 1. Armstrong, T. W. and C. N. Haas. 2007. Quantitative Microbial Risk Assessment Model for Legionnaires' Disease: Assessment of Human Exposures for Selected Spa Outbreaks. Journal of Occupational and Environmental Hygiene. 4(8): 634 - 646. 2. Barbaree, J. M., B. S. Fields, J. C. Feeley, G. W. Gorman and W. T. Martin. 1986. Isolation of protozoa from water associated with a legionellosis outbreak and demonstration of intracellular multiplication of Legionella pneumophila. Applied and Environmental Microbiology. 51(2): 422-424. 3. Barbeau, J. and T. Buhler. 2001. Biofilms augment the number of free-living amoebae in dental unit waterlines. Research in Microbiology. 152(8): 753-760. 4. Bartram, J., Y. Chartier, J. Lee, K. Pond and S. Surman-Lee. 2007. Legionella and the prevention of Legionellosis. Geneva, World Health Organisation 5. Behets, J., P. Declerck, Y. Delaedt, L. Verelst and F. Ollevier. 2007. Survey for the presence of specific free-living amoebae in cooling waters from Belgian power plants. Parasitology Research. 100(6): 1249-1256. 6. Bentham, R. H. 1993. Environmental factors affecting the colonization of cooling towers by Legionella spp. in South Australia. International Biodeterioration & Biodegradation. 31(1): 55-63. 7. Bentham, R. H. 2000. Routine sampling and the control of Legionella spp. in cooling tower water systems. Current Microbiology. 41(4): 271-275. 8. Berk, S. G., J. H. Gunderson, A. L. Newsome, A. L. Farone, B. J. Hayes, K. S. Redding, N. Uddin, E. L. Williams, R. A. Johnson, M. Farsian, A. Reid, J. Skimmyhorn and M. B. Farone. 2006. Occurrence of infected amoebae in cooling towers compared with natural aquatic environments: implications for emerging pathogens. Environmental Science & Technology. 40(23): 7440-7444. 9. Breiman, R. F., B. Fields, G. N. Sanden, L. Volmer, A. Meier and J. Spika. 1990. Association of shower use with Legionnaires' disease. Possible role of amoebae. Journal of the American Medical Association. 263(21): 2924 - 2926. 10. erva, L. 1971. Studies of limax amoebae in a Swimming Pool. Hydrobiologia. 38(1): 141- 161. 11. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634. 12. Corsaro, D., G. S. Pages, V. Catalan, J.-F. Loret and G. Greub. 2010. Biodiversity of amoebae and amoeba-associated bacteria in water treatment plants. International Journal of Hygiene and Environmental Health. 213(3): 158-166. 13. Declerck, P., J. Behets, A. Margineanu, V. van Hoef, B. De Keersmaecker and F. Ollevier. 2009. Replication of Legionella pneumophila in biofilms of water distribution pipes. Microbiological Research. 164(6): 593-603. 14. Declerck, P., J. Behets, V. van Hoef and F. Ollevier. 2007. Detection of Legionella spp. and some of their amoeba hosts in floating biofilms from anthropogenic and natural aquatic environments. Water Research. 41(14): 3159-3167. 15. Den Boer, J. W., P. F. Yzerman, J. Schellekens, K. D. Lettinga, H. C. Boshuizen, J. E. V. Steenbergen, A. Bosman, S. V. d. Hof, H. A. V. Vliet, M. F. Peeters, R. J. V. Ketel, P. Speelman, J. L. Kool, A. E. Marina and C. V. Spaendonck. 2002. A large outbreak of Legionnaires’ disease at a flower show, the Netherlands, 1999. Emerging Infectious Diseases. 8(1): 37-43. 16. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925.

189 Chapter 6. Heated annular reactors 17. Esterman, A., I. Calder, S. Cameron, D. Roder, R. Walters, P. Christy and B. Robinson. 1987. Determinants of the microbiological characteristics of spa pools in South Australia. Water Research. 21(10): 1231-1235. 18. Euser, S. M., M. Pelgrim and J. W. den Boer. 2010. Legionnaires' disease and Pontiac fever after using a private outdoor whirlpool spa. Scandinavian Journal of Infectious Diseases. 42(11-12): 910-916. 19. Eyring, A., F. St. Denis and B. Gary. 2008. Changes in water quality in premise plumbing: cold water verse hot water. American Water Works Association Water Quality and Technology Conference, Cincinnati, Ohio, American Water Works Association. 20. Falkinham, J., M. Iseman, P. de Haas and D. van Soolingen. 2008. Mycobacterium avium in a shower linked to pulmonary disease. Journal of Water and Health. 06(2): 209- 213. 21. Fallon, R. J. and T. J. Rowbotham. 1990. Microbiological investigations into an outbreak of Pontiac fever due to Legionella micdadei associated with use of a whirlpool. Journal of Clinical Pathology. 43(6): 479-483. 22. Feazel, L. M., L. K. Baumgartner, K. L. Peterson, D. N. Frank, J. K. Harris and N. R. Pace. 2009. Opportunistic pathogens enriched in showerhead biofilms. Proceedings of the National Academy of Sciences of the United States of America. 106: 16393- 16399. 23. Fields, B., R. Benson and E. Besser. 2002. Legionella and Legionnaires' disease: 25 years of investigation. Clinical Microbiology Reviews. 15(3): 506-526. 24. Fields, B. S., G. N. Sanden, J. M. Barbaree, W. E. Morrill, R. M. Wadowsky, E. H. White and J. C. Feeley. 1989. Intracellular multiplication of Legionella pneumophila in amoebae isolated from hospital hot water tanks. Current Microbiology. 18(2): 131-137. 25. García-Fulgueiras, A., C. Navarro, D. Fenoll, J. García, P. González-Diego, T. Jiménez-Buñuales, M. Rodriguez, R. Lopez, F. Pacheco, J. Ruiz, M. Segovia, B. Baladrón and C. Pelaz. 2003. Legionnaires’ disease outbreak in Murcia, Spain. Emerging Infectious Diseases. 9(8): 915-921. 26. Heath, T. C., C. Roberts, B. Jalaludin, I. Goldthrope and A. G. Capon. 1998. Environmental investigation of a legionellosis outbreak in western Sydney: the role of molecular profiling. Australian and New Zealand Journal of Public Health. 22(4): 428- 431. 27. Hill, G., E. Pring and P. Osborn. 1990. Cooling towers, principles and practice. London, Butterworth-Heinemann. 28. Jernigan, D. B., J. Hofmann, M. S. Cetron, J. P. Nuorti, B. S. Fields, R. F. Benson, R. F. Breiman, H. B. Lipman, R. J. Carter, C. A. Genese, S. M. Paul, P. H. Edelstein and I. C. Guerrero. 1996. Outbreak of Legionnaires' disease among cruise ship passengers exposed to a contaminated whirlpool spa. The Lancet. 347(9000): 494-499. 29. Joly, P., P.-A. Falconnet, J. Andre, N. Weill, M. Reyrolle, F. Vandenesch, M. Maurin, J. Etienne and S. Jarraud. 2006. Quantitative real-time Legionella PCR for environmental water samples: data interpretation. Applied and Environmental Microbiology. 72(4): 2801-2808. 30. Kool, J. L. M. D., D. M. P. H. Bergmire-Sweat, J. C. M. D. Butler, E. W. Brown, D. J. M. D. Peabody, D. S. M. D. Massi, J. C. P. E. Carpenter, J. M. B. A. Pruckler, M. S. Robert F. Benson and B. S. P. Fields. 1999. Hospital characteristics associated with colonization of water systems by Legionella and risk of nosocomial Legionnaires' Disease: a cohort study of 15 hospitals. Infection Control and Hospital Epidemiology. 20(12): 798-805. 31. Lin, Y.-s. E., R. D. Vidic, J. E. Stout and V. L. Yu. 1998. Legionella in water distribution systems. Journal American Water Works Association. 90(9): 11. 32. Lindahl, P. 2004. "Operating strategies and water treatment " Cooling tower operation and maintenance for improved efficiency Retrieved 03 November 2008, from http://www.cti.org/cgi-bin/download.pl. 190 Chapter 6. Heated annular reactors 33. Ménard-Szczebara, F., N. Berthelot, D. Cavereau, S. Oberti, Y. Héchard, V. Sarroca, D. Rivière and S. Mazoua. 2008. Ecology of free-amoebae in real in-house water networks. European Journal of Water Quality. 39(1): 65-75. 34. Miyamoto, H., H. Yamamoto, K. Arima, J. Fujii, K. Maruta, K. Izu, T. Shiomori and S. Yoshida. 1997. Development of a new seminested PCR method for detection of Legionella species and its application to surveillance of Legionellae in hospital cooling tower water. Applied and Environmental Microbiology. 63(7): 2489-2494. 35. NSW Department of Health. 2004. NSW code of practice for the control of Legionnaires' disease. NSW Department of Health, NSW Department of Health,: 75. 36. Okada, M., K. Kawano, K. Fumiaki, J. Amemura-Maekawa, H. Watanabe, K. Yagita, T. Endo and S. Suzuki. 2005. The largest outbreak of legionellosis in Japan associated with spa baths : Epidemic curve and environmental investigation. Kansenshogaku zasshi 79(6): 365-374. 37. Ollos, P., P. Huck and R. Slawson. 2003. Factors affecting biofilm accumulation in model distribution systems. Journal of American Water Works Association. 95: 87-97. 38. Packroff, G., J. Lawrence and T. Neu. 2002. In situ confocal laser scanning microscopy of protozoans in cultures and complex biofilm communities. Acta Protozoologica. 41: 245- 253. 39. Rohr, U., S. Weber, R. Michel, F. Selenka and M. Wilhelm. 1998. Comparison of free- living amoebae in hot water systems of hospitals with isolates from moist sanitary areas by identifying genera and determining temperature tolerance. Applied and Environmental Microbiology. 64(5): 1822-1824. 40. Sanden, G. N., W. E. Morrill, B. Fields, R. F. Breiman and J. M. Barbaree. 1992. Incubation of water samples containing amoebae improves detection of Legionellae by the culture method. Applied and Environmental Microbiology. 58(6): 2001-2004. 41. Shoff, M., A. Rogerson, S. Kessler, S. Schatz and D. Seal. 2008. Prevalence of Acanthamoeba and naked amoebae in South Florida domestic water. Journal of Water and Health. 6(1): 5. 42. Storey, M., N. Ashbolt and T. A. Stenström. 2004. Biofilms, thermophilic amoebae and Legionella pneumophila - a quantitative risk assessment for distributed water. Water Science & Technology. 50(1): 77-82. 43. Storey, M. and N. J. Ashbolt. 2001. Persistence of two model enteric viruses (B40-8 and MS-2 bacteriophages) in water distribution pipe biofilms. Water Science and Technology. 43(12): 133-138. 44. Stout, J. E., V. L. Yu, P. Muraca, J. Joly, N. Troup and L. S. Tompkins. 1992. Potable water as a cause of sporadic cases of community-acquired Legionnaires' Disease. New England Journal of Medicine. 326(3): 151-155. 45. Thomas, V., K. Herrera-Rimann, D. S. Blanc and G. Greub. 2006. Biodiversity of amoebae and amoeba-resisting bacteria in a hospital water network. Applied and Environmental Microbiology. 72(4): 2428-2438. 46. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 47. van der Kooij, D., H. R. Veenendaal and W. J. Scheffer. 2005. Biofilm formation and multiplication of Legionella in a model warm water system with pipes of copper, stainless steel and cross-linked polyethylene. Water Research. 39(13): 2789 - 2798. 48. Volk, C. J. and M. W. LeChevallier. 1999. Impacts of the reduction of nutrient levels on bacterial water quality in distribution systems. Applied and Environmental Microbiology. 65(11): 4957-4966. 49. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 191 Chapter 6. Heated annular reactors 50. Wéry, N., V. Bru-Adan, C. Minervini, J.-P. Delgénes, L. Garrelly and J.-J. Godon. 2008. Dynamics of Legionella spp. and bacterial populations during the proliferation of L. pneumophila in a cooling tower facility. Applied and Environmental Microbiology. 74(10): 3030-3037. 51. Wullings, B. A., G. Bakker and D. van der Kooij. 2011. Concentration and diversity of uncultured Legionella spp. in two unchlorinated drinking water supplies with different concentrations of natural organic matter. Applied and Environmental Microbiology. 77(2): 634-641. 52. Yee, R. B. and R. M. Wadowsky. 1982. Multiplication of Legionella pneumophila in unsterilized tap water. Applied and Environmental Microbiology. 43(6): 1330-1334.

192 CHAPTER 7

FLA INFECTION WITH LEGIONELLA PNEUMOPHILA 7 TABLE OF CONTENTS

7.1 INTRODUCTION______195 7.1.1 Environmental FLA isolates infected with ARM______195 7.1.2 Laboratory cultures of FLA infected with ARM ______196 7.1.3 Influence of environmental characteristics______196 7.2 AIMS ______197 7.3 METHODOLOGY ______197 7.3.1 FLA and Legionella inoculation and incubation ______197 7.3.2 Preliminary optimization and controls ______199 7.4 RESULTS ______200 7.4.1 FLA controls infected with ARB ______200 7.4.2 Experimental controls ______202 7.4.3 Modified Robbins Device FLA isolates______208 7.4.4 Annular reactor FLA isolates ______214 7.4.5 Garden hoses FLA isolates______216 7.5 DISCUSSION ______217 7.5.1 Control FLA infected with ARB ______217 7.5.2 Infection rates______218 7.5.3 L. pneumophila numbers ______220 7.5.4 Trophozoite numbers ______221 7.5.5 Conclusions ______222 7.6 REFERENCES ______223

Chapter 7. L. pneumophila infection LIST OF TABLES

Table 7.1 FLA used in the uptake experiments______198 Table 7.2 Identity of partial 16S rDNA sequences from ARB infected ATCC FLA ______202

LIST OF FIGURES

Figure 7.1 Inoculation of tissue culture plates with FLA and Legionella ______199 Figure 7.2 Infected A. castellanii ATCC 30234 trophozoites ______201 Figure 7.3 Video stills of an infected cyst of A. polyphaga ATCC 30461 ______201 Figure 7.4 Gel electrophoresis of 16S rDNA PCR products______202 Figure 7.5 Individual FLA and L. pneumophila controls over seven days ______204 Figure 7.6 Fluorescent microscopy of A. castellanii (ATCC 30234) ______205 Figure 7.7 Control A. castellanii (ATCC 30234) infected with L. pneumophila ______206 Figure 7.8 Control H. vermiformis (ATCC 50237) infected with L. pneumophila ______208 Figure 7.9 Fluorescent microscopy of MRD drinking water H. vermiformis (pfcii) ______209 Figure 7.10 MDR drinking water H .vermiformis (pfcii) infected with L. pneumophila _____ 210 Figure 7.11 Fluorescent microscopy with depth scans of Acanthamoeba sp. (rfbii) ______211 Figure 7.12 MRD recycled water Acanthamoeba sp. (rfbii) infected with L. pneumophila ___ 212 Figure 7.13 Fluorescent microscopy of MRD recycled water Acanthamoeba sp. (rfbiih) ____ 212 Figure 7.14 MRD recycled water Acanthamoeba sp. (rfbiih) infected with L. pneumophila __ 214 Figure 7.15 Annular reactor Acanthamoeba sp. (SDB4) infected with L. pneumophila _____ 215 Figure 7.16 Garden hose H. vermiformis (SDB7) infected with L. pneumophila______217

194 Chapter 7. L. pneumophila infection

7.1 INTRODUCTION FLA can actively uptake ARM via phagocytosis during feeding (Scheid et al., 2008). Hence, uptake is known to occur when the FLA are in the trophozoite feeding form, which often occurs within biofilms where they are in close proximity to their prey (Thomas et al., 2008). After uptake ARM are contained with membrane-bound vacuoles within the FLA cells and avoid digestion (Snelling et al., 2006). Aside from the active FLA driven uptake it is also possible that ARM enter the FLA via infective mechanisms. Bacterial pathogens enter and infect eukaryotic cells via injecting bacterial effectors into the cell cytoskeleton which causes engulfment of the bacteria into an entry vacuole (Ray et al., 2009). Although these mechanisms are likely they have not been shown specifically for FLA. It would however, explain how FLA are rapidly infected by ARM (Cirillo et al., 1994). Legionella spp. are to date the only genera that have a number of species that are able to infect a range of FLA consistently (Lau and Ashbolt, 2009), which is likely due to the evolution of the bacteria (Albert-Weissenberger et al., 2007). Legionella pneumophila utilises 27 dot/icm genes to avoid digestion and replicate successfully in both FLA and human macrophage cells (Heidtman et al., 2009; Isberg et al., 2009). After growth in FLA, pathogenic Legionella cells are more infective for human epithelial cells (Cirillo et al., 1994) and human monocyte cell lines (Neumeister et al., 2000) (Section 1.3.1.1). However, the infection and growth rates of L. pneumophila are specific for different species and FLA (Neumeister et al., 2000; Molmeret et al., 2001) and therefore need to be determined for the FLA of interest.

7.1.1 Environmental FLA isolates infected with ARM In a review of the literature covering FLA in drinking water systems six different genera of FLA were identified as naturally hosting ARM; Acanthamoeba, Echinamoeba, Hartmannella, Naegleria, Platyamoeba and Vannella (Table 2.1). Recent environmental FLA isolates have significantly different responses to chemical stresses when compared to culture collection FLA (Rowbotham, 1980; Srikanth and Berk, 1993). There is also evidence that similar differences exists for FLA and ARM interactions. For example, Mycobaterium avium infects a significantly lower portion of environmental Acanthamoeba spp. (< 2 months since initial isolation) with less cells compared to laboratory cultured Acanthamoeba spp. (Berry et al., 2010). Therefore it is essential that experiments with ARM are conducted with environmental isolates of FLA from the water system of interest to determine what interactions and therefore risks FLA may present.

195 Chapter 7. L. pneumophila infection 7.1.2 Laboratory cultures of FLA infected with ARM Acanthamoeba followed by Naegleria are the genera of choice for laboratory based research looking at FLA and ARM interactions (Greub and Raoult, 2004). This is due in part to the greater availability of these pathogenic genera through culture collections. However, given the diversity of environmental FLA (Table 2.10) and the difference in responses to ARM (Berry et al., 2010) the results with culture collection FLA may have little relevance to the actual environment. Although, techniques utilising FLA co-culture and enrichment are valuable and have enabled isolation of a specific group of water-based ARM that would not grow as pure cultures on traditional culture media (Rowbotham, 1980; Pagnier et al., 2008). Novel microorganisms such as Amoebosporidium minutum (microsporidian organism) (Hoffmann et al., 1998), Rhabdochlamydia spp. (Corsaro et al., 2009), Parachlamydia acanthamoebae (Birtles et al., 1997), and the giant mimivirus (La Scola et al., 2003) and its virophage (La Scola et al., 2008) have all been identified with the aid of amoebae co-culture.

7.1.3 Influence of environmental characteristics The type of ARM and environmental characteristics such as temperature and cell ratios will determine if FLA are infected. As environmental conditions are key factors it is critical that research is conducted under environmental conditions that are representative of what is likely to occur in situ in the water system of interest. Infection of FLA by Legionella spp. is optimal above 20 °C (Rowbotham, 1980) and specifically L. pneumophila has been shown to infect A. castellannii (ATCC 30234) at 35 °C. For A. polyphaga, it has been found that the bacterium Burkholderia cepia is endosymbiotic at temperatures below 30 °C but causes FLA lysis at 37 °C (Marolda et al., 1999). Similarly, A. polyphaga infected with Parachlamydia acanthamoeba is endosymbiotic at 25 - 30 °C degrees but lytic at 32 - 37 °C probably due to optimal growth temperatures (Greub et al., 2003). If FLA lyse then large numbers of pathogenic ARMs are released into the water which then allows for rapid infection of other FLA and potentially humans (Rowbotham, 1980). The rate of infection varies with A. castellannii shown to be infected after only 2 hr of incubation with L. pneumophila (Neumeister et al., 2000) while H. vermiformis infection was reported after eight days of incubation with Legionella spp. (Wadowsky et al., 1991). More frequently described are infections observed over one to three days for FLA and Legionella spp. (Neumeister et al., 2000; Molmeret et al., 2001). The rate of infection is likely linked to the ARB to FLA ratios used in the experiments. Higher inoculation ratios of bacteria to FLA are often used to achieve a result quickly, with a Legionella to FLA ratio of 100:1 frequently used (Moffat, 1992)(Neumeister et al., 2000; Molmeret et al., 2001).

196 Chapter 7. L. pneumophila infection Nutrient conditions are another important factor. L. pneumophila has been shown to be able to infect Acanthamoeba spp. in both low nutrient saline media (Molmeret et al., 2001) and sterile tap water (Wadowsky et al., 1991) as well as high nutrient human cell line media (Neumeister et al., 2000). In this research low nutrient drinking and recycled water conditions from which the FLA were isolated environmental characteristics were mimicked in the experiment. Understanding of the interactions between the environmental FLA isolated and pathogenic Legionella is critical to determining what risk the FLA may present. Knowledge of infection and growth rates of L. pneumophila facilitated by the environmental FLA will allow more detailed and specific risk assessments to be developed for the sampled water systems and applications.

7.2 AIMS The aim of this research was to establish if the environmental FLA isolated from the environmental sampling could be infected with Legionella pneumophila. Furthermore, the rates of infections, any increase in L. pneumophila numbers and effects on FLA cells numbers with respect to temperature were to be determined. This data was essential in estimating the relationship between FLA and Legionella in drinking and recycled water systems and was a critical in-put into the risk assessment model (Chapter 8).

7.3 METHODOLOGY

7.3.1 FLA and Legionella inoculation and incubation FLA and Legionella cultures were enumerated and Legionella was stained (Vybrant® CFDA SE Cell Tracer Kit) as described (Section 3.7.3). The ATCC® L. pnuemophila species selected were ATCC® 33152 isolated from a human lung infection and ATCC® 33155 isolated from creek water (Table 3.3). In total three FLA controls and five FLA isolates were used in the experiments (Table 7.1).

197 Chapter 7. L. pneumophila infection Table 7.1 FLA used in the uptake experiments FLA Isolate # Species Source location 30234 Acanthamoebae castellanii Yeast culture (London, UK) Control 30461 Acanthamoeba polyphaga Eye infection (Houston, USA) 50237 Hartmannella vermiformis Cooling tower (South Dakota, USA) pfcii Hartmannella vermiformis MRD drinking water rfbii Acanthamoeba sp. MRD recycled water Environmental rfbiih Acanthamoeba sp. MRD recycled water isolates SDB4 Acanthamoeba sp. Annular reactor heated biofilm SDB7 Hartmannella vermiformis Garden hose biofilm Volumes of cells were added to achieve a Legionella to FLA ratio of 5:1 (Section 7.3.2) in 500 mL of sterile filtered water (Milli-Q) in 24 well tissue culture plates (Sarstedt). Specifically, 1  106 Legionella cells were added to 2 x 105 FLA cells (both trophozoites and cysts). Two plates for each temperature (22 and 37 °C), four in total, were inoculated with FLA and Legionella combinations in duplicate (Figure 7.1) with FLA and Legionella only controls on the second plate for each temperature. All plates were wrapped in film (Parafilm) and incubated at either 22 or 37 °C in the dark. Samples (10 μL) were taken every one to two days from each well and enumerated and imaged as described (Section 3.7.3 and 3.7.4). In total the whole experiment was repeated three times.

198 Chapter 7. L. pneumophila infection

       "&&$(%  "&&$((

             ! !

                  !!

                 !!

              '!'!

       )!

     "&#%&'!

Figure 7.1 Inoculation of tissue culture plates with FLA and Legionella. Plates were inoculated with FLA and Legionella combinations in 500 mL of sterile filtered water (Milli-Q) in duplicate. 7.3.2 Preliminary optimization and controls Initial optimisation experiments were conducted prior to the main experiments to determine the ideal L. pneumophila to FLA ratios and the validity of the stain. FLA and L. pneumophila were cultured and stained as described (Section 3.7.3). Experiments were set up as described for the main experiment (Section 7.3.1) except only with ratios of L. pneumophia (ATCC®33152) to A. castellannii (ATCC®30234) of 5:1, 50:1 and 500:1 at 22 °C. From these optimisation experiments (data not shown) it was found that the lowest concentration at which infection was observed was 5:1 and hence this ratio was used for further experiments as the lower concentrations of L. pneumophila were more representative of the water systems from which the FLA were isolated.

7.3.2.1 Calculation of L. pneumophila numbers From the optimisation experiments it was determined that the dimensions of the observed infected vacuoles were 2 -3 μm3 and it was estimated that 10 L. pneumophila cells were present per infected vacuole. Heavily infected trophozoites generally had greater then 10 vacuoles or one

199 Chapter 7. L. pneumophila infection very large vacuole that dominated the entire trophozoite (Figure 7.6 B - C) and it was estimated that they contained 100 L. pneumophila cells in total. As it was not possible to count individual cells directly this estimation was partly guided by other research which quantified the number of L. pneumophila in a heavily infected FLA as 100 CFU (Kuiper et al., 2004). The number of infected vacoules was used to calculate the number of intracellular L. pneumophila present within the trophozoites.

7.3.2.2 Heat killed negative control To ensure that infected vacuoles being observed were due to living ingested L. pneumophila and not diffusion of the cell stain from digested L. pneumophila, a heat killed negative control was designed. A. castellanii and H. vermiformis have both been shown to produce the same number of new trophozoites when fed with living or stained heat killed bacteria although the growth rate of each trophozoite is significantly lower when feeding on heated killed bacteria (Pickup et al., 2007). For this experiment L. pneumophila bacteria were stained as described then heated killed by incubating at 90 °C for 5 min (Wang et al., 2010). After heat killing the CDFA stain was still viable in the heat killed L. pneumophila (Wang et al., 2010). Heat killed L. pneumophila were incubated with the FLA and enumerated as described. Non-motile L. pneumophila could be observed but with little fluorescence present however no L. pneumophila infected vacuoles could be observed in any of the FLA used in the experiment. This preliminary control indicates that the FLA were either not being infected by heat killed L. pneumophila or that once ingested the stain present in the L. pneumophila was not being activated by the esterase activity within the FLA. Either way this control gave confidence that the experimental methods used identified living infecting L. pneumophila in the vacuoles of the infected FLA trophozoites.

5.4 RESULTS

7.4.1 FLA controls infected with ARB

7.4.1.1 Microscopy of FLA infected with ARB During the preliminary stages of the experiment it was observed that the ATCC control FLA trophozoites and cysts, A. castellanii (ATCC 30234) and A. polyphaga (ATCC 30461), both contained motile ARB no more than 1 μm in length or width (Figure 7.2 and Figure 7.3). The high motility of the ARB within the FLA could be clearly observed using video imaging (Figure 7.3).

200 Chapter 7. L. pneumophila infection

Figure 7.2 Infected A. castellanii ATCC 30234 trophozoites. Image A and B: A. castellanii trophozoites with selected visible ARB indicated. The image was taken using bright field light microscopy at 1000  magnification (DM400B, Lieca) with a 10 μm scale bar.



Figure 7.3 Video stills of an infected cyst of A. polyphaga ATCC 30461. A - D: images taken in sequence at 1 s intervals. The ARB can be seen to move around within the cyst. Video was taken using phase contrast light microscopy at 1000  magnification (DM400B, Lieca) with a 10 μm scale bar. 201 Chapter 7. L. pneumophila infection 7.4.1.2 Molecular detection of ARB infecting FLA Partial 16S rRNA gene PCR of the axenic FLA ATCC cultures was conducted as described (Section 3.6.2.4) and confirmed the microscopy observations that there were contaminating bacteria. Bacterial 16S rRNA genes were not detected in the H. vermiformis (ATCC 50237) cultures (lanes 1 - 4, Figure 7.4) but was amplified from both the one and three week old cultures of A. polyphaga (ATCC 30461) (lanes 5 - 8, Figure 7.4).

Figure 7.4 Gel electrophoresis of 16S rRNA PCR products for A. polyphaga (ATCC 30461) and H. vermiformis (ATCC 50237). Lane L: ladder. P: positive control E. coli DNA. Lane 1 and 2: H. vermiformis one week old culture. Lane 3 and 4: H. vermiformis three week old culture. Lane 5 and 6: A. polyphaga one week old culture. Lane 7 and 8: three week old A. polyphaga culture. Lane N: negative control nuclease free water. PCR products were purified, cloned, sequenced and the BLAST searched as described (Section 3.6.5 and 3.6.6.2). For A. castellanii (ATCC 30234) one PCR product and two clones were sequenced. For A. polyphaga (ATCC 30461) two PCR products and eight clones in total were sequenced. The ARB were identified as two different species of bacteria (Figure 7.2). Table 7.2 Identity of partial 16S rRNA sequences from ARB infected ATCC FLA Length BLAST search best hits, RDP % identity Culture (bp) and phylogentic (accession #) A. castellanii 100 % 708 Sphingomonas rhizogenes (ATCC 30234) (JF276901) A. polyphaga 93 % 722 Uncultured bacterium clone (ATCC 30461) (GU078602) 7.4.2 Experimental controls FLA and Legionella only controls were conducted for each experiment. Over three replicated experiments at 22 °C the number of FLA trophozoites generally stayed at similar concentrations as inoculated or slightly increased over the seven days (Figure 7.5 A-B). Specifically, MRD drinking water H. vermiformis (pfcii) mean concentration increased from 493 to 700 trophozoites.μL-1 while recycled water Acanthamoeba sp. (rfbii) decreased slightly from 434 to 305 trophozoites.μL-1. Both L. pneumophila (ATCC 33152 and 33155) mean

202 Chapter 7. L. pneumophila infection concentration increased from 7  103 to 1.3  104 Legionella.μL-1 for ATCC 33152 and 1.2  104 Legionella.μL-1 for ATCC 33155 over seven days. Under warmer experimental temperatures of 37 °C all mean concentrations of the FLA trophozoites decreased to zero over the seven days (Figure 7.5 C-D). Trophozoite numbers either disappeared within the first two days or decreased gradually over the seven days. Garden hose H. vermiformis isolate (SDB7) decreased quickly from an inoculating concentration of 367 to 0 trophozoites.μL-1 in just two days. While, other FLA such as recycled water Acanthamoeba sp. (rbiih) had a more gradual decrease from an inoculation mean of 559 to zero trophozoites.μL- 1 at day seven. L. pneumophila were not adversely affected by temperature with mean concentrations increased in a similar pattern as observed at 22 °C. Mean concentrations increased from 7  103 to 1.1  104 Legionella.μL-1 for ATCC 33152 and 1.3  104 Legionella.μL-1 for ATCC 33155 over seven days.

203 Chapter 7. L. pneumophila infection

ABA Controls for L. pnuemophila (ATCC 33152) BB Controls for L. pnuemophila (ATCC 33155) and MRD FLA at 22 °C and other FLA at 22 °C Legionella.

16000 16000 -1 16000 16000 -1 L Legionella. L

14000  14000 14000  14000 12000 12000 12000 12000 10000 10000 10000 10000 8000 8000

8000 8000  L

6000  6000

6000 -1

6000 L Legionella. Legionella.

4000 -1 4000 4000 4000 2000 2000 2000 2000 -1 -1 1600 1600 L Trophozoites. L 1600 Trophozoites. 1600   1400 1400 1400 1400 1200 1200 1200 1200 1000 1000 1000 1000 800 800 800 800 600 600 600 600   Trophozoites. Trophozoites. L 400 400 L 400 400 -1 200 200 -1 200 200 0 0 0 0 01234567 01234567 Days Days

L. pneumophila (33152) H. vermiformis (pfcii) L. pneumophila (33155) Acanthamoeba sp. (SDB4) H. vermiformis (SDB7) A.castellannii (30234) Acanthamoeba sp. (rfbii) Acanthamoeba sp. (rfbiih) H. vermiformis (50237)

CDControls for L. pnuemophila (ATCC 33152) Controls for L. pnuemophila (ATCC 33155) and MRD FLA at 37 °C and other FLA at 37 °C

16000 16000 Legionella.

16000 16000 -1 Legionella. L -1 14000 14000  L 14000 14000

 12000 12000 12000 12000 10000 10000 10000 10000 8000 8000 8000

8000   6000 6000 L L 6000 6000 -1 Legionella. -1 4000

Legionella. 4000 4000 4000 2000 2000 2000 2000 Trophozoites. -1 1600 1600 1600 Trophozoites. -1 1600 L L 1400  1400 1400  1400 1200 1200 1200 1200 1000 1000 1000 1000 800 800 800 800 600 600 600  600  L 400 L 400 400 -1 400 Trophozoites. -1 Trophozoites. 200 200 200 200 0 0 0 0 01234567 01234567 Days Days

Figure 7.5 Individual FLA and L. pneumophila controls over seven days. A: L. pneumophila (ATCC 33152) graphed with MRD isolated FLA at 22 °C. B: L. pneumophila (ATCC 33155) graphed with other FLA at 22 °C. C: L. pneumophila (ATCC 33152) graphed with MRD isolated FLA at 37 °C. D: L. pneumophila (ATCC 33155) graphed with other FLA at 37 °C. Line graphs for each control with error bars displaying ranges over the three replicates experiments with 15 observations per microorganism type per day.

The A. castellanii cultures inoculated with L. pneumophila (ATCC 33152) at 22 °C revealed that trophozoites were infected at a mean infection rate of 65 % after one day and this increased to 100 % by day three (Figure 7.7 - A). A similar infection pattern was observed with L. pneumophila (ATCC 33155) with no significant differences between the two (paired t-test p = 0.42). Infection rates started with a single visible vacuole of fluorescent L. pneumophila but by day five to six multiple vacuoles (5 - 10) could be observed (Figure 7.6 - B) or the cell appeared 204 Chapter 7. L. pneumophila infection to be overtaken by one large vacuole of infecting L. pneumophila (Figure 7.6 - C). The intracellular positions were these vacuoles were confirmed by depth scan using the CSLM. The numbers of extracellular L. pneumophila cells were significantly lower by day seven than the control (paired t-test p = 0.048) but this would be expected as the extracellular L. pneumophila were now intracellular infecting 100 % of the 250 trophzoites.uL-1 present. Hence, the mean number of intracellular L. pneumophila for this experiment was estimated at 2.5  104 Legionella.μL-1. Therefore the combined means of intracellular and extracellular L. pneumophila at day seven was 1.0  108 Legionella.μL-1 which was 7.7  103 times higher than the control (1.3  104 Legionella.μL-1). During the L. pneumophila (ATCC 33152) infections trophozoite numbers fluctuated and decreased to a mean of 250 trophozoites.μL-1 at day seven but this was found not to be significantly different from the control (paired t-test p = 0.65) over the whole seven days. The infection rates at 37 °C for A. castellanii infected with L. pneumophila (ATCC 33152) where significantly lower (paired t-test, p = 0.003) to the experiments at 22 °C. The infection rates did not rise above 10 % and at the end of the seven days no infected FLA could be observed (Figure 7.7 - C). The infection of A. castellanii showed a similar pattern for both L. pneumophila and any differences were found not to be significant (paired t-test, p = 0.63) (Figure 7.7 - D). The number of trophozoites also decreased over the seven days to zero but this was not significantly different (paired t-test, p = 0.21) from the control. Also the extracellular L. pneumophila numbers decreased to a mean of 1.8  103 Legionella.μL-1 at day seven but this was found not to be significantly different from the control (paired t-test, p = 0.24). It was observed that the small infecting ARB identified in the A. castellanii (ATCC 30234) were more active at 37 °C compared to 22 °C and appeared to be swarming the L. pneumophila present.  



 

Figure 7.6 Fluorescent microscopy of A. castellanii (ATCC 30234) infected with L. pneumophila (ATCC 33152) at 22 °C. A: Trophozoite at day three with the start of visible 205 Chapter 7. L. pneumophila infection FLA infection and a fluorescent L. pneumophila cell visible in the bottom left corner. B: Infected trophozoite at day six showing multiple L. pneumophila filled vacuoles. C: Infected trophozoite at day six with one large infected vacuole dominating. Overlays of fluorescent (FITC) and DIC images at 1000  magnification (CSLM, FV1000) with a 10 μm scale bar.

Control A. castellannii (ATCC 30234) and Control A. castellannii (ATCC 30234) and AB L. pneumophila (ATCC 33152) at 22 °C L. pneumophila (ATCC 33155) at 22 °C

100 16000 Legionella. 100 16000 Legionella. 14000 14000 90 90 12000 12000 80 80 10000 10000 8000

8000  70 70 L  6000

6000 -1 L

-1 60 60 4000 4000 2000 2000 50

50 Trophozoites. 1600 Trophozoites. 1600 1400 40 1400 40 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% 20 % trophozoites infected infected trophozoites % 20 600 600

400 

400  10

10 L L 200

200 -1 -1 0 0 0 0 01234567 01234567 Days Days

% infected # legionella¬ # trophzoites

C Control A. castellannii (ATCC 30234) and Control A. castellannii (ATCC 30234) and L. pneumophila (ATCC 33152) at 37 °C D  100 16000 Legionella. L. pneumophila (ATCC 33155) at 37 °C 100 16000 14000 Legionella. 90 14000 12000 90

12000 80 10000 80 10000

8000 

70 L 8000 70 -1 6000 6000  L

60 -1 4000 60 4000 50 2000 Trophozoites. 2000 50 Trophozoites. 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800

% trophozoites infected% 800 20 infected trophozoites % 20

600  600 L 400 

-1 400 10 10 L

200 200 -1 0 0 0 0 01234567 01234567 Days Days

Figure 7.7 Control A. castellanii (ATCC 30234) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments with 15 observations per variable per day. 7.4.2.1 Control H. vermiformis (ATCC 50237) H. vermiformis (ATCC 50237) infection rates with L. pneumophila (ATCC 33152) at 22 °C were significantly lower (pair t-test, p = 0.009) compared to the same experiment using A. castellanii (ATCC 30234). The maximum mean infection rate observed for both L. pneumophila experiments was 34 % on day seven (Figure 7.8 A-B). Generally, only one infected vacuole was observed in the infected trophozoites. Unusually, on day seven two cysts 206 Chapter 7. L. pneumophila infection were observed to contain L. pneumophila (ATCC33152) in the lining of the cyst wall. The trophozoite numbers decreased steadily from a mean inoculation concentration of 779 to 125 trophozoites.μL-1 with L. pneumophila (ATCC 33152) and 37 trophozoites.μL-1 with L. pneumophila (ATCC33155). However, this decrease was found not to be significantly different from the control over the seven days (pair t-test, p = 0.31). Extracellular L. pneumophila cell numbers peaked at day six for both experiments with 1.5  104 for L. pneumophila (ATCC 33152) and 1.6  104 for L. pneumophila (ATCC 33155) before decreasing sharply however this was found not to be significantly different from the control (pair t-test, p = 0.10). The drop in extracellular L. pneumophila numbers corresponded with a rise in the number of intracellular L. pneumophila with 30 -34 % of trophozoites infected. As the trophozoites generally only had a single vacuole of infecting L. pneumophila it was estimated that there would be approximately 10 cells per trophozoite. Therefore for L. pneumophila (ATCC 33152) intracellular (3.8  102 legionella.μL-1) combined with extracellular (3.9  103 legionella.μL-1) would gave a total of 7.7  105 legionella.μL-1 which was 57 times higher than the control (1.3  104 legionella.μL-1). No infected trophozoites were observed in the experiments at 37 °C but this was not significantly different (paired t-test, p = 0.20) compared to same experiments at 22 °C. The decline in trophozoite numbers to zero after one day was comparable to the control. Furthermore, the L. pneumophila numbers were found not to be significantly different (paired t-test, p = 0.19) from the control.

207 Chapter 7. L. pneumophila infection

A Control H. vermiformis (ATCC 50237) and B Control H. vermiformis (ATCC 50237) and L. pneumophila (ATCC 33152) at 22 °C L. pneumophila (ATCC 33155) at 22 °C Legionella. 100 16000 100 16000 Legionella. 14000 14000 90 12000 90

12000

80 10000 80 10000 8000 8000  70 6000  70 L L 6000 -1 4000 -1 60 60 4000 2000 Trophozoites.

2000 Trophozoites. 1600 50 50 1600 1400 40 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% % trophozoites infected infected trophozoites % 20 600 20 600   L 400 L 10 400

10 -1 200 -1 200 0 0 0 0 01234567 01234567 Days Days

% infected # legionella¬ # trophzoites

Control H. vermiformis (ATCC 50237) and Control H. vermiformis (ATCC 50237) and C L. pneumophila (ATCC 33152) at 37 °C D L. pneumophila (ATCC 33155) at 37 °C 100 16000 Legionella. 100 16000 Legionella. 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000

70 

70  L

6000 6000 L -1 60 4000 60 4000 -1 2000 2000 Trophozoites. 50 Trophozoites. 50 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected infected trophozoites % 20 infected trophozoites % 600 20 600  400  400 L 10 L 10 -1 200 -1 200 0 0 0 0 01234567 01234567 Days Days

Figure 7.8 Control H. vermiformis (ATCC 50237) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments 15 observations per variable per day. 7.4.3 Modified Robbins Device FLA isolates Three different isolates from the MRD were infected with L. pneumophila. The drinking water H. vermiformis (pfcii) had a low infection rate with the highest mean infection rate of 25 % over the seven days for both L. pneumophila (Figure 7.10 A-B) These infection rates were very similar to the control H. vermiformis (ATCC 50237) and any differences observed were found not to be significant (pair t-test, p = 0.74). When infections were observed in the H. vermiformis (pfcii) they generally consisted of one to two vacuoles only (Figure 7.9 B-C) which was a different pattern than that observed for H. vermiformis (ATCC 50237). At 22 °C the numbers of trophozoites decreased from the inoculation mean of 493 to 175 trophozoites.μL-1 208 Chapter 7. L. pneumophila infection with L. pneumophila (ATCC 33152) and 150 trophozoites.μL-1 with L. pneumophila (ATCC 33155) but this was found not to be significant (pair t-test, p = 0.36). Extracellular L. pneumophila mean concentrations peaked at day four or five but this was not significantly different from the control (pair t-test, p = 0.09). At day six for L. pneumophila (ATCC 33152) it was observed that 25 % of the 200 trophozoites.μL-1 present were infected with one vacuole containing approximately 10 cells. Therefore intracellular L. pneumophila (5  102 Legionella.μL-1) combined with extracellular L. pneumophila (8.8  103 Legionella.μL-1) gave a total of 1.4  106 legionella.μL-1 which was 108 times higher than the control (1.3  104 Legionella.μL-1). No trophozoites were observed to be infected in the experiments conducted at 37 °C which is similar to both the control H. vermiformis (ATCC 50237) and the experiments conducted at 22 °C with any difference observed found to be not significant (pair t-test, p = 0.15) (Figure 7.10 A-B). Again the numbers of trophozoites quickly decreased to zero after two days and the L. pneumophila concentrations were not significantly different (pair t-test, p = 0.70) from the control.  



 

Figure 7.9 Fluorescent microscopy of MRD drinking water H. vermiformis (pfcii) infected with L. pneumophila (ATCC33152) at 22 °C. A: Trophozoite at day seven without visible FLA infection. B: Infected trophzoite at day seven showing multiple L. pneumophila filled vacuoles. C: Infected trophozoite at day seven with one large infected vacuole. Overlays of fluorescent (FITC) and DIC images at 1000  magnification (CSLM, FV1000) with a 10 μm scale bar.

209 Chapter 7. L. pneumophila infection

MRD drinking water H. vermiformis (pfcii) and  MRD drinking water H. vermiformis (pfcii) and L. pneumophila (ATCC 33152) at 22 °C  AB100 16000 L. pneumophila (ATCC 33155) at 22 °C Legionella.

# trophzoites 100 16000 Legionella. 14000 90 14000 12000 90

12000 80 10000 80 10000 8000 8000

70  70  L 6000 6000 L -1 60 -1 4000 60 4000 2000 2000

50 Trophozoites. 1600 50 1600 Trophozoites. 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected infected trophozoites %

20 trophozoites infected% 600 20 600 

400 400  10 L

10 L -1

200 200 -1 0 0 0 0 01234567 01234567 Days Day

% infected # legionella¬ # trophzoites

MRD drinking water H. vermiformis (pfcii) and MRD drinking water H. vermiformis (pfcii) and C L. pneumophila (ATCC 33152) at 37 °C D L. pneumophila (ATCC 33155) at 37 °C

100 16000 Legionella.

100 16000 Legionella. 14000 14000 90 90 12000

12000

80 10000 80 10000 8000 70  8000 70  L

6000 L

-1 6000 -1 60 4000 60 4000 2000 Trophozoites.

Trophozoites. 2000 50 50 1600 1600 1400 40 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites% infected

20 infected trophozoites % 600 20 600  400  L 10 L 10 400 -1 200 -1 200 0 0 0 0 01234567 01234567 Days Days

Figure 7.10 MDR drinking water H .vermiformis (pfcii) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments 15 observations per variable per day. The recycled water Acanthamoeba sp. (rfbii) mean infection rates at 22 °C after one day were 100 % with L. pneumophila (ATCC 33152) and 76 % with L. pneumophila (ATCC 33155) (Figure 7.12 A-B). The infection rates for the two L. pneumophila were found not to be significantly different (paired t-test, p = 0.31). However, these infection rates were found to be significantly different (paired t-test, p < 0.0001) from the H. vermiformis (pfcii) infection rates but similar (paired t-test, p = 0.38) to the control A. castellannii (ATCC 30234). For both experiments at 22 °C trophozoite numbers decreased over the seven days. Extracellular L. pneumophila numbers peaked at day five for both experiments. At day seven L. pneumophila (ATCC 33152) was observed to heavily infect 100 % of the 187 trophzoites.μL-1 with about 10

210 Chapter 7. L. pneumophila infection vacuoles per trophozoite (Figure 7.11), which gave an intracellular concentration of 1.9  104 Legionella.μL-1. Extracellular L. pneumophila (6.3  103 Legionella.μL-1) combined with intracullar (1.9  104 Legionella.μL-1) gave an estimated total concentration of 8.2  107 Legionella.μL-1 which was 6.3  103 times higher than the control (1.3  104 Legionella.μL-1). The infection rates at 37 °C were significantly different (paired t-test, p = 0.0011) to those observed at 22 °C. The highest mean infection rate observed was 16.5 % in the first 3 days after which time no infected FLA were observed (Figure 7.12 C-D). After day three a small number (one or two per 15 FLA) of fluorescent L. pneumophila containing vesicles (3 - 5 μm in diameter) were observed extracellular to the FLA in the experiment. Trophozoites concentration decreased to zero within the first three days. L. pneumophila mean concentrations decreased to 3.1  103 Legionella.μL-1 (ATCC 33152) and 3.0  103 Legionella.μL-1 (ATCC 33155) and this was not significantly different (paired t-test, p = 0.06) from the L. pneumophila only control.  

 



Figure 7.11 Fluorescent microscopy with depth scans of MRD recycled water Acanthamoeba sp. (rfbii) infected with L. pneumophila (ATCC 33152) at 22 °C. A: trophozoite at day three with visible FLA infection. B: depth image for horizontal plane. C: depth image for vertical plane. Overlays of fluorescent (FITC) and DIC images at 1000  magnification (CSLM, FV1000) with a 10 μm scale bar.

211 Chapter 7. L. pneumophila infection

MRD recycled water Acanthamoeba sp. (rfbii) and MRD recycled water Acanthamoeba sp. (rfbii) and AB L. pneumophila (ATCC 33152) at 22 °C L. pneumophila (ATCC 33155) at 22 °C 100 16000 Legionella.

100 16000 Legionella. 14000 14000 90 90 12000

12000 80 10000 80 10000 8000 8000 70  70 6000  L 6000 L -1 60 4000 -1 60 4000 2000 50 Trophozoites. 50 2000 Trophozoites. 1600 1600 1400 40 1400 40 1200 1200 30 1000 30 1000 800

800 infected trophozoites % % trophozoites infected% 20 20 600 600 

 400 400 10 L

10 L 200 -1 200 -1 0 0 0 0 01234567 01234567 Days Days

% infected # legionella¬ # trophzoites

MRD recycled water Acanthamoeba sp. (rfbii) and MRD recycled water Acanthamoeba sp. (rfbii) and C L. pneumophila (ATCC 33152) at 37 °C D L. pneumophila (ATCC 33155) at 37 °C Legionella. 100 16000 100 16000 Legionella. 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000  70 70  L 6000 6000 L -1 -1 60 4000 60 4000

2000 Trophozoites. 2000 50 50 Trophozoites. 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% 20 600 infected trophozoites % 20 600   400 L 400 10 10 L -1 200 200 -1 0 0 0 0 01234567 01234567 Days Days

Figure 7.12 MRD recycled water Acanthamoeba sp. (rfbii) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments 15 observations per variable per day.  



Figure 7.13 Fluorescent microscopy of MRD recycled water Acanthamoeba sp. (rfbiih) infected with L. pneumophila (ATCC 33152) at 22 °C. A and B: Trophozoite at day three with a single vacuole of L. pneumophila. Overlays of fluorescent (FITC) and DIC images at 1000  magnification (CSLM, FV1000) with a 10 μm scale bar. 212 Chapter 7. L. pneumophila infection The second recycled water Acanthamoeba sp. isolate (rfbiih) was infected quickly with 100 % of trophozoites infected by day two or three at 22 °C (Figure 7. 14 A-B). This pattern of infection was similar to both Acanthamoeba sp. (rfbii) and the control A. castellannii (ATCC30234) and differences observed between them were found not to be significant (paired t-test, p = 0.84 and p = 0.32 respectively). All cysts were heavily infected by day seven with three to four large vacuoles dominating the cells. The intracellular L. pneumophila (ATCC33152) was estimated at 2  104 Legionella.μL-1 (100 % infection of 200 trophzoites.μL- 1) which combined with extracellular counts (4.1  103 Legionella.μL-1) gives a total of 6.1  107 Legionella.μL-1 which is 4.7  103 times higher than the control (1.3  104 Legionella.μL-1). At 37 °C Acanthamoeba sp. (rfbiih) trophzoites were infected at a maximum rate of 70 % at day three but then decreased with no infected trophozoites observed after this time (Figure 7. 14 C-D). The infection rates were significantly lower (paired t-test, p = 0.05) than for the same experiments at 22 °C but comparable to Acanthamoeba sp. (rfbii) at 37 °C with any differences not being significant (paired t-test, p = 0.39). The higher infection rate is most likely linked to the greater number of trophozoites available to infect. After day two a small number (1 or 2 per 15 FLA) of fluorescent L. pneumophila containing vesicles (3 - 5 μm in diameter) were observed in the experiment which is again similar to Acanthamoeba sp. (rfbii). Trophozoite numbers decreased from an inoculation concentration of 559 trophozoites.μL-1 to zero at day six with A. pneumophila (ATCC33152) and day seven with L. pneumophila (ATCC33155). This decrease was significantly (paired t-test, p = 0.01) less rapid than observed for the other Acanthamoeba sp. (rfbii) in the same experiment. Also, in the FLA only controls Acanthamoeba sp. (rfbiih) trophozoite numbers decreased at a significantly slower rate (paired t-test, p = 0.02) than Acanthamoeba sp. (rfbii).

213 Chapter 7. L. pneumophila infection

MRD recycled water Acanthamoeba sp. (rfbiih) and MRD recycled water Acanthamoeba sp. (rfbiih) and AB L. pneumophila (ATCC 33152) at 22 °C L. pneumophila (ATCC 33155) at 22 °C

100 16000 Legionella.

100 16000 Legionella. 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000  70 70  L 6000 6000 L -1 60 4000 60 4000 -1 2000 2000 Trophozoites.

50 50 Trophozoites. 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% 20 600 trophozoites infected% 20 600  400 400  L 10 10 L -1 200 200 -1 0 0 0 0 01234567 01234567 Days Days

% infected # legionella¬ # trophzoites

MRD recycled water Acanthamoeba sp. (rfbiih) and MRD recycled water Acanthamoeba sp. (rfbiih) and L. pneumophila (ATCC 33152) at 37 °C L. pneumophila (ATCC 33155) at 37 °C Legionella.

C 100 16000 Legionella. D 100 16000 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000  70  70 L 6000 L 6000 -1 -1 60 4000 60 4000 2000 2000 Trophozoites.

50 50 Trophozoites. 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected infected trophozoites % % trophozoites infected infected trophozoites % 20 600 20 600  400 400  L

10 10 L -1

200 200 -1 0 0 0 0 01234567 01234567 Days Days

Figure 7. 14 MRD recycled water Acanthamoeba sp. (rfbiih) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments (n=15) 7.4.4 Annular reactor FLA isolates The Acanthamoeba sp. (SDB4) isolated from the annular reactors were infected quickly with 100 % of trophozoites infected after three days at 22 °C (Figure 7.15 A-B). The pattern of infection matched all the other Acanthamoeba spp. observed and there were no significant differences (paired t-test, p = 0.30) between Acanthamoeba sp. (SDB4) and the recycled water isolate Acanthamoeba sp. (rfbii) for this experiment. After day four a small number (one or two per 15 FLA) of fluorescent L. pneumophila containing vesicles (3 - 5 μm in diameter) were observed in the experiment which is similar to the two MRD recycled water isolates Acanthamoeba sp. (rfbii and rfbiih) except they were observed at a higher temperature (37 °C). 214 Chapter 7. L. pneumophila infection At day seven extracellular L. pneumophila concentrations (7  103 Legionella.μL-1) were combined with intracellular concentrations (100 % infection of 250 trophozoites.μL-1 equals 2.5  104 Legionella.μL-1) to give a total concentration of 9.5  107 Legionella.μL-1 which was 7.3  103 times higher than the control (1.3  104 Legionella.μL-1). Low number of inoculating trophozoites (32 trophozoites.μL-1) may be responsible for the lack of infection observations in the first few days. In the experiments at 37 °C there were no infections of Acanthamoeba sp. (SDB4) which is likely due to the negligible numbers of trophzoites present (Figure 7.15 C-D). Although there were no infection observed at 37 °C this was not quite significantly different (paired t-test, p = 0.07) to the same experiment at 22 °C over the full seven days.

Annular reactor Acanthamoeba sp. (SDB4) and Annular reactor Acanthamoeba sp. (SDB4) and AB L. pneumophila (ATCC 33152) at 22 °C L. pneumophila (ATCC 33155) at 22 °C

100 16000 Legionella. 100 16000 Legionella. 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000  70  70 L 6000 L 6000 -1 60 4000 -1 60 4000 2000 2000 Trophozoites. 50 Trophozoites. 50 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% % trophozoites infected infected trophozoites % 20 600 20 600  400  400 L 10 L 10 -1 200 -1 200 0 0 0 0 01234567 01234567 Days Days

% infected # legionella¬ # trophzoites

Annular reactor Acanthamoeba sp. (SDB4) and Annular reactor Acanthamoeba sp. (SDB4) and C L. pneumophila (ATCC 33152) at 37 °C D L. pneumophila (ATCC 33155) at 37 °C

100 16000 100 16000 Legionella. Legionella. 14000 14000 90 90 12000 12000

80 10000 80 10000 8000 8000 70 70   6000 6000 L L -1

60 4000 -1 60 4000 2000 2000 Trophozoites. 50 Trophozoites. 50 1600 1600 40 1400 40 1400 1200 1200 30 1000 30 1000 800 800 % trophozoites infected% 20 600 trophozoites infected% 20 600  400  400 L 10 L 10 -1 200 -1 200 0 0 0 0 01234567 01234567 Days Days

Figure 7.15 Annular reactor Acanthamoeba sp. (SDB4) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments with a total of 15 observations per variable per day. 215 Chapter 7. L. pneumophila infection 7.4.5 Garden hoses FLA isolates Garden hose H. vermiformis (SDB7) isolates were infected at a maximum mean of 14 % for both L. pneumophila experiments at 22 °C (Figure 7.16 A-B). This pattern of infection was consistent with all the other H. vermiformis infections experiments and there was no significant difference (paired t-test, p = 0.48) with the MRD drinking water H. vermiformis (pfcii) infection rates nor the control H. vermiformis (ATCC 50237) (paired t-test, p = 0.17). As with the other H. vermiformis experiments only one infected vacuole was generally observed in the trophozoites. The L. pnuemophila numbers both peaked at day six but was significantly (paired t- test, p = 0.04) higher for L. pneumophila (ATCC 33155) (1.6  104 Legionella.μL-1) compared to L. pneumophila (ATCC 33152) (9.3  103 Legionella.μL-1). The numbers of extracellular L. pneumophila (ATCC 33152) in the experiment were not significantly different (paired t-test, p = 0.35) from the corresponding L. pneumophila control, however, the total L. pneumophila concentration were higher. At day seven extracellular L. pneumophila (ATCC 33152) concentrations (5  103 Legionella.μL-1) were combined with intracellular concentrations (13 % infection of 200 trophozoites.μL-1 which gave 2.6  102 Legionella.μL-1) to give a total concentration of 7.6  105 Legionella.μL-1 which was 58 times higher than the control (1.3  104 Legionella.μL-1). At 37 °C few infected trophozoites were observed over the seven days for both L. pneumophila (Figure 7.16 C-D). This is consistent with the trophozoites numbers falling to zero by day three after an initial inoculation mean of 367 trophozoites.μL-1. Additionally this infection rate was similar to that observed for MRD drinking water H. vermiformis (pfcii) and any differences observed were found not to be significant for both L. pneumophila (paired t-test, p = 0.36).

216 Chapter 7. L. pneumophila infection

Water hose H. vermiformis (SDB7) and AB L. pneumophila (ATCC 33152) at 22 °C Water hose H. vermiformis (SDB7) and

100 16000 Legionella. L. pneumophila (ATCC 33155) at 22 C 14000 ° 90 100 16000 Legionella. 12000 control # Legionella 33155

14000 90 80 10000 12000

8000 80 10000

70  6000 L 8000 -1 70  60 4000 6000 L -1 2000 60 4000 50 Trophozoites. 1600 2000

50 Trophozoites. 40 1400 1600 1200 40 1400 30 1000 1200 800 30 1000 % trophozoites infected infected trophozoites % 20 600 800 % trophozoites infected% 20 400  600 10 L 400  -1

200 10 L 0 0 200 -1 01234567 0 0 01234567 Days Days

% infected # legionella¬ # trophzoites

D C Water hose H. vermiformis (SDB7) and Water hose H. vermiformis (SDB7) and   L. pneumophila (ATCC 33155) at 37 °C L. pneumophila (ATCC 33152) at 37 °C 100 16000 100 16000 Legionella.

% infected pfcii 37CLegionella. 152 % infected pfcii 37C 152 14000 14000 90 90 12000 12000 80 80 10000 10000 8000 8000 70

70  6000  6000 L L -1

60 -1 60 4000 4000 2000 2000 50 Trophozoites.

50 Trophozoites. 1600 1600 1400 40 1400 40 1200 1200 30 1000 30 1000 800 800 % trophozoites infected%

% trophozoites infected% 20 20 600 600 400   400 10 L

10 L 200 -1

200 -1 0 0 0 0 01234567 01234567 Days Days

Figure 7.16 Garden hose H. vermiformis (SDB7) infected with L. pneumophila over seven days. A: With L. pneumophila (ATCC 33152) at 22 °C. B: With L. pneumophila (ATCC 33155) at 22 °C. C: With L. pneumophila (ATCC 33152) at 37 °C. D: With L. pneumophila (ATCC 33155) at 37 °C. Line graphs for each treatment with error bars displaying ranges over the three replicates experiments 15 observations per variable per day. 7.5 DISCUSSION

7.5.1 Control FLA infected with ARB ATCC can supply FLA cultures free of other micro-organisms in axenic form. Axenic cultures are used extensively in research on ARM (Greub and Raoult, 2004) because of the absence of other contaminating micro-organisms. However, the axenically supplied A. castellanii (ATCC 30234) and A. polyphaga (ATCC 30461) were confirmed microscopically and via molecular methods to contain infecting ARB. The bacteria were confirmed as Sphingomonas rhizogenes and an uncultured bacteria. ATCC FLA cultures have been found to

217 Chapter 7. L. pneumophila infection contain undetected ARM before screening nine axenically grown ATCC FLA two cultures (22 %), Acanthamoeba polyphaga ATCC 30871 (axenic culture) and Acanthamoeba castellanii ATCC 30868 (supplied with Escherichia coli as a food source), were identified by microscopy to contain ARB with gram negative rod morphology (Fritsche et al., 1993). Similarly, axenic A. polyphaga HN-3 ATCC 30173 was also found to contain ARB (Hall and Voelz, 1985). Given the frequency of detection of previously unidentified ARB in ATCC FLA cultures all axenic FLA cultures should be screened for potential infection. The ARB present within the ATCC FLA were visibly more active at 37 °C compared to 22 °C (data not shown). Greater activity and growth at higher temperatures is an indicator of pathogenicity (De Jonckheere, 1980). Hence, it is likely that the ARB identified could be human pathogens based on their ability to readily infect FLA and show increased activity at 37 °C. Attempts should be made to sequence more of the infecting ARB genes and evaluate their potential human pathogenicity. The ARB may have been inhibitory to L. pneumophila in the experiments as at 37 °C with A. castellannii (ATCC 30234) as L. pneumophila numbers decreased to a mean of 1.8  103 Legionella.μL-1 at day seven which was seven times lower than the control (1.3  103 Legionella.μL-1). The ATCC FLA ARB were observed to swarm around the L. pneumophila cells and may have been predatory but this was not directly quantified. FLA are known to contain multiple ARM (Heinz et al., 2007) and the presence of the other ARB at 22 °C did not inhibit infection by L. pneumophila of A. castellannii (ATCC 30234).

7.5.2 Infection rates

7.5.2.1 Acanthamoeba spp. verses H. vermiformis There was a distinct difference in the infection rates for Acanthamoeba spp. and H. vermiformis. For all of the Acanthamoeba spp. (ATCC 30234, rfbii, rfbiih, and SDB4) at 22 °C by no later than day three 100 % of the trophozoites present were infected for both L. pneumophila. This is in stark contrast to H. vermiformis (ATCC 50237, pfcii and SDB7) where the maximum infection rate over the entire seven days was only 34 % (ATCC 50237 with L. pneumophila ATCC 33152). This difference could be in due to the smaller size of H. vermiformis (< 5 μm in width) compared to the Acanthamoba spp. ( > 10 μm in width). This size difference could mean that the L. pneumophila cells are less of an optimal food choice for the smaller H. vermiformis. It also possible that the H. vermiformis have some innate characteristic which prevents them from being infected by L. pneumophila. Other research has shown that H. vermiformis increases the growth and survival of L. pneumophila in drinking water (Fields et al., 1989; Wadowsky et al., 1991) and drinking water biofilm (Donlan et al., 2005). Using fluorescent in situ hybridisation (FISH) L. pneumophila was shown to infect 26 % 218 Chapter 7. L. pneumophila infection of H. vermiformis trophozoites in an artificial drinking water biofilm after 10 days (Kuiper et al., 2004). These infection rates are similar to those observed over seven days in the experiments conducted. Temperature was the other main factor in the infection rates observed. For every FLA at 37 °C there were significantly less trophozoites infected compared to the same experiment at 22 °C. For all the four Acanthamoeba spp. (ATCC 30234, rfbii, rfbiih, and SDB4) the highest infection rates observed at 37 °C were 71 % (rfbiih with ATCC 33152) compared to 100 % at 22 °C. For the three H. vermiformis (ATCC 50237, pfcii and SDB7) the highest infection rates observed at 37 °C were 13 % (SDB7 with ATCC 33155) compared to 34 % (ATCC 50237) at 22 °C. The lower infection rates may be due to the trophozoites not feeding due to the increased temperature compared to initial incubation. Of interest the Acanthamoeba sp. (rfbiih) that did have the highest L. pneumophila infection at 37 °C was isolated from recycled water on NNA with E. coli at 37 °C. Hence although the FLA had been incubated at 22 °C prior to the experiment it may have had an innate temperature tolerance and fed at 37 °C. While the other environmental Acanthamoeba sp. were isolated at 22 °C and the control ATCC 30234 grew at 30 °C hence they may not have been able to survive effectively at 37 °C. Another reason for the reduced FLA infection rates is likely linked to the absence of FLA trophozoites available to infect after the first few days (Section 7.5.4) which has been reported elsewhere to account for reduced infection rates for H. vermiformis (Kuiper et al., 2004).

7.5.2.2 Legionella filled vacuoles As FLA were infected by the L. pneumophila they were seen in distinct single vacuoles that ranged in size from 2 -3 μm3 initially. As the infection progressed more L. pneumophila filled vacuoles (between 5-10) were observed and some vacuoles become very large in diameter (> 5 μm) and started to dominate the trophozoite. This is consistent with other research where L. pneumophila was shown to infect a range of FLA and the infections were first contained within vacuoles (10-15 in number) (Rowbotham, 1980) and then in some trophozoites spread resulting in heavily infected trophozoites with up to 100 CFU of infecting L. pneumophila (Kuiper et al., 2004).

7.5.2.3 Legionella filled vesicles Vesicles were observed rarely for all the environmental Acanthameoba spp. (rfbii, rfbiih, and SDB4) at both 22 and 37 °C. However, no vesicles were observed for the H. vermiformis isolates but is consistent with other research. Acanthamoeba sp. was observed producing L. pneumophila filled vesicles occasionally during experiments (Rowbotham, 1980). Vesicles are often 2.1 - 6.4 μm in size when expelled by Acanthamoeba spp. and were confirmed as

219 Chapter 7. L. pneumophila infection containing live L. pneumophila which protects them from environmental stresses (Berk et al., 1998).

7.5.2.4 Legionella infected cysts In all the experimental observations only cysts of H. vermiformis (ATCC 50237) at day seven were observed to contain L. pneumophila (ATCC 33152) infecting the lining of the cyst wall. Research has shown L. pneumophila infected A. polyphaga trophozoites when induced to form cysts using pH encystment media still contain viable L. pneumophila in the cysts (Kilvington and Price, 1990). Furthermore, A. castellannii produced cysts containing L. pneumophila after 16 days of incubation (Storey et al., 2004). Therefore incubation periods longer than seven days appear to be required to produce more infected cysts.

7.5.3 L. pneumophila numbers Extracellular L. pneumophila nearly doubled in number over the seven days in the control. L. pneumophila has been reported not to grow readily in low nutrient environments such as tap water (Wadowsky et al., 1991). Hence, the increase observed may have been due to residual nutrients from the growth media being carried through to the experiment. To reduce this effect a duplicate washing steps in the cell staining process should be introduced to ensure complete removal of the high nutrient media after centrifugation. The intracellular concentrations of L. pneumophila were estimated based on the size and number of the infecting vacuoles at 10 cells per small vacuole. This estimation may be an underestimate compared with other research were the maximum number of L. pneumophila cells in an Acanthamoeba trophzoite that was entirely infected was estimated at 1000 cells and an expelled vesicle was estimated to contain 50 cells (Rowbotham, 1980). Total L. pneumophila cell concentrations consisted of both intracellular and extracellular counts at day seven of the experiments. Infection with Acanthamoeba sp. increased the number of L. pneumophila (ATCC 33152) cells at 22 °C compared to the L. pneumophila only controls. The specific increases of L. pneumophila cells were 7.7  103 times with A. castellannii (ATCC 30234), 6.3  103 times with Acanthamoeba sp. (rfbii), 4.7  103 with Acanthamoeba sp. (rfbiih) and 7.3  103 with Acanthamoeba sp. (SDB4). Across all the Acanthamoeba spp. the number of L. pneumophila increased by greater than three orders of magnitude. Compared to other research these increases are comparable with environment isolates of L. pneumophila being found to increase by up to 1.2  103 times in the presence of Acanthamoeba lenticulata (Molmeret et al., 2001). L. pneumophila increase in the presence of H. vermiformis at 22 °C was consistently two orders of magnitude lower than that observed for Acanthamoeba spp. Mean increase in 220 Chapter 7. L. pneumophila infection L. pneumophila numbers compared to the control were 57 times with H. vermiformis (ATCC50237), 108 times with H. vermiformis (pfcii) and 58 times with H. vermiformis (SDB7). The lower increase of L. pneumophila are consistent with other research where increases due to H. vermiformis were found to average 16 times greater over seven different Legionella spp. (Wadowsky et al., 1991). After seven days at 37 °C there was no significant increase in the number of extracellular L. pneumophila present in any of the experiments compared to the L. pneumophila only control and there was no intracellular L. pneumophila observed. The numbers of L. pneumophila approximately doubled between inoculation (7  103 Legionella.μL-1) and day seven (1.1  104 and 1.3  104 Legionella.μL-1 for ATCC33152 and ATCC33155 respectively). By day seven L. pneumophila numbers actually reduced to below 5  103 Legionella.μL-1 for three Acanthamoeba spp. (ATCC30234rfbii and rfbiih) and one H. vermiformis (ATCC30237). This number of L. pneumophila was lower than the actual inoculating concentration and could be due to predation by ARB (Section 7.5.1). Alternatively, there may have been some inhibitory response from the encysting trophozoites.

7.5.4 Trophozoite numbers In the FLA only controls at 22 °C trophozoite numbers remained relatively constant throughout the seven days (Figure 7.5). However, within the experiments with L. pneumophila the number of trophozoites all decreased from the inoculating concentrations except for Acanthamoeba sp. (SDB4) which had only very low concentrations of trophozoites present at inoculation (33 trophozoites.μL-1). The gradual decline in the number of trophozoites is likely due to two factors: trophozoite lysis by infecting L. pneumophila and encystment to avoid infection. L. pneumophila are known to lyse heavily infected trophozoites releasing a large number of L. pneumophila to infect the other trophozoites (Rowbotham, 1980). Encysting is a response by trophozoites to contact with L. pnuemophila which is thought to protect them from infection (Rowbotham, 1980; Kuiper et al., 2004). Temperature appears to play a role in the number of trophozoites present as at 37 °C a different pattern was observed. At 37 °C the number of control trophozoites decreased to zero for all of the FLA within the seven days. This is likely part of the encysting stress response to the higher temperatures compared to culturing temperature. The only exception was for Acanthamoeba sp. (rfbiih) where trophozoite numbers decreased at a significantly slower rate (paired t-test, p = 0.02) compared to Acanthamoeba sp. (rfbii). A phylogenetic tree of the partial 18S rRNA sequences from Acanthamoeba sp. rfbii and rfbiih reveal that they are very closely related (Figure 4.13). They were however isolated at different temperatures: 22 °C for Acanthamoeba sp. (rfbii) and 37 °C

221 Chapter 7. L. pneumophila infection for Acanthamoeba sp. (rfbiih). This may explain the differences in their temperature tolerances. A higher temperature tolerance also meant that Acanthamoeba sp. (rfbiih) was able to maintain the highest number of infected trophozoites at 37 °C (71 %) (Figure 7. 14 -C). Temperature tolerance appears to be critical a factor in maintaining feeding trophozoite population that L. pneumophila can infect and replicate within. The steady reduction in trophozoite numbers and consequential increase in cysts across all FLA when in the presence of L. pneumophila highlights the dynamic relationship that exist between ARM and FLA populations and the influence that temperature can have.

7.5.5 Conclusions Environmental Acanthamoeba sp. isolates were infected at a greater rate and increased L. pneumophila concentrations by 4.7  103 to 7.3  103 times over seven days at 22 °C. Environmental H. vermiformis increase L. pnuemophila concentrations by 58 - 108 times which is two fold lower than Acanthamoeba sp. under the same conditions. At higher temperatures of 37 °C the number of infections and L. pnuemophila decreased due to absence of trophozoites to infect in the last few days of the experiment. This indicated that spikes in temperatures may actually be slightly prohibitive to L. pneumophila proliferation in FLA. However, if FLA are present which are tolerant to the higher temperatures infection, proliferation is still possible. The FLA isolated from the drinking and recycled water systems and applications are all able to be infected by, and increase the number of L. pneumophila to levels that could cause human health concerns. Overall this experiment described the infection rates of L. pneumophila for FLA isolates and the corresponding variations in L. pneumophila, trophozoite and cyst densities. It would be advantageous to extend the experiment beyond seven days to be able to track more infected cysts and this is possible using this experimental design as the stain (Vybrant® CFDA SE Cell Tracer Kit) has been reported to be traceable for up to eight weeks (Weston and Parish, 1990). Future experiments should focus on quantification of virulence up-regulation in the L. pneumophila during FLAL infection. This could be achieved using qPCR targeting virulence genes such as the mip gene RNA (Wellinghausen et al., 2001). Alternatively, infection rates with human cell lines would be a useful tool to determine any increases in infectivity (Cirillo et al., 1994; Neumeister et al., 2000). Additionally, introducing the infected FLA into an annular reactor and tracking them would give further knowledge about the growth. Estimating the impact of FLA interaction on the virulence of L. pneumophila is a critical element needed to design realistic QMRA models for Legionella infection.

222 Chapter 7. L. pneumophila infection

7.6 REFERENCES 1. Albert-Weissenberger, C., C. Cazalet and C. Buchrieser. 2007. Legionella pneumophila - a human pathogen that co-evolved with fresh water protozoa. Cellular and Molecular Life Sciences. 64(4): 432-448. 2. Berk, S., R. Ting, G. Turner and R. Ashburn. 1998. Production of respirable vesicles containing live Legionella pneumophila cells by two Acanthamoeba spp. Applied and Environmental Microbiology. 64(1): 279-286. 3. Berry, D., M. Horn, C. Xi and L. Raskin. 2010. Mycobacterium avium infections of Acanthamoeba strains: host strain variability, grazing acquired infections, and altered dynamics of inactivation with monochloramine. Applied and Environmental Microbiology. 76(19): 6685-6688. 4. Birtles, R. J., T. J. Rowbotham, C. Storey, T. J. Marrie and D. Raoult. 1997. Chlamydia- like obligate parasite of free-living amoebae. The Lancet. 349(9056): 925-926. 5. Cirillo, J. D., S. Falkow and L. S. Tompkins. 1994. Growth of Legionella pneumophila in Acanthamoeba castellanii enhances invasion. Infection and Immunity. 62(8): 3254-3261. 6. Corsaro, D., V. Feroldi, G. Saucedo, F. Ribas, J.-F. Loret and G. Greub. 2009. Novel Chlamydiales strains isolated from a water treatment plant. Environmental Microbiology. 11(1): 188-200. 7. De Jonckheere, J. F. 1980. Growth characteristics, cytopathic effect in cell culture, and virulence in mice of 36 type strains belonging to 19 different Acanthamoeba spp. Applied and Environmental Microbiology. 39(4): 681-685. 8. Donlan, R., T. Forster, R. Murga, E. Brown, C. Lucas, J. Carpenter and B. Fields. 2005. Legionella pneumophila associated with the protozoan Hartmannella vermiformis in a model multi-species biofilm has reduced susceptibility to disinfectants. Biofouling. 21(1): 1-7. 9. Fields, B. S., G. N. Sanden, J. M. Barbaree, W. E. Morrill, R. M. Wadowsky, E. H. White and J. C. Feeley. 1989. Intracellular multiplication of Legionella pneumophila in amoebae isolated from hospital hot water tanks. Current Microbiology. 18(2): 131-137. 10. Fritsche, T. R., R. K. Gautom, S. Seyedirashti, D. L. Bergeron and T. D. Lindquist. 1993. Occurrence of bacterial endosymbionts in Acanthamoeba spp. isolated from corneal and environmental specimens and contact lenses. Journal of Clinical Microbiology. 31(5): 1122-1126. 11. Greub, G., B. La Scola and D. Raoult. 2003. Parachlamydia acanthamoeba is endosymbiotic or lytic for Acanthamoeba polyphaga depending on the incubation temperature. Annals of the New York Academy of Sciences. 990: 628-634. 12. Greub, G. and D. Raoult. 2004. Microorganisms resistant to free-living amoebae. Clin Microbiol Rev. 17: 413 - 433. 13. Greub, G. and D. Raoult. 2004. Microorganisms resistant to free-living amoebae. Clinical Microbiology Reviews. 17(2): 413-433. 14. Hall, J. and H. Voelz. 1985. Bacterial endosymbionts of Acanthamoeba sp. The Journal of Parasitology. 71(1): 89-95. 15. Heidtman, M., E. J. Chen, M.-Y. Moy and R. R. Isberg. 2009. Large-scale identification of Legionella pneumophila Dot/Icm substrates that modulate host cell vesicle trafficking pathways. Cellular Microbiology. 11(2): 230-248. 16. Heinz, E., Irina Kolarov, Christian K‰stner, Elena R. Toenshoff, Michael Wagner and Matthias Horn. 2007. An Acanthamoeba sp. containing two phylogenetically different bacterial endosymbionts. Environmental Microbiology. 9(6): 1604-1609.

223 Chapter 7. L. pneumophila infection 17. Hoffmann, R., R. Michel, E. N. Schmid and K.-D. Müller. 1998. Natural infection with microsporidian organisms (KW19) in Vannella spp. (Gymnamoebia) isolated from a domestic tap-water supply. Parasitology Research. 84(2): 164-166. 18. Isberg, R. R., T. J. O'Connor and M. Heidtman. 2009. The Legionella pneumophila replication vacuole: making a cosy niche inside host cells. Nature Reviews Microbiology. 7(1): 13-24. 19. Kilvington, S. and J. Price. 1990. Survival of Legionella pneumophila within cysts of Acanthamoeba polyphaga following chlorine exposure. Journal of Applied Microbiology. 68(5): 519-525. 20. Kuiper, M. W., B. A. Wullings, A. D. L. Akkermans, R. R. Beumer and D. van der Kooij. 2004. Intracellular proliferation of Legionella pneumophila in Hartmannella vermiformis in aquatic biofilms grown on plasticized polyvinyl chloride. Applied and Environmental Microbiology. 70(11): 6826-6833. 21. La Scola, B., S. Audic, C. Robert, L. Jungang, X. de Lamballerie, M. Drancourt, R. Birtles, J.-M. Claverie and D. Raoult. 2003. A giant virus in amoebae. Science. 299(5615): 2033. 22. La Scola, B., C. Desnues, I. Pagnier, C. Robert, L. Barrassi, G. Fournous, M. Merchat, M. Suzan-Monti, P. Forterre, E. V. Koonin and D. Raoult. 2008. The virophage as a unique parasite of the giant mimivirus. Nature. 455(7209): 4. 23. Lau, H. Y. and N. J. Ashbolt. 2009. The role of biofilms and protozoa in Legionella pathogenesis: implications for drinking water. Journal of Applied Microbiology. 207(2): 368-378. 24. Marolda, C., B. Hauroder, M. John, R. Michel and M. Valvano. 1999. Intracellular survival and saprophytic growth of isolates from the Burkholderia cepacia complex in free-living amoebae. Microbiology 45(7): 1509-1517. 25. Molmeret, M., S. Jarraud, J. Pierre Morin, P. Pernin, F. Forey, M. Reyrolle, F. Vandenesch, J. Etienne and P. Farge. 2001. Different growth rates in amoeba of genotypically related environmental and clinical Legionella pneumophila strains isolated from a thermal spa. Epidemiology and Infection. 126(2): 231-239. 26. Neumeister, B., G. Reiff, M. Faigle, K. Dietz, H. Northoff and F. Lang. 2000. Influence of Acanthamoeba castellanii on intracellular growth of different Legionella species in human monocytes. Applied and Environmental Microbiology. 66(3): 914-919. 27. Pagnier, I., D. Raoult and B. La Scola. 2008. Isolation and identification of amoeba- resisting bacteria from water in human environment by using an Acanthamoeba polyphaga co-culture procedure. Environmental Microbiology. 10(5): 1135-1144. 28. Pickup, Z. L., R. Pickup and J. D. Parry. 2007. Growth of Acanthamoeba castellanii and Hartmannella vermiformis on live, heat-killed and DTAF-stained bacterial prey. FEMS Microbiology Ecology. 61(2): 264-272. 29. Ray, K., B. Marteyn, P. J. Sansonetti and C. M. Tang. 2009. Life on the inside: the intracellular lifestyle of cytosolic bacteria. Nature Reviews Microbiology. 7(5): 333-340. 30. Rowbotham, T. 1980. Preliminary report on the pathogenicity of Legionella pneumophila for freshwater and soil amoebae. Journal of Clinical Pathology. 33: 1179-1183. 31. Scheid, P., L. Zöller, S. Pressmar, G. Richard and R. Michel. 2008. An extraordinary endocytobiont in Acanthamoeba sp. isolated from a patient with keratitis. Parasitology Research. 102(5): 945-950. 32. Snelling, W., J. Moore, J. McKenna, D. Lecky and J. Dooley. 2006. Bacterial-protozoa interactions; an update on the role these phenomena play towards human illness. Microbes and Infection. 8: 578-587. 33. Srikanth, S. and S. G. Berk. 1993. Stimulatory effect of cooling tower biocides on amoebae. Applied and Environmental Microbiology. 59(10): 3245-3249.

224 Chapter 7. L. pneumophila infection 34. Storey, M., J. Winiecka-Krusnell, N. Ashbolt and T. A. Stenström. 2004. The efficacy of heat and chlorine treatment against thermotolerant Acanthamoebae and Legionellae. Scandinavian Journal of Infectious Diseases. 36(9): 656-622. 35. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 36. Wadowsky, R. M., T. M. Wilson, N. J. Kapp, A. J. West, J. M. Kuchta, S. J. States, J. N. Dowling and R. B. Yee. 1991. Multiplication of Legionella spp. in tap water containing Hartmannella vermiformis. Applied and Environmental Microbiology. 57(7): 1950-1955. 37. Wang, Y., L. Claeys, D. van der Ha, W. Verstraete and N. Boon. 2010. Effects of chemically and electrochemically dosed chlorine on Escherichia coli and Legionella beliardensis assessed by flow cytometry. Applied Microbiology and Biotechnology. 87(1): 331-341. 38. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 39. Weston, C. A. and C. R. Parish. 1990. New fluorescent dyes for lymphocyte migration studies. Analysis by flow cytometry and fluorescence microscopy. Journal of Immunological Methods. 133(1): 87 - 97.

225

CHAPTER 8 HUMAN HEALTH RISK ASSESSMENT

8. TABLE OF CONTENTS

8.1 INTRODUCTION ______227 8.1.1 Legionella infection______227 8.1.2 Quantitative microbial risk assessment (QMRA) ______227 8.2 AIMS ______228 8.3 METHODS ______228 8.3.1 QMRA model ______228 8.3.2 Parameter estimation ______230 8.3.3 Test scenarios ______237 8.4 RESULTS ______238 8.5 DISCUSSION ______240 8.5.1 Risk of Legionella infection ______240 8.5.2 Reducing the risk of infection ______242 8.5.3 Conclusion and future research ______244

LIST OF TABLES

Table 8.1 Percentage of FLA as trophzoites______231 Table 8.2 Estimation of total trophozoites in water and biofilm ______231 Table 8.3 Percentage of trophozoites infected with pathogenic Legionella ______232 Table 8.4 Number of native Legionella bacteria ______235 Table 8.5 Test scenarios used in the risk assessment model calculations______238 Table 8.6 Risk assessment results for garden hose and shower scenarios ______239

Chapter 8. Risk assessment

8.1 INTRODUCTION 8.1.1 Human infections with Legionella Outbreaks and sporadic cases of human infections with Legionella from treated drinking water and applications of drinking water cause hundreds of cases and scores of deaths every year (Table 1.1, Section 1.1.1). Presently, Legionella cases account for 29 % of all reported drinking water disease outbreaks in the United States (Craun et al., 2010). Outbreaks of legionellosis represent the peaks of the disease within a community and there is likely to be a constant background level of infection existing below the detectable outbreak level (Frost et al., 1996). Determining what the likely level of Legionella infection is within the community is important as it allows for estimations as to what portion of community acquired pneumonia (CAP) is caused by Legionnaires disease and what portion of flu cases are actually Pontiac Fever. Conservative past estimates in Australia are that Legionella infection is responsible for 3 -5 % of all the CAP admissions to hospital annually (Broadbent, 1996). But this is likely to be an under- estimate due the lack of systematic testing for the pathogen (Wilson and Ferguson, 2005). In the absence of epidemiological data probabilistic models can be used to determine the probability of Legionella infection in a community after exposure via common activities such as garden hose use and showering.

8.1.2 Quantitative microbial risk assessment (QMRA) Quantitative microbial risk assessment (QMRA) is a tool to help quantify the risk of infection by a particular pathogen after given exposure events. QMRA involves five steps: hazard assessment, exposure assessment, dose response analysis, risk characterisation and risk management (Haas et al., 1999). This tool has been applied to determine the risk proposed by a number of microbial pathogens including Legionella (Storey et al., 2004; Armstrong and Haas, 2008; Schoen and Ashbolt, 2011). However, initial risk assessments over estimated the risk of infection using a maximum risk model for drinking water (Storey et al., 2004). Post this initial Legionella QMRA there has been further development and validation of the guinea-pig dose response model which appears to have increased the validity of the model but which neglects the role of FLA and biofilm (Armstrong and Haas, 2007). In both of these initial models it was recommended that low level sources of Legionella infection in the community needed to be explored further (Storey et al., 2004; Armstrong and Haas, 2008). Furthermore, a number of researchers in the field have recommended that FLA need to be considered in any risk assessment for Legionella and other drinking water pathogens (Breiman et al., 1990; Szewzyk et al., 2000; Storey et al., 2004; Bichai et al., 2008; Loret et al., 2008; Thomas et al., 2008; Loret

227 Chapter 8. Risk assessment and Greub, 2010; Thomas et al., 2010). Further, it has been recommended that FLA be proactively targeted for reduction in order to lower drinking water Legionella and other pathogen numbers (Srikanth and Berk, 1993; Critchley and Bentham, 2009; Loret and Greub, 2010).

8.2 AIMS The aim of this research was to improve the current QMRA models for Legionella infection to include FLA and then determine the probability of infection after exposure via garden hose use and showering scenarios. Based on the probability of infection determined recommendations to reduce the risk of infection were made.

8.3 METHODS 8.3.1 QMRA model

8.3.1.1 Total pathogenic Legionella in water Based on the results of the L. pneumophila uptake experiments conducted with isolates from the drinking water applications (Chapter 7) an equation (Equation 8.1) was developed to describe the number of total pathogenic Legionella (LpT ) in a given volume of water (1 L). Pathogenic Legionella can be present in water in two forms, either as intracellular Legionella infecting a FLA host (LpF) or as extracellular Legionella residing as free cells in the water (LpW) possibly in vesicles expelled from FLA (Berk et al., 1998). The total quantity of intracellular Legionella present in a given volume is dependent on how many FLA are infected with the bacteria (FLp).

LpT =F LpLpF +LpW (Equation 8.1) Where:

-1 LpT = total Legionella concentration (cells.L )

FLp = number of FLA (trophozoites or cysts) infected with pathogenic Legionella in water (trophzoites.L-1)

LpF = the number of intracellular pathogenic Legionella per infected FLA -1 LpW= the number of extracellular pathogenic Legionella in the water (cells.L )

8.3.1.2 Legionella in aerosols Legionella infection occurs when aerosols of the optimal size ( 7μm) are inhaled and reach the aveolar region of the human lung (Wilkes, 1999). Hence, the number of pathogenic Legionella in water that are aersolised and present in the air was determined (Equation 8.2) (Armstrong and Haas, 2007). The total number of pathogenic Legionella present in the water

(LpT) is multiplied by the partitioning coefficient (PC). The partitioning coefficient is unique to

228 Chapter 8. Risk assessment water application and is based on estimations of the formation of aerosols from the water in a given volume of air. Partitioning coefficients have been used to estimate the number of bacteria in volumes of air from both wastewater (Bauer et al., 2002) and irrigation spray (Teltsch et al., 1980).

LpAir = LpT PC (Equation 8.2) Where:

-3 Lpair= the number of pathogenic Legionella in a cubic meter of air (cells.m ) -1 LpT = total Legionella concentration (cells.L ) from Equation 8.1 PC = the partitioning coefficient for the volume of water aerosolised in a volume of air (L.m-3)

8.3.1.3 Exposure dose The total exposure dose of Legionella bacteria was calculated using standard inhalation dose calculations used by the US EPA (1997). The dose (d) is a function of the pathogenic

Legionella within aerosols in the air (LpAir) by the volume of that air is inhaled (IH) over a given exposure time (TEx). Finally the fraction of Legionella particles remaining in the lung (R) was applied (Equation 8.3).

. d = LpAir I H T ExR (Equation 8.3) Where: d = the dose of pathogenic Legionella received

-3 Lpair= the number of pathogenic Legionella in a cubic meter of air (cells.m ) from Equation 8.2 3 -1 IH = inhalation rate (m .h )

TEx= exposure time (h) R = fraction of Legionella aerosols retained in the lungs

8.3.1.4 Dose-response The dose-response model was estimated based on a animal model (Baskerville et al., 1983; Fitzgeorge et al., 1983). The guinea pig animal dose response model shows infection of macrophage cells which is similar to human lung macrophage infection and was therefore determined to be the best available model for human risk assessments (Armstrong and Haas, 2008). This model was also evaluated against three outbreaks at spas of human L. pneumophila infection and was found to be accurate to within one log of actual infections observed (Armstrong and Haas, 2008). Dose-response for sub-clinical and clinical infections was best estimated with an exponential model (Equation 8.4) (Armstrong and Haas, 2007). The probability of infection (P1(d)) as an exponential function of dose (d) and a model parameter for severity of infection (r).

229 Chapter 8. Risk assessment (rd ) P1(d) =1 e (Equation 8.4) Where:

P1 (d) = probability of infection d = the dose of pathogenic Legionella received, from Equation 8.3 r = model parameter with set values for severity of infections

8.3.1.5 Annual probability of infection The annual probability of infection was calculated based on the number of estimated exposures during a year. The calculation for annual probability of infection was developed previously (Haas et al., 1993) (Equation 8.5). n P1(d)annual =1 (1  P1(d)) (Equation 8.5) Where:

P1 (d)annual = annual probability of infection

P1 (d) = probability of infection from Equation 8.4 n = number of exposures in a year

8.3.2 Parameter estimation

8.3.2.1 Proportion of FLA as trophozoites Within a given FLA population the microorganism exists as both cysts and trophozoites (Rodriguez-Zaragoza, 1994). The proportion of cysts to trophozoites depends on the availability of food and if any adverse stress factors are present (Rodriguez-Zaragoza, 1994). However, the exact proportions present in the environment is not known (Rodriguez-Zaragoza, 1994) because culture and molecular techniques do not reveal the form that the FLA detected is in. In situ microscopy is the only technique that will allow cyst and trophozoites to be enumerated but just two studies were identified which used microscopy and neither of them reported on the ratio of cysts to trophzoites (Amblard et al., 1996; Berk et al., 2006). Cysts are infrequently infected with L. pneumophila as established for the FLA isolated from the drinking water system (Chapter 7) and in other lab research (Rowbotham, 1980; Kuiper et al., 2004). Therefore estimating what portion of the FLA population was in trophozoite form was important to determine how many of the FLA detected were actually able to host pathogenic Legionella. In the experiments reported in Chapter 7, after seven days of incubation with food source (E. coli) and L. pneumophila in water the percentages of trophozoites and cysts were recorded (Table 8.1). The mean trophozoite percentages for H. vermiformis isolates were slightly lower (39 %) than the mean percentages for Acanthamoeba sp. isolates (52 %).

230 Chapter 8. Risk assessment Table 8.1 Percentage of FLA as trophozoites after 6 - 7 days of incubation with L. pneumophila. Environmental % FLA as number SD FLA trophozoites samples

H. vermiformis 39 % 5.6 30 Acanthamoeba sp. 52 % 15 50

The ratios of trophozoites to cysts were applied to the total FLA concentrations from the respective systems (garden hose or heated annular reactor as a shower surrogate) to estimate how many FLA were present as trophozoites and available to be infected. The total number of trophozoites was based on only the qPCR results for Acanthamoeba sp. and H. vermiformis (Table 8.2) and does not take into account the other species of FLA that were present such as E. exudans (Chapter 6). Although this is likely to be an under estimation of the total trophozoites it must be kept in mind that the qPCR detected all FLA DNA some of which would have been from non-viable FLA and therefore over estimating of the viable FLA population density. Table 8.2 Estimation of total trophozoites in water and biofilm samples. Acanthamoeba sp. H. vermiformis Total System Sample type Total Trophs Total Trophs trophs Garden water 18 9 31 12 21 hose (amoebae.mL-1) (green biofilm 231 120 0.5 0.2 120.2 type) (amoebae.cm-2) water Garden 324 168 300 117 285 (amoebae.mL-1) hose (lilac biofilm type) 81 42 71 28 70 (amoebae.cm-2) Showers water 8 4 21 8 12 (heated (amoebae.mL-1) ambient biofilm 0.3 0.2 1 0.4 0.6 reactor) (amoebae.cm-2)

8.3.2.2 Proportion of trophozoites infected with pathogenic Legionella In the drinking water system and applications sampled in this research in total 17 % (4/24) of the FLA screened were naturally infected with Legionella (Chapters 4 - 6). Furthermore, one of those native Legionella bacteria was positively identified as the pathogenic L. micdadei (Table 6.4). These results correlate with other studies where FLA isolated from cooling towers were found to be infected with pathogenic Legionella in 10 % (4/40) of samples (Berk et al., 2006). Further, 7 % (1/15) FLA isolated from hot and cold water hospital network contained pathogenic Legionella (Thomas et al., 2006). In an additional study of drinking water

231 Chapter 8. Risk assessment distribution system biofilm 50 % (1/2) FLA contained pathogenic Legionella (Corsaro et al., 2010) but the very small number of samples was taken into account with this percentage. Based on the available literature the percentage of FLA observed to be infected with Legionella in this research was higher than reported for pathogenic Legionella in other studies. Therefore as the portion of pathogenic Legionella were not confirmed in this research a more conservative estimate of 8 % (2/24) was used to determine the number of FLA trophozoites present at any point infected with native pathogenic Legionella capable of causing infection in humans

(parameter FLp in Equation 8.1). In the water system sampled no L. pneumophila was detected (Chapters 4 - 6). L. pneumophila is highly pathogenic and responsible for the majority of legionellosis infections (Fields et al., 2002; Bartram et al., 2007). Hence, a selection of isolated FLA were infected with L. pneumophila under different conditions (Chapter 7). From these experiments it was found that after seven days 95 % ( = 7, n = 30) of Acanthamoeba sp. and 18 % ( = 6, n = 20) of H. vermiformis were heavily infected with L. pneumophila (Chapter 7). Based on the proportions of Acanthamoeba sp. and H. vermiformis in the garden hoses and heated annular reactor (water and biofilm) and assuming that the other FLA present were infected at similar rates then the total proportion of trophozoites that could be infected were estimated. The estimated trophozoite infection rates of L. pneumophila (variable FLp in Equation 8.1) in the experiments are indicative of what would happen if a more pathogenic Legionella was introduced into the drinking water system. Particularly as the presence of other bacteria has been found not to compete with L. pnueumophila's ability to infect FLA (Declerck et al., 2005). Table 8.3 Percentage of trophozoites infected with pathogenic Legionella (native and susceptible to). Legionella infected trophozoites System type native susceptible Garden hose (green) ~ 8 % ~ 86 %

Garden hose (lilac) ~ 8 % ~ 58 % Showers ~ 8 % ~ 40 %

8.3.2.3 Number of pathogenic Legionella in infected trophozoites An infected trophozoite contains a number of vacuoles filled with the infecting L. pneumophila (Chapter 7). In the initial stages of infection only one L. pneumophila filled vacuole was observed but as days passed eventually the entire trophozoite was filled with L. pueumophila (Chapter 7). The number of infecting L. pneumophila in a single heavily infected trophozoite has been reported as approximately 100 colony forming units (Kuiper et al., 2004). The dimensions and number of the infected vacuoles observed in the infection experiments 232 Chapter 8. Risk assessment (Chapter 7) support this estimation for the number of L. pneumophila bacteria contained in the trophozoite. Hence, the number of L. pneumophila infecting a susceptible trophozoite was estimated at 100 infecting bacteria (variable LpF in Equation 8.1). For trophozoites naturally infected with a native pathogenic Legionella that was not L. pneumophila a lower estimate was used. L. pneumophila is the most infective Legionella species and responsible for 90 - 95 % of human infections (Bartram et al., 2007) but other pathogenic Legionella have lower infection rates (Fields et al., 2002). Pathogenic L. micdadei was identified in an FLA isolate and observation of this FLA and others infected with Legionella found that the trophozoites were not lysed nor were they filled to the same extent as with L. pneumophila (Chapter 6 and 7). Therefore, it was assumed that only two vacuoles were infected at any point of time, where one vacuole was estimated to contain 10 Legionella bacteria (Chapter 7). This is consistent with other research which reported that L. pneumophila was able to infect and lyse FLA more effectively compared to less pathogenic species L. micdadei, which is able to infect at a lower rate but not lyse the host FLA at room temperatures (Fallon and Rowbotham, 1990; Gao et al., 1999). For this reason, the number of pathogenic Legionella (not including L. pneumophila) infecting a single trophozoite naturally was assumed to be 20 infecting bacteria (variable LpF in Equation 8.1). The size of trophozoites are generally greater than 10 μm which is at the upper limit of particle size that can be inhaled (U.S. Environmental Protection Agency, 2004). However, it must be considered that once FLA are infected with ARM their cell membrane becomes weaker and they are more likely to lyse when passing through a spray device such as shower head or hose nozzle (Rowbotham, 1980). Even without cell lysis FLA could still transport their infecting pathogenic Legionella to the lung aveoli by movement via their pseudopodia. Additionally, once at the higher human body temperature (37 °C) pathogenic Legionella growth and lysis of the host FLA is more likely to occur (Gao et al., 1999) thus releasing more virulent Legionella in closer proximity to the target lung cells (Cirillo et al., 1997; Nora et al., 2009). Therefore, pathogenic Legionella infecting trophozoites in the water are estimated to partition into aerosols at a similar rate to the Legionella free cells in the water.

8.3.2.4 Number of native pathogenic Legionella as independent cells in water Pathogenic Legionella can exist as independent extracellular cells outside of FLA in water. The quantification of Legionella in water and biofilm samples using culture and molecular methods includes both the bacteria as independent cells and those infecting FLA. To estimate the proportion of Legionella which reside outside of FLA in a given population the results from the up-take experiments were used (Chapter 7). In these experiments the number of extracellular Legionella was found not to significantly change in the presence of FLA compared to the no 233 Chapter 8. Risk assessment FLA control (Chapter 7). Here, the increases in the Legionella population were observed within the infected FLA. The rates of Legionella infection was distinct for Acanthamoeba sp. and H. vermiformis isolates. In this model the differences in the susceptibility of the trophozoite population to an infection with L. pneumophila has already been incorporated based on the number of Acanthamoeba sp. and H. vermiformis detected in the samples (Table 8.3). In a water sample where the native densities of Legionella are known the proportion present independent of FLA can be estimated using the number of total trophozoites present and a 17 % infection rate with Legionella other than L. pneumophila (Section 8.3.2.2). The number of Legionella residing as individual cells is the difference between the total Legionella density and the number of trophozoites naturally infected with 20 Legionella bacteria per trophozoite (Section 8.3.2.3) (Table 8.4). It is important to make the distinction between Legionella infecting FLA and those Legionella as independent cells to be able to run the model scenarios in the absence of FLA in order to describe the influence FLA have on the probability of infection. The proportion of the Legionella that exist as independent cells that were pathogenic to humans in the drinking water systems was estimated. From the identification of the Legionella qPCR products only 11 % (2/18) were identified as pathogenic Legionella but not L. pneumophila (Chapter 4 and 6). Two thirds (12/18) of the Legionella sequenced from the qPCR products were identified as unclassified Legionella sp. or Legionellales bacteria with unknown pathogenicity. Some of the unclassified Legionella sp. were closely matched with LLAPs on the phylogenetic trees. LLAPs are able to replicate within FLA and are phylogenetically similar to Legionella sp. (Rowbotham, 1986). Research looking at 12 LLAP isolates revealed that collectively they were associated as a co-infecting agent or the sole infection agent in 9.8 % (n = 255) of hospitalised CAP infections (Marrie et al., 2001). Another consideration is that presently there are 48 different identified species of Legionella and 42 % (20/48) of them have been associated with disease (Fields et al., 2002). Applying that same percentage to the unclassified Legionella proportion gives 25 % of the native Legionella cells independent of FLA can be estimated as human pathogens (variable LpW in equation 8.1) (Table 8.4).

234 Chapter 8. Risk assessment Table 8.4 Number of native Legionella bacteria in FLA or as independent free cells Legionella Sample System Free cells type Total In trophs Total Pathogenic (LpW) water 39 72 - - Garden hose (cells.mL-1) (drinking type) biofilm 22 409^ - - (cells.cm-2) water 6546 970 5576 1394 Garden hose (cells.mL-1) (recycled type) biofilm 295 237 58 15 (cells.cm-2) water Showers 80 43 37 9 (cells.mL-1) (heated ambient biofilm reactor) 13 2 11 3 (cells.cm-2) ^ number of Legionella estimated in trophozoites is greater than total FLA as in the drinking water hose an anti-bacterial compound was present in the hose material which reduced the Legionella concentrations independent of the FLA present. 8.3.2.5 Introduction of pathogenic L. pneumophila In the event of a pathogenic L. pneumophila being introduced into the drinking water system the concentration of the introduced pathogen needs to be estimated. For garden hoses the concentration of L. pneumophila entering was based on the reported concentrations in tap water and distribution systems. L. pneumophila was detected at concentrations from 0 - 4 cell.mL-1 in tap water (Devos et al., 2005) and 0 - 0.6 cell.mL-1 in distribution systems (Wullings et al., 2011). Based on this data a probable (0.06 cells.mL-1) and high concentration (4 cells.mL-1) of L. pneumophila entering the garden hoses was selected for use in the model scenarios. Similarly in the literature the concentration of L. pneumophila in shower water has been reported between 0.05 - 80 cells.mL-1 (Devos et al., 2005) and in an alternative study with a mean of 0.04 cells.mL-1 (Chen and Chang, 2010). Based on this data a probable (0.05 cells.mL-1) and high (80 cells.mL-1) introduced concentration of L. pneumophila were selected for use in the model. Within Legionella populations it has been reported that as L. pneumophila populations increase the other Legionella in the population decrease in number despite being previously dominant (Wéry et al., 2008). Hence, the estimated L. pneumophila is likely to out compete the native Legionella in the water type and dominate the system. Furthermore, for both exposure scenarios the number of trophozoite cells that are able to be infected by L. pneumophila have already been projected (Table 8.3). and it was estimated that all the growth of the introduced L. pneumophila will be due to infecting the FLA population in the water and biofilm of the sample. However, in the model only the infected trophozoites in the water fraction were used to calculate the dose of Legionella in the water as this was a true account of the 'equilibrium'

235 Chapter 8. Risk assessment between FLA in the water and biofilm that was already established in the system. Assuming that all the replication of the L. pnuemophila are occurring in the trophozites is consistent with the results observed in the experiments conducted with the FLA isolates (Chapter 7). The literature supports these assumptions also, as L. pneumophila has fastidious growth requirements (Fields et al., 2002), a large proportion of L. pneumophila as free cells in drinking water systems are in a viable but non culturable (VBNC) state (Diederen et al., 2007) and L. pneumophila numbers increase significantly after the introduction of FLA (Declerck et al., 2009) including VBNC cultures of Legionella (Steinert et al., 1997).

8.3.2.6 Partitioning co-efficient The partitioning co-efficient is the parameter that captures what volume of the drinking water, and its Legionella, is converted into aerosols in a volume of air during a particular aerosolization event. Aerosols < 7 μm in diameter are small enough to reach the aveolar region of the lung and be retained (Wilkes, 1999). For showering a partitioning co-efficient was calculated based on available research. Measuring the size of aerosols present after showering using a water saving shower head with water at 7 L.min -1 at 42 °C for 15 min the concentration of aerosols (< 7 μm) was 3.3  105 aerosols.m-3 in a 1.7 m3 enclosed shower recess (O'Toole et al., 2009). The total volume of water as aerosols was calculated using the volume of sphere and the radius of the aerosols detected giving a total volume of 1.35  10-2 L of aerosols < 7 μm in diameter. The partitioning co-efficient was calculated based on the total volume of water used during showering (105 L) and the total volume of inhalable aerosols present (1.35  10-2 L) to give a partitioning co-efficient of 1.29  10-4 L.m-3 for showering (parameter PC in Equation 8.2). This estimation is consistent with other reported partitioning coefficients for showering which range from < 5  10-5 to 3  10-4 L.m-3 (Rose et al., 1998). For the use of garden hoses a separate partitioning co-efficient was calculated in a similar manner to the shower scenario. Concentration of aerosols (<7 μm) were reported for the use of a garden hose while washing a car with a spray nozzle using water at 7.3 L.min-1 for 5 min in a 6 m3 enclosed space gave a total of 1.9  106 aerosols.m-3 (O'Toole et al., 2009). The total volume of inhalable aerosols present was calculated at 9.8  10-3 L, therefore the partitioning co-efficient for the 36.5 L used was 2.68  10-4 for garden hose use (parameter PC in Equation 8.2). No previously reported partitioning co-efficient for garden hoses could be identified however the estimations made here are in the same order of magnitude as those estimated for showering which gives some validity. For both showering and hose use the number of pathogenic Legionella bacteria in the aerosols was estimated to be a direct function of the density of pathogenic Legionella in the

236 Chapter 8. Risk assessment water samples. Research examining aerosols during showering report that > 90 % of the L. pneumophila detected were in the small aerosols (< 5 μm) (Bollin et al., 1985). Hence, pathogenic Legionella is estimated to be present in the shower and hose aerosols as equal proportions to that of the supply water assuming homogeneous distribution of the bacteria.

8.3.2.7 Exposure dose parameters

3. -1 The inhalation dose (IH) was estimated at 0.6 - 1.5 m .h (United States Environmental Protection Agency, 1997) and has been used in other Legionella QMRA (Armstrong and Haas, 2008). For this model the mid-point in the range of the inhalation rate 1.05 m3..h-1 was used in the model (parameter IH in Equation 8.3). Exposure time for showering was estimated using the results of an Australian a study which reports that people shower on average for 8 min per day (Juan, 2006). However, there was no literature identified which reported on the frequency and duration of garden hose use by individuals. Garden hoses are used to wash cars, water gardens, wash pets, hose paths or hard surfaces, cleaning outdoors areas and for children to play under in hot weather. Based on this broad list of uses it is estimated that each person in a house with a yard would use a garden hose for 10 min once a week (parameter TEx in Equation 8.3). Human retention rates of inhaled aerosols containing bacteria have been recorded at 50 % (Harper and Morton, 1953) and used in other Legionella QMRA's (Armstrong and Haas, 2008) (parameter R in Equation 8.3).

8.3.2.8 Dose-response Within the exponential dose-response model there is a parameter that estimates the severity of infection (parameter r in Equation 8.4). Based on the available data for the exponential dose response model, Armstrong and Haas (2008) estimated the severity of infection parameter as r = 0.06 for sub-clinical infections with no mortality and full recovery. For infections that resulted in clinical infections or mortality the severity of infection parameter was determined to be r = 1.07  10-4 (Armstrong and Haas, 2008). These two dose response models were tested against data from Legionella outbreaks in spas and reported to be accurate to within one order of magnitude (Armstrong and Haas, 2008).

8.3.3 Test scenarios Test scenarios for the risk assessment were conducted for exposure during garden hose use and showering (Table 8.5). For both exposure events the model was first run using parameters recorded for the state of the drinking water type when sampled and calculated the risk of the native pathogenic Legionella and FLA present. The second part of the model was based on the event of pathogenic L. pneumophila being introduced into the drinking water system and resulting probability of infection was calculated for FLA presence and absence. FLA 237 Chapter 8. Risk assessment presence and absence was included in the model to provide a first approximation to the potential significance of FLA presence on the probability of infection with L. pneumophila. Table 8.5 Test scenarios used in the risk assessment model calculations Legionella pathogen FLA Scenario System type type presence number Garden hose use Native pathogenic green hose FLA H-1 Legionella lilac hose FLA H-2 green hose FLA H-3 Introduced lilac hose FLA H-4 L. pneumophila any hose No FLA H-5 Showering Native pathogenic FLA S-1 Legionella water saving Introduced shower head FLA S-2 L. pneumophila No FLA S-3 8.4 RESULTS Deterministic model scenarios were run for the probability of infection models and results reported for an individual per exposure event and per year of 'normal' activity (Table 8.6). For garden use the probability after one exposure for native Legionella conditions was higher for the lilac hose (subclinical 0.93 and clinical 0.005) compared to the green hose (subclinical 0.048 and clinical 8.7  10-5). This difference was even more pronounced over a year of exposures with a probability of clinical infection after lilac hose use estimated at 0.219 compared to 0.004 for green hose use. Both the low and high densities of introduced L. pneumophila resulted in the same probability of infection but again were greater for the lilac hose with the model predicting at least one clinical infection for an individual over a year (> 1). The higher probability of infections from the lilac hose was directly related to the high number of trophozoites (165.3 amoebae.mL-1) susceptible to infection with L. pneumophila. The absence of FLA reduced the risk of clinical infection by three to five orders of magnitude to 5.3  10-4 for a high concentration (80 cells.mL-1) of introduced L. pneumophila. The risks estimated for showering were generally lower than garden hose use because of the lower densities of FLA. Using water heated to 42 °C with native pathogenic Legionella present the risk of clinical infection after one year of exposure was 0.015. While, if low densities (0.05 cells.mL-1) of L. pneumophila were introduced that risk may have increased to 0.224 in the presence of FLA but only 2.6  10-5 in the absence of FLA.

238 Chapter 8. Risk assessment

Table 8.6 Risk assessment results for garden hose and shower scenarios with and without FLA present Pathogenic Probability Probability Probability of Legionella of clinical of clinical Legionella System FLA Scenario #trophs subclinical infected infections infection over pathogen type type presence number in a infections after (amoebae.mL-1) as cells after one one year of FLA one exposure exposure exposures Garden hose use -5 Native pathogenic green hose FLA H-1 1.7 20 0 0.048 8.7  10 0.004 Legionella lilac hose FLA H-2 23 20 1394 0.930 0.005 0.219 H-3a 0.06 0.924 0.005 0.213 green hose FLA 18 100 H-3b 4 0.925 0.005 0.213 Introduced H-4a 0.06 1.00 0.041 0.889 lilac hose FLA 165.3 100 L. pneumophila H-4b 4 1.00 0.041 0.889 H-5a 0.06 8.6  10-5 1.5  10-7 7.9  10-6 Any hose No FLA - - H-5b 4 0.006 1.0  10-5 5.3  10-4 Showering Native pathogenic -5 water FLA S-1 1 20 9 0.015 2.7  10 0.01 Legionella saving -4 shower S-2a 0.05 0.224 4.5  10 0.152 FLA 4.8 100 Introduced head - S-2b 80 0.256 5.2  10-4 0.175 L. pneumophila water at -5 -8 -5 S-3a 0.05 2.6  10 5.0  10 1.7  10 42 °C No FLA - - -5 S-3b 80 0.041 7.5  10 0.027

239 Chapter 8. Risk assessment

8.5 DISCUSSION

8.5.1 Risk of Legionella infection The Australian Drinking Water Guidelines, which are consistent with the World Health Organization guidelines, recommends that the probability of annual clinical infection for a given pathogen should be less than 1  10-4 (Australian Government, 2004). In risk modeling for the different scenarios only the introduced L. pneumophila at the lowest density in the absence of FLA for both garden hose usage and showering were below the recommended threshold. Therefore exposure during showering and garden hoses use when Legionella pathogens and FLA are present resulted in an estimated risk of clinical infection that was above the recommended guidelines. In particular, estimations for the lilac hose type produced the maximum infection probability of 0.889 over a year, even if pathogenic L. pneumophila were introduced at low concentrations (0.06 cells.mL-1). The probability of Legionella infection if the exposure scenarios were supplied with recycled water were estimated based on the density and diversity of FLA and Legionella in the MRD samples (Chapter 4). There was not a significant difference (Kruskal-Wallis test, p = 0.39) between the densities and diversities of FLA in drinking compared to recycled water (Section 4.4.2.2). However, recycled water samples did have positive detects of pathogenic Legionella anisa in two samples while no Legionella was detected in the drinking water samples. Based on this data it is estimated that the recycled water would not present a greater risk of Legionella infection compared to drinking water when applied to garden hose use or showering. This is primarily because it is the FLA densities are comparable between the two water types and FLA densities that have a large impact on the probability of infection in the QMRA model. When recycled water is treated to a standard comparable with drinking water, as in the case for the dual distribution system sampled (Chapter 4), then it is estimated that recycled water does not to present any increased risk of Legionella infection compared to drinking water. However, not all recycled water is treated to the same standard and therefore the estimation of risk is specific for the recycled water system sampled. As with all models there are limitations on the ability to predict probabilities of infections which need to be considered. In particular the dose-response model was developed based on infections of healthy guinea pigs with L. pneumophila (Baskerville et al., 1983; Fitzgeorge et al., 1983). Therefore the probability of infection does not take into account the human risk factors for Legionella infection nor the protective immune response after initial exposure. Human risk factors for Legionella infection for humans include being elderly, diabetic, a smoker,

240 Chapter 8. Risk assessment immuno-suppressed or suffering chronic heart disease (Bartram et al., 2007). These risk factors are not taken into consideration in the dose response model and therefore the model is likely to underestimate the probability of infection for these susceptible sub-populations. At a time when the demographics of developed nations are changing to an increasingly older populace understanding the risk of infections for this group will become even more important as attempts are made to reduce the health burden. Already the rates of CAP requiring hospitalisation have increased over the last couple of decades (Fry et al., 2005) and Legionnaires disease is a portion of those infections. Although current epidemiology can not determine what fraction is caused by Legionella infections because the causative agent is not identified in approximately half of the CAP infections presenting at hospital (Wilson and Ferguson, 2005) and sporadic cases are largely unreported. Conservative estimates are that 3 - 5 % of hospitalised CAP infections are due to Legionella (Broadbent, 1996). The high probability of infections estimated for exposure to Legionella via showering and garden hose highlight the distinct possibility that Legionella infection may account for a higher portion of CAP than presently recognised. Controlling Legionella in drinking water and its applications is estimated to result in significant health benefits to the population and saving in health care costs. In contrast, for the portion of the population that does not have risk factors for Legionella infection then the dose-response model is likely to be an over estimate of the probability of infection as it does not take into account resistance and immune response. From a survey of the literature it was identified that pathogenic Legionella are isolated from the majority of drinking waters and applications sampled (Table 2.5 and 2.11). People are exposed to Legionella through their use of drinking water on a daily basis and there is estimated to be a high level of sub- clinical infections within the population that would result immune protection from infection from subsequent exposures. The deficiency of the model to account for immune response means that it is likely an over estimate of the probability of infection for healthy people without risk factors for Legionella infection and who have already been exposed to pathogenic Legionella through out their life. The does-response model did not consider Pontiac Fever which is a flu like illness characterised by fever, headache, myalgia, malaise and is generally self limiting (Glick et al., 1978). After exposure to L. pneumophila it has been reported that less at risk people will contract Pontiac Fever over Legionnaires disease (Euser et al., 2010). Additionally, Legionella pathogens, such as L. micdadei, have been reported to be the cause of outbreaks of the more mild Pontiac Fever from drinking water applications (Fallon and Rowbotham, 1990). Hence, in this model the clinical infections for people not at risk and from the native Legionella pathogens in the water system is likely to have been Pontiac Fever rather than pneumonia as part of 241 Chapter 8. Risk assessment Legionnaires disease. As Pontiac Fever would have been passed off as a seasonal flu by most of the population the probability of infection estimated within this model could well be occurring within the population but Legionella is never identified as the etiological agent because it is never tested for as the infections are mild and self limiting. Flu like illnesses account for a huge number of lost days of work and productivity. If Pontiac Fever caused by exposure to pathogenic Legionella is a large proportion of the flu cases each year then taking steps to reduce exposure to Legionella could result in considerable improvement to public health. Although, data on the contribution of Pontiac Fever to overall flu statistics is not available and this is one area of the epidemiology of Legionella infection that needs to be explored further. The up-regulation of Legionella virulence factors after growth within FLA and subsequent increased infectivity for human cells is well documented (Cirillo et al., 1997; Nora et al., 2009). The dose-response model does not presently accommodate this increased virulence in the probability of infection. Considering that the increase in Legionella infectivity can be two orders of magnitude greater after growth within FLA (Cirillo et al., 1994) then it could alter the probability of infection estimated by the dose-response model considerably. However, at present the work on infectivity of Legionella for human cells lines has not been translated to any observations about dose response (Armstrong and Haas, 2008) and more work needs to occur in this area in order to quantify the full impact that FLA have on the probability of infection. Although the model uses FLA as the protozoan host any other protozoa where the infecting numbers of Legionella and the protozoan densities are known could be substituted. Ciliates can be infected with Legionella (Fields et al., 1984) and are isolated from drinking water applications (Barbaree et al., 1986). However, at present their is limited data available for the critical densities needed for an accurate risk model.

8.5.2 Reducing the risk of infection The QMRA reported here provides estimates of Legionella infections at probabilities higher than the recommended guidelines. Even though these infection rates are likely to be an overestimate for the portion of the population who is not at risk but for the sub-population which is at risk the probabilities of infections are likely to be relatively accurate. Therefore a series of risk reducing practices for garden hose and showering are recommended for people who are at risk of Legionella infection. The risk reductions are based on changing behaviour rather than trying to remove the Legionella from the systems with biocides or antibiotics. In part because FLA cysts are resistant to biocides and disinfectants at the concentrations typically used (Critchley and Bentham, 2009). Additionally, there is evidence that biocides when applied to water applications such as cooling towers may actually stimulate FLA growth (Srikanth and

242 Chapter 8. Risk assessment Berk, 1993). A further benefit of behavioural changes over the use of chemicals is that introducing additional chemicals would create another health risk that may be greater than the Legionella risk that it is trying to reduce.

8.5.2.1 Garden hoses The higher probability of infections in the lilac hose type compared to the common green hose type was due to the higher densities of FLA observed (Chapter 5). There were more FLA in the lilac hose type as it was likely lacking the anti-microbial compounds, such as triclosan, in the hose material (Smith and Lourie, 2009). The use of anti-microbial compounds such as antibiotics is an attractive solution to microbial growth because of the convenience of use. However, the widespread use of antibiotics is directly linked to the rapid increase of antibiotic resistant bacteria which is a health crisis worldwide. However, if consumers are presented with a choice of products then the standard green garden hose should be used over the lilac hose type recommended for use with recycled water. To reduce FLA and Legionella densities more naturally, it is recommended to drain garden hoses after use and hang them up as microorganisms require moisture for growth. Another means of reducing the number of Legionella filled aerosols inhaled is to leave a hose to run water on the garden or grass for three minutes to flush out the stagnant water present and remove sloughing biofilm before switching to hold the hose and use it at a spray setting. This will be effective if the bulk of the Legionella growth was occurring within the stagnant water of the garden hose as was observed in this research (Chapter 5). However, flushing dose not guarantee to reduce the Legionella and FLA (Rice et al., 2006) especially if the supplying water is the source of the contamination. Finally for those people most at risk of Legionella infection such as the elderly or those with prior Legionella infections, a safety precaution would be not to use the hose on a spray setting at all but rather use a watering can for gardening purposes or a bucket for cleaning. For those that do wish to continue using garden hoses then a safeguard would be to covering the nose and mouth with a handkerchief or scarf which should be sufficient to prevent the Legionella filled aerosol droplets from entering the lungs.

8.5.2.2 Showering The source of Legionella and FLA within a shower is likely to be a combination of the supplying drinking water and growth or survival (depending on the temperature) in the hot water tank plus the colonised pipes and shower head which are predominantly maintained at moist ambient temperatures except for a flush with hot water during shower use. Due to the number of locations where the Legionella and FLA can reside and grow there are not many structural

243 Chapter 8. Risk assessment changes that can be made to a shower system to reduce that growth. Increasing the temperature of water in the hot water tank to 60 °C is likely to reduce growth in that part of the system but will have little impact on the Legionella in the pipes or shower head and may actually result in increased chances of scolding. Burns for a person not at risk of Legionella infection would be a more serious heath issue than the chance of Pontiac Fever that they are likely already to have immunity to. Hence, for these reasons increasing the temperature of the hot water system over 50 °C is not recommend. For that portion of the population who are at risk of Legionella infection it is recommended they take baths instead of showering or at the very least reduce the frequency of showering. For those people who do not have access to a bath then it is possible to purchase and install a point of use water filter for the shower head. Point of use water filters have been shown to effectively remove Legionella and Mycobacterium present in the supplying water in hospitals (Exner et al., 2005; Sheffer et al., 2005). It was also recommended that for hospitals where Legionella is known to colonise the hot water systems that the point of use filter be used selectively for at risk patients (Sheffer et al., 2005). Hospital systems present a special problem for Legionella growth in hot water systems and a variety of disinfection and control mechanisms have been recommended including reducing stagnant areas in the water system (Lin et al., 1998). Some disinfectants such as chlorine dioxide display more effectiveness against Legionella than others and warrant further research (Zhang et al., 2009).

8.5.3 Conclusion and future research The finding that there is a high probability of infection after exposure to pathogenic Legionella during garden hose use and showering presents a significant largely unrecognised health burden that needs to be addressed. However, in order to determine the exact size of that health burden better dose-response models need to be developed that accommodate different susceptibility of people to infection, protective immune responses and the increased infectivity of Legionella after growth in FLA. Furthermore, more epidemiological research is needed to determine what portion of CAP and flu like infections are due to Legionella infections. Once this information is available then it will be evident if a health campaign needs to be raised to inform people of the risks and what behavioural changes they can make to reduce their own risk of infection.

244 Chapter 8. Risk assessment

8.6 REFERENCES 1. Amblard, C., G. Bourdier, J. Carrias, N. Maurin and C. Quiblier. 1996. Seasonal evolution of microbial community structure in drinking water reservoir. Water Research. 30: 613-624. 2. Armstrong, T. and C. Haas. 2008. Legionnaires' disease: evaluation of a quantitative microbial risk assessment model. Journal of Water and Health. 6(2): 149-166. 3. Armstrong, T. W. and C. N. Haas. 2007. A Quantitative Microbial Risk Assessment Model for Legionnaires' Disease: Animal Model Selection and Dose-Response Modeling. Risk Analysis: An International Journal. 27(6): 1581-1596. 4. Armstrong, T. W. and C. N. Haas. 2007. Quantitative Microbial Risk Assessment Model for Legionnaires' Disease: Assessment of Human Exposures for Selected Spa Outbreaks. Journal of Occupational and Environmental Hygiene. 4(8): 634 - 646. 5. Australian Government. 2004. Australian Drinking Water Quality Guidelines. National Health And Medical Research Council. 6. Barbaree, J. M., B. S. Fields, J. C. Feeley, G. W. Gorman and W. T. Martin. 1986. Isolation of protozoa from water associated with a legionellosis outbreak and demonstration of intracellular multiplication of Legionella pneumophila. Applied and Environmental Microbiology. 51(2): 422-424. 7. Bartram, J., Y. Chartier, J. Lee, K. Pond and S. Surman-Lee. 2007. Legionella and the prevention of Legionellosis. Geneva, World Health Organisation 8. Baskerville, A., R. B. Fitzgeorge, M. Broster and P. Hambleton. 1983. Histopathology of experimental Legionnaires' disease in guinea pigs, rhesus monkeys and marmosets. The Journal of Pathology. 139(3): 349-362. 9. Bauer, H., M. Fuerhacker, F. Zibuschka, H. Schmid and H. Puxbaum. 2002. Bacteria and fungi in aerosols generated by two different types of wastewater treatment plants. Water Research. 36(16): 3965-3970. 10. Berk, S., R. Ting, G. Turner and R. Ashburn. 1998. Production of respirable vesicles containing live Legionella pneumophila cells by two Acanthamoeba spp. Applied and Environmental Microbiology. 64(1): 279-286. 11. Berk, S. G., J. H. Gunderson, A. L. Newsome, A. L. Farone, B. J. Hayes, K. S. Redding, N. Uddin, E. L. Williams, R. A. Johnson, M. Farsian, A. Reid, J. Skimmyhorn and M. B. Farone. 2006. Occurrence of infected amoebae in cooling towers compared with natural aquatic environments: implications for emerging pathogens. Environmental Science & Technology. 40(23): 7440-7444. 12. Bichai, F., P. Payment and B. Barbeau. 2008. Protection of waterborne pathogens by higher organisms in drinking water: a review. Canadian Journal of Microbiology. 54(7): 509-524. 13. Bollin, G. E., J. F. Plouffe, M. F. Para and B. Hackman. 1985. Aerosols containing Legionella pneumophila generated by shower heads and hot-water faucets. Applied and Environmental Microbiology. 50(5): 1128-1131. 14. Breiman, R. F., B. Fields, G. N. Sanden, L. Volmer, A. Meier and J. Spika. 1990. Association of shower use with Legionnaires' disease. Possible role of amoebae. Journal of the American Medical Association. 263(21): 2924 - 2926. 15. Broadbent, C. 1996. Guidance for the control of Legionella Adelaide National Environmental Health Forum. 16. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634.

245 Chapter 8. Risk assessment 17. Cirillo, J., S. Falkow, L. Tompkins and L. Bermudez. 1997. Interaction of Mycobacterium avium with environmental amoebae enhances virulence. Infection and Immunity. 65(9): 3759-3767. 18. Cirillo, J. D., S. Falkow and L. S. Tompkins. 1994. Growth of Legionella pneumophila in Acanthamoeba castellanii enhances invasion. Infection and Immunity. 62(8): 3254-3261. 19. Corsaro, D., G. S. Pages, V. Catalan, J.-F. Loret and G. Greub. 2010. Biodiversity of amoebae and amoeba-associated bacteria in water treatment plants. International Journal of Hygiene and Environmental Health. 213(3): 158-166. 20. Craun, G. F., J. M. Brunkard, J. S. Yoder, V. A. Roberts, J. Carpenter, T. Wade, R. L. Calderon, J. M. Roberts, M. J. Beach and S. L. Roy. 2010. Causes of outbreaks associated with drinking water in the United States from 1971 to 2006. Clinical Microbiology Reviews. 23(3): 507-528. 21. Critchley, M. and R. Bentham. 2009. The efficacy of biocides and other chemical additives in cooling water systems in the control of amoebae. Journal of Applied Microbiology. 106(3): 784-789. 22. Declerck, P., J. Behets, Y. Delaedt, A. Margineanu, E. Lammertyn and F. Ollevier. 2005. Impact of non-Legionella bacteria on the uptake and intracellular replication of Legionella pneumophila in Acanthamoeba castellanii and Naegleria lovaniensis. Microbial Ecology. 50(4): 536-549. 23. Declerck, P., J. Behets, A. Margineanu, V. van Hoef, B. De Keersmaecker and F. Ollevier. 2009. Replication of Legionella pneumophila in biofilms of water distribution pipes. Microbiological Research. 164(6): 593-603. 24. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925. 25. Diederen, B., C. d. Jong, I. Aarts, M. F. Peeters and A. v. d. Zee. 2007. Molecular evidence for the ubiquitous presence of Legionella species in Dutch tap water installations. Journal of Water and Health. 5(3): 375 - 383. 26. Euser, S. M., M. Pelgrim and J. W. den Boer. 2010. Legionnaires' disease and Pontiac fever after using a private outdoor whirlpool spa. Scandinavian Journal of Infectious Diseases. 42(11-12): 910-916. 27. Exner, M., A. Kramer, L. Lajoie, J. Gebel, S. Engelhart and P. Hartemann. 2005. Prevention and control of health care - associated waterborne infections in health care facilities. American Journal of Infection Control. 33(5): S26-S40. 28. Fallon, R. J. and T. J. Rowbotham. 1990. Microbiological investigations into an outbreak of Pontiac fever due to Legionella micdadei associated with use of a whirlpool. Journal of Clinical Pathology. 43(6): 479-483. 29. Fields, B., R. Benson and E. Besser. 2002. Legionella and Legionnaires' disease: 25 years of investigation. Clinical Microbiology Reviews. 15(3): 506-526. 30. Fields, B. S., E. B. Shotts, Jr, J. C. Feeley, G. W. Gorman and W. T. Martin. 1984. Proliferation of Legionella pneumophila as an intracellular parasite of the ciliated protozoan Tetrahymena pyriformis. Applied and Environmental Microbiology. 47(3): 467-471. 31. Fitzgeorge, R. B., A. Baskerville, M. Broster, P. Hambleton and P. J. Dennis. 1983. Aerosol infection of animals with strains of Legionella pneumophila of different virulence: comparison with intraperitoneal and intranasal routes of infection. Journal of Hygiene (London). 90(1): 81-89. 32. Frost, F., G. F. Craun and R. Claderon. 1996. Waterborne disease surveillance. Journal of American Water Works Association 88(9): 66-75. 33. Fry, A. M., D. K. Shay, R. C. Holman, A. T. Curns and L. J. Anderson. 2005. Trends in hospitalizations for pneumonia among persons aged 65 years or older in the United States, 1988-2002. Journal of the American Medical Association. 294(21): 2712-2719. 246 Chapter 8. Risk assessment 34. Gao, L. Y., M. Susa, B. Ticac and Y. A. Kwaik. 1999. Heterogeneity in intracellular replication and cytopathogenicity of Legionella pneumophila and Legionella micdadei in mammalian and protozoan cells. Microbial Pathogenesis. 27(5): 273-287. 35. Glick, T., M. Gregg, B. Berman, G. Mallison, W. Rhodes and I. Kassanoff. 1978. Pontiac Fever; An epidemic of unknown etiology in a health department. American Journal of Epidemiology. 107(2): 149 -160. 36. Haas, C. N., J. B. Rose and C. P. Gerba. 1999. Quantitative microbial risk assessment. New York, John Wiley. 37. Haas, C. N., J. B. Rose, C. P. Gerba and S. Regli. 1993. Risk assessment of virus in drinking water. Risk Analysis. 13(5): 545 - 552. 38. Harper, G. and J. Morton. 1953. The respiratory retention of bacterial aerosols: experiments with radioactive spores. The Journal of Hygiene. 51(3): 372-385. 39. Juan, S. 2006. "Do we really need a daily shower or bath to stay healthy?" The Register, from http://www.theregister.co.uk/2007/02/17/the_odd_body_daily_shower/. 40. Kuiper, M. W., B. A. Wullings, A. D. L. Akkermans, R. R. Beumer and D. van der Kooij. 2004. Intracellular proliferation of Legionella pneumophila in Hartmannella vermiformis in aquatic biofilms grown on plasticized polyvinyl chloride. Applied and Environmental Microbiology. 70(11): 6826-6833. 41. Lin, Y.-s. E., R. D. Vidic, J. E. Stout and V. L. Yu. 1998. Legionella in water distribution systems. Journal American Water Works Association. 90(9): 11. 42. Loret, J.-F. and G. Greub. 2010. Free-living amoebae: Biological by-passes in water treatment. International Journal of Hygiene and Environmental Health. 213(3): 167-175. 43. Loret, J. F., M. Jousset, S. Robert, G. Saucedo, F. Ribas, V. Thomas and G. Greub. 2008. Amoebae-resisting bacteria in drinking water: risk assessment and management. Water Science and Technology. 58(3): 571-577. 44. Marrie, T., D. Raoult, B. La Scola, R. Birtles and E. Carolis. 2001. Legionella-like and other amoebal pathogens as agents of community-acquired pneumonia. Emerging Infectious Diseases. 7(6): 1026. 45. Nora, T., M. Lomma, L. Gomez-Valero and C. Buchrieser. 2009. Molecular mimicry: an important virulence strategy employed by Legionella pneumophila to subvert host functions. Future Microbiology. 4: 691-701. 46. O'Toole, J., M. Keywood, M. Sinclair and K. Leder. 2009. Risk in the mist? Deriving data to quantify microbial health risks associated with aerosol generation by water- efficient devices during typical domestic water-using activities. Water Science and Technology. 60(11): 2913-2920. 47. Rice, E., W. Rich, C. Johnson and D. Lye. 2006. The role of flushing dental water lines for the removal of microbial contaminants. Public Health Reports. 121(3): 270-274. 48. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241. 49. Rose, C. S., J. W. Martyny, L. S. Newman, D. K. Milton, T. E. King and J. L. Beebe. 1998. “Lifeguard lung”: Endemic granulomatous pneumonitis in an indoor swimming pool. American Journal of Public Health. 88(12): 1795 -1800. 50. Rowbotham, T. 1980. Preliminary report on the pathogenicity of Legionella pneumophila for freshwater and soil amoebae. Journal of Clinical Pathology. 33: 1179-1183. 51. Rowbotham, T. 1986. Current views on the relationships between amoebae, legionellae and man. Israel Journal of Medical Sciences. 22: 678-689. 52. Schoen, M. and N. Ashbolt. 2011. An in-premise model for Legionella exposure during showering events. Water Research. In press. 53. Sheffer, P. J., J. E. Stout, M. M. Wagener and R. R. Muder. 2005. Efficacy of new point-of-use water filter for preventing exposure to Legionella and waterborne bacteria. American Journal of Infection Control. 33(5, Supplement 1): S20-S25.

247 Chapter 8. Risk assessment 54. Smith, R. and B. Lourie. 2009. Slow death by rubber duck. How the toxic chemistry of everyday life affects our health. Ontario, Random House. 55. Srikanth, S. and S. G. Berk. 1993. Stimulatory effect of cooling tower biocides on amoebae. Applied and Environmental Microbiology. 59(10): 3245-3249. 56. Steinert, M., L. Emody, R. Amann and J. Hacker. 1997. Resuscitation of viable but nonculturable Legionalla pneumophila Philadelphia JR32 by Acanthamoeba castellanii. Applied and Environmental Microbiology. 64(5): 2047-2053. 57. Storey, M., N. Ashbolt and T. A. Stenström. 2004. Biofilms, thermophilic amoebae and Legionella pneumophila - a quantitative risk assessment for distributed water. Water Science & Technology. 50(1): 77-82. 58. Szewzyk, U., R. Szewzyk, W. Manz and K. H. Schleifer. 2000. Microbiological safety of drinking water. Annual Review of Microbiology. 54: 81-127. 59. Teltsch, B., S. Kedmi, L. Bonnet, Y. Borenzstajn-Rotem and E. Katzenelson. 1980. Isolation and identification of pathogenic microorganisms at wastewater-irrigated fields: ratios in air and wastewater. Applied and Environmental Microbiology. 39(6): 1183- 1190. 60. Thomas, V., K. Herrera-Rimann, D. S. Blanc and G. Greub. 2006. Biodiversity of amoebae and amoeba-resisting bacteria in a hospital water network. Applied and Environmental Microbiology. 72(4): 2428-2438. 61. Thomas, V., J. F. Loret, M. Jousset and G. Greub. 2008. Biodiversity of amoebae and amoebae-resisting bacteria in a drinking water treatment plant. Environmental Microbiology. 10(10): 2728-2745. 62. Thomas, V., G. McDonnell, S. P. Denyer and J. Y. Maillard. 2010. Free-living amoebae and their intracellular pathogenic microorganisms: risks for water quality. FEMS Microbiology Reviews. 34(3): 231-259. 63. U.S. Environmental Protection Agency. 2004. Air quality criteria for particulate matter. U.S. Environmental Protection Agency. Washington DC. II. 64. United States Environmental Protection Agency. 1997. Exposure Factors Handbook Volume 1. O. o. R. a. Development. Washington DC, United States Environmental Protection Agency. 65. Wéry, N., V. Bru-Adan, C. Minervini, J.-P. Delgénes, L. Garrelly and J.-J. Godon. 2008. Dynamics of Legionella spp. and bacterial populations during the proliferation of L. pneumophila in a cooling tower facility. Applied and Environmental Microbiology. 74(10): 3030-3037. 66. Wilkes, C. R. 1999. Exposure to contaminants in drinking water. S. S. Olin. Washington DC, ISLI Press: 183-224. 67. Wilson, P. A. and J. Ferguson. 2005. Severe community-acquired pneumonia: an Australian perspective. Internal Medicine Journal. 35(12): 699-705. 68. Wullings, B. A., G. Bakker and D. van der Kooij. 2011. Concentration and diversity of uncultured Legionella spp. in two unchlorinated drinking water supplies with different concentrations of natural organic matter. Applied and Environmental Microbiology. 77(2): 634-641. 69. Zhang, Z., C. McCann, J. Hanrahan and A. Jencson. 2009. Legionella control by chlorine dioxide in hospital water systems. Journal American Water Works Association. 101(5): 12.

248 CHAPTER 9

CONCLUSION

TABLE OF CONTENTS

9.1 SUMMARY OF MAIN RESEARCH FINDINGS ______250 9.2 RESEARCH METHOD IMPROVEMENTS ______251 9.2.1 Sampling ______251 9.2.2 Culture identification techniques______251 9.2.3 Molecular identification techniques ______252 9.2.4 Microscopic identification techniques______253 9.3 FUTURE RESEARCH NEEDS ______253 9.4 REFERENCES ______256

Chapter 9. Conclusions

9.1 SUMMARY OF MAIN RESEARCH FINDINGS

Overall the research met the aims identified and contributes novel knowledge about the density and diversity of Legionella and FLA in drinking and recycled water. Further, combining the infection of the isolated FLA by L. pneumophila to produce a quantitative microbial risk assessment for common exposure scenarios of showering and garden hose use is valueable. Each component of the research has additional unique findings. Analysing the available literature on Legionella and FLA density and diversity in drinking and recycled systems (Chapter 2) initially highlighted the pervasiveness of FLA in water systems across the world. This research filled a gap in the present knowledge available on FLA in drinking water and established the need for further research on FLA interactions with pathogenic ARM. FLA were detected using both culture, molecular and microscopic methods (Chapter 3). Application of these methods to drinking and recycled water samples drew attention to the need for further method development because of the biases present in culture work, limited number of published qPCR primers and challenges in developing additional qPCR primers due to availability of viable positive control cultures. FLA and Legionella density and diversity were assessed in a recycled water system from water recycling plant (WRP) through to the distribution system using a modified Robins device (MRD) and appears to be the first study of its kind (Chapter 4). In comparison to the drinking water system, sampled in parallel from the MRD using qPCR, the recycled water system had slightly lower mean detection frequencies of Acanthamoeba sp. and H. vermiformis but the difference was not significant (Kruskal-Wallis test, p = 0.39) (Figure 4.12). The source of the H. vermiformis was likely due to cysts breaking through the WRP treatment process (Figure 4.11), emphasising the role of water treatment plants as a source of FLA in recycled water systems. Garden hoses have been neglected to date in research examining any microbial pathogens, despite being a common exposure pathway for infection. This research revealed that garden hose types and FLA presence played a significant role in the Legionella concentrations detected by qPCR (Figure 5.11). Specifically, the lilac hose type recorded mean Legionella densities (6.5  103 cells.mL-1) that were as high as any reported Legionella densities associated with Legionella outbreaks in the literature (Chapter 5). Heated annular reactors were set at water temperatures commonly used for showering (42 °C) and the finding drew primary attention to the limited roll of biofilm in accommodating Legionella and FLA despite containing comparable biofilm quantities to the ambient reactor (Chapter 6). 250 Chapter 9. Conclusions For all studies the density and diversity of FLA revealed that H. vermiformis was a dominant organism in drinking water systems that demands as much attention as Acanthamoeba sp. when researching with pathogenic ARM. Particularly, as at least one FLA isolate from each genera/species detected (Acanthamoeba sp., H. vermiformis and E. exudans) were found to be naturally infected with Legionella months after isolation. The density and diversity of Legionella revealed a large portion of the population were unclassified Legionella bacteria of unknown pathogenicity that need to be investigated. Infection of FLA isolated from the drinking and recycled water system demonstrated that Acanthamoeba sp. and H. vermiformis are infected at distinctly different rates with L. pneumophila (Chapter 7). These infection rates were unique to the particular drinking water system and were used in the QMRA scenarios to estimate the probability of human infection should L. pneumophila be introduced into the distribution system. The QMRA undertaken is a novel improvement on previous risk assessments as it incorporated biofilm as well as Legionella and FLA density and diversity from existent drinking water systems and applications (showering and garden hose use) (Chapter 8). Furthermore, the garden hose use scenarios and infection probabilities are unique, no comparable work has previously been published on this risk pathway. The behavioural changes recommended, in order to reduce the risk of Legionella infection for at risk groups, present a practical and low intervention solution to a potentially widespread problem.

9.2 RESEARCH METHOD IMPROVEMENTS

With the benefit of hindsight a number of changes to the methods and experimental set- ups would have been employed to improve the quality of this research while remaining within the original resource limitations.

9.2.1 Sampling

Sampling frequency during the garden hose and annular reactor experiments would have been improved by more even spacing over the course of the experiment and by adding two additional sampling periods. The additional data generated would have enabled better understanding of the population dynamics in the long time gap leading up to the last sampling point. Also, given the variability of FLA and Legionella densities more replicates should have been taken during sampling. 9.2.2 Culture identification techniques Detection of FLA by culture with NNA - E. coli overlay is the most common detection method used (Thomas and Ashbolt, 2011). However, in the current research the method 251 Chapter 9. Conclusions appeared to be positively biased for H. vermiformis. Furthermore, it has been estimated that between 20-30 % of FLA present in a sample are lost during isolation using this method (Rodriguez-Zaragoza, 1994). Alternative culture methods have been successfully used to isolate FLA; where water samples are incubated with a low nutrient source to allow for FLA and other protozoa to enrich by feeding on their natural prey also present in the water sample (Barbeau and Buhler, 2001; Rice et al., 2006). These latter approaches could effectively increase the diversity of FLA detected although quantification would be more time consuming as it would rely on counting of the protozoa present in the sample after initial inoculation. For Legionella detection by culture larger volumes maybe be required as no bacteria were successfully detected in 0.5 mL of water or re-suspended biofilm sample despite being detected by qPCR, however the qPCR could have been detecting DNA from non-viable cells. However, culture methods are frequently known to be less sensitive for the detection of Legionella than molecular methods (Wellinghausen et al., 2001; Devos et al., 2005; Joly et al., 2006; Diederen et al., 2007; Chen and Chang, 2010; Wullings et al., 2011) given that a number of Legionella are present in a VBNC state in drinking water (Lee et al., 2011).

9.2.3 Molecular identification techniques

For the detection of FLA targeted qPCR is still assessed as the best method as other molecular techniques, such as DGGE and tRFLP based on the 18S rRNA gene, are problematic as FLA are too phylogenetically diverse to design a single primer set to target them exclusively (Chapter 3). Additionally, the 18S rRNA gene does not present a good option for targeting FLA and it is recommended that other loci be explored such as the cytocrome c oxidase subunit I (COI) loci (Nassonova et al., 2010). In the short term a broader range of qPCR primers targeting common FLA, such as Echinamoeba and Vannella, would present a means to increase the diversity of FLA detected. However, it remains difficult to validate and optimise novel qPCR primers as culture collections of FLA from ATCC are not well stocked and contain organisms from more extreme environments and that require complex medium for cultivation. Also it is recommended that axenic cultures not be used for standard curve generation as they have been reported to contain ten times more DNA than xenic cultures (Mukherjee et al., 2008). The Legionella detection by qPCR may have been an underestimate of the actual densities as the primer set used (LEG 225 and LEG 858) where shown to be significantly less sensitive than other Legionella primers with environmental samples (Devos et al., 2005). As not all the qPCR products could be amplified there could be non-specific amplification present also. Another useful application of qPCR would have been targeting genes responsible for infection to determine possible pathogenicity of the Legionella bacteria present (Merault et al., 2011).

252 Chapter 9. Conclusions Determining the viability of the cells detected by qPCR in environmental samples is frequently achieved using propidium monoazide (PMA) (Wahman et al., 2008) or ethidium monoazide (EMA) (Chang et al., 2009; Chen and Chang, 2010) to bind any extracellular DNA. More information about the state of cells detected by qPCR could be determined using PMA or EMA pre-treatment and should provide an improvement on the method when used in conjunction with standard qPCR.

9.2.4 Microscopic identification techniques

With knowledge of the diversity and density of FLA and Legionella in the water and biofilm samples, cell staining and fluorescent microscopy could be used to target high density samples. Fluorescent in situ hybridisation (FISH) techniques have been used to detect ARM within FLA in environmental samples (Grimm et al., 2001) and experimental set-ups (Kuiper et al., 2004; Axelsson-Olsson et al., 2005; Declerck et al., 2009). The advantage of FISH is that it would allow visualisation of Legionella and FLA in situ and may assist with determining if the Legionella was infecting the FLA. However, a large quantity of optimisation for FISH would be required for biofilm samples and biofilm sectioning is likely to be required to gain sections that are thin enough for successful staining and imaging (Flood et al., 2000). Aside from improvements to the methods already employed in the current research there are a number of new technologies available that could be utilised in future research to attain much needed information on Legionella and FLA.

9.3 FUTURE RESEARCH NEEDS

This research has confirmed that Legionella and FLA are pervasive microorganisms in treated drinking water systems and applications, and QMRA supported the view that they could present a significant risk to human health. However, more research is needed to better understand FLA and Legionella populations as well as the epidemiology of Legionella infection in order to develop effective risk reduction strategies. Furthermore, it must be kept in mind that we know very little about the other eukaryotic microorganisms, such as other protists, fungi and nematodes, that may also be infected with a diverse range of pathogenic ARM and also prey on FLA within our treated water systems. To gather data critical to understanding FLA and Legionella populations future research needs to increase both the sampling breadth and depth. The benefits of sampling both water and biofilm from within drinking water distribution system and applications were elucidated in this research. Although, future research could gain breadth by sampling from a number of independent drinking water systems and applications. Additionally collecting larger samples

253 Chapter 9. Conclusions sizes (> 100) over a period of time would give the research the depth of data required to correlate Legionella and FLA populations with water quality characteristics and other distribution system variables. The cooperation of water utilities will be required to achieve the sampling desired. However, water utilities are corporate entities who are careful to manage the consumer perception of their product - drinking and recycled water. This position may be directly in conflict with the scientific researcher who wants to publish their research findings. Negotiation between these two positions needs to be established at the initial outset of the research planning, specifically acknowledging how to deal with results that may be too sensitive to be published. The methods used in this research to detect and identify FLA and Legionella had their disadvantages (Section 9.2). With advancements in molecular technology there now exists a number of next-generation sequencing platforms for high through-put DNA sequencing including Roche 454 sequencing, Illumina Solexa and Life Technologies SOLiD(Metzker, 2010). The advantages of these technologies are that thousands to millions of sequences reads can be produced to capture an entire population of microorganisms in water without the biases and time consuming methods of conventional PCR, cloning and automated Sanger sequencing chemistry, however the disadvantages include cost and short sequence lengths (Metzker, 2010; Shanks et al., 2011). Next generation sequencing technology has already been applied successfully to drinking water biofilm samples (Hong et al., 2010) and surface water samples (Rooks et al., 2010) and in both these studies a large number of unclassified microorganisms were revealed. The data produced from next generation sequencing results would allow for the full populations of FLA, other protozoa, Legionella and other pathogenic ARM to be determined from samples. This information would generate much needed understanding of the diversity of ARM and their potential hosts in drinking water. Furthermore, if a metagenomic approach was taken then the diversity of functional genes present for pathogenicity is likely to further highlight the range of ARM present in drinking water. The application of next generation sequencing does not permit the association of the detected ARM and their hosts to be determined. To ascertain that association it is proposed that fluorescence activated cell sorting (FACS) flow cytometry technology be utilised. Multi-channel FACS allows for fluorescently labeled cells to be sorted and captured based on their cell size and fluorescence. The technology has already been successfully applied to adenovirus detection in wastewater effluent (Li et al., 2010). In water samples FLA and select ARM could be labeled with fluorescent antibodies and then gated into populations of FLA cells that contained the select ARM and those that did not. The advantages of the technology is that the protozoa would be still viable after sorting and the location of the intracellular ARM could be visualised with microscopy. Furthermore, the sorted population could be sequenced using the next-generational 254 Chapter 9. Conclusions sequencing platforms to determined the exact composition of the ARM infecting their protozoan hosts. Obviously, a large volume of method development and optimisation needs to occur before this proposed application could be successful but the technique has great potential for this proposed application. Knowledge of the host and pathogenic ARM populations in drinking water does not directly translate into a realisation of the human health risks. The probability of Legionella infection is complex as emphasised by the large number of estimations and assumptions that were made during the risk assessment (Chapter 8). Furthermore, there are limitations with the animal dose-response model used particularly given the absence of infection severity variables for at risk populations groups and people with acquired immunity (Chapter 8). An alternative approach would be to link known pathogen densities in drinking water with actual recorded instances of infection for that population from that particular pathogen traced back to drinking water. Yet, this level of epidemiological detail is presently not collected except when an outbreak is detected, such as for Legionella infections (Table 1.1). There are some cases where a small number of infections have been traced to a source in the home using genotyping techniques (Euser et al., 2010) but that has been the exception. It is projected that if the etiological agents in CAP and flu cases were identified and then traced to the environmental sources of exposure then the health risks of water-based pathogenic ARM could be determined with greater confidence. Only after direct links to significant disease instances in the populations will there be the required incentives to introduce health campaigns, in-premise microbial reduction or point-of-use interventions to decrease exposure to pathogenic Legionella and other pathogenic ARM. To achieve these direct links it is recommended that more research collaborations be established between water microbial pathogen researchers and epidemiologists to produce the required data to determine the actual risk of water-based pathogenic ARM infections to advance upon the probabilistic risk of infection generated by QMRA.

255 Chapter 9. Conclusions

9.4 REFERENCES

1. Axelsson-Olsson, D., J. Waldenstrom, T. Broman, B. Olsen and M. Holmberg. 2005. Protozoan Acanthamoeba polyphaga as a potential reservoir for Campylobacter jejuni. Applied and Environmental Microbiology. 71(2): 987-992. 2. Barbeau, J. and T. Buhler. 2001. Biofilms augment the number of free-living amoebae in dental unit waterlines. Research in Microbiology. 152(8): 753-760. 3. Chang, B., K. Sugiyama, T. Taguri, J. Amemura-Maekawa, F. Kura and H. Watanabe. 2009. Specific detection of viable Legionella cells by combined use of photoactivated ethidium monoazide and PCR/Real-Time PCR. Applied and Environmental Microbiology. 75(1): 147-153. 4. Chen, N. T. and C. W. Chang. 2010. Rapid quantification of viable legionellae in water and biofilm using ethidium monoazide coupled with real-time quantitative PCR. Journal of Applied Microbiology. 109(2): 623-634. 5. Declerck, P., J. Behets, A. Margineanu, V. van Hoef, B. De Keersmaecker and F. Ollevier. 2009. Replication of Legionella pneumophila in biofilms of water distribution pipes. Microbiological Research. 164(6): 593-603. 6. Devos, L., K. Clymans, N. Boon and W. Verstraete. 2005. Evaluation of nested PCR assays for the detection of Legionella pneumophila in a wide range of aquatic samples. Journal of Applied Microbiology. 99(4): 916-925. 7. Diederen, B., C. d. Jong, I. Aarts, M. F. Peeters and A. v. d. Zee. 2007. Molecular evidence for the ubiquitous presence of Legionella species in Dutch tap water installations. Journal of Water and Health. 5(3): 375 - 383. 8. Euser, S. M., M. Pelgrim and J. W. den Boer. 2010. Legionnaires' disease and Pontiac fever after using a private outdoor whirlpool spa. Scandinavian Journal of Infectious Diseases. 42(11-12): 910-916. 9. Flood, J. A., N. J. Ashbolt and P. J. Beatson. 2000. Investigation of biofilms. Lancaster, Technomic Publishers. 10. Grimm, D., W. Ludwig, B. Brandt, R. Michel, K.-H. Schleifer, J. Hacker and M. Steinert. 2001. Development of 18S rRNA-targeted oligonucleotide probes for specific detection of Hartmannella and Naegleria in Legionella - positive Environmental Samples. Systematic and Applied Microbiology. 24(1): 76-82. 11. Hong, P.-Y., C. Hwang, F. Ling, G. L. Andersen, M. W. LeChevallier and W.-T. Liu. 2010. Pyrosequencing analysis of bacterial biofilm communities in water meters of a drinking water distribution system. Applied and Environmental Microbiology. 76(16): 5631-5635. 12. Joly, P., P.-A. Falconnet, J. Andre, N. Weill, M. Reyrolle, F. Vandenesch, M. Maurin, J. Etienne and S. Jarraud. 2006. Quantitative real-time Legionella PCR for environmental water samples: data interpretation. Applied and Environmental Microbiology. 72(4): 2801-2808. 13. Kuiper, M. W., B. A. Wullings, A. D. L. Akkermans, R. R. Beumer and D. van der Kooij. 2004. Intracellular proliferation of Legionella pneumophila in Hartmannella vermiformis in aquatic biofilms grown on plasticized polyvinyl chloride. Applied and Environmental Microbiology. 70(11): 6826-6833. 14. Lee, J. V., S. Lai, M. Exner, J. Lenz, V. Gaia, S. Casati, P. Hartemann, C. Lück, B. Pangon, M. L. Ricci, M. Scaturro, S. Fontana, M. Sabria, I. Sánchez, S. Assaf and S. Surman-Lee. 2011. An international trial of quantitative PCR for monitoring Legionella in artificial water systems. Journal of Applied Microbiology. 110(4): 1032-1044. 15. Li, D., M. He and S. C. Jiang. 2010. Detection of infectious adenoviruses in environmental waters by fluorescence-activated cell sorting assay. Applied and Environmental Microbiology. 76(5): 1442-1448. 256 Chapter 9. Conclusions 16. Merault, N., C. Rusniok, S. Jarraud, L. Gomez-Valero, C. Cazalet, M. Marin, E. Brachet, P. Aegerter, J. L. Gaillard, J. Etienne, J. L. Herrmann, the DELPH-I Study Group, C. Lawrence and C. Buchrieser. 2011. Specific Real-Time PCR for simultaneous detection and identification of Legionella pneumophila serogroup 1 in water and clinical samples. Applied and Environmental Microbiology. 77(5): 1708-1717. 17. Metzker, M. L. 2010. Sequencing technologies - the next generation. Nature Reviews Genetics. 11(1): 31-46. 18. Mukherjee, C., C. G. Clark and A. Lohia. 2008. Entamoeba shows reversible variation in under different growth conditions and between life cycle phases. PLoS Neglected Tropical Diseases. 2(8): e281. 19. Nassonova, E., A. Smirnov, J. Fahrni and J. Pawlowski. 2010. Barcoding amoebae: comparison of SSU, ITS and COI genes as tools for molecular identification of naked lobose amoebae. Protist. 161(1): 102-115. 20. Rice, E., W. Rich, C. Johnson and D. Lye. 2006. The role of flushing dental water lines for the removal of microbial contaminants. Public Health Reports. 121(3): 270-274. 21. Rodriguez-Zaragoza, S. 1994. Ecology of free-living amoebae. Critical Reviews in Microbiology. 20(3): 225-241. 22. Rooks, D. J., D. L. Smith, J. E. McDonald, M. J. Woodward, A. J. McCarthy and H. E. Allison. 2010. 454-pyrosequencing: a molecular battiscope for freshwater viral ecology. Genes. 1(2): 210-226. 23. Shanks, O., S. McLellan, S. Huse and M. Sogin. 2011. Characterization of microbial community structures in recreational waters and primary sources of faecel pollution with a next-generation sequencing approach. Environmental Microbiology. Current technology and water applications. K. Sen and N. Ashbolt. Norfolk, Caister Academic Press: 203. 24. Thomas, J. M. and N. J. Ashbolt. 2011. Do free-living amoebae in treated drinking water systems present an emerging health risk? Environmental Science & Technology. 45(3): 860-869. 25. Wahman, D., K. Kleier and J. Pressman. 2008. Monochloramine disinfection kinetics of Nitrosomonas europea using propidium monoazide quantitative real-time PCR. American Water Works Association Water Quality and Technology Conference, Cincinnati, Ohio, American Water Works Association. 26. Wellinghausen, N., C. Frost and R. Marre. 2001. Detection of Legionellae in hospital water samples by quantitative real-time light cycler PCR. Applied and Environmental Microbiology. 67(9): 3985-3993. 27. Wullings, B. A., G. Bakker and D. van der Kooij. 2011. Concentration and diversity of uncultured Legionella spp. in two unchlorinated drinking water supplies with different concentrations of natural organic matter. Applied and Environmental Microbiology. 77(2): 634-641.

257