1 Harmful Algal Blooms in Small Lakes

Harmful Algal Blooms in Small Lakes: Causes, Health Risks, and Novel Exposure Prevention

Strategies

Dissertation

Presented in Partial Fulfillment of the Requirements for the Degree Doctor of Philosophy in the

Graduate School of The Ohio State University

Igor Mrdjen, B.S.

Graduate Program in Public Health

The Ohio State University

2018

Dissertation Committee

Jiyoung Lee, Advisor

Christopher M. Weghorst

C.K. Shum

Qinghua Sun

Copyrighted by

Igor Mrdjen

2018

Abstract

The increasing frequency and severity of harmful algal blooms (HABs) has quickly

become an environmental health concern worldwide. As rates of anthropogenic nutrient use rise

with growing food demand, and as extreme precipitation events are expected to increase with the

changing climate, rates of nutrient influx into watersheds are expected to increase. With

increasing nutrient loads and temperatures in watersheds, many of the world’s water bodies and reservoirs are becoming eutrophic, establishing optimal conditions for HAB formation. HABs may change the ecology of water bodies; produce hypoxic zones resulting in fish deaths; and produce cyanotoxin compounds toxic to the liver, nervous system, and reproductive system of most eukaryotic organisms. Human and animal exposure to the most commonly occurring cyanotoxin, microcystin (MC), has been linked with hepatotoxicity, nausea, vomiting, and death in extreme circumstances. Within the MC family of toxins, microcystin-LR (MC-LR) is the most

common, possesses the highest toxicity, and is cited as a suspected carcinogen.

While HAB mitigation and MC exposure prevention efforts have often focused on large lakes and bodies of water, small lakes and ponds (SLaPs) remain understudied and unmonitored.

SLaPs are the most numerous lentic bodies of water worldwide, providing vital ecosystem services and biodiversity support. Due to their location, low volume, and seasonal water level changes, SLaPs are at an increased risk for eutrophication and HAB formation. SLaPs have many uses including recreation, aquaculture, irrigation, and even drinking water sources in

ii economically stressed areas of the world. The utility of SLaPs presents a potential exposure pathway to HABs and MC compounds. While our knowledge base of MC-related health

outcomes continues to grow, several gaps exist, including: low dose, chronic exposure outcomes;

differences in toxicity between pure toxic compounds and crude cyanobacterial extracts; and the

cancer promoting role of MCs. Finally, current satellite-based remote sensing methods

demonstrate several limitations in the study and monitoring of HAB formation in SLaPs,

including: land adjacency effect contamination, prohibitive costs, and inadequate spatial and

temporal resolutions. The overall goal of this work is to address the existing knowledge gaps

pertaining to HAB appearance in SLaPs, MC toxicity from chronic and acute exposures, and to improve the current monitoring methods.

Chapter 1 reviews the current knowledge base of HABs and cyanobacteria, the

production of toxins, and the mechanisms of MC toxicity and potential mechanisms of cancer

promotion. Various cyanobacterial genera were discussed in terms of different types of

cyanotoxin production. Contributions of environmental triggers of cyanotoxin production are

described and detailed.

Chapter 2 explores the impacts of anthropogenic land use and nutrient runoff on HABs in

24 SLaPs in central Ohio. Statistical analyses showed several relationships connecting

anthropogenic land use to increased nutrient concentrations in sampled SLaPs. In one of those 24

lakes, the impact of tile drainage was explored. Changes in prokaryotic and eukaryotic

communities, HAB occurrence, and high MC concentrations were noted.

Chapter 3 and Chapter 4 focus on the potential health outcomes in acute and chronic

exposure scenarios, respectively, using mice. Chapter 3 explores the changes in pathological,

iii

clinical chemistry, and gap junction intercellular communication (GJIC) attributed to acute MC-

LR exposures. Male and female CD-1 mice exposed to MC-LR via oral gavage were evaluated

for measures of changes in liver health. Findings show increased MC-LR toxicity in female mice

compared to males, and show that GJIC does not play a prominent role in MC-LR toxicity under

the experimental condition. Similarly, Chapter 4 explores chronic exposure to 10 µg/L MC-LR

and a crude M. aeruginosa Lysate in a two-stage mouse model of cancer promotion. Mice were initiated with a known carcinogen, and administered MC-LR or cyanobacterial Lysate via ad libitum drinking water consumption. Results show increased toxicity of cyanobacterial Lysate and the potential promotion of liver cancer by MC compounds.

Chapter 5 summarizes a proof-of-concept of a novel approach to HAB monitoring in near-shore lake environments based on remote sensing technologies. An unmanned aerial vehicle

(UAV) equipped with multispectral cameras was deployed for remote sensing of near-shore areas of Lake Erie and Buckeye Lake. Image derived Normalized Difference Vegetation Index

(NDVI) values were compared to lab analyzed water samples. The feasibility of UAV-based remote sensing is discussed and solutions to future implementations are proposed.

Dedication

This work is dedicated to my family, friends, and mentors, who have invested a great deal of time and energy to allow me the opportunity to follow my passions.

Acknowledgments

This work would not have been possible without the guidance of my advisor and lab- mates, support of my mentors, and the help of my family and friends.

I would like to acknowledge the incredible contributions of my advisor Dr. Jiyoung Lee.

Her patience, support and guidance were invaluable during my time at The Ohio State University

(OSU). Dr. Lee is an incredible advisor and role model who ensured a professional, friendly, and educational environment in lab and in the classroom. Additionally, I would like to thank my past and current peers & lab-mates: Seungjun Lee, Lindsay Collart, Tyler Gorham, Claire Bollinger,

Alba Mayta, Manjunath Manubolu, Chenlin Hu, Xuewen Jiang, Alina Yang, Yuehan Ai,

Yuanyuan Jia, and Matt McCrink. The training, experience and friendships they provided were essential to my development.

I am forever grateful for the support and patience provided to by my committee and various collaborators during my studies at OSU. I would personally like to thank Dr. Christopher

M. Weghorst for his wiliness to mentor me and introduce me to areas of research which were previously unknown to me. Special thanks also go out to my collaborators Dr. C.K. Shum, Dr.

Jim Gregory, Dr. Kellie Archer, Dr. Michael Pennell, Dr. Mark Morse, and Dr. Randall J. Ruch, for all of their contributions toward my education and the completion of my research. I wish to thank Dr. Michael Bisesi and Dr. Amy Ferketich for mentoring me and supporting me in my development as an instructor. I would also like to acknowledge Dr. Thomas J. Knobloch, who

vi was there at every step of the way to offer guidance, advice, friendly conversation, and much- needed coffee breaks.

Finally, special thanks go to my parents, who sacrificed everything imaginable to provide me with the opportunity at a great education and the chance to follow my dreams. Their love and support will forever be welcomed. Finally, I would like to thank my brother and friends for always pushing me to succeed and providing an outlet for the stress that comes with ambition.

vii

Vita

2010...... Normandy High School

2014...... B.S. Biology

Baldwin Wallace University

2014 to present ...... Graduate Program in Public Health

Specialization: Environmental Health Sciences

The Ohio State University

Publications

Mrdjen, I, Fennessy, S, Schaal, A, Slonchzewski, J, Lee, J. Tile Drainage and Anthropogenic

Land Use Contribute to Harmful Algal Blooms and Microbiota Shifts in Inland Water

Bodies, Environmental Science and Technology, 2018, in press.

Mrdjen, I. & Lee, J. Simple and Practical On-Site Treatment of High Microcystin Levels in

Water Using Polypropylene Plastic for Protection of Human, Animal, and Ecosystem

Health, Journal of Environmental Science and Health Part A, (2018) in press.

Mrdjen, I & Lee, J. High volume hydraulic fracturing operations: potential impacts on surface

water and human health, International Journal of Environmental Health Research, 2016:

26:4, 361-380, DOI: 10.1080/09603123.2015.1111314

viii

Field of Study

Major Field: Public Health

Table of Contents

Abstract ...... ii Dedication ...... v Acknowledgments...... vi Vita ...... viii List of Tables ...... xiv List of Figures ...... xv Chapter 1. Introduction ...... 1 1.1 Abstract ...... 1 1.2 Cyanobacteria and Harmful Algal Blooms ...... 2 1.3 Cyanotoxin Production ...... 6 1.4 Microcystins ...... 8 1.5 MC Toxicity ...... 12 1.6 Other Cyanotoxins ...... 15 1.7 MC Exposures ...... 21 1.8 The Future of Harmful Algal Blooms ...... 23 1.9 Conclusions ...... 23 Chapter 2. Tile Drainage and Anthropogenic Land Use Contribute to Harmful Algal Blooms and Microbiota Shifts in Inland Water Bodies ...... 25 2.1 Abstract ...... 25 2.2 Introduction ...... 27 2.3 Materials & Methods ...... 29 2.3.1 Sample Collection ...... 29 2.3.2 Site Characterization ...... 31 2.3.3 Water Sample Processing ...... 31 2.3.4 Water Chemistry ...... 32 x

2.3.5 Cyanotoxin Measurement ...... 32 2.3.6 DNA Extraction ...... 32 2.3.7 PCR Inhibition Testing ...... 33 2.3.8 PCR detection of Microcystis aeruginosa and Planktothrix sp...... 33 2.3.9 Microbial Source Tracking (MST) ...... 38 2.3.10 Microbial Community Analysis ...... 38 2.3.11 Statistical Analysis ...... 39 2.4 Results ...... 39 2.4.1 Analysis of Factors Linked to HAB-friendly Conditions ...... 39 2.4.2 HAB Conditions, Toxin concentrations, and Microbial Communities in FC ...... 43 2.4.3 Water Chemistry ...... 47 2.4.5 Microbial Source Tracking ...... 47 2.4.6 Microbial Community Analysis ...... 49 2.5 Discussion ...... 55 Chapter 3 Comprehensive Determination of Acute MC-LR Exposure on Male and Female Murine Liver Health ...... 61 3.1 Abstract ...... 61 3.2 Introduction ...... 62 3.3 Methods...... 65 3.3.1 Chemical Compounds ...... 65 3.3.2 Animals ...... 66 3.3.3 Animal Husbandry ...... 66 3.3.4 Experimental Protocol ...... 67 3.3.5 Animal Observations ...... 70 3.3.6 Euthanasia and Necropsy ...... 70 3.3.7 Unscheduled Deaths...... 70 3.3.8 Clinical Chemistry Parameters ...... 71 3.3.7 Tissue Collection, Preservation, and Histopathology ...... 73 3.3.8 Florescent Dye Cut-Loading ...... 73 3.3.9 Statistical Analysis ...... 75 3.4 Results ...... 75 3.4.1 Mouse Mortality, Health, and Clinical Observations ...... 75 3.4.2 Clinical Chemistry ...... 76 xi

3.4.3 Histopathology ...... 83 3.4.4 Gap Junction Intercellular Communication ...... 85 3.5 Discussion ...... 87 Chapter 4. Chronic Exposure to Environmentally Relevant Levels of Microcystin and its Role in Liver Cancer Promotion: A Two-Staged Carcinogenesis Model Study ...... 91 4.1 Abstract ...... 91 4.2 Introduction ...... 92 4.3 Methods...... 94 4.3.1 MC-LR ...... 94 4.3.2 Lysate Preparation ...... 95 4.3.3 M. aeruginosa Lysate Characterization ...... 97 4.3.4 Microcystin Adsorption to Plastic Assessment ...... 97 4.3.5 Animal Husbandry ...... 98 4.3.6 Experimental Design ...... 98 4.3.7 Gross and Histopathological Examination...... 100 4.3.8 Statistical Analyses ...... 100 4.4 Results ...... 100 4.4.1 M. aeruginosa Lysate Characterization ...... 100 4.4.2 Microcystin Adsorption by Plastic...... 103 4.4.3 Health Metrics and Dosing ...... 105 4.4.4 Lesion Counts ...... 108 4.4.5 Mortality ...... 114 4.5 Discussion ...... 116 Chapter 5. Early Warning of Near-Shore Harmful Algal Blooms Using Unmanned Aerial Vehicle (UAV) Technology: A Proof-of-Concept Study ...... 119 This chapter has been submitted for review and publication in Harmful Algae...... 119 5.1 Abstract ...... 119 5.2 Introduction ...... 120 5.3 Methodology ...... 123 5.3.1 UAV ...... 123 5.3.2 Multispectral Cameras ...... 126 5.3.3 Sample Collection Device...... 127 5.3.4 Sampling and Flight Paths ...... 127 xii

5.3.5 Image Processing and Analysis ...... 128 5.3.6 Water Filtration ...... 128 5.3.7 DNA Extraction ...... 128 5.3.8 Presence of PCR Inhibitors ...... 129 5.3.9 qPCR Quantification of Cyanobacterial Genes ...... 129 5.3.10 Quantification of Total Microcystins ...... 130 5.3.11 Quantification of Cyanobacterial Pigments ...... 130 5.3.12 Data Analysis ...... 130 5.4 Results ...... 131 5.5.1 Sample Collection and Processing ...... 131 5.4.2 NDVI and Cyanobacterial Pigment Concentrations ...... 134 5.4.3 NDVI and PCR Determined Cyanobacterial Gene Concentrations ...... 136 5.4.4 Concentrations of Microcystin ...... 139 5.5 Discussion ...... 139 5.5.1 UAV Image Collection ...... 139 5.5.2 Image Processing ...... 140 5.5.3 NDVI and Lab Data ...... 141 Chapter 6. Conclusion ...... 143 Bibliography ...... 145 Appendix A. Supplementary Information...... 161

xiii

List of Tables

Table 1.Overall costs of cyanobacterial bloom management in Australia, in $ millions per year. Adapted from Atech Group & Muray-Darling Basin Commission (2000). 24 ...... 5 Table 2. Cyanotoxins and factors regulating their production. Adapted from Boopathi & Ki (2014).40 ...... 9 Table 3. qPCR Primer and Probe Sequences ...... 34 Table 4. qPCR Reaction Composition ...... 35 Table 5. qPCR Assay Conditions...... 36 Table 6. qPCR Assay Quantification Curves ...... 37 Table 7. Microcystis aeruginosa and cyanotoxin concentrations of FC samples (Jun-Aug 2015)...... 46 Table 8. Experimental Design for Phase A ...... 68 Table 9. Experimental Design for Phase B ...... 69 Table 10. Clinical Chemistry Parameters Examined During Study ...... 72 Table 11. Male clinical chemistry parameters...... 78 Table 12. Female clinical chemistry parameters...... 80 Table 13. Maximal Magnitude of Change in Clinical Chemistry Parameters ...... 82 Table 14. CT media composition. All ingredients were mixed and autoclaved simultaneously. . 96 Table 15. Lesion count and incidence as determined by gross examination...... 109 Table 16. Mean lesion counts per area of tissue (lesion count/mm2)...... 110 Table 17. Operational parameter demands of UAV used for sample collection...... 124 Table 18.Mean microcystin (MC) concentrations (µg/L), gene concentrations (log [gene copies/mL]), and NDVI values across locations...... 138 Table 19. Water chemistry data of samples collected in various Knox County lakes over a 3- month period...... 162

xiv

List of Figures

Figure 1. Bloom coverage area (percentage by county) in the US in 2005 as estimated by MERIS. Adapted from Zhang et al. 2015.14 ...... 3 Figure 2. Common cyanobacterial genera responsible for cyanotoxin production. Adapted from Zurawell et al (2011) and Boopathi & Ki (2014).32,33 ...... 7 Figure 3.Microcystin chemical structure with variable amino acid positions X and Z. Adapted from Nishizawa et al (2001). 33 ...... 10 Figure 4. Interaction of MC-LR and PP1. Cys273 of PP1 binds covalently to the Mdha residue of MC-LR. Adapted from MacKintosh et al (1995).50 ...... 13 Figure 5. Nodularin chemical structure and variable region. Adapted from Kelker et al (2009). 70 ...... 16 Figure 6. Chemical structure of cylindrospermopsin. Adapted from Pearson et al (2010). 77...... 18 Figure 7. Chemical structure of saxitoxin (A) and anatoxin (B). Adapted from Araoz et al (2010).83 ...... 20 Figure 8. Sources of nutrient input into small lakes and ponds (SLaPs).115 ...... 26 Figure 9. Map of sampling sites in Knox County, OH. Locations were chosen to be representative of urban and agriculturally impacted bodies of water. Ohio State Plane North, NAD1983 (Arc Map 10.4.1, ESRI 2016).115 ...... 30 Figure 10. Comparison of microcystin-producing gene concentrations in SLaPs with tile drainage or animal presence (Y), and those without either factor (N). Sites were compared based on concentrations of the mcyE gene (gene copies/mL)...... 40 Figure 11. Principal component analysis of data sets showing relationships between land use, bacterial communities, and environmental parameters in the sampled locations. The first two principle components explain 50.4% of the variability. A). SLaPs were cluster by land use intensity. B) Strength of relationships between variables (vectors) used in the PCA; vector length indicates strength of correlation. Nutrient concentrations were correlated along the first principle component (x-axis) with HAB-forming cyanobacteria of the Microcystis, Planktothrix, and Aphanizomenon genera. Sites 8-11 were indicative of FC transects, and were therefore not used for PC analysis, instead a representative sample (#12) of FC on July 28th was used...... 42 Figure 12. Mean Microcystis aeruginosa concentration and maximal MC production from sampled locations. Samples were analyzed using qPCR methods for total Microcystis (PC-IGS) and Microcystis capable of synthesizing MC toxin (mcyE). All MC concentrations were reported as maximal levels detected during the sampling period (Jun – Aug 2015) apart from FC whose maximal MC was 876 µg/L, an extreme outlier...... 44 Figure 13. Mean Planktothrix mcyE gene concentrations by site. Samples analysis was conducted via ddPCR for Planktothrix sp. capable of synthesizing MC toxin (mcyE). While some variation

exists across samples, mcyE concentrations were similar throughout the samples, mcyE concentrations were similar throughout the sampling sites...... 45 Figure 14. qPCR determined host-specific fecal bacterial load by location. Nearly all locations contained human- and goose-specific fecal markers. The canine fecal marker was not found in any of the sampled locations, while only one contained the ruminant marker...... 48 Figure 15. The taxonomic composition of the Eukaryotic microbial community in FC at three time-points was shown to vary drastically. Figure A corresponds to the June 18th, figure B corresponds to July 7th, and figure C corresponds to the July 28th sampling date. The largest changes were seen in the Eukaryotic class Oligohymenophorea, a class of ciliated organisms.115 ...... 50 Figure 16. Prokaryotic microbial community structure of FC during the sampling period, as determined by 16S and 18S Barcoding analysis. (A) June 18, 2015; (B) July 7, 2015; and (C) composite result from the four transect samples on July 28, 2015.115 ...... 52 Figure 17. MC concentration and microbial communities of FC by transect. Samples are representative of the total microbial community in each transect, as determined by 16S/18S sequencing. The prokaryotic community was largely dominated by the cyanobacterial genus Planktothrix, while the dammed transect of the lake also included a large proportion of Aphanizomenon (22.7%).115 ...... 54 Figure 18. Total Microcystis and toxin-producing Microcystis concentrations during a 2-month period (2015) in FC. Samples were collected at three time points, during which HABs were reported. The final sampling date (July 28th) is a mean of four transect samples of the lake...... 58 Figure 19. Quantification of fluorescent dye diffusion across cells via cut-loading assay...... 74 Figure 20. CD-1 male mouse clinical chemistry parameters by exposure group. Groups were compared to the 0 µg/kg/bw group using Dunn’s test. Differences in results were significant at a levels of 0.05 (*), 0.01(**), and 0.001(⁂)...... 79 Figure 21. CD-1 female mouse clinical chemistry parameters by exposure group. Groups were compared to the 0 µg/kg/bw group using Dunn’s and Dunnett’s tests. Differences found via Dunn’s test were significant at a levels of 0.05 (*) and 0.01(**)...... 81 Figure 22. Pathological findings in mice exposed to varying MC-LR concentrations. Four pathological parameters were noted across groups: A) Hypertrophy B) Necrosis C) Degeneration D) Hemorrhage...... 84 Figure 23. Hepatocellular gap junction intercellular communication (GJIC) as measured by mean dye-positive cell count. No significant differences in gap junction communication were seen across treatments or sexes...... 86 Figure 24. Experimental design and timeline...... 99 Figure 25. Chemical composition of M. aeruginosa Lysate as Determined via LC-MS. Proportions of MCs in the cyanobacterial Lysate were as follows: MC-RR (37%); [D-Asp3] MC- RR (24%); MC-LR (22%); and MC-YR (17%)...... 102 Figure 26. Adsorption of microcystins onto drip bottle plastic. We noted no significant loss or adsorption of MC-LR or total MCs in the cyanobacterial Lysate in the drip bottles, over a 7-day period...... 104 Figure 27. Mean water consumption, per mouse, over study duration. There were no significant differences in water consumption across treatment groups...... 106

xvi

Figure 28. Mouse liver weight/ body weight ratios by treatment group. No significant differences in liver weight to body weight ratios were seen across treatment groups...... 107 Figure 29. Mean lesion density (lesion/mm2) across treatments. There were no significant differences in lesion counts across treatment...... 111 Figure 30. Representative gross lesions in the DEN & water only (A,B) treatment group; Liver Lesions from the Lysate (C) and MC-LR (D) exposure groups, respectively...... 112 Figure 31. Lesion proportions by classification across treatment methods. Histopathological lesion classification showed a significantly higher proportion of lesions classified as “carcinoma” in the Lysate (55%; p< 0.01) and MC-LR (44.5%; p<0.05) groups, compared to the ingestion of water (14.9%)...... 113 Figure 32. Kaplan-Meier determined mouse survivability estimate by treatment. Chronic ingestion of cyanobacterial Lysate produced a significantly higher mortality rate (35.6%) than ingestion of MC-LR (4.3%) and water (3.2%), in mice...... 115 Figure 33. Design of UAV, imaging and sampling instruments. The DJI S1000+ carbon fiber frame (A) was fitted with four multispectral cameras (B) for image data collection. Upon image collection, the UAV was fitted with a water sampling apparatus (C) and used for water sample collection...... 125 Figure 34. Two successive images collected during a random sampling flight (Buckeye Lake). Images were intended to represent the same area of water sampled, yet were shown to be significantly different in practice...... 132 Figure 35. Pitch and roll dynamics during a random sampling flight. A graph of the flight dynamics seen during a typical image collection flight. The random nature of wind interference makes image stabilization difficult...... 133 Figure 36. Influence of the time variable, phycocyanin, and chlorophyll-a concentrations on NDVI values of Buckeye Lake and Lake Erie. Increased phycocyanin concentrations were correlated with increased NDVI values in Buckeye Lake. Increased chlorophyll-a and decreased phycocyanin were correlated with increasing NDVI values in Lake Erie...... 135 Figure 37. Influence of time, M. aeruginosa mcyE, and Planktothrix sp. mcyE on NDVI values of Buckeye Lake and Lake Erie. No significant correlation was seen between NDVI values and genetic markers of cyanobacteria...... 137

xvii

Chapter 1. Introduction

1.1 Abstract

Harmful algal blooms (HABs) are classified as extreme proliferations of cyanobacteria

which can produce negative impacts on ecological, human, and animal health. Cyanobacterial

HABs most readily form in nutrient rich fresh water bodies. The production of cyanotoxins by

some of the cyanobacteria adds further complexity to the impacts of HABs as cyanotoxins are

capable of generating various health outcomes, such as neurotoxicity and hepatotoxicity.

Amongst the commonly seen cyanotoxins (i.e. nodularin, cylindrospermopsin, etc.), the most

prevalent toxin produced in HABs is the family of hepatotoxic cyclic peptides: microcystins

(MC). MCs are transported across the cellular membrane and bind to protein phosphatases 1 and

2A (PP1/2A). In acute exposures, the inhibition of PP1/2A results in damage to cellular

structures and activation of apoptotic pathways. Additionally, MC produces reactive oxygen

species (ROS), initiates cytochrome-c release, and activates caspases. Chronic exposure to MCs

has been linked to cancer promotion due to increased cellular proliferation via the activation of

the diacylglycerol (DAG) pathway, leading to the activation of the B cell lymphoma 2 (Bcl-2)

protein and protein kinase C (PKC). With the occurrences of HABs expected to rise due to

changing climate conditions and increased anthropogenic land use, a more complete

understanding of HABs, MC toxicity, and exposure prevention methods is needed.

1.2 Cyanobacteria and Harmful Algal Blooms

Cyanobacteria are a morphologically diverse phylum of Gram-negative oxygenic prokaryotes found in various environments around the world.1,2 The first appearance of cyanobacteria was thought to have been 2.3–2.9 billion years ago, but some studies have identified fossilized cyanobacteria dating back 3.5 billion years.1,3,4 These photoautotrophic organisms played a vital role in the development of an oxygen-rich atmosphere on Earth due to their hardiness.5,6 Most crucially, their ability to fix atmospheric nitrogen, solubilize phosphorous and sequester iron allows cyanobacteria to thrive in nearly all environments.7 Cyanobacteria have been found to thrive and colonize in aquatic, terrestrial, and even aerial environments, but are most commonly found proliferating in bodies of water rich in nutrients.8 The nutrient–rich conditions necessary to support cyanobacterial blooms are most commonly established in the presence of anthropogenic influences such as nutrient deposition from agricultural sources.9,10

Blooms of this nature have been noted on every continent aside from Antarctica and in countries including the USA, Canada, Brazil, Serbia, and China.7,11–13 In fact, 62% of USA counties showed signs of blooms when measured using satellite remote sensing technology (Figure 1).14

Figure 1. Bloom coverage area (percentage by county) in the US in 2005 as estimated by MERIS. Adapted from Zhang et al. 2015.14

Cyanobacterial harmful algal blooms (HABs) present various environmental health and

socio-economic challenges. The formation of large blooms can impact the ecology of affected lakes during growth and death stages.11 As cyanobacteria compete for resources with established phytoplanktonic organisms, changes in microbial community structures are expected to occur.15

This trend has already been reported with Daphnia magna, a grazer whose population health is

negatively impacted by bloom occurrence.16 Further, during bloom deaths, decomposition of

cyanobacterial cells can deprive waters of oxygen content, resulting in large hypoxic zones in

water bodies.17 Hypoxic zones have been shown to result in the death of large organisms such as

fish.18

Occurrences of such large death events influences public perception regarding the health and integrity of water bodies.19 In regions experiencing blooms, economic impacts of bloom

management, opportunity loss, and water treatment carry high costs (Table 1). Due to HABs in

the Maumee River and the western basin of Lake Erie, the city of Toledo, OH, was forced to

invest over $250 million in improved water treatment methods to address cyanobacterial matter

in source waters.20 These increased measures aim to curb presence of HAB produced odorants,

address the need for higher quantities of disinfection agents, and reduce public exposure to

cyanotoxins.21–23

Table 1.Overall costs of cyanobacterial bloom management in Australia, in $ millions per year. Adapted from Atech Group & Muray-Darling Basin Commission (2000). 24

Cost Category Cost ($)

Joint Management costs 9 Urban extractive users 35 Rural extractive users 30 Non-extractive users 76-136 Total 180-240

1.3 Cyanotoxin Production

Cyanotoxins are a group of secondary metabolites produced by various cyanobacteria,

which can result in toxicity in humans, animals, and plants.22,25,26 Increasing eutrophication of

the world’s water bodies by anthropogenic impacts has led to an increase in frequency and

severity of HABs producing cyanotoxins.2,27 The production of cyanotoxins has been postulated to aid in reducing grazing pressures from zooplankton and protozoa on cyanobacterial colonies, allowing for greater survival rates.28–30 Studies have also found that MC binds to RuBisCO in cyanobacteria, acting as a form of protection for oxidative stress generated due to high UV exposure.31 Multiple species of cyanobacteria are responsible for the production of cyanotoxins,

with each compound possessing various toxicities and impacts on eukaryote systems (Figure 2).

While the most commonly produced cyanotoxins are microcystins, others have also been of

concern.32

Figure 2. Common cyanobacterial genera responsible for cyanotoxin production. Adapted from Zurawell et al (2011) and Boopathi & Ki (2014).32,33

1.4 Microcystins

Microcystins (MCs) are a family of cyclic peptides produced by several genera of cyanobacteria (Figure 2; Figure 3). Each unique congener of MC has differing toxicities based

on amino acids incorporated into the variable (X) and (Z) positions (Figure 3).33As a secondary

metabolite, MC production depends on many genetic and environmental factors.34 Genetic

variation across species and strains of cyanobacteria play a vital role in determining the congener

and toxicity of MC produced. The genes necessary for MC synthesis can be found in multiple

genera, including: Microcystis, Planktothrix, Anabaena, and Nostoc.35–37 Genetic sequencing of these genera has shown that the mcy region is responsible for the production of MC compounds.

The mcy gene is comprised of 10 open reading frames, mcy A to J, which are divided into two operons (mcy A to C and mcy D to J) by a promoter region.38,39 The expression of these genes is driven by environmental stimuli, as many strains of cyanobacteria possess these genes, but do not produce MC (Table 2).39

Table 2. Cyanotoxins and factors regulating their production. Adapted from Boopathi & Ki (2014).40

Regulatory Gene Toxin Up-Regulators Down-Regulators Cluster active photosynthesis N-limitation MCs mcy Ferrous Iron increased nitrate concentration high light intensity Nitrogen fixation Ammonia supplementation Phosphate NODs nda limitation High salinity light stress high inorganic nitrogen high temperatures lack of fixed N ammonia as sole N-source source phosphate short period of high light CYNs cyr/aoa limitation intensity long periods of phosphate limitation high light inensity N-starvation sub-optimal light sub-optimal ATXs ana high temperatures temperatures green algal extract presence high light intensity high temperatures High Nitrogen sub-optimal STSx sts Dark conditions temperatures Extracellular salt

(NaCl)

Variable Regions Figure 3.Microcystin chemical structure with variable amino acid positions X and Z. Adapted from Nishizawa et al (2001). 33

Many studies have explored how various environmental factors influence mcy gene

expression and the production of microcystin toxins in various genera of cyanobacteria. A

benchtop study by Kaarina Sivonen (1990)41 explored the impacts of bacteria, temperature, light

and nutrient concentrations on MC-RR production and growth of Oscillatoria agardhii cultures.

The study found two of three toxic cultures produced more MC in axenic conditions. Cultures

with low light conditions and high nitrogen concentrations also produced high MC content.

Below optimal concentrations of phosphorus also increased MC production, but significantly higher concentrations did not impact MC production. Optimal temperature for growth and MC production was 25°C, with the lowest toxin production at 30°C.

Environmental studies have also shown MC concentrations were correlated with warmer water temperatures and increased nitrogen and phosphorus content.42,43 Further, MC has been

found to bind to RuBisCO as a defense mechanism for oxidative stress in high light intensity settings.31 Although correlations between MC content and nutrient concentrations are present,

most are driven by the increase of biomass concentration, which naturally leads to higher MC

production.42 Additionally, studies have found that the form of nitrogen present in the

environment influences cyanobacterial bloom composition and, therefore, changes the congener

and concentration of MC produced.44,45 Microcystis and Planktothrix species can use various

forms of nitrogen for the production of MC compounds. While our current understanding of MC

production is expanding, there are many conflicting findings across genera, species, and strains

of cyanobacteria. Therefore, our understanding of HAB-MC production dynamics needs to be

expanded.

1.5 MC Toxicity

The toxicity of MC compounds is attributed to two characteristics: 1) the transport of MC

across cellular membranes via organic anion transport polypeptides (OATP); and 2) unique

chemical structure of MC. Upon entering the gastrointestinal tract, MC is absorbed through the

small intestine and reabsorbed into the portal blood stream. The portal blood stream transports

the compound to the liver to be metabolized. Upon reaching the liver, MC compounds are

transported across the hepatocellular membrane by the OATP superfamily of polypeptides

(OATP1B2, OATP1B1, and OATP1B3).46,47 A similar transport mechanism (OATP1A) also allows the compound to cross the blood-brain barrier.48

MC incorporated into hepatocytes produces cellular damage via several pathways. The most prominent pathway is the inhibition of protein phosphatases 1 and 2A (PP1/2A) via the binding of the MC Mdha carbonyl group to PP1/2A, and the added interaction of the ADDA moiety, and the glutamyl carboxylate group of MC with PP1/2A structures. (Figure 4).49 This

interaction is initially reversible, but quickly becomes a covalent bond.50,51 The inhibition of

PP1/2A results in the hyperphosphorylation of microtubule units in the cell, causing damage leading to apoptosis.52 The most toxic congener of MC is MC-leucine, arginine (MC-LR), which incorporates the amino acids leucine and arginine into the X and Z variable positions, respectively (Figure 3).32

Figure 4. Interaction of MC-LR and PP1. Cys273 of PP1 binds covalently to the Mdha residue of MC-LR. Adapted from MacKintosh et al (1995).50

Other pathways of MC toxicity also exist. The production of reactive oxygen species

(ROS) due to MC exposure has also been reported.53 ROS is capable of damaging cellular

structure and activating intracellular mechanisms of apoptosis.54 The production of ROS has

been linked to the interaction of MC and the mitochondrial membrane, resulting in

mitochondrial-DNA damage and leakage of mitochondrial contents into the cytoplasm.55,56 MC-

LR has also been implicated in the disruption of p53 function, which has been shown to be vital in initiating apoptosis in MC exposed cells.57,58 Finally, MC interaction with the intrinsic

apoptotic pathway during acute exposures triggers apoptosis by altering expression of Bcl-2 proteins, proapoptotic genes, activating caspases and initiating lysosome involvement.54,59,60

In low-dose, chronic exposures, MC and the inhibition of PP1/2A activates the

diacylglycerol (DAG) pathway.61 The resulting interaction favors cellular survival and

proliferation by activating Bcl-2 proteins and protein kinase C (PKC).51 The favoring of cellular

survival can lead to proliferation of mutated cells and cancer promotion.54,62 Furthermore,

chronic exposure to MC-LR has been shown to promote hepatic inflammation and cause

nonalcoholic steatohepatitis disease.63 As a result of the production of ROS, increased

inflammation in exposed organs, and induction of cellular proliferation pathways, MC-LR has

being classified as a potential human carcinogen by the International Agency for Research on

Cancer (IARC).64 However, these impacts are not localized to the hepatic cells as MC has also

been found to cause toxic responses in reproductive, pulmonary, and neural cells.65–68

Our understanding of MC toxicity and exposure outcomes is growing but requires

substantial expansion. While MC-LR is the most potent and prevalent congener of MC, exposure

to multiple MC congeners is rarely explored.32 While studies have hypothesized higher toxicities

of MC-LR compared to complex mixtures containing multiple MCs69,70, others have reported

increased toxicity of cyanobacterial Lysates.71 These findings indicate that our understanding of

outcomes due to combinatorial exposure to MCs and bioactive compounds is incomplete and

requires further expanding.

1.6 Other Cyanotoxins

Besides MC, various other cyanotoxins with unique properties and toxicities are

produced by cyanobacterial species (Figure 2). Nodularin (NOD) is a hepatotoxin produced by

Nodularia and Nostoc species, and is perhaps the most similar to MC in terms of toxicity, structure, and function (Figure 5).72,73 Much like MC, NOD’s arginine variant (NOD-R) has an

74 LD50 range of 50-150 µg/kg, and a similar mechanism of hepatotoxicity. Expression of the nda

region in these cyanobacterial species promotes NOD production (Figure 2).75 The up-regulation

and down-regulation of NOD production is controlled by several environmental factors, but in

general is dependent on phosphate and nitrogen concentrations, light intensity and temperature

(Table 2).40

Variable Region

Figure 5. Nodularin chemical structure and variable region. Adapted from Kelker et al (2009). 70

Unlike the cyclic peptides of MC and NOD, cylindrospermopsin (CYN) is a water

soluble, polar alkaloid.76 CYNs are most readily produced by Aphanizomenon and Anabaena sp.

and are regulated by the cyr gene cluster. The up- and down-regulation of cyr is related to light

intensity, nutrient type and nutrient availability (Table 2). CYNs’ toxicity is attributed to a tricyclic guanidine unit and a uracil region in the compound (Figure 6).77 Some studies have

shown that CYN have a relatively short half-life of 10 days in high purity water, and only 3 days when exposed to sunlight, while others attest to minimal degradation even after 40 days.76,78

CYNs mainly impact the liver and kidney, but have shown lung and intestine toxicity as well.79

The main mechanism of toxicity is cytochrome P450 and glutathione inhibition which leads to improper metabolism of toxic compounds.79–81 CYN has also been shown to be genotoxic and

result in fetal toxicity.81,82

Figure 6. Chemical structure of cylindrospermopsin. Adapted from Pearson et al (2010). 77

Finally, two commonly produced cyanotoxins with neurotoxic properties are anatoxin

(ATX) (Figure 7A) and saxitoxin (STX) (Figure 7B).83 While both alkaloids vary in chemical

composition, both are most commonly produced by cyanobacteria in the Planktothrix,

Microcystis, Anabaena, and Aphanizomenon genera (Figure 2). ATX’s are produced with the expression of the anaA-anaG gene cluster, while STX is produced by a biosynthesis process involving 33 unique genes present across various clusters.84–86 The expression of gene clusters

encoding for either toxin is dictated by optimal light and temperature conditions (Table 2).40

A B

Variable Region

Figure 7. Chemical structure of saxitoxin (A) and anatoxin (B). Adapted from Araoz et al (2010).83

1.7 MC Exposures

Due to the prominence of MC compounds and the variety of cyanobacteria capable of producing MC, exposures to MC may be more likely, compared to other cyanotoxins. Routes of

MC exposures are crucial for understanding their toxicity. Health outcomes associated with MC

exposures vary from skin and eye irritation due to physical contact, liver toxicity and

gastrointestinal distress due to ingestion, and death in extreme circumstances.60,87 Preclinical

models of study have also noted a vast difference in toxicity when MC compounds are injected

compared to oral administration.88 Studies have also shown the accumulation of MC toxins in edible portions of vegetables, and their soils when irrigated with HAB contaminated waters, opening up new routes of human and animal exposure which could exceed WHO guidelines.89,90

Therefore, while studies have focused on single-congener exposures, real-world exposures are

much more likely to involve complex mixtures of toxins and various means of MC and

cyanobacterial ingestion.

Various instances of human and animal toxicity associated with MCs and cyanobacteria

have been reported worldwide. Most exposures occur due to ingestion of improperly treated

water or during recreational activity.91,92 Exposures of this nature have been reported in various

regions of China, and are linked to increased rates of liver toxicity in children93 and decreased

serum insulin levels in fishermen.94 Epidemiologic studies in the various regions of Serbia have

also found increased rates of primary liver cancer in HAB affected regions.95 Regions which rely

on bloom impacted source waters for drinking water supplies showed a two- to three-fold

increase in primary liver cancer incidence in these districts compared to non-impacted districts.95

Other epidemiological studies of Serbia show that areas impacted by HABs have increased

incidence of various cancers including: ovarian, testicular, gastric, colorectal, primary liver

cancer and others.13

In Brazil, where HABs are common, cyanobacterial exposures have resulted in various toxicities and outcomes in exposed individuals. In 1988, 2000 cases of gastroenteritis resulting in

88 deaths were attributed to cyanobacterial exposure in drinking water, after all other

possibilities including chemical, bacterial, and heavy metal exposures were ruled out.96 In the

Brazilian city of Caruaru, in 1996, 76 mortalities were reported in dialysis patients treated with

MC contaminated waters.87,97 Of the 131 exposed patients, 116 experienced nausea, headaches,

vomiting, visual impairment, and muscle weakness.97 Studies have shown that of the exposed

population, older males were the most susceptible.87,98 A similar, less severe, exposure event

occurred just two years later, in 2001. A hemodialysis center in Rio de Janeiro saw 44 patients

exposed to MCs, with 90% testing positive for MCs in serum.99 MC concentrations in serum

persisted for 57 days after exposure.

In the United States, outcomes related to cyanobacterial exposures were first documented

in Charleston, WV, in 1931. An estimated 5,000–8,000 people developed gastroenteritis after

ingesting Ohio River water experiencing a Microcystis bloom.100 Later, in 1974, 23 cases of chills, fever, and hypotension in hemodialysis patients in Washington, D.C., were linked to endotoxin exposures of cyanobacterial origin.101 More recent studies in Florida have found 54%

of recreational and surface source waters to contain cyanobacterial species capable of producing

cyanotoxins.102 Of these, 90% contained cyanotoxins at detectable levels. While the USA

specifically monitors for cyanobacterial toxins and blooms, the possibility of MC exposure

persists and may be on the rise due to changing climate conditions. Specifically, an increasing

number of publications have shown that MCs accumulate in plant matter and fish

tissue.25,90,103,104 Therefore, crops and seafood contaminated with MC present further human and

animal exposure pathways which may be overlooked by monitoring agencies. This would

suggest that while efforts to monitor MC in drinking water supplies are working, our

understanding of exposure pathways should be expanded.

1.8 The Future of Harmful Algal Blooms

Not only are occurrences of HABs becoming more frequent, but overall severity duration, and distribution of HABs have also risen.105–108 Overall, global temperatures are expected to

increase, specifically in higher latitude areas.109 Winters in Norther Europe, North America and

Asia are expected to become warmer, with heat wave frequency increasing.110 Additionally,

extreme variability in precipitation is expected to increase around the globe, with more extreme

precipitation events becoming more regular.109,111 Increased precipitation events are likely to lead

to greater point and non-point surface runoff events, leading to increased nutrient content in

watersheds and soils.112 With increasing nutrient content, warmer temperatures, and increases in sunlight, HABs occurrences are expected to rise worldwide, leading to increased cyanotoxin exposure hazards. The dominance of cyanobacteria in freshwater environments is also expected to increase with changing climate conditions.113,114

1.9 Conclusions

In conclusion, past and current literature tells us that many factors impact the formation

of HABs and the toxins which they produce. While the types of toxins produced vary, the most

23 common and prevalent toxins belong to the MC family of cyclic polypeptides. MC toxicity is limited to cells possessing OATP structures which transport the toxins over the cellular membrane. These cells include liver, neural, and gonadal cells. Upon entering the cells, MCs bind to PP1/2A causing microtubular collapse and ROS production, leading to cellular death.

MCs have also been proposed as promoters of carcinogenesis, via the production of ROS, increased cellular proliferation, and disruption of p53 function. Various MC and cyanotoxin exposure events have been reported worldwide, leading to a variety of outcomes ranging from fever and liver toxicity, to death and increased cancer rates. With changing climate conditions, the rates of HABs are expected to increase. Therefore, timely and novel research is necessary to: better understand the causes and impacts of HABs in understudied matrices; further the understanding of health outcomes due to cyanotoxin exposure; and develop novel monitoring strategies for HAB study and exposure prevention.

Chapter 2. Tile Drainage and Anthropogenic Land Use Contribute to Harmful Algal Blooms and Microbiota Shifts in Inland Water Bodies

The following chapter has been accepted for publication, with full rights and permissions, in Environmental Science and Technology (2018).

2.1 Abstract

Freshwater harmful algal blooms (HABs), driven by nutrient inputs from anthropogenic sources, pose unique risks to human and ecological health worldwide. A major nutrient contributor is agricultural land use, specifically tile drainage discharge. Small lakes and ponds

(SLaPs) are at elevated risk for HAB appearance, as they are uniquely sensitive to nutrient input

(Figure 8). HABs introduce exposure risk to microcystin (MC), hepatotoxic and potentially carcinogenic cyanotoxins. To investigate the impact of anthropogenic land use on SLaPs, 24 sites in central Ohio were sampled from June to August of 2015. MC concentration, microbial community structure and water chemistry were analyzed. Land use parameters including presence of tile drainage systems or animals were significantly correlated with concentrations of microcystin-producing Microcystis aeruginosa throughout the sampling period (p= 0.03).

Relative abundance of HAB-forming genera was correlated with elevated concentrations of nitrate and soluble reactive phosphate. One location (FC) showed MC concentrations exceeding

875 µg/L and large community shifts in ciliates (Oligohymenophorea) associated with hypoxic conditions and coral disease. The prokaryotic community at FC was dominated by Planktothrix sp. at the July 28th sampling date. These results demonstrate the impact of HABs in SLaPs and

that prevailing issues extend beyond cyanotoxins, such as cascading impacts on other trophic levels.

Figure 8. Sources of nutrient input into small lakes and ponds (SLaPs).115

2.2 Introduction

Emerging issues related to worldwide occurrences of harmful algal blooms (HABs),

which produce negative impacts on ecosystem and human health, have garnered much attention

from the scientific community.116 Eutrophication is the primary cause of HABs and is often

attributed to nutrient runoff from anthropogenic sources.117,118 Other factors such as sunlight,

water temperature, and turbidity also impact HAB occurrence.7,27

Freshwater HABs are products of extreme proliferation of cyanobacteria, with

Microcystis and Planktothrix genera being the most common in the Midwestern United States.119

The dominant species of cyanobacteria and toxin production during HABs are determined by

factors including community characteristics, seasonality, and nutrient abundance.118,120

HAB production of cyanotoxins has previously been linked with phosphorus and nitrogen

concentrations, with nitrogen being the primary driver of toxin production.121 Of the various

cyanotoxins (microcystin, anatoxin-a, saxitoxin) reported in the Midwestern U.S., microcystins

(MCs) are most prevalent .60 Human exposure to MCs has been linked to symptoms ranging

from skin and eye irritation, carcinogenicity, liver failure, and death in extreme

circumstances.60,62,64,87,122 Over 100 naturally occurring congeners of MC have been identified,

each possessing unique chemical structures, dictating toxicity.91,123 Of these congeners, MC-LR

is the most common and toxic, and can persist in the environment for months at a time.119,124

Outcome severity and potential of exposure to cyanotoxins have been topics of many

publications.125,126 However, HABs have been least studied in small lakes and ponds (SLaPs),

which are often overlooked by monitoring efforts due to their high numbers and sizes.127

Currently the Ohio EPA Inland Lake Program monitors 16 lakes per year, well below the state’s

total lake count.128 There are an estimated 304 million natural lakes in the world, covering

4.2×106 km2, most of which are SLaPs.127 SLaPs are important providers of ecosystem services

such as biodiversity support, hosting more species per unit of area than large lakes .127,129

Further, SLaPs have irrigational and recreational uses and are often contacted by human and

animals directly. Agricultural ponds comprise a large area of land worldwide, covering 77,000

km2 and are increasing with growing food demand.127

As HABs seem to be more severe in low volume water bodies, SLaPs may be more

sensitive to nutrient influx from agricultural sources.130 One major nutrient contributor is tile

drainage effluent.131 Due to decreased opportunity for nutrient adsorption into soils, subsurface

transport of water from agriculture via tile drainage results in a rapid nutrient transport

mechanism. This is most problematic during heavy-rains and snow melts.10,132 Tile drainage has been shown to displace 80% more phosphorus and 43% more nitrogen in subsurface drained systems compared to managed systems, with concentrations varying with precipitation and soil composition.10,131,133 This quality has lead us to identify tile drainage as a factor of interest for

this study.134

In addition to tile drainage, nutrient runoff into watersheds occurs from non-point sources.135 Non-point source runoff acts as a transport system for fecal matter, and may

contribute to nutrient loading in watersheds.136–138 Runoff can also transport enteric pathogens,

including Cryptosporidium, Campylobacter, Shiga-toxin producing E. coli and various enteric

viruses.139,140

For the above reasons, we investigated the microbial communities and water chemistry of

24 SLaPs in central OH. We focused on small water bodies proximate to agricultural lands and

those receiving tile drainage effluent. SLaPs were chosen to represent those likely to be used for

agriculture or recreation. Our goal was to determine whether human influence promoted the

formation of HABs, and whether MC concentrations posed human health risks in sampled

SLaPs. Further, structures of eukaryotic and prokaryotic microbial communities were

investigated. Eukaryotic and prokaryotic communities are often impacted by changing oxygen

levels during bloom deaths and selective pressures exerted by HABs.116 Therefore, findings from

this study imply consequences to ecosystem and human health in HAB impacted SLaPs.

2.3 Materials & Methods

2.3.1 Sample Collection

Water samples were collected from 24 public and private SLaPs (size range 0.06–50.5

ha) in Knox County, an agriculturally dense rural county in central Ohio (Figure 9). Water samples were collected at a depth of 0–20 cm using a Nasco Swing Sampler and placed in sterile

710 mL Whirl-Pak® stand-up bags. Two bags were filled per site. Each site was sampled twice

between June and August 2015. At the time of sample collection, water pH, conductivity and

temperature measurements were recorded using a combined field pH and conductivity meter

(Hanna Instruments HI98129). Dissolved oxygen (DO) was measured using an YSI ODO

Optical Meter (Yellow Springs Instruments). Water samples were transported on ice and stored

at 4oC until filtration within 24 hours of collection.

Figure 9. Map of sampling sites in Knox County, OH. Locations were chosen to be representative of urban and agriculturally impacted bodies of water. Ohio State Plane North, NAD1983 (Arc Map 10.4.1, ESRI 2016).115

2.3.2 Site Characterization

The presence or absence of septic tanks on properties housing SLaPs was determined by

referencing property records at the Knox Co. Ohio Auditor’s office. Land use was determined in

the 100m buffer surrounding each site using ArcGIS (version 10, ESRI, Inc.) and the 2011

National Land Cover Database.141 The study considered agricultural land and developed land

use to be human land uses for the purposes of categorizing the land use setting of each site. The

proportion of the area of the 100m buffer around each site was used to indicate whether they

were surrounded by low, medium, or high intensity land uses. Sites with 0–30% human land use

were ranked as low, 31–75% was considered moderate land use, and 76–100% was considered

high intensity land use. If a site was in a low or medium land use setting, but was receiving

water directly from drain tiles from crop fields that were more than 100 m distance (thus

circumventing the function of the land buffer), the site was considered to have high human

impacts and so was put into the high intensity category.

2.3.3 Water Sample Processing

Samples were vacuum filtered using sterile Microfilm® V Filtration devices (Millipore

Sigma) and membrane filters (Isopore TM membrane filter; 0.4 µm for water samples and 0.22

µm for DNA samples; Millipore). Four subsamples were filtered from each water sample.

Between 10–100 mL of water was passed through each filter, depending on the microbial concentrations. Filters were folded, placed in 2-mL sterile centrifuge tubes, and stored at -80°C.

Samples for determination of MC concentrations were stored frozen in amber glass containers to

prevent degradation of MC.

2.3.4 Water Chemistry

Each sample (150 mL) was filtered within 24 hours of collection using a 0.4 µm pore size

IsoporeTM polycarbonate membrane, stored at -20°C, and then used to analyze water chemistry.

Nutrient concentrations, including nitrate (NO3-N), ammonia (NH3-N) and soluble reactive phosphate (PO4-P) were measured colorimetrically using EPA approved Hach (Ames, IA) water quality procedures and a DR890 colorimeter (Hach: Loveland, CO).

2.3.5 Cyanotoxin Measurement

Concentrations of total MC were determined utilizing the Abraxis microcystins/nodularin

(ADDA) ELISA colorimetric immunoassay kit (ABRAXIS Inc. Prod #520011). Samples were tested in duplicate. Sample dilutions were necessary to align toxin concentrations within kit detection limits (0.15 to 5 µg/L on average). A Molecular Devices SPECTRAmax PLUS 384

(Molecular Devices, Silicon Valley, CA) was used for plate reading purposes. Anatoxin-a

(ABRAXIS Inc. Prod. #520060) and saxitoxin (ABRAXIS Inc. Prod. #52255B) concentrations were tested in SLaPs with detectable MC levels.

2.3.6 DNA Extraction

DNA was extracted from membrane filters using two methods. Cyanobacterial DNA was extracted using a modified xanthogenate-sodium dodecyl sulfate (XS) DNA extraction method with a modified DNeasy Blood & Tissue Kit (QIAGEN Group, Cat #69504).142,143 For microbial community analysis and microbial source tracking, microbial DNA was extracted using the

QIAamp DNA Stool Kit (Qiagen, Cat #51504). Sample extraction was conducted using

manufacturer instructions, followed by -80oC storage.

2.3.7 PCR Inhibition Testing

A Sketa22 assay was conducted on all samples to screen for the presence of PCR

inhibition. The method previously described by Haugland et al (2005) was used.144 Samples

exhibiting PCR inhibition were diluted and re-tested. This process was repeated until inhibition was within acceptable levels.

2.3.8 PCR detection of Microcystis aeruginosa and Planktothrix sp.

qPCR analysis was conducted to quantify Microcystis aeruginosa concentrations by

targeting the PC-IGS region (total M. aeruginosa) and mcyE gene (microcystin-producing M.

aeruginosa). Both assays were conducted as previously established (Table 3 - Table 6).120,145

Due to the recent the adoption and validation of droplet digital PCR (ddPCR) methods for

cyanobacterial gene quantification, concentrations of MC-producing Planktothrix were quantified by targeting mcyE using previously established primers and PCR cycling conditions, and the QX200 ddPCR system (Bio-Rad; Des Plaines, IL).146 The ddPCR mixture (20 μL total

volume) contained 2× QX200 ddPCR EvaGreen Supermix (Bio-Rad), 200 nM of each primer,

DNA template, and RNase-/DNase free water. Droplets were generated using the Droplet

Generator (Bio-Rad) with 70 μL of QX200 Droplet Generation Oil for EvaGreen (Bio-Rad) with each 20 μL ddPCR PCR mixture. PCR was conducted via thermal cycler (C1000 TouchTM

Thermal Cycler, Bio-Rad). After completion of PCR, droplets were quantified using the Droplet

Reader and QuantaSoft software version 1.7 (Bio-Rad).

Table 3. qPCR Primer and Probe Sequences

Gene Target Assay Sequences Target Ref.

F Primer TTTTCAGCCCCGTTGTTTCG

DG3 R Primer TGAGCGGGCATGGTCATATT Bacteroides plebius 41

Probe [FAM]-AGTCTACGCGGGCGTACT-[MGB]

F Primer TTTTCTCCCACGGTCATCTG

DG37 R Primer CTTGGTTATGGGCGACATTG Lacnospiraceae 41 [FAM]- TTGAACGTTTAAAGGAGCAGGTGGCAG- Probe [TAMRA]

Unclassified GFD GFDF TCGGCTGAGCACTCTAGGG 42 Helicobacter sp. GFDR GCGTCTCTTTGTACATCCCA ATCATGAGTTCACATGTCCG Hf183 F Human Feces (HF8 Hf183 BFDR CGTAGGAGTTTGGACCGTGT 39 Cluster) [FAM]-TGAGAGGAAGGTCCCCCACATTGGA- Probe [MGB] 188F GCTACTTCGACCGCGCC M. M. aeruginosa aeruginosa 254R TCCTACGGTTTAATTGAGACTAGCC phycocyanin 7 PC-IGS [FAM]-CCGCTGCTGTCGCCTAGTCCCTG- intergenic spacer Probe [MGB] 127F AAGCAAACTGCTCCCGGTATC M. 247R CAATGGGAGCATAACGAGTCAA M. aeruginosa aeruginosa microcystin 37 [FAM]- mcyE CAATGGTTATCGAATTGACCCCGGAGAAAT- synthesis gene Probe [MGB] BacB2- 590F ACAGCCCGCGATTGATACTGGTAA Ruminant-specific Rum2Bac Bac708Rm CAATCGGAGTTCTTCGTGAT 40 Bacteroidales BacB2- [FAM]-ATGAGGTGGATGGAATTCGTGGTGT- 626P [BHQ1]

Table 4. qPCR Reaction Composition

Assay Component Volume (µL) Final Concentration TaqMan Environmental MasterMix 10 1× F Primer 0.05 200 nM R Primer 0.05 200 nM DG3/DG37 Probe (100 pmole/µL) 0.04 100 nM DNA Template 2 Water 9.86 2× SYBR MasterMix 10 1× GFDF (100 pmole/µL) 0.04 200 nM GFD GFDR (100 pmole/µL) 0.04 200 nM DNA Template 2 Water 7.92 2× TaqMan MasterMix 10 1× Hf183F (100 pmole/µL) 0.04 200 nM Hf183R (100 pmole/µL) 0.04 200 nM Hf183 Probe (100 pmole/µL) 0.04 200 nM DNA Template 2 Water 7.88 2× TaqMan MasterMix 10 1× 188F (100 pmole/µL) 0.06 200 nM 254R (100 pmole/µL) 0.06 200 nM M.aeruginosa PC-IGS Probe (100 pmole/µL) 0.02 200 nM Mg2+ (25 mM) 0.5 DNA Template 2 Water 7.36 2× SYBR MasterMix 10 1× 127F (100 pmole/µL) 0.06 200 nM M.aeruginosa 247R (100 pmole/µL) 0.06 200 nM mcyE Probe (100 pmole/µL) 0.05 200 nM DNA Template 2 Water 7.82 2× TaqMan MasterMix 10 1× BacB2-590F (100 pmole/µL) 0.04 200 nM Bac708Rm (100 pmole/µL) 0.04 200 nM Rum2Bac BacB2-626P (100 pmole/µL) 0.04 200 nM DNA Template 2 Water 7.88 35

Table 5. qPCR Assay Conditions

Assay Step Temp (°C) Time Cycles Initialization 94 5 min Denaturation 94 40 sec DG3/DG37 Annealing 60 1 min Extension 72 30 sec 40 Initialization 95 10 min Denaturation 95 15 sec

GFD Annealing & Extension 60 1 min 50 Melting Curve 95 15 sec 60 1 min 95 15 sec Initialization 95 10 min Hf183 Denaturation 95 15 sec Annealing & Extension 60 1 min 50 50 2 min Initialization 95 10 min M.aeruginosa PC-IGS Denaturation 95 30 sec Annealing 56 1 min Extension 72 30 sec 50 50 2 min M.aeruginosa Initialization 95 10 min mcyE Denaturation 95 30 sec Annealing & Extension 62 1 min 50 Initialization 95 10 min Rum2Bac Denaturation 95 15 sec Annealing & Extension 60 1 min 50

Table 6. qPCR Assay Quantification Curves

Range of Quantification Assay Equation R2 (copies/µL) DG3 y=-3.633x + 37.904 0.996 5.0x100 - 5.0x104 DG37 y=-3.604x + 38.150 0.894 5.0x100 - 5.0x104 GFD y=-3.8719x + 53.435 0.982 2.8 x101 - 2.8x109 Hf183 y=-3.3897x + 40.766 0.989 2.8 x101 - 2.8x106 M.aeruginosa PC-IGS y=-3.199x+20.793 0.988 3.6x101 - 3.6x106 M.aeruginosa mcyE y=-2.9375x + 21.645 0.996 3.1x101 - 2.1x105 Planktothrix sp. mcyE y=-3.4212x + 37.588 0.991 1.0x101 - 1.0x106 Rum2Bac y=-3.3977x + 42.735 0.998 1.1x102 - 1.1x106

2.3.9 Microbial Source Tracking (MST)

Potential sources of fecal bacteria were determined at each site, targeting human

(HF183), bovine (Rum2Bac), canine (DG3/DG37), and avian sources (GFD). Protocols used for

each assay can be found in the Supporting Information section (Table 3. qPCR Primer and Probe

Sequences - Table 6. qPCR Assay Quantification Curves), and rely on previously published literature.147–150 All MST was conducted using instrumentation identical to the Microcystis mcyE

assay, apart from DG3/DG37 markers which were analyzed using the QuantStudio 6 Flex Real-

Time PCR System (Applied Biosystems; Foster City, CA), in collaboration with Dr. Hodon Ryu

(U.S. EPA, Cincinnati, OH).

2.3.10 Microbial Community Analysis

Barcoding analysis was conducted by MR DNA (www.mrdnalab.com, MR DNA,

Molecular Research LP; Shallowater, TX). For barcoding, the 16S primers were 515F

(GTGCCAGCMGCCGCGGTAA) and 806R (GGACTACHVGGGTWTCTAAT). 18S primers

were Euk7F (AACCTGGTTGATCCTGCCAGT) and Euk570R

(GCTATTGGAGCTGGAATTAC). A modified barcoded amplicon sequencing (bTEFAP®)

procedure was used. Following PCR, all amplicon products from different samples were mixed

in equal concentrations and purified using Agencourt Ampure beads (Agencourt Bioscience

Corporation, MA, USA). Samples were sequenced using Illumina sequencing platforms

according to standard protocols, as described previously.151 The 16S sequences were obtained

with Hiseq 2x250bp, and the 18S sequences were obtained with Miseq 2x300bp.

Q25 sequence data derived from sequencing was processed by MR DNA using a

proprietary analysis pipeline. Sequences were depleted of barcodes and primers. Q25 merged

sequences were clustered and operational taxonomic units were defined, clustering at 3%

divergence (97% similarity), with removal of singleton sequences and chimeras and operational

taxonomic units were defined.152,153 Final operational taxonomic units (OTUs) were

taxonomically classified using BLASTn against a curated database derived from GreenGenes,

RDPII (http://rdp.cme.msu.edu), and NCBI (www.ncbi.nlm.nih.gov), and compiled into each

taxonomic level into both “counts” and “percentage” files.

2.3.11 Statistical Analysis

One-way analysis of variance (ANOVA) was used to evaluate site characteristics based

on land use. We used Principal Components Analysis (PCA) to reduce data complexity and

investigate patterns of bacterial species distribution as a function of water quality and land use

using the statistical package JMP 12 (SAS Institute, Inc.; Cary, NC).

2.4 Results

2.4.1 Analysis of Factors Linked to HAB-friendly Conditions

Concentrations of M. aeruginosa MC synthesizing gene mcyE (gene copies/mL) were correlated with presence of tile drains and/or animals within a 100m vicinity (Figure 10). SLaPs that received tile drainage or had animals present contained significantly higher (p = 0.03) concentrations of mcyE.

Figure 10. Comparison of microcystin-producing gene concentrations in SLaPs with tile drainage or animal presence (Y), and those without either factor (N). Sites were compared based on concentrations of the mcyE gene (gene copies/mL).

Data were characterized using PCA analysis to investigate the relationship between nutrient content and bacterial community structure in the analyzed samples (Figure 11). Results show that SLaPs were clustered by level of anthropogenic impact into three categories: low

(blue), moderate (grey), and high (red) anthropogenic land use. The first two principle components explain 50.4% of the variability in the data, while the first three explain 64%. Three cyanobacterial genera Microcystis, Planktothrix, and Aphanizomenon, were correlated with nutrient concentrations in the sampled SLaPs. These results are in line with past findings which have shown nitrate and phosphorus to be drivers of HABs.43

A B 42

Figure 11. Principal component analysis of data sets showing relationships between land use, bacterial communities, and environmental parameters in the sampled locations. The first two principle components explain 50.4% of the variability. A). SLaPs were cluster by land use intensity. B) Strength of relationships between variables (vectors) used in the PCA; vector length indicates strength of correlation. Nutrient concentrations were correlated along the first principle component (x-axis) with HAB-forming cyanobacteria of the Microcystis, Planktothrix, and Aphanizomenon genera. Sites 8-11 were indicative of FC transects, and were therefore not used for PC analysis, instead a representative sample (#12) of FC on July 28th was used.

2.4.2 HAB Conditions, Toxin concentrations, and Microbial Communities in FC

No unique trend was observed between M. aeruginosa mcyE, PC-IGS, and MC concentrations in the samples (Figure 12). Similarly, concentrations of Planktothrix mcyE gene were relatively consistent throughout the sampling sites (Figure 13). While MC content of all samples was below the Ohio guideline limit for MC-LR in recreational waters (4 µg/L), we found detectable levels of MC in 42% of samples (Figure 9), and one location (FC) which showed a maximum MC concentration of 876 µg/L on July 7th (Table 7).154. Anatoxin-a (4.35

µg/L) and saxitoxin (0.34 µg/L) levels also peaked on this date. All FC samples had detectable

MC levels.

Figure 12. Mean Microcystis aeruginosa concentration and maximal MC production from sampled locations. Samples were analyzed using qPCR methods for total Microcystis (PC-IGS) and Microcystis capable of synthesizing MC toxin (mcyE). All MC concentrations were reported as maximal levels detected during the sampling period (Jun – Aug 2015) apart from FC whose maximal MC was 876 µg/L, an extreme outlier.

Figure 13. Mean Planktothrix mcyE gene concentrations by site. Samples analysis was conducted via ddPCR for Planktothrix sp. capable of synthesizing MC toxin (mcyE). While some variation exists across samples, mcyE concentrations were similar throughout the samples, mcyE concentrations were similar throughout the sampling sites.

Table 7. Microcystis aeruginosa and cyanotoxin concentrations of FC samples (Jun-Aug 2015).

PC-IGS (gene mcyE (gene MC Anatoxin Saxitoxin Date copy/mL) copy/mL) (µg/L) (µg/L) (µg/L) June 18th 8.00E+02 5.09E+01 0.17 0.07 0.01 July 7th 4.20E+05 2.64E+02 876.64 4.35 0.34 July 28th 2.55E+04 1.15E+02 1.24 0.06 0.01

2.4.3 Water Chemistry

Nutrient concentrations were highest on June 18th at the FC location. During this

sampling period, concentrations of overall phosphorus (21.7 mg/L) and nitrate (30.33 mg/L)

were 10-fold higher than the cumulative sampling mean. Based on phosphate and nitrate

concentrations, FC could be classified as both mesotrophic and eutrophic respectively (Appendix

A. Table 19).155 The FC location was of great interest as it received direct tile drainage effluent.

On July 7th, FC experienced a spike in levels of total Microcystis and MC (876 µg/L) toxin levels

which exceeded recreational water guidelines.154

2.4.5 Microbial Source Tracking

Human-associated marker (HF183) was detected in nearly all samples (Mean: 3.9 ± 1.3

log[gene/mL]; Range: 1.1 - 6.8 log[gene/mL]) (Figure 14). This may be attributed to use of septic tanks in the area. Goose-specific marker (GFD) was also high (Mean: 3.4 ± 1.5

log[gene/mL]; Range: 0 – 5.8 log[gene/mL]), which is consistent with our observations, as several SLaPs were surrounded by grassy areas ideal for grazing by geese. While deer and dogs were sighted during sampling, canine markers (DG3/DG37) were not detected in any sample,

and the ruminant marker (Rum2Bac) was detected in only one sample.

Figure 14. qPCR determined host-specific fecal bacterial load by location. Nearly all locations contained human- and goose- specific fecal markers. The canine fecal marker was not found in any of the sampled locations, while only one contained the ruminant marker.

2.4.6 Microbial Community Analysis

The eukaryotic community of FC experienced substantial shifts during individual sampling periods. During the June 18th sampling date (Figure 15A), the eukaryotic community was made up of various eukaryotic classes, with the highest proportion being Colpodea (16%) and Chrysophyceae (12%). A substantial shift in the eukaryotic community occurred on July 7th, as Oligohymenophorea (82%) became dominant (Figure 15B). July 28th, saw a diminished proportion of Oligohymenophorea (4%), but the appearance and dominance of new classes such as Chytridiomycetes (23%) and Cryptophyta (17%) (Figure 15C).

Figure 15. The taxonomic composition of the Eukaryotic microbial community in FC at three time-points was shown to vary drastically. Figure A corresponds to the June 18th, figure B corresponds to July 7th, and figure C corresponds to the July 28th sampling date. The largest changes were seen in the Eukaryotic class Oligohymenophorea, a class of ciliated organisms.115

The prokaryotic community composition also fluctuated, although to a lesser extent. For the June 18th sampling date, the notable cyanobacterial genus was Planktothrix (Figure 16A). On

July 7th, a similar percentage of Planktothrix (5%) was noted, and was accompanied by

Microcystis (3%) (Figure 16B). At this point, the community shifted from one dominated by

Candidatus Planktophilla to one dominated by Aquaspirillum (18%), Clostridium (10%), and

Comamona (9%). On July 28th, the prokaryotic community was dominated by Planktothrix

(52%) and Aphanizomenon (8%) (Figure 16C). A significant change in community structure was seen over time.

Figure 16. Prokaryotic microbial community structure of FC during the sampling period, as determined by 16S and 18S Barcoding analysis. (A) June 18, 2015; (B) July 7, 2015; and (C) composite result from the four transect samples on July 28, 2015.115

FC, which showed high levels of MC, anatoxin-a, and saxitoxin, (Table 7) was chosen for further examination. Four transects of FC were sampled on July 28, 2015 and the distribution of

MC and microbial community were investigated (Figure 17). The highest concentration of MC

(3.95 µg/L) was observed near the dammed portion of the lake. Microbial community analysis showed that the eukaryote community was dominated by Cryptophyta near FC’s inlet (38.7%) and dammed portion (21.2%), while Chrysophyceae (36.7%, 26.9%) dominated the middle portions. The prokaryotic community was dominated by Planktothrix throughout all four transects, which accounted for 36.0% to 57.7% of the prokaryotic community. The dammed portion of the lake also contained a notable community of Aphanizomenon (22%).

Figure 17. MC concentration and microbial communities of FC by transect. Samples are representative of the total microbial community in each transect, as determined by 16S/18S sequencing. The prokaryotic community was largely dominated by the cyanobacterial genus Planktothrix, while the dammed transect of the lake also included a large proportion of Aphanizomenon (22.7%).115

54 2.5 Discussion

The objective of the study was to evaluate the influence of anthropogenic activity on

HAB-related water quality of small water bodies. These lakes are overlooked by state and local

agencies in terms of water quality monitoring, yet are some of the most numerous bodies of

water worldwide, providing vital services to humans and ecosystems.127,128 The study found that

42% of sampled lakes had detectable levels of MC.

The study design also tested multiple variables linked to HABs and water quality in 24

unique SLaPs. First, sources of fecal contamination in the sites were analyzed using MST

methodologies. Results show that intrusion of fecal bacteria originate from two main

contributors: avian, specifically geese, and human sources. This holds true to on-site

observations, where geese were often seen grazing near SLaPs. Further, in Ohio, over 250,000

septic tanks discharge waste without meeting water quality requirements and are a likely source

of human fecal contamination.156 Fecal bacteria associated with canine source were not detected,

while one site (BD) showed presence of ruminant-specific bacteria. Intrusions of such fecal

wastes were previously shown to increase nitrogen loads in Lake Erie beaches, leading to

increased concentrations of MC synthase genes.137 Although incomplete, the data show seven sites with septic tanks in use, and five without.

The relative abundance of MC-producing Microcystis was significantly higher (p=0.03) in areas where either animal life was detected during sampling or where tile drainage was present

(Figure 10). This was reiterated by water chemistry data (Appendix A. Table 19), which showed that lakes impacted by tile drainage or animal presence had higher nutrient content than non- impacted lakes. Results suggest that areas with high nutrient runoff are more likely to experience

55 HAB development. This finding goes in line with past studies citing tile drainage, and animal wastes as sources of nutrient loading.132,134,157

To better understand the influence of anthropogenic impacts on water quality, a cluster analysis was conducted. Sites were clustered based on human land use: the most developed land was classified as “high” and the least developed was termed “low”. Results show that human land use impacts concentrations of HAB-driving nutrients, ammonia, nitrate and phosphate

(Appendix A. Table 19).43 This relationship was also seen within the SLaPs’ cyanobacterial community structure, as nutrient content was correlated with higher proportions of cyanobacterial cells within SLaPs (Figure 11). The first three components of the PCA explain

64% of the variability in the data, and indicate that the concentrations of ammonia and phosphate were most important in explaining species composition.

These results indicate that anthropogenic influence increases the deposition of nutrients in SLaPs, shifting microbial community structures and establishing conditions more favorable to cyanobacteria (i.e. Planktothrix, Microcystis, and Aphanizomenon). These shifts are exacerbated the deposition of nutrient-rich waters near dams, a trend also noted by Pearce et al (2017).158

Nutrient rich conditions drive cyanobacterial proliferation and MC production, shifting SLaP microbial communities and allowing cyanobacteria to dominate.121,158

Analysis of MC concentrations showed one location of particular interest: FC. While other SLaPs contain tile drainage systems in the studied buffer area, FC was the only SLaP to receive direct tile drainage effluent from nearby agricultural fields. Surrounding FC is an area with large trees and dense vegetation, minimizing surface runoff intrusion. Therefore, FC is a unique system which can be used as a model to study direct impacts of tile drainage on small

56 water bodies. This type of relationship is not readily seen in larger watershed studies as they

often involve large areas with high volumes of surface runoff and confounders such as industrial

and wastewater effluent.159 While several studies have linked the influence of tile drainage to the elevated nutrient concentrations in large water bodies, a direct link between tile drainage and

HABs is difficult to establish.10,12,132 Further, the impact of tile drainage may be elevated in

SLaPs, as highly concentrated nutrients are introduced into a smaller volume of water, making

SLaPs more sensitive to nutrient influx.

FC was the site noted as having the largest blooms amongst those sampled, and had

detectable MC levels at every sampling point. The maximal MC level reached 876 µg/L MC in

early July, exceeding the State of Ohio recreational water guideline of 4 µg/L for MC, while

saxitoxin and anatoxin-a were also detected (Table 7).154 It is important to note the limitations of

ELISA determined MC concentrations and that more precise means of toxin measurement (i.e.

LC-MS) should be used in future works. Analysis reveals that the presence of toxin producing

Microcystis genes was relatively low during this event, indicating that the cyanobacteria

responsible for MC production is most likely outside of the Microcystis genus (Figure 18).

However, MC synthesis gene abundance does not necessarily represent the relative expression of

the targeted genes, a relationship which should be explored in the future. Ingestion of water with

such MC concentrations could cause human and animal illness, and be especially worrisome in

seldom monitored SLaPs.26 Further, elevated nitrate levels (30.33 mg/L), were detected on June

18th, and were likely responsible for elevated MC concentrations on July 7th.121

57 Figure 18. Total Microcystis and toxin-producing Microcystis concentrations during a 2-month period (2015) in FC. Samples were collected at three time points, during which HABs were reported. The final sampling date (July 28th) is a mean of four transect samples of the lake.

58 Due to the unique hydrodynamic nature, nutrient influx source, and high level of toxin

detected in FC, the July 28th sampling was separated into 4 transects. Levels of MC were lowest

near the tile drainage inlet, and highest, near the constructed dam (Figure 17). The nature of this

MC concentration distribution is most likely due to water flow created by the inlet and the

stagnation of water created by the dam. On July 7th, the cyanobacterial community of FC was

dominated by Planktothrix, with a lesser, yet substantial Microcystis community. A shift in the

community was seen on July 28th, as the community was largely dominated by Planktothrix

(Figure 4).

Results show nutrient content of FC on June 18th was 10 times higher than the

cumulative sampling mean (Appendix A. Table 19), reaching mesotrophic levels of phosphorus

(21.7 mg/L) and eutrophic levels of nitrogen (30.33 mg/L). Such nutrient concentrations indicate that bloom-friendly conditions were highest in June, but returned to low concentrations during

July 7th. It is likely that nutrient concentrations were depleted by cyanobacteria, and dissipated following bloom death. Lysis of cyanobacteria has been shown to release intracellular MC into waters producing high MC accompanied by low cyanobacterial DNA concentrations, such as those seen in early July.160 Further, the emergence of Planktothrix as the dominant prokaryotic genus on July 28th could be attributed to competitive pressures exerted by cyanobacteria, establishing favorable conditions after dissipation of the July 7th bloom (Figure 16).

We noted a sizable shift from the initial (June 18th) proportion of the Eukaryote

Oligohymenophorea (8%) to 82% during the suspected bloom (July 7th) (Figure 15). On July

28th, proportions of Oligohymenophorea reverted to June 18th levels, hinting that selective pressure from bloom-forming cyanobacteria may have influenced the prokaryotic community.

59 Oligohymenophorea have previously been linked with coral disease in marine, and hypoxic conditions in freshwater environments.161,162 We hypothesize that Oligohymenophorea, a class of ciliated, predatory, eukaryotes are migrating to sites of algal blooms, to prey upon HAB associated microbes and take advantage of potentially hypoxic conditions.163 This finding should be further investigated, as HAB-driven shifts in microbial community structure, particularly in

Eukaryotic species, are poorly understood and could have cascading impacts on SLaPs’ food chains.

Our findings indicate that HAB conditions in FC are of potential concern. Due to the site’s status as a popular lake for fishing and shoreline gatherings, we recognize the possibility of human and animal exposure exists. If MC levels seen during the July 7th sampling date occur in the future, without a proper monitoring system, exposures could result in human and animal illness. Further studies should focus on microbial community shifts seen during HABs and monitoring methods for SLaPs used for recreation, irrigation, or drinking water.

60 Chapter 3 Comprehensive Determination of Acute MC-LR Exposure on Male and Female Murine Liver Health

3.1 Abstract

Exposures to microcystin-LR (MC-LR), a cyanotoxin with hepatotoxic properties in humans and animals have been noted worldwide. Exposure risks continue to be of concern in areas with inadequate drinking water treatment and those experiencing high rates of harmful algal blooms. Acute toxicity of MC-LR has been widely studied and has been linked to mild outcomes such as skin and eye irritation, and more severe outcomes such as liver toxicity, and death in extreme cases. This study’s goal was to provide an accurate and comprehensive description of the pathological, clinical chemistry, and gap junction intercellular communication

(GJIC) changes attributed to MC-LR exposure in CD-1 male and female mice. Mice were exposed to 0, 3000, and 5000/4000 µg/kg/day pure MC-LR, daily for 7 days. Mice were necropsied and liver samples were collected on Day 8. Blood samples were processed to serum and used for clinical chemistry analysis. Sections of liver tissues were fixed for histopathological examination. Hepatocellular GJIC was analyzed using fluorescent dye cut-loading. Results show a dose-dependent relationship with MC-LR exposure and incidence of hepatocellular hypertrophy, degradation, and necrosis. Clinical chemistry parameters alanine aminotransferase

(ALT), aspartate aminotransferase (AST), alkaline phosphatase (ALP), total bilirubin (TBIL), and cholesterol (CHL) increased significantly in mice exposed to MC-LR. Serum glucose

(GLUC) levels decreased with dose. Clinical chemistry parameters showed increased MC-LR

61 toxicity in female CD-1 mice compared to male CD-1 mice. We saw no changes in (GJIC)

across liver sections. This study provides a comprehensive description of the liver health

outcomes associated with MC-LR exposure between male and female. Future murine studies

involving microcystin compounds should consider toxicity differences across sexes and should

explore impacts of MC-LR on GJIC.

3.2 Introduction

Exposure to cyanotoxins, specifically microcystins (MCs), has been shown to produce

various health outcomes in humans and animals. Physical contact with high concentrations of

MCs has been shown to produce rashes as well as skin and eye irritation.164 Ingestion of MCs has

been shown to result in nausea and gastrointestinal distress.91,164 Extreme cases of MC ingestion

have been shown to result in animal and human liver toxicity and even death.87,97,165 While

proper treatment methods can curb exposure risks to MC toxins, economically stressed areas still

do not possess the tools needed to prevent MC exposures.166 As shown in Chapter 2,

consumption of water, recreational use, aquaculture, and drinking water use of SLaPs in

agricultural areas may prove to be an additional exposure hazard.

For these reasons, it is imperative to understand the health outcomes resulting from

mammalian MC ingestion. This study focuses on investigating the liver health outcomes

associated with acute exposure to microcystin-LR (MC-LR) via ingestion, which is the most common and toxic congener of MC.119 Ingestion of MC-LR has been widely studied, and has

shown several trends in pre-clinical, animal models of study. While several animal models of

exposure have been employed, mice and rats are the most common mammalian models of

62 study.71,119,167 These studies note increased sensitivity of mouse models compared to rats, taking

into account dose and body weight differences.88 Further, variations in exposure times and time

to sacrifice, vary across study designs. Studies have varied in time-to-sacrifice, ranging from examining the immediate impacts of MC-LR ingestion (30 minutes post-exposure)168, to sub-

chronic impacts of exposure (up to a 21-day treatment period).56 Dosage and frequency of

exposure also vary across studies, ranging from one dose of MC-LR, to daily exposures for

extended periods of time.169,170

Along with variations in study design and model use, the existing knowledge base of

published preclinical studies describes varied outcomes related to the acute exposure to MC.

Studies have reported changes in various clinical chemistry parameters, including alterations in:

aspartate aminotransferase (AST) to alanine aminotransferase (ALT) ratio, and levels of

glutathione-S-transferase (GST).171,172 AST/ALT ratios are often used as markers of liver

damage, with outcomes such as non-alcoholic liver disease producing 2-fold increases in

AST/ALT ratios.173 Other studies, however, report decreases in AST levels in mice, leading to a

lower AST/ALT ratio than that of non-alcoholic liver disease.174 Clinical chemistry markers of

liver toxicity have been shown to increase as early as 30 minutes after MC exposure.171

Studies have also shown changes in GST concentrations in serum, during MC exposure.

GST is a Phase II enzyme responsible for the detoxification of MC-LR via use of glutathione

(GSH) as a conjugation substrate.175 Fluctuations of the GST enzyme in serum are expected

during exposure as GST is depleted initially as MC-LR is detoxified in the liver.171,176

Pathological analysis during toxicological studies have shown mortality and liver toxicity to be

63 attributed to hemorrhaging, hepatocellular necrosis, and the presence of inflammatory

responses.169,177

The transport of MCs into the liver and uptake into the hepatocyte by the organic anion-

transporting polypeptide 1B2, make the liver the primary target organ of MC-LR toxicity.178

Clinical and pathological outcomes can be attributed to the covalent binding of MC compounds

to the catalytic subunits on protein phosphatases 1 (PP1) and 2A (PP2A).59 This binding results in inhibition of PP1/2A leading to hyperphosphorylation of cellular microtubules, loss of cellular structure, and production of oxidative stress leading to cellular apoptosis.55

Toxicities and lethal doses of MC-LR can vary across exposure methods and animal

models. MC exposure methods have generally trended toward the use of intraperitoneal (i.p.)

injection and oral gavage administration, ensuring accurate dosing and controlled

delivery.67,68,167,177,179 Results show that these two standard practice exposures for MC-LR toxicity produce drastically different lethal dose (LD50) estimates as i.p. injection, on average,

produces 10-fold higher toxicity than oral gavage.88 The large variations in findings demonstrate the necessity for better understanding of MC-LR toxicity and show that it is necessary to gather further wellness metrics, as well as evidence of histopathological and biochemical alterations with near-lethal and sub-lethal exposures to MC-LR.

Finally, while many studies have focused on clinical chemistry and pathological outcomes of MC-LR exposure, few have investigated the impacts of MC-LR on gap junction intercellular communication (GJIC) in hepatic cells. GJIC involves the transport of sugars, nucleotides, water, amino acids, and other small molecules across gap junction channels.180 GJIC

channels are found in all vertebrate cells, are limited to transport of molecules up to 1kDa in size.

64 Increases in GJIC have been found to increase the toxicity of compounds in previous studies.180

Further alterations or disruptions in GJIC have been linked to diabetes, autoimmune disorders,

cancers, and neuropathy.181 Findings have shown that interaction with non-genotoxic agents have

shown to inhibit GJIC.182,183 As MC-LR has been linked to increased cancer incidence, yet have

been shown to be non-genotoxic, it is expected that GJIC would be inhibited in vivo.95,184

Additionally, MC-LR mediated production of ROS may also play a role in hepatocellular

GJIC.185

The goal of this descriptive study is to examine the liver health outcomes of acute MC-

LR ingestion between male and female mice. The study aims to investigate the impacts of MC-

LR ingestion on the pathology, clinical chemistry parameters, and GJIC in CD-1. We aim to

accurately describe the dose-response relationship between MC-LR and histopathological, clinical chemistry, and GJIC outcomes across doses and sexes.

3.3 Methods

3.3.1 Chemical Compounds

MC-LR (Lot No. MC LR-2021, MC LR-2022) was purchased from Beagle Bioproducts

Inc. (Columbus, OH). The compound was stored in amber glass vials at -20°C. Aliquots were resuspended in deionized water before exposure.

65 3.3.2 Animals

Phase A of the study was conducted during May–June of 2016 using 8 male and 8 female

CD-1 mice (Charles River Laboratories; Spencerville, OH). The animals were 8 weeks of age with body weights ranging from 23.3– 32.9 grams at the time of MC-LR dosing.

Phase B of the study was conducted during January of 2017, and utilized 30 male and 30 female CD-1 mice received from (Charles River Laboratories). Animals were 7 weeks of age and weighed 21.6–31.5 grams at start of MC-LR dosing.

A stratified randomization scheme was used to assign animals to dosing groups with a goal of achieving similar group mean body weights. Males and females were randomized separately for both phases.

3.3.3 Animal Husbandry

All animal husbandry, handling, treatment administration and mouse sample collection was conducted by professionally trained staff at Charles River Laboratories (Spencerville, OH).

Individual animals were housed in wire mesh floor cages, were allowed 6 days of acclimation, and identified using numbered metal ear tags. Cages were housed in rooms on a 12-hour light/dark cycle and temperatures from 21°C–23°C, and with relative humidity of 47%–58%.

Mice received PMI Nutrition International Certified Rodent Chow No. 5CR4 (Purina Mills,

LLC; St. Louis, MO) (14% protein) and municipal water treated by reverse osmosis and ultraviolet irradiation water ad libitum via an automatic watering valve.

66 3.3.4 Experimental Protocol

For Phase A, animals were randomly assigned to three groups and administered 2000

µg/kg/day of MC-LR via oral gavage (Table 8). Group 1 received one treatment daily from days

1–7. Group 2 treatments were administered once daily from days 4–7. Group 3 treatments were

administered once on Day 7.

For Phase B, animals were randomly assigned to three groups and administered concentrations of MC-LR varying by assigned group (Table 9). Treatment was administered

once daily from days 1–7.

67 Table 8. Experimental Design for Phase A

MC-LR Dose Dose Group Dose Level Volume Concentration Number of Animals No. (µg/kg/day) (mL/kg) (µg/mL) Males Females 1a 2000 10 200 2 2 2b 2000 10 200 2 2 3c 2000 10 200 2 2 a Group 1 was dosed once daily on Days 1–7. b Group 2 was dosed once daily on Days 4–7. c Group 3 was dosed once on Day 7.

68 Table 9. Experimental Design for Phase B

MC-LR Dose Group Dose Level Dose Volume Concentration Number of Animals No. (µg/kg/day) (mL/kg) (µg/mL) Males Females 1 0a 10 0 10 10 2 3000 10 300 10 10 3 5000/4000b 10/8b 500 10 10 a Reverse osmosis deionized (RODI) water. b Due to mortality observed on Day 1, the dose level was changed to 4000 µg/kg/day on Day 2.

69 3.3.5 Animal Observations

General health, mortality, and moribundity checks were conducted twice daily, throughout the study. A further, once daily cage side observation was conducted throughout the dosing period (1–3 hours post-dose) without animal removal. A detailed clinical observation was conducted on the first day of randomization, the first day of dosing, and day 8 for each phase and group. During clinical observations, body weights were also recorded.

3.3.6 Euthanasia and Necropsy

All animals were euthanized by isoflurane inhalation followed by exsanguination.

Animals were euthanized rotating across dose groups such that similar numbers of animals from each group, including controls, were necropsied through the day. Animals were subjected to a complete necropsy examination which evaluated the musculoskeletal system, all external orifices and surfaces; the cranial cavity and external surfaces of the brain; and the thoracic, abdominal, and pelvic cavities with their associated organs and tissues.

3.3.7 Unscheduled Deaths

Necropsies were conducted for animals that died on study during Phase B. Animals were refrigerated prior to necropsy to minimize autolysis. Following death, animals were necropsied and specified tissues were collected and stored. One female (3F) was also euthanized prior to study conclusion, during Phase B due to deteriorating health per removal/exclusion criteria.

Clinical chemistry parameters and tissue samples were retained according to previously planned protocol.

3.3.8 Clinical Chemistry Parameters

Blood samples were collected from all animals (Phase A and Phase B) on Day 8 via a vena cava blood draw under isoflurane anesthesia at gross necropsy. Samples were also collected for animals that were euthanized moribund. Blood samples were processed to serum and evaluated for several clinical parameters (Table 10).

71 Table 10. Clinical Chemistry Parameters Examined During Study

Alanine aminotransferasea Albumin Aspartate aminotransferasea Globulin (calculated) Alkaline phosphatasea Albumin/globulin ratio Gamma-glutamyltransferasea Glucose Creatine Kinasea Cholesterol Total bilirubin Triglycerides Urea nitrogen Sodium Creatinine Potassium Calcium Chloride Phosphorus Sample Quality Total protein a Priority for collection.

72 3.3.7 Tissue Collection, Preservation, and Histopathology

During Phase A, one-half of the right liver lobe was collected and stored into a tube containing 2 mL of RNAlater (Thermo Fischer Scientific; Waltham, MA). Samples were stored overnight at 5°C, and transferred to -20°C after 24 hours. Due to the relative absence of effects found in Phase B, these samples were not analyzed.

In Phase B, livers were grossly examined, weighed and portions of each liver were collected for histopathology, fluorescent dye cut-loading, and gene expression analysis. Tissues for histopathology were fixed for 24 hours in 10% neutral buffered formalin (NBF), trimmed, embedded in paraffin, sectioned, and mounted on glass slides. Mounted samples were stained with hematoxylin and eosin. Histopathological evaluation was conducted by a board-certified veterinary pathologist.

3.3.8 Florescent Dye Cut-Loading

Immediately following removal and weighing, a portion of the right lobe of each mouse liver was rinsed in phosphate buffered saline (PBS), submerged in a solution of 0.04% Lucifer

Yellow CH in PBS (Millipore Sigma, Burlington, MA) and cut in half. Cut liver pieces remained immersed in dye for 5 minutes to permit uptake and transfer. Liver samples were rinsed in PBS and fixed by immersion in 10% NBF. After a 4-day fixation, tissues were clarified by soaking in

DMSO for 48 hours, and embedded in paraffin. Tissues were sectioned transversely to the dye- loaded, cut surface (5mm) by standard methods. Fluorescence microscopy was utilized to evaluate dye perfusion into cut-loaded tissue. Dye-coupled cells were counted at six random

73 points on each fixed liver sample (Figure 19). A representative mean was determined for each liver sample.

Figure 19. Quantification of fluorescent dye diffusion across cells via cut-loading assay.

74 3.3.9 Statistical Analysis

For Phase A, data were presented as values individual to each animal. For Phase B, all

statistical tests were conducted at the a = 0.05 significance level, while pairwise comparisons

used 2-sided tests reporting at a values of 0.001, 0.01, and 0.05.

Levene’s test was used to assess homogeneity of group variances. Comparisons across

these groups were done using ANOVA F-test or Kruskal-Wallis test, depending on normality. If

ANOVA or Kruskal-Wallis tests were found to be significant, pairwise comparisons were conducted using Dunnett’s or Dunn’s test, respectively. Datasets with two groups were compared using t-test or Wilcoxon Rank-Sum test, depending on normality determined by Levene’s test.

Differences in cellular GJIC across treatments and sexes were examined using ANOVA.

3.4 Results

3.4.1 Mouse Mortality, Health, and Clinical Observations

During Phase A, no MC-LR related mortalities, nor clinical changes, were recorded. All males, with one exception (Group 3), experienced slight net body weight losses during the study.

Female Groups 1 and 2 gained weight within normal limits, while Group 3 females gained only

0.15 g. Based on Charles River Laboratories historical data, male and female CD-1 mice are

expected to gain 1.5 g and 1.0 g body weight per week, respectively.

In Phase B, multiple mortalities related to MC-LR were noted in female Group 3 (3F)

(5000/4000 µg/kg/day). The first mortality occurred less than 3 hours following the only

injection of 5000 µg/kg/day, with the second coming prior to dosing on Day 2, in the same

group. Consequently, MC-LR dosage of Group 3 was lowered from 5000 to 4000 µg/kg/day,

75 beginning on Day 2. On Day 7, a third Group 3 female (3F) was euthanized moribund with

clinical signs including: abnormal gait, decreased activity, apparent hypothermia, labored

breathing, and decreased fecal output. Examination of the euthanized moribund female showed

moderate necrosis and degeneration in the liver. Only one male mortality was noted in Phase B,

occurring on Day 4 in Group 2 (2M) (3000 µg/kg/day). Histopathological analysis of deceased

mice found signs of marked perisinusoidal hemorrhaging and liver necrosis.

Apart from the clinical signs associated with the moribund euthanasia of the Group 3

female (3F), clinical signs were only found on Day 8. Clinical signs on Day 8 consisted of

decreased fecal output in one Group 2 female, and hunched posture and decreased fecal output

another Group 2 female. These signs were regarded as MC-LR related.

Group 3 males showed a decrease in body weight gain compared to Group 1, resulting in

~7% lower body weight than controls. Four Group 3 females (3F) showed a decrease in body

weight (0.3–3.8 g of body weight lost from Day 1–8). Females in Group 2 (2F) had lower mean

body weight gain than Group 1(1F) throughout the study, leading to a ~5% lower body weight

than Group 1. The mentioned changes in body weight were minimal and not dose related. No

clear MC-LR effects on body weight gain in Group 2 males or Group 3 females were seen.

3.4.2 Clinical Chemistry

Clear alterations in clinical chemistry parameters were seen in males and female clinical chemistry parameters. In males (2M) and female (2F) groups administered 3000 µg/kg/day MC-

LR elevations, at varying degrees of significance, elevation of AST, ALT, ALP, and bilirubin

were seen (Table 11 & Table 12). Elevations in cholesterol levels were also seen in group 2F,

76 while blood glucose levels declined 0.8-fold (p≤0.001) and 0.7-fold (p≤0.01) in groups 2M and

2F, respectfully. Changes in clinical chemistry were generally dose related in males (Figure 20), but not in females (Figure 21). Additionally, variations in clinical chemistry across individual mice were seen, showing elevations in bilirubin and/or serum enzymes beyond the previously mentioned parameters. Overall, female mice experienced higher magnitudes of clinical chemistry change than males (Table 13).

77 Table 11. Male clinical chemistry parameters.

AST ALT ALP TBIL CHOL GLUC Group (U/L) (U/L) (U/L) (mg/dL) (mg/dL) (mg/dL) 1M µ 49 39.1 77.2 0.148 157 199.9 s.e. 2.15 3.34 6.01 0.01 8.30 9.23

2M µ 91.67a 137.56b 116.56 0.203 205.11 b 159.44 b s.e. 8.77 22.38 8.03 0.01 12.37 8.33

3M µ 203.00c 419.00c 223.67c 0.600c 166.22 140.89 c s.e. 80.17 219.08 25.78 0.30 11.03 9.13

a,b, c Significantly different from group 1 value: a=p≤0.05; b=p≤0.01; c=p≤0.001 (Dunn)

3.5 ⁂ 3 ⁂ ⁂ [MC] 2.5 ** (µg/kg/bw) * 0 2 3000 1.5 5000/4000

(U/L)] log[parameter 1

0.5

Figure 20. CD-1 male mouse clinical chemistry parameters by exposure group. Groups were compared to the 0 µg/kg/bw group using Dunn’s test. Differences in results were significant at a levels of 0.05 (*), 0.01(**), and 0.001(⁂).

79 Table 12. Female clinical chemistry parameters.

AST ALT ALP TBIL CHOL GLUC Group (U/L) (U/L) (U/L) (mg/dL) (mg/dL) (mg/dL) 1F µ 73.50 63.50 103.70 0.14 109.50 206.40 s.e. 11.25 17.43 5.21 0.01 2.99 8.62

2F µ 952.40b 1574.50b 226.70c 0.330a 151.70 b 136.78 e s.e. 553.54 946.57 29.25 0.12 10.23 14.96

3F µ 891.43 857.00a 184.57a 0.226c 163.86 b 172.14 s.e. 793.15 731.71 17.39 0.01 12.97 24.44 a,b,c Significantly different from group 1 value: a=p≤0.05; b=p≤0.01; c=p≤0.001 (Dunn) d Significantly different from group 1 value: d=p≤0.01 (Dunnett)

Figure 21. CD-1 female mouse clinical chemistry parameters by exposure group. Groups were compared to the 0 µg/kg/bw group using Dunn’s and Dunnett’s tests. Differences found via Dunn’s test were significant at a levels of 0.05 (*) and 0.01(**).

81 Table 13. Maximal Magnitude of Change in Clinical Chemistry Parameters Maximal Fold Change

Observed Marker Abbrev. Males Females Aspartate aminotransferase AST 4.1 13 Alanine amniotransferase ALT 10.7 24.8 Alkaline Phosphatase ALP 2.9 2.2 Cholesterol N/A 1.5 Bilirubin 4.1 2.3

Mean Glucose (-0.8) (-0.7)

82 3.4.3 Histopathology

During Phase A, minimal hepatocellular hypertrophy was noted in exposed male mice, but not in female mice. Minimal to mild hepatocellular hypertrophy occurred in a dose- dependent manner in males and females. Hypertrophy was characterized by enlarged hepatocytes around central veins and midlobular areas, and was more pronounced in males than females

(Figure 22A).

Hepatocellular degeneration was seen in minimal to mild severity in both sexes.

Hepatocellular degeneration was characterized by the presence of swollen and individualized hepatocytes with pale nuclei and vacuolated cytoplasm. Hepatocellular necrosis ranged from minimal to moderate across sexes, and was characterized by hepatocytes exhibiting hypereosinophilic cytoplasm, nuclear pyknosis, karyorhexis, and karyolysis. Generally, these changes were dose dependent and clustered around central veins and midlobular areas, with degeneration progressing to necrosis. In severely affected livers, massive necrosis was characterized by dissociation of hepatic cords, with abundant hemorrhage, necrotic debris, degenerate neutrophils, and Kupffer cells in place (Figure 22B). Degeneration increased with dose in both sexes (Figure 22C). Mild hemorrhaging, characterized by presence of extravasated erythrocytes around the sinusoids and within the liver parenchyma, was seen in the 3M group

(Figure 22D).

Figure 22. Pathological findings in mice exposed84 to varying MC-LR concentrations. Four pathological parameters were noted across groups: A) Hypertrophy B) Necrosis C) Degeneration D) Hemorrhage. 3.4.4 Gap Junction Intercellular Communication

Microscopic examination of cut-load dyed hepatic tissue showed no statistically significant dose-dependent changes in GJIC across sexes or doses (Figure 23).

85 12

Dye Dye Positive Cells 4

Male Female Dose (µg/kg/day) Figure 23. Hepatocellular gap junction intercellular communication (GJIC) as measured by mean dye-positive cell count. No significant differences in gap junction communication were seen across treatments or sexes.

86 3.5 Discussion

The goal of this observational pilot study was to provide a comprehensive description of

the pathological, clinical chemistry, and GJIC related outcomes attributed to MC-LR ingestion in mice across sex and exposure dose. Phase A of the study aimed to determine a tolerable, sub-

lethal MC-LR dose. Upon exposure at varying intervals, data and observations of two male and

two female mice were used to determine that exposure to concentrations near 2000 µg/kg/day

would not result in excess mortality. Hepatocellular hypertrophy was observed in both daily-

exposed (Group 3) male mice in Phase A, but not the female mice. Based on findings of Phase

A, and past literature, doses of 3000 µg/kg/day and 5000 µg/kg/day were chosen to represent

88,164 sub-lethal and near-LD50 MC-LR exposures respectivelly. While the LD50 values cited in

past studies range from 3000–10,000 µg/kg/day varying by mouse strain, sex, and exposure

duration, we found the LD50 for orally administered MC-LR in CD-1 mice to be greater than

5000 µg/kg/day.32,88,164,169 However, due to 3 observed mortality events in the 3F group (5000

µg/kg/day MC-LR), the MC-LR dose was adjusted to 4000 µg/kg/day on the second day of

dosing.

Upon necropsy on Day 8, histopathologic examination of collected liver tissue showed

that premature death in 3 of 4 mice was attributed to marked intrahepatic hemorrhaging related

to MC-LR exposure.178 Hepatocellular hypertrophy, similar to Phase A, was observed in groups

2M (70%) and 2F (20%), exposed to 3000 µg/kg/day (Figure 22). Mice exposed to 5000/4000

µg/kg/day demonstrated higher rates of hypertrophy, with groups 3M and 3F showing 100% and

60% incidence of hypertrophy, respectively (Figure 22). While hypertrophy is a non-adverse

87 adaptive change in mammalian models, it is seemingly the earliest histopathologic alteration,

attributed to MC-LR ingestion, in the liver.

Microscopically determined differences in GJIC across MC-LR treatments and sexes were not statistically significant. GJIC has been shown to play a role in toxicity of agents, and has been previously shown to decrease in hepatocytes exposed to non-genotoxic compounds.180,182,186 We previously hypothesized that MC-LR would decrease GJIC, but results

show no significant changes amongst exposed mice (Figure 23). This finding may be explained

by the tendency of MC-LR to covalently bind to PP1/2A. The covalent binding of the MC-LR molecule to PP1/2A found in the nucleus and the cytoplasm, may make it unavailable for

transport via GJIC.187 In contrast, oxidative stress, one of the main causes of cellular damage

associated with MC-LR toxicity, has been shown to reduce GJIC in murine hepatocytes, in vitro,

therefore, some change in GJIC was expected.188 However, our results indicate that GJIC does

not play a role in increasing the hepatotoxic properties of MC-LR. Similar findings were noted

by Novakova et al (2011)189 and Blaha et al (2010)190 who showed no inhibition of GJIC by MC-

LR, in vitro. However, these studies did show that GJIC was impacted by exposure to

cyanobacterial Lysates, indicating that exposure to whole cyanobacterial cellular extracts may

play a role in cancer promotion. Therefore, the relationship between MCs and GJIC should be explored further in future works, using higher sample numbers and dose variations.

Difference in MC-LR related mortality and toxicity across sexes in CD-1 mice is the most notable finding of this study. Pathological examination noted a higher incidence of hepatocellular degradation and necrosis in male than female mice (Figure 22), while clinical chemistry parameter changes were much more severe in exposed females (Table 12) than

88 exposed males (Table 13). Cellular damage was clustered near central veins and midlobular areas, indicating that MC-LR is taken up by hepatocytes immediately upon transfer into the liver.

Upon uptake by the hepatocyte, MC-LR is metabolized or bound to cellular components,

preventing its spread beyond the immediate hepatocytes outside of central veins and localizing

its damage. This mechanism may also explain the unchanging rates of GJIC in exposed animals.

Exposure to MC-LR in males produced dose-dependent increases in ALT, AST, ALP, cholesterol, and bilirubin, while serum glucose decreased dose-dependently (Figure 20. CD-1

male mouse clinical chemistry parameters by exposure group. Groups were compared to the 0

µg/kg/bw group using Dunn’s test. Differences in results were significant at a levels of 0.05 (*),

0.01(**), and 0.001(⁂).; (Figure 20; Table 11). ALT, AST, ALP, cholesterol, and bilirubin levels also increased, while glucose decreased in female mice (Table 12). However, female effects were not dose-dependent and increases occurred at a much higher magnitude compared to male mice

(Figure 21). Variations in individual chemistry are hypothesized to be responsible for the lack of a dose-response relationship in females. The observed changes in clinical chemistry parameters indicate the localization of MC-LR toxicity in the liver, with AST/ALP ratios lower than those related to liver cell apoptosis due to non-alcoholic liver disease.173

Varying oral MC-LR toxicity across sexes of CD-1 mice can most likely be attributed to

difference in activity of GST and resulting GSH levels in male and female CD-1 mice. As MC-

LR detoxification is mainly catalyzed by GST enzymes, and relies on the production of GSH conjugates for the elimination of MC-LR metabolites, differences in GST activity and GSH concentration can impact toxicity.171 Previous studies have shown that CD-1 female mice have a

lower constitutive GST activity and, therefore, also produce lower levels of GSH

89 conjugates.175,191,192 Therefore, our results indicate that CD-1 female mice experience more negative liver health outcomes and are more likely to experience MC-LR related hepatotoxicity, than male mice.

Although, human ingestion of MC-LR at such high concentrations is unlikely to occur in nature, animals continue to be at risk of toxicity. Several past events have shown mortalities in dogs and cattle consuming large quantities of MC-LR contaminated water.22,26 These events can often occur in areas heavy in recreational use, small lakes, and agricultural ponds experiencing

HABs. Furthermore, while this study is not replicative of human ingestion, it lends insight into the toxicity of MC-LR in humans. While such a scenario is rare, cases of human mortality in individuals receiving dialysis treatment with MC contaminated water would be most at risk of exposure to MC-LR at analogous concentrations.96,97,101

In conclusion, this study successfully describes the impacts of MC-LR exposure on hepatocyte histopathology, clinical chemistry, and GJIC in vivo. Importantly, this appears to be the first study to report on variations in MC-LR toxicity across sexes while investigating GJIC alteration in CD-1 mice. Future works will explore the impact of MC-LR exposure on hepatocellular gene expression and study the adaptive response of exposed hepatocytes.

Prospective studies should take into consideration possible differences in MC-LR toxicity across sexes, and further investigate the relationship between MC-LR exposure and GJIC.

90 Chapter 4. Chronic Exposure to Environmentally Relevant Levels of Microcystin and its Role in Liver Cancer Promotion: A Two-Staged Carcinogenesis Model Study

4.1 Abstract

Microcystins (MCs) are hepatotoxins produced by over 40 cyanobacterial species found in harmful algal blooms (HABs) reported in fresh waterbodies worldwide. MC-LR, the most common and potent congener of MC can readily be found in eutrophic freshwater, including western Lake Erie. Exposure to MCs and other cyanobacterial hepatotoxins have previously been linked to promotion of liver carcinogenesis and increased liver cancer incidence in pre-clinical and epidemiologic studies. In this pilot study, we hypothesized that chronic ingestion, via drinking water, of MC-LR (via drinking water or food) or a complex bioactive mixture, extracted from Microcystis aeruginosa (Lysate) would promote liver cancer development in mice. Three groups of C3H/HeJ mice received one intraperitoneal (i.p.) injection of diethylnitrosamine

(DEN, complete chemical carcinogen) at 3 weeks of age. Three weeks later mice were administered ad libitum drinking water containing one of: 1) reverse osmosis, deionized water,

2) water with MC-LR (10 µg/L), 3) water with Lysate (10 µg/L total MC). Exposure

concentrations were based on environmentally relevant concentrations and previously

established Ohio EPA recreational water MC guidelines. Over the course of the study, mouse

weights, food consumption and water consumption were not significantly impacted by toxin

ingestion. Upon analysis, we found no significant differences in the number of gross and

histopathologic liver lesion counts across treatment groups. However, the proportion of lesions

91 classified as hepatocellular carcinomas in the MC-LR group (44%; p= 0.02) and Lysate (56%;

p=0.0073) were significantly higher compared to the pure water group (13%). Mice ingesting

Lysate also had a significantly increased mortality (35.6%; p=0.0009), compared to the other

groups, over the course of the study. This study uses an ingestion model analogous to real-world

exposure. Here, we note the cancer promoting effects of combinatorial exposures to multiple

MCs and bioactive compounds of cyanobacterial cells, and mortality associated with chronic

relatively low-dose exposure to MCs.

4.2 Introduction

Exposure to cyanotoxins, which may be produced during cyanobacterial harmful algal

blooms (HABs), has been a subject of several studies.60,63,169 The most common toxin formed

during freshwater HABs are the cyanobacterial toxins: microcystins (MCs). Microcystins are a

family of hepatotoxins with over 100 congeners, the most potent and common of which is

microcystin- leucine, arginine (MC-LR).119 The toxicity of MC-LR initiates with its transport

across cellular membranes via the organic anion transporting polypeptides (OATPs).46 Once

internalized into the hepatocyte, the ability of MC-LR to inhibit protein phosphatase 1 and 2A

(PP1/2A) disrupts cellular structures, produces reactive oxygen species (ROS), promotes cellular

survival, leads, necrosis to apoptosis and causes tissue inflammation.119,193 Previous cell-line studies have also linked exposure to cyanobacterial Lysates to GJIC inhibition, which is associated with diseases such as diabetes, neuropathy, and cancer.181,190,194 Overall, the

production of oxidative stress, the promotion of cellular proliferation, induction of inflammation,

and potential GJIC inhibition establish a cascading mechanism for hepatocarcinogenic

92 promotion.53 Additionally, MC-LR caused oncotic necrosis can compromise liver health, and

decrease liver function in impacted organisms.195

Epidemiological findings have also investigated the role of MC-LR in liver cancer

promotion. A study in the Three Gorges Dam area of China found higher rates of liver cancer

and childhood liver damage in areas with detected MC in food and water supplies.93,196 Similarly,

rates of primary liver cancer were higher in agricultural regions of Serbia impacted by HABs,

compared to those without.95 Further epidemiologic examination of Serbian HABs showed rates

of gonadal, gastric, colorectal, skin and primary liver cancer were all linked to presence of HABs

in surveyed regions.13 MC concentrations in some Serbian lakes exceeded 100 µ/L total MC,

well above World Health Organization (WHO) guideline levels.91,197 Epidemiological studies

have also found that HAB appearance was significantly related to non-alcoholic liver disease rates in the US.198 Echoing these findings, the International Agency for Research on Cancer

classifies MC-LR as a possible human carcinogen, while data on the carcinogenicity of other

MCs is limited.64,198 Data from these studies demonstrate the need for accurate evaluation of the

cancer promoting ability of these naturally occurring cyanotoxins.32,119

To address this need, several past studies have explored the liver cancer promoting ability

of MCs. These studies rely on mouse and rat models, as well as controlled exposure delivery

methods: intraperitoneal injection and oral gavage.68,199 Findings in animal studies exploring

carcinogenicity and cancer promotion of MCs have varying results primarily based on study

design. For example, Nishiwaki-Matsushima et al (1992)199 found tumor promotion in Fischer

rats initiated with diethylnitrosamine (DEN) and exposed to MC-LR via i.p. injection. Other

studies have noted increases in neoplastic lesions and induction of liver nodules during long-term

93 exposures.200,201 Labine & Milik (2015) found that exposure to 1 µg/L MC-LR did not result in

cancer development in mice when a cancer initiator was not used.202 This result is consistent with

our understanding of MC as a weak initiator of carcinogenesis, especially at such low

concentrations.176

The goal of this study is to explore the role of MC-LR and cyanobacterial Lysate as

potential liver tumor promoters in a two-stage mouse model of carcinogenesis, and to propose a novel means of animal exposure which better simulates ingestion of MCs in water. We chose to compare the potential promoting ability of MC-LR, the most potent and common congener of

MCs, to a cyanobacterial Lysate (“Lysate,” a complex mixture of MCs and bioactive cellular compounds) due to findings showing increased toxicity of Lysates versus pure toxins and potential GJIC inhibition in hepatocytes.190,203 As cyanobacterial concentrations during HABs

can exceed 100,000 cells/mL the added toxicity has been attributed to presence of multiple MC

congeners, cellular components, and endotoxins.203 Such exposures are expected to occur in

areas with insufficient water treatment methods and economically disadvantaged areas relying on

at-home water treatment. This study aims to simulates such real-world, chronic exposure

conditions, via at libitum ingestion of drinking water containing MC-LR and cyanobacterial

Lysate, bypassing the need for invasive exposure delivery techniques.

4.3 Methods

4.3.1 MC-LR

Purified MC-LR was purchased from Beagle Bioproducts (Columbus, OH). MC-LR was

stored at -20°C in glass amber vials until usage.

4.3.2 Lysate Preparation

A strain of Microcystis aeruginosa, isolated from Lake Erie, was cultured in CT media

204 supplemented with K(NO3), to maximize production of MCs (Table 14). Cyanobacterial cells

were cultured en batch at 22°C, for 21 days at 12-hour cycles of light and dark. Throughout the culturing period, cells were monitored for colony formation and overall density. On the 21st day, colonies were sonicated on ice and cell density was determined using light microscopy and hemocytometer counts. Cultures were harvested with varying cell densities (range: 1.19×105 –

1.48×106, mean: 8.5×105).

95 Table 14. CT media composition. All ingredients were mixed and autoclaved simultaneously.

Component Concentration Volume Water -- 941.92 mL

Ca(NO3)2, 4H2O 50 g/L 3 mL KNO5 100 g/L 1 mL

MgSO4 36.6 g/L 0.53 mL

P-Na2 glycocophosphate, 5H2O 100 g/L 0.25 mL Vitamin B12* 0.1 µ/L 100 µL Biotin* 0.1 µ/L 100 µL Thiamine HCl* 10 µ/L 100 µL T.A.P. 0.4 g/L 0.4 g

Na2EDTA H2O 0.75 g/L

MnCl2, 4H2O 0.4 g/L ZnCl , 7H O 0.005 g/L PIV 2 2 3 mL NaMoO4, 2H2O 0.004 g/L

FeCl3, 6H2O 0.097 g/L

CoCl2, 6H2O 0.002 g/L

Supplemental Nitrate [K(NO3), 4H2O] 1000 mM 25 mL *Ingredients were added post-sterilization to avoid degradation. Media was normalized to pH8.2, before inoculation with cyanobacteria.

96 Cyanobacterial cells were removed from culture solution using centrifugation (30 min ×

2,000g, 2°C), and re-suspended in deionized water to minimize possible ingestion of the

nutrient-rich culture medium. To ensure a uniform mixture, all culture concentrates were

consolidated into one batch prior to processing and analysis. Cells were freeze/thaw lysed five

times and total toxin concentrations were measured using Abraxis microcystins/nodularins

(ADDA) ELISA colorimetric immunoassay kit (ABRAXIS Inc. Prod #520011). Weekly aliquots

were created and stored at -20°C in 40 mL amber glass vials. Aliquots were normalized to equal

concentrations of total MCs.

4.3.3 M. aeruginosa Lysate Characterization

A sample of the M. aeruginosa Lysate was analyzed via liquid chromatography – mass

spectrometry (LC-MS), to determine the chemical composition of the complex mixture. The

sample was additionally screened for the presence of the neurotoxin β-N-methylamino-L-alanine

(BMAA) using the β-N-methylamino-L-alanine (BMAA) ELISA (Microtiter Plate) colorimetric

immunoassay kit (ABRAXIS Inc. Prod #520040).

4.3.4 Microcystin Adsorption to Plastic Assessment

To study the potential loss of MC via plastic bottle adsorption, a 7-day experiment was

conducted to monitor changes in MC concentrations over time. Clear plastic drip bottles were

wrapped in foil and filled with RO water containing 10 µg/L MC-LR or cyanobacterial Lysate containing 10 µg/L total MCs. Bottles were stored at room temperature with a 12hr light/dark cycle to replicate a typical week of toxin storage. Samples were collected daily and stored in -

97 20°C, in amber glass vials. Total MC concentrations were determined using the Abraxis ELISA

kit as previously described.

4.3.5 Animal Husbandry

Ninety-Six C3H/HeJ mice were purchased from Jackson Laboratories and randomized into three groups. C3H/HeJ mice were chosen due to their susceptibility for liver lesion development and resistance to lipopolysaccharide (LPS) endotoxin.205 Mice were received at 3 weeks of age and

given 1 week to acclimatize. All animals were housed in The Ohio State University (Columbus,

Ohio), Association for Assessment and Accreditation of Laboratory Animal Care International

accredited University Live Animal Resources vivarium, at standard conditions. Animals were ear

tagged, and housed 5 animals per cage, on average. All animals received AIN76A standardized

chow diet (Dyets, Inc., Bethlehem, PA). Animal weights, food and water consumption were

tracked weekly. Health checks were conducted daily by ULAR staff, veterinarians, and

investigators.

4.3.6 Experimental Design

At 4 weeks of age, all mice were i.p. injected with 5 µg/mg body weight of DEN, to initiate liver carcinogenesis. After a 3-week recovery period, drinking water exposure was administered via drip bottle (Figure 24). Bottles containing MC-LR and Lysate were wrapped in aluminum foil to prevent potential UV degradation. At 37 weeks of age, animals were necropsied and samples were collected.

Figure 24. Experimental design and timeline.

99 4.3.7 Gross and Histopathological Examination

At necropsy lives were excised, weighted and examined grossly for lesion count and

measurement. If present, lesions were enumerated and measured in two dimensions (length ×

width). Following gross examination, liver samples were sectioned, placed in pathology

cassettes, and fixed in 10% neutral buffered formalin for 24 hours. Liver samples were subsequently embedded in paraffin, sectioned, and stained using hematoxylin-eosin.

Histopathological examination was conducted by Dr. Krista LaPerle, a veterinary pathologist at

The Ohio State University Comprehensive Cancer Center Comparative Pathology and Mouse

Phenotyping Shared Resource. Lesion counts and classifications were normalized to the surface area of tissue observed using ImageJ software (National Institutes of Health, Version 1.52d).206

4.3.8 Statistical Analyses

All statistical analyses were conducted using Stata 10.4 software (StataCorp, LLC;

College Station, TX).207 Time series data were examined using linear regression methodology, lesion counts were analyzed using Kruskal-Wallis one-way analysis of variance, and lesion proportions were analyzed using a two-sided t-test for proportions. A Kaplan-Meier survivability analysis was conducted to analyze longitudinal mortality rates.

4.4 Results

4.4.1 M. aeruginosa Lysate Characterization

The cultured M. aeruginosa produced a mixture of toxins predominantly consisting of

MC-RR (37%), with smaller portions of MC-LR (22%), MC-YR (17%) and a variant of MC-RR

100 demethylated at the third amino acid position ([D-Asp3] MC-RR) (24%) (Figure 25). BMAA concentrations were below the limit of detection of the used kit (Abraxis Inc.), and were therefore not explored further.

101

6.0e6

MC-RR 5.5e6

5.0e6

4.5e6

4.0e6

3.5e6 3.0e6

2.5e6 2.0e6

1.5e6 MC-LR

1.0e6

[D-Asp3] MC-RR3.35

5.0e5

0.0

3.57

0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5 6.5 7.0 7.5 8.0 8.5 9.0 9.5 10.0 10.5 11.0 MC-YR 6.0 Ti me , mi n

Figure 25. Chemical composition of M. aeruginosa Lysate as Determined via LC-MS. Proportions of MCs in the cyanobacterial Lysate were as follows: MC-RR (37%); [D-Asp3] MC-RR (24%); MC-LR (22%); and MC-YR (17%).

102 4.4.2 Microcystin Adsorption by Plastic

No significant adsorption of MC compounds was seen over a 1-week period, across treatment methods (Figure 26). Slight variations in MC levels throughout the experiment can be attributed to limitations of the ELISA kit method and amplification of variations due to dilution.

103 10.0

8.0

6.0

4.0 [MC] (µg/L) 2.0

0.0 0 1 2 3 4 5 6 Day Lysate MC-LR Figure 26. Adsorption of microcystins onto drip bottle plastic. We noted no significant loss or adsorption of MC-LR or total MCs in the cyanobacterial Lysate in the drip bottles, over a 7-day period.

104 4.4.3 Health Metrics and Dosing

Over the course of the experiment, mouse body weights, food consumption and water consumption did not vary significantly across treatment groups. Equivalent consumption of water over the course of the study was crucial for assessing delivered dose and health assessment metrics (Figure 28). Additionally, mouse liver weight to body weight ratios did not differ significantly across treatments (Figure 27).

105

Figure 27. Mean water consumption, per mouse, over study duration. There were no significant differences in water consumption across treatment groups.

106

0.070

0.060

0.050

0.040

0.030

Liver Wegith/BodyLiver weight 0.020

0.010

0.000 Water MC-LR (10µg/L) Lysate (10µg/L MCs)

Figure 28. Mouse liver weight/ body weight ratios by treatment group. No significant differences in liver weight to body weight ratios were seen across treatment groups.

107 4.4.4 Lesion Counts

There were no significant differences neither in gross (Table 15) nor histopathologic

(Table 16; Figure 29) lesion count across treatment groups. Gross examination of livers showed varying sizes and stages of lesion development across treatment methods (Figure 30).

Histopathological lesion counts also showed no significant differences in lesion density across groups. However, we did note significantly higher proportions of lesions classified as

“carcinomas” in both the MC-LR and Lysate groups, compared to the water-only controls

(Figure 31).

108 Table 15. Lesion count and incidence as determined by gross examination.

Mice With Grossly Visible Lesions/Total Mice Lesions per Mouse Group (%) (Mean ± S.D.) Water 21/30 (70) 1.23 ± 0.27 MC-LR (10 µg/L) 9/22 (41) 0.68 ± 0.21 Lysate (10 µg/L Total MCs) 10/15 (67) 2.60 ± 0.93 *No significant differences in gross lesion counts were found across groups.

109 Table 16. Mean lesion counts per area of tissue (lesion count/mm2).

Preneoplastic Hepatocellular Group Foci Adenoma Carcinoma Total Lesion Water 0.018 ± 0.028 0.004 ± 0.018 0.003 ± 0.008 0.026 ± 0.034 MC-LR (10 µg/L) 0.008 ± 0.012 0.001 ± 0.003 0.007 ± 0.011 0.015 ± 0.019 Lysate (10 µg/L MCs) 0.015 ± 0.019 0.004 ± 0.009 0.023 ± 0.058 0.042 ± 0.068 *All values reported as "mean ± s.d."

110 0.060

) 2

0.040

0.020 Lesion Density (lesion/mm Density Lesion

0.000 Water MC-LR (10 µg/L) Lysate (10 µg/L MCs)

Preneoplastic Foci Adenoma Hepatocellular Carcinoma

Figure 29. Mean lesion density (lesion/mm2) across treatments. There were no significant differences in lesion counts across treatment.

111

Figure 30. Representative gross lesions in the DEN & water only (A,B) treatment group; Liver Lesions from the Lysate (C) and MC-LR (D) exposure groups, respectively.

112

Water Preneoplastic Foci 82.3%

Adenoma Carcinoma 2.8% 14.9%

MC-LR (10 µg/L) Lysate (10 µg/L Total MC) Preneoplastic Foci Preneoplastic Foci 51.0% 35.1%

Carcinoma Carcinoma Adenoma Adenoma 55.0% 44.5% 9.9% 4.5% (P <0.05) (P <0.01) Figure 31. Lesion proportions by classification across treatment methods. Histopathological lesion classification showed a significantly higher proportion of lesions classified as “carcinoma” in the Lysate (55%; p< 0.01) and MC-LR (44.5%; p<0.05) groups, compared to the ingestion of water (14.9%).

113 4.4.5 Mortality

During the study mortality events were observed in all three groups (Figure 32). The

Lysate group produced a significantly higher mortality rate (35.4%; p=0.0009) over the course of the experiment than mice ingesting MC-LR (4.3%) or water (3.2%).

114 100 96.8 95.7 90

80 p < 0.001 70 Survival Esitmate (%) Esitmate Survival 64.4 60 7 12 17 22 27 32 37 Age (Weeks)

Water MC-LR Lysate

Figure 32. Kaplan-Meier determined mouse survivability estimate by treatment. Chronic ingestion of cyanobacterial Lysate produced a significantly higher mortality rate (35.6%) than ingestion of MC-LR (4.3%) and water (3.2%), in mice.

115 4.5 Discussion

The goal of this pilot study was to simulate chronic, low-dose ingestion of MCs via a

drinking water route of exposure. Using MC ingestion via drinking water the study replicated

ingestion scenarios analogous to real-world conditions. During the study no significant variation

in food consumption and mouse weight attributable to the differences exposure conditions were

seen. Results also demonstrated equivalent water consumption (Figure 27) across treatment groups, indicating matched dosing over the course of the experiment. Aversion to neither the ad libitum consumption of MC-LR nor the cyanobacterial Lysate was noted over the course of the study.

Gross examination of mouse livers showed no significant differences in lesion counts across treatment methods (Table 15). Similarly, no significant changes in histopathologic lesion density normalized to liver tissue area were detected across treatments (Table 16; Figure 29).

However, proportions of lesions classified as carcinomas were significantly greater in the MC-

LR (44.5%) and Lysate (55.0%) treatment groups, compared to the water only exposure group

(15.9%) (Figure 31). Significant differences in proportions of carcinomas between the MC-LR and Lysate groups were not detected.

The appearance of carcinomas in Lysate treated mice may be underreported as we saw a significantly greater mortality (35.6%; p< 0.001) in the group, compared to the water (3.2%) and

MC-LR (4.3%) groups (Figure 32). Early mortalities often occurred before lesion promotion had progressed enough to produce lesions visible under gross and histopathological analysis. Limited observation of mice experiencing mortality showed no grossly visible lesions, therefore samples were not sent for histopathological analysis. Additionally, the mortalities seen during exposure

116 negatively impacted the statistical power of the study, making direct comparisons of lesion

counts difficult.

While others have noted a greater toxicity of cyanobacterial lysates compared to pure

toxins, this is the first study, to our knowledge, to demonstrate such a high mortality rate, in a

relatively low-dose chronic exposure model. Analysis of the Lysate shows that multiple MC

compounds including two forms of the MC-RR congener, MC-YR, and MC-LR exist in the mixture (Figure 25). Some previous works suggested that toxicity of such a cocktail is expected to be lower than similar concentrations of MC-LR.70 However, this was not seen during the

experiment.

To account for possible contributors to the increased Lysate toxicity, levels of the

cyanobacterial neurotoxin BMAA was assessed and showed to be below the level of detection in

Lysate. The presence of other common cyanotoxins was excluded as the isolated strain of M.

aeruginosa do not synthesize anatoxin-a and saxitoxin (Figure 2).22 Due to a mutation to the toll-

like receptor responsible for LPS toxicity, C3H/HeJ mice are hyporesponsive to LPS endotoxin,

supporting a negligible attribution of LPS to mortality rates.203,205,208 Therefore, the increase in

toxicity seen during this study can most likely be attributed to the presence of multiple congeners

of MC as well as pigments, acids, salts, and additional ROS which may assist in the toxicity of

MCs.71 Other studies have also postulated that increases in toxic response may be attributed to

unidentified metabolites in cyanobacteria, which has previously been shown to affect tumor

promotion.190 Therefore, the chemical make-up of cyanobacterial Lysates should be investigated

further, to better understand the interactions of bioactive compounds found in cyanobacterial

Lysates.

117 Future studies should also focus on expanding the study population, including use of both sexes of mice, and increasing statistical power to ensure subsequent analysis methods are sufficient at detecting differences in lesion counts across groups. A similar exposure method should be used. We successfully demonstrated the application of the drinking water exposure method. Further investigations should explore the pathological mechanisms attributed to combinatorial exposure to multiple MC compounds and bioactive cellular compounds. In particular, the bioactive compounds responsible for increases in toxicity should be studied.

118 Chapter 5. Early Warning of Near-Shore Harmful Algal Blooms Using Unmanned Aerial Vehicle (UAV) Technology: A Proof-of-Concept Study

This chapter has been submitted for review and publication in Harmful Algae.

5.1 Abstract

As reports of harmful algal blooms (HABs) increase worldwide, timely monitoring of

HAB severity in freshwater bodies used for drinking water, recreation and food production is urgently needed to protect human, animal and ecosystem health. Satellite remote sensing (RS) methods are useful for HAB monitoring on a specific temporal and large spatial scale.

Depending on methodology, temporal and spatial resolution, cloud cover, cost of deployment, and timeliness of deployment are all limiting factors in satellite RS. Additionally, monitoring of

HABs in nearshore areas and in small lakes and ponds continues to be challenging due to land adjacency effects and coarse spatial resolution of satellite imageries. This proof-of-concept pilot study proposes a novel method of image and water sample collection using unmanned aerial vehicle (UAV) technology and multispectral imaging for nearshore HAB-impacted lakes. Water color images and water samples from western Lake Erie and Buckeye Lake (Ohio, USA) were collected from June to August of 2016 using a custom designed UAV, camera, and water collection apparatus that was attached to the UAV. The designed system achieved a ground spatial resolution of 1.43 cm, at a 50m flight altitude. The method avoids common limitations of satellite imaging including cloud cover interference, land adjacency effects, deployment time, and cost. Normalized Difference Vegetation Index (NDVI) values were calculated from the

119 collected images, while phycocyanin pigment, cyanobacteria, and microcystin (MC) concentrations were determined from water samples. NDVI values in Buckeye Lake were correlated with increased phycocyanin content, while Lake Erie NDVI values were correlated with increases in chlorophyll-a and decreases in phycocyanin. Data suggest NDVI values in

Buckeye Lake are driven by cyanobacteria while Lake Erie values are driven by other photosynthesizes (i.e. diatoms). Future optimization of this model will use higher altitude image collection, gimbal mount systems for image stabilization, additional spectral bands exceeding those of the Moderate Resolution Imaging Spectroradiometer (MODIS) on NASA’s Terra and

Aqua orbiting platforms, and ground truth calibration data sets.

5.2 Introduction

Warm temperatures and eutrophication of freshwater sources around the world have led to the establishment of ideal conditions for the appearance of harmful algal blooms (HAB).27,43

HABs are a result of extreme proliferation of cyanobacterial species, often producing toxins, depleting oxygen content, and outcompeting natural fauna which can be harmful to humans, animals, and ecosystems.11,209,210 The family of cyanotoxins called microcystins, which may be produced during HABs have been shown to be cause liver toxicity in mammals and are suspected carcinogens.60,64 Consumption of these toxins has been linked to human and animal deaths in extreme circumstances, while skin and eye exposure can result in rashes and irritation.26,87,97,211

Due to the risk factors associated with exposures to HABs and their toxins, extensive research and monitoring methods have been devoted to better understand bloom ecology, track

120 occurrence, and establish warning systems.43,151,212,213 The use of satellite remote sensing (RS)

technology has been employed for these purposes.214–216 RS methodologies utilize the unique

reflectance characteristics of cyanobacteria manifest into water color pigments to estimate bloom intensity. Relative bloom intensity is calculated based on the relative reflectance of light corresponding to the presence of chlorophyll-a (670 nm) and phycocyanin (620 nm) pigments, and by calculating the Normalized Difference Vegetation Index (NDVI).217 NDVI is generated

using Equation 1, and relies on the intensity of red and near-infrared (NIR) reflectance bands.

= + 푁푁푁 − � 푁푁푁푁 Equation 1. Normalized Difference푁푁푁 � Vegetation Index

However, satellite remote sensing methods are limited in application due to several

factors. Firstly, cost of deploying and maintaining hardware is extensive, as all deployed

hardware needs to be positioned in orbit.218 Another limitation existent in satellite based remote

sensing is deployment time of instruments.219 Due to technological development times, costs of

launch, and a need for hardware deployment into orbit, developing new or updating existing

hardware can be time consuming. This lag time in terms of advancement may result in years

between hardware updates or fixes if a failure is to occur.

In terms of image collection, further limitations exist. For one, cloud cover can obscure

targeted sites and result in areas without data.220 Other limitations include the spatial and

temporal resolutions associated with the satellite systems used for data collection. One of the

most utilized remote sensing systems is the Moderate Resolution Imaging Spectroradiometer

121 (MODIS) onboard NASA’s Earth Observing System platforms, Terra and Aqua.221 The Terra- and Aqua-carried MODIS instruments produce daily to weekly sampled imageries with a spatial resolution of 1000 m and spectral bands adequate for HAB monitoring.221,222 This coarse resolution greatly limits MODIS imaging of water bodies smaller than 1 hectare in size. Even in large lakes, such as Lake Erie, MODIS imaging cannot be reliably applied in nearshore and coastal areas due to land adjacency effects, and inclusion of land into pixels which are meant to represent water.223 Land adjacency effects can introduce large variations in signal and impact reflectance in critical bands. While higher resolution imaging systems exist, access to this data often requires purchase or subscription, and still falls victim to the other limitations of satellite remote sensing.224 Even with higher resolution at 300 m (and 15 spectral bands, which is greater than MODIS), European Space Agency’s MEdium Resolution Imaging Spectrometer (MERIS) sensor onboard of the Environmental Satellite (Envisat) has a similar coastal contamination problem. This greatly limits its use in imaging of smaller and moderate sized lakes.

Finally, all satellites have a temporal resolution which considers the amount of time spent travelling around the Earth in order to revisit the same location twice. Temporal resolution varies greatly, and typically requires 1 to 3 days for satellites to image the same area twice.225

Combined with cloud cover, this could result in a large or serious gap in data when imaging a specific location during critical times of HAB occurrence.

For these reasons, a novel method of image collection is needed to ensure accurate, reliable, cost-effective, timely and high-resolution monitoring of small-to-moderate lakes and nearshore areas. The goal of this pilot study is to propose such a method, utilizing unmanned aerial vehicles (UAV) and multispectral imaging, and compare it to laboratory analysis methods.

122 Using UAV technology, we aim to develop a means of remote sensing which bypasses the traditional downfalls of satellite RS including: cloud interference; land adjacency effects in near shore areas; operation costs; deployment time and collection of water samples for validation. To accomplish this aim, we propose the use of UAV’s due to their relatively low costs, low altitude necessary for image collection, speed of deployment, and ease of hardware update and alteration.

5.3 Methodology

5.3.1 UAV

A notional image gathering mission was designed to demonstrate the effectiveness of

UAVs for near-shore sensing and sample collection. The UAV was based on a commercially available bare platform with performance metrics outlined in Table 17. The chosen frame was the DJI S1000+ carbon fiber frame (Figure 33A), consisting of eight motors with integrated speed controllers and power distribution board suitable for high-lift, and long endurance missions. An open source Pixhawk autopilot system was selected to perform the guidance, control, and sample collection and was integrated into the DJI S1000+ frame. Use of an open- source autopilot allows for rapid development and allows for inclusion of new path planning algorithms, and flight control modes for simultaneous imaging and physical sampling of the affected bodies of water.

123 Table 17. Operational parameter demands of UAV used for sample collection.

Parameter Value Flight velocity 10 m/s Maximum flight duration 20 minutes Maximum payload capacity 5 kg Payload volume 200 cm3 Onboard systems required Autopilot Telemetry link 4 Cameras (NDVI)

124 A B C

Figure 33. Design of UAV, imaging and sampling instruments. The DJI S1000+ carbon fiber frame (A) was fitted with four multispectral cameras (B) for image data collection. Upon image collection, the UAV was fitted with a water sampling apparatus (C) and used for water sample collection.

125 To meet flight duration requirements, a 10,000 mAh battery, with a maximal flight time of 18 min, was used. If needed, battery replacement could be done between flights in a rapid and simple way. The completed vehicle (frame, battery, and camera equipment) weighs 8 kg, with a maximal takeoff weight of 15 kg total. A Certificate of Authorization (COA) was issued by the

FAA to ensure all sampling flights were conducted in a safe and legal manner.

5.3.2 Multispectral Cameras

The design of the drone sampling system was predicated by a need to provide low-level multispectral imagery and rapid physical sampling of HAB. Four cameras, mounted on a plate, were used to mimic the multi-spectral bands found onboard MODIS (Figure 33B). The bands of interest consist of: visible spectrum (RGB); 650 nm; 680 nm; and 750 nm wavelengths. Four

MapIR cameras were utilized, along with three custom, narrow-band filters in place of IR block filter normally found on the MapIR system. One camera remained unmodified for RGB image collection. The filters used have a +/- 10 nm cutoff around the target wavelengths.

Cameras were mounted near the UAV’s center of gravity and aligned using ground based feature detection code scripted in Matlab (MathWorks 2018, Torrence, CA). Cameras were designed to take a single frame image every 3 seconds once triggered. Image collection was conducted using flight paths that were calibrated to conform to 80% side and 75% front overlap in successive images, as recommended by the manufacturer with 60° field of view. Camera resolution was 12 megapixels, with ground resolution of 1.43 cm, from a 50m flight altitude.

126 5.3.3 Sample Collection Device

A water sample collection device was required to sample areas of lakes containing the

HAB, with minimal disturbance of water surface due to rotor downwash, sample contamination, and no compromise to the flight characteristics of the UAV. To meet these criteria, a novel sample collection device was designed and employed (Figure 33C). A sling line, 20m in length, was used to avoid water disturbance. A novel cap system was developed to ensure sample contamination and spillage did not occur. Upon deployment, the sampler would fill with water, and would be lifted up by the UAV, triggering a plug valve to close and seal the sample. The buoyancy of the sampler was designed to be altered, as necessitated by sampling depth requirements. Three metal legs were used to ensure upright landing of the sampler upon return to shore.

5.3.4 Sampling and Flight Paths

Two sampling locations were chosen for UAV imaging to simulate potential real-world use. The first location sampled Lake Erie near public beaches at Maumee Bay State Park

(41°41'18.35" N, 83°23'13.00" W), near the Maumee River Inlet. The second location aimed to sample Gibson Island (39°55'24.80" N, 82°28'12.08" W), an area of Buckeye Lake, OH often used for recreation. Sample collection took place from June to August in 2016. For safety purposes, sampling did not take place when winds over water exceeded 15 mph.

A grid pattern was chosen as the most appropriate means of achieving an 80% side image overlap at a 50m flight altitude. The maximum UAV flight speed was limited to 10 m/s as required by MapIR cameras. Water sampling flights were carried out at three discrete sampling

127 points within the chosen grid, to produce a representative, composite water sample for each flight.

5.3.5 Image Processing and Analysis

Collected images were georeferenced using flight log data and timestamps generated by the onboard computer. Ideally, the images gathered could be used to construct a single orthomosaic for each sampling trip, showing the spatial distribution of the HAB. However, several issues made the construction of orthomosaics difficult and restrictive. Therefore, 20 representative images were taken from each sampling date, and a composite NDVI was generated using the red channel and the 750 nm reflectance bands. The representative NDVI was then compared to the representative water samples to validate the image collection method.

5.3.6 Water Filtration

Water samples collected using the UAV catchment apparatus were filtered using sterile

Microfilm V Filtration devices (Milipore Sigma) in conjunction with sterile membrane filters

(Isopore TM membrane filter; 0.22 um; HTTP, Milipore). This process was done in duplicate.

Filters were stored in a -20°C freezer until DNA extraction.

5.3.7 DNA Extraction

DNA was extracted from membrane filters using the xanthogenate-sodium dodecyl sulfate (XS) DNA extraction method.142,143 This method utilizes a modified DNeasy Blood &

Tissue Kit (QIAGEN group, Cat #69504) to maximize the extraction of cyanobacterial

128 DNA.142,143 DNA eluted from extraction columns was stored at -80°C until PCR analysis was

performed.

5.3.8 Presence of PCR Inhibitors

To test for the presence of PCR inhibitors in extracted DNA, a Sketa22 assay was conducted. We followed the method described by Haugland et al (2005).144 Samples which showed presence of PCR inhibition were diluted and retested until inhibition fell within appropriate limits. Dilutions were considered in calculations of gene copy numbers.

5.3.9 qPCR Quantification of Cyanobacterial Genes

Microcystis aeruginosa concentrations were quantified using qPCR methods targeting the

PC-IGS (total M. aeruginosa) and mcyE (M. aeruginosa capable of synthesizing MC) genes.

Previously published protocols, also used in Chapter 1 (Table 3 - Table 6),were utilized to ensure accuracy.213

Established protocols were also used to quantify Planktothrix sp. capable of synthesizing

MC by targeting the mcyE gene.226 Quantification was accomplished using a QX200 droplet

digital PCR (ddPCR) system (Bio-Rad). Each well (20 µL total volume) contained 2X QX200

ddPCR EvaGreen Supermix (Bio-Rad), 200 nM of each primer, sample DNA, and RNase-DNase

free water. The Droplet Generator (Bio-Rad) with 70 µL of QX200 Droplet Generation Oil for

EvaGreen (Bio-Rad) and 20 µL of the ddPCR mixture were used to generate droplets. A thermal

cycler (C1000 touch thermal cycler, Bio-Rad), was used to conduct standard PCR protocol.

129 Upon completion of thermal cycling, droplets were read using the Droplet Reader and

QuantaSoft software (Version 1.7; Bio-Rad).

5.3.10 Quantification of Total Microcystins

Total MCs concentrations were measured using the microcystins/nodularin (ADDA)

ELISA colorimetric immunoassay kit (ABRAXIS Inc. Prod #520011). Samples were tested in duplicate and mean concentrations of total MCs were reported. Some samples required dilutions to fall within the detection range of the ELISA kit (0.15 to 5 µg/L on average), which was considered when quantifying final toxin measurements. For plate reading purposes, a

SPECTRAmax PLUS 384 instrument was used (Molecular Devices, Silicon Valley, CA).

5.3.11 Quantification of Cyanobacterial Pigments

Measurements of the cyanobacterial pigments chlorophyll-a and phycocyanin were determined, in triplicate, using the Aquaflour 8000-010 spectroflourometer (Turner Designs, San

Jose, CA).

5.3.12 Data Analysis

Correlations between pigment, genetic measurements and NDVI values were investigated using multiple linear regression, conducted using Stata version 14 (College Station, TX).

Pairwise comparisons of means were conducted using sample t-tests for unknown variance.

130 5.4 Results

5.5.1 Sample Collection and Processing

Image collection was successfully conducted for all sampling dates, except for the

August 8th date (Buckeye Lake). During the August 8th sampling trip, a malfunction in the multispectral camera apparatus resulted in no image data collection. Water sample data collection was successfully conducted for all sampling times. In total, sample collection lasted 30 minutes or less, as limited by battery capacity. The specially designed sample collection apparatus functioned as intended in all cases. Further, substantial variation in image collection exists (Figure 34) due to UAV roll and pitch (Figure 35) during sampling.

131 Figure 34. Two successive images collected during a random sampling flight (Buckeye Lake). Images were intended to represent the same area of water sampled, yet were shown to be significantly different in practice.

132 60 Pitch Roll 40

0 Attitude (deg)

-20

-40 0 100 200 300 400

Figure 35. Pitch and roll dynamicsTime (s)during a random sampling flight. A graph of the flight dynamics seen during a typical image collection flight. The random nature of wind interference makes image stabilization difficult.

133 5.4.2 NDVI and Cyanobacterial Pigment Concentrations

Comparison of NDVI values and lab analyzed water samples was done using multiple

linear regression to study correlations between NDVI values and cyanobacterial pigment

concentrations (Figure 36). Throughout the sampling period, phycocyanin concentration had a

significant, positive correlation with NDVI values in Buckeye Lake. In Lake Erie, phycocyanin

had a significant, negative correlation with NDVI values, while chlorophyll-a had a significant, positive correlation.

134 Time

Phycocyanin

Chlorophyll-a

Figure 36. Influence of the time variable, phycocyanin, and chlorophyll-a concentrations on NDVI values of Buckeye Lake and Lake Erie. Increased phycocyanin concentrations were correlated with increased NDVI values in Buckeye Lake. Increased chlorophyll-a and decreased phycocyanin were correlated with increasing NDVI values in Lake Erie.

135 5.4.3 NDVI and PCR Determined Cyanobacterial Gene Concentrations

Genetic analysis of collected water samples showed that mean concentrations of genes

for total M. aeruginosa (PC-IGS; p<0.01) and MC-synthesizing M. aeruginosa (mcyE; p<0.01) were greater in Lake Erie samples than those from Buckeye Lake (Table 18). Concentrations of the MC synthesizing Planktothrix gene mcyE, were greater in Buckeye Lake when compared to

Lake Erie (p<0.001) (Table 18). Throughout the study, we saw a higher concentration of MC toxins and a beach closure (due to high MC content) during the first three sampling periods in

Buckeye Lake, however due to large variance in MC concentrations and lack of population data, differences were not significant (Table 18). Genetic markers of Planktothrix sp. and Microcystis aeruginosa, were not correlated with observed NDVI values in either study location (Figure 37).

136 Time

M. aeruginosa mcyE

Planktothrix sp. mcyE

Figure 37. Influence of time, M. aeruginosa mcyE, and Planktothrix sp. mcyE on NDVI values of Buckeye Lake and Lake Erie. No significant correlation was seen between NDVI values and genetic markers of cyanobacteria.

137 Table 18.Mean microcystin (MC) concentrations (µg/L), gene concentrations (log [gene copies/mL]), and NDVI values across locations. M. aeruginosa M. aeruginosa Planktothrix [MC] NDVI PC-IGS mcyE sp. mcyE

Buckeye x̅ 4.51 2.30 3.13* 24.33 0.349 Lake variance 0.12 0.31 0.15 727.14 0.002 x̅ 5.86* 4.12* 0.70 5.31 0.290 Lake Erie variance 0.733 1.35 0.61 7.88 0.007 *Indicates significantly greater value compared to other location (p<0.01)

138 5.4.4 Concentrations of Microcystin

ELISA determined, overall mean total MC concentrations were not significantly higher

(a = 0.05) in Buckeye Lake than Lake Erie samples. However, we noted beach closures and algal toxin advisories during the first three sampling periods at Buckeye Lake. No such signage was posted during Lake Erie samplings.

5.5 Discussion

5.5.1 UAV Image Collection

UAV based image collection was timely and successful. Average time of UAV deployment for each sampling attempt was 30 min or less. While this method is promising, several barriers to real-world implementation still exist. First, the use of 12-megapixel cameras, in theory, produces a ground resolution of 1.43 cm, at a 50-m flight altitude. However, in practice, due to variability in water surface this was not realized, proving problematic as water surface fluctuated over time. Another issue is the density of data collected. Due to such a fine resolution images were pixel dense, yet covered a relatively small surface area. This resulted in increased time of data handling, manipulation, and processing.

Further complications were encountered during UAV flight. Pitch and roll upsets to UAV flight were on the order of 35 degrees, resulting in an uncertainty level of 35 m for ground level images from a 50-m height (Figure 35). Consequently, two successive images (Figure 34) collected over the same location were shown to be significantly different due to pitch and roll attributed to high winds over the body of water. These complications can be solved in future applications using a gimbal mount system to stabilize cameras during unfavorable conditions.

139

5.5.2 Image Processing

As with image collection, several barriers to implementation exist with image processing.

Perhaps the most troublesome is the lack of identifying landmarks over water areas. While an orthomosaic was constructed for nearshore areas, the inclusion of 25% of images was not possible due to lack of identifying features over water. This is problematic as at an altitude of

50m the surface area coverage of each produced orthomosaic is not sufficient for concrete conclusions to be drawn. Furthermore, significant delay and clock bias was seen between cameras, resulting in a spatial error of uncertainty on the order of +/- 10 m. An overlap comparison of images with distinct features narrowed this error down to +/- 3 m, but the same could not be done for water only images due to drift and movement of the water surface.

In terms of image comparison, for UAV data to be directly compared to MODIS collected data, ground reflectance calibration needs to be developed. A ground proofing protocol, using the same proofs utilized by MODIS should be administered. This ensures collected data are interchangeable across methods. The currently employed method utilizes calibrated reflectance estimates for the four targeted wavelengths. The data presented here rely on a normalized index in lieu of a ground truth. This approach is sufficient in analyzing the qualitative spatial distribution of HABs, but produces reflection artifacts and “black spots” in the orthomosaic, making quantification of all HAB values difficult. This issue can be addressed using ground proofs and more intensive post-processing methods.

140 5.5.3 NDVI and Lab Data

Due to the difficulties of image processing and mosaic formation discussed previously,

NDVI was calculated based on a mean NDVI of 20 randomly selected images from each date and location. Multiple linear regression analysis showed that composite NDVI values in Lake

Erie were significantly impacted by phycocyanin and chlorophyll-a concentrations (Figure 36).

The most novel finding of this pilot study was the correlation of lab data with observations during sampling and NDVI values. We noted that, higher phycocyanin levels translated to lower

NDVI, while increases in chlorophyll-a concentrations correlated with increased NDVI in Lake

Erie (Figure 36). These data suggest that NDVI values in Lake Erie are driven by non- cyanobacterial primary producers, most likely diatoms. Findings also show that the opposite is true in Buckeye Lake, where higher NDVI values correlated with increased phycocyanin

(cyanobacteria) presence (Figure 36). These data agree with the observation of HABs in Buckeye

Lake during sampling periods, while none were detected in Lake Erie. Further investigation of this phenomenon is needed, as our dataset is limited and should be expanded further.

Overall, concentrations of total M. aeruginosa (PC-IGS) and MC-producing M. aeruginosa (mcyE) were greater in Lake Erie than in Buckeye Lake (Table 18). However, concentrations of the MC-producing Planktothrix sp. were higher in Buckeye Lake than Lake

Erie. We also noted higher mean concentrations of MC toxins in the Buckeye Lake samples

(Table 18). However, due to large sample variance and small dataset, differences in MC concentrations were not significant. The relative increase in MCs was attributed to the lowered water levels in Buckeye Lake, resulting from the Army Corps of Engineers’ drainage of the lake for dam construction, and the dominance of toxin producing Planktothrix spp.227

141 In summary, this proof-of-concept pilot study provides valuable insight on the potential application of UAV technology for remote sensing of HABs in nearshore environments. The goal of this study was to propose and evaluate the feasibility of UAV remote sensing methods for

HAB monitoring purposes. Our findings demonstrate that UAV remote sensing can provide rapid data collection, meaningful imaging data which correlates to laboratory findings, and be deployed in small lakes and near-shore areas. Further research will focus on optimization of this method, including: camera stabilization; spatial image linkage; and data standardization to satellite proofs.

142 Chapter 6. Conclusion

As frequency and intensity of cyanobacterial harmful algal blooms (HABs) increases worldwide, a deeper, more thorough understanding of causes of HABs, the risks they pose, and exposure prevention is required. Chapter 1 explored the history of the cyanobacteria, their ecological roles, and our current understanding of the impacts which drive HAB formation. The chapter also discusses several cyanotoxins, with special focus on the most abundant toxin: microcystin. Chapter 1 also details the mechanisms of MC toxicity, historic exposure events, and potential exposure pathways. Chapter 2 addresses the appearance and formation of HABs in small lakes and ponds (SLaPs), in central Ohio. It describes the impact of tile drainage on HABs and the changes in water chemistry and microbial community in blooming SLaPs. The study found not only high concentrations of MC in one SLaP, but drastic impacts of HABs on microbial communities. These factors were driven by tile drainage, and present a risk for human and animal exposure to MC in SLaPs.

Due to the observed exposure risks noted in Chapter 2, Chapter 3 focuses on examining the health outcomes in CD-1 mice associated with acute MC-LR exposure. MC-LR is the most potent and common congener of MC, therefore an animal study was conducted to describe the impacts of MC-LR on hepatic health between male and female. The study found significant changes in pathological and clinical chemistry parameters of health. Gap junction intracellular communication was not impacted by MC-LR exposure. Similarly, Chapter 4 explored the cancer

143 promoting ability of chronically ingested MC-LR and a Microcystis aeruginosa Lysate containing a complex mixture of toxins and cyanobacterial cell matter, in a C3H/HeJ mouse model. The study found significantly higher mortality rates in mice consuming the M. aeruginosa Lysate, compared to the other groups. The study also hints at a role of MCs as liver tumor promoters and demonstrates their ability to further the development of lesions from foci to carcinomas.

To address a need for exposure prevention to these toxins, a novel means of monitoring and sampling near-shore waters and SLaPs was proposed in Chapter 5. The aim of the study was to address the limitations of current remote sensing methods. To this end, an unmanned aerial vehicle (UAV) equipped with multispectral cameras and water sampling device was developed and deployed for sample collection. Image data was collected and NDVI values were calculated to quantify water color at each sampling site. Lab data findings were compared to calculated

NDVI values and found correlations between cyanobacterial pigments and NDVI values. UAV application promises to improve on the limitation of current satellite remote sensing methods, yet requires further optimization. Overall, this work detail the impacts of human influences on, and the consequences of, HABs. Concerning the ecological and health impacts of MC, the study furthers our current understanding of the microbial community and health impacts of MC exposure. Finally, this work proposes a novel method to prevent human and animal exposure to

MC using UAV methodology, in near-shore and small lake settings.

144 Bibliography

1. Knoll A. The Cyanobacteria: molecular biology, genomics, and evolution. In: Herrero A, Flores E, eds. Cyanobacteria and Earth History. Norfolk, UK: Caister Academic Press; 2008:484. 2. Paerl HW, Paul VJ. Climate change: Links to global expansion of harmful cyanobacteria. Water Res. 2012;46(5):1349-1363. doi:10.1016/J.WATRES.2011.08.002 3. Schirrmeister BE, Antonelli A, Bagheri HC. The origin of multicellularity in cyanobacteria. BMC Evol Biol. 2011;11(1):45. doi:10.1186/1471-2148-11-45 4. Schopf JW. The fossil record of cyanobacteria. In: Ecology of Cyanobacteria II: Their Diversity in Space and Time. Dordrecht: Springer Netherlands; 2013:15-36. doi:10.1007/978-94-007-3855-3_2 5. Olson JM. Photosynthesis in the Archean Era. Photosynth Res. 2006;88(2):109-117. doi:10.1007/s11120-006-9040-5 6. Blank C, Sanchez-Baracaldo P. Timing of morphological and ecological innovations in the cyanobacteria - a key to understanding the rise in atmospheric oxygen. Geobiology. 2010;8(1):1-23. doi:10.1111/j.1472-4669.2009.00220.x 7. Paerl HW, Xu H, Hall NS, et al. Controlling cyanobacterial blooms in hypertrophic Lake Taihu, China: Will nitrogen reductions cause replacement of non-N2 Fixing by N2 fixing taxa? PLoS One. 2014;9(11). doi:10.1371/journal.pone.0113123 8. Michalak AM, Anderson EJ, Beletsky D, et al. Record-setting algal bloom in Lake Erie caused by agricultural and meteorological trends consistent with expected future conditions. Proc Natl Acad Sci U S A. 2013;110(16):6448-6452. doi:10.1073/pnas.1216006110 9. Van Esbroeck CJ, Macrae ML, Brunke RI, McKague K. Annual and seasonal phosphorus export in surface runoff and tile drainage from agricultural fields with cold temperate climates. J Great Lakes Res. 2016;42(6):1271-1280. doi:10.1016/j.jglr.2015.12.014 10. Smith DR, King KW, Johnson L, et al. Surface runoff and tile drainage transport of phosphorus in the midwestern United States. J Environ Qual. 2015;44(2):495-502. doi:10.2134/jeq2014.04.0176 11. Berry MA, Davis TW, Cory RM, et al. Cyanobacterial harmful algal blooms are a biological disturbance to Western Lake Erie bacterial communities. Environ Microbiol. 2017;19(3):1149-1162. doi:10.1111/1462-2920.13640 12. Muenich RL, Kalcic M, Scavia D. Evaluating the Impact of Legacy P and Agricultural Conservation Practices on Nutrient Loads from the Maumee River Watershed. Environ Sci Technol. 2016;50(15):8146-8154. doi:10.1021/acs.est.6b01421 13. Svirčev Z, Drobac D, Tokodi N, et al. Epidemiology of Cancers in Serbia and Possible Connection with Cyanobacterial Blooms. J Environ Sci Heal Part C. 2014;32(4):319-337.

145 doi:10.1080/10590501.2014.967053 14. Zhang F, Lee J, Liang S, Shum CK. Cyanobacteria blooms and non-alcoholic liver disease: evidence from a county level ecological study in the United States. Environ Health. 2015;14:41. doi:10.1186/s12940-015-0026-7 15. Xie L, Rediske RR, Hong Y, et al. The role of environmental parameters in the structure of phytoplankton assemblages and cyanobacteria toxins in two hypereutrophic lakes. Hydrobiologia. 2012;691(1):255-268. doi:10.1007/s10750-012-1077-1 16. Lürling M. Effects of microcystin-free and microcystin-containing strains of the cyanobacterium Microcystis aeruginosa on growth of the grazer Daphnia magna. Environ Toxicol. 2003;18(3):202-210. doi:10.1002/tox.10115 17. Burns NM, Rockwell DC, Bertram PE, Dolan DM, Ciborowski JJH. Trends in Temperature, Secchi Depth, and Dissolved Oxygen Depletion Rates in the Central Basin of Lake Erie, 1983–2002. J Great Lakes Res. 2005;31:35-49. doi:10.1016/S0380- 1330(05)70303-8 18. Weinke AD, Biddanda BA. From Bacteria to Fish: Ecological Consequences of Seasonal Hypoxia in a Great Lakes Estuary. Ecosystems. 2017:1-17. doi:10.1007/s10021-017-0160- x 19. Steffensen D. Economic Cost of Cyanobacterial Blooms. In: Advances in Experimental Medicine and Biology. Vol 619. New York, NY: Springer Science+Business Media, LLC; 2008:855-865. doi:10.1007/ 978-0-387-75865-7 20. City of Toledo. 2018 Proposed Annual Operating Budget. Toledo, OH; 2017. https://toledo.oh.gov/media/4786/2018-proposed-for-website.pdf. Accessed May 3, 2018. 21. Cho M, Chung H, Yoon J. Disinfection of Water Containing Natural Organic Matter by Using Ozone-Initiated Radical Reactions. Appl Environ Microbiol. 2003;69(4):2284-2291. doi:10.1128/AEM.69.4.2284 22. Carmichael WW, Boyer GL. Health impacts from cyanobacteria harmful algae blooms: Implications for the North American Great Lakes. Harmful Algae. 2016;54:194-212. doi:10.1016/j.hal.2016.02.002 23. Pantelić D, Svirčev Z, Simeunović J, Vidović M, Trajković I. Cyanotoxins: Characteristics, production and degradation routes in drinking water treatment with reference to the situation in Serbia. Chemosphere. 2013;91(4):421-441. doi:10.1016/j.chemosphere.2013.01.003 24. Atech Group., Land and Water Resources Research and Development Corporation (Australia), Murray-Darling Basin Commission (Australia). Cost of Algal Blooms : Final Report. Land & Water Resources Research & Development Corporation; 1999. https://catalogue.nla.gov.au/Record/2336311. Accessed May 3, 2018. 25. Cao Q, Rediske RR, Yao L, Xie L. Effect of microcystins on root growth, oxidative response, and exudation of rice (Oryza sativa). Ecotoxicol Environ Saf. 2018;149(October 2017):143-149. doi:10.1016/j.ecoenv.2017.11.020 26. Hilborn ED, Beasley VR. One health and cyanobacteria in freshwater systems: Animal illnesses and deaths are sentinel events for human health risks. Toxins (Basel). 2015;7(4):1374-1395. doi:10.3390/toxins7041374 27. Paerl HW, Huisman J. Blooms like it hot. Science. 2008;320(5872):57-58. doi:10.1126/science.1155398

146 28. Lambert TW, Holmes CFB, Hrudey SE. Microcystin class of toxins: health effects and safety of drinking water supplies. http://www.nrcresearchpress.com/doi/pdf/10.1139/a94- 011. Accessed May 3, 2018. 29. Lampert W. Further studies on the inhibitory effect of the toxic blue-green Microcystis aeruginosa on the filtering rate of zooplankton. Arch Hydrobiol. 1982;95:207-220. 30. Lampert W. Inhibitory and Toxic Effects of Blue-green Algae on Daphnin. Int Revile gas Hydrobiol ~. 1981;66(1):285-298. https://onlinelibrary.wiley.com/doi/pdf/10.1002/iroh.19810660302. Accessed May 3, 2018. 31. Kaplan A, Harel M, Kaplan-Levy RN, Hadas O, Sukenik A, Dittmann E. The languages spoken in the water body (or the biological role of cyanobacterial toxins). Front Microbiol. 2012;3:138. doi:10.3389/fmicb.2012.00138 32. Gupta N, Pant SC, Vijayaraghavan R, Rao PVL. Comparative toxicity evaluation of cyanobacterial cyclic peptide toxin microcystin variants (LR, RR, YR) in mice. Toxicology. 2003;188(2-3):285-296. doi:10.1016/S0300-483X(03)00112-4 33. Nishizawa T, Asayama M, Shirai M. Cyclic heptapeptide microcystin biosynthesis requires the glutamate racemase gene. Microbiology. 2001;147(5):1235-1241. doi:10.1099/00221287-147-5-1235 34. Carmichael WW. Cyanobacteria secondary metabolites-the cyanotoxins. J Appl Bacteriol. 1992;72(6):445-459. doi:10.1111/j.1365-2672.1992.tb01858.x 35. Meissner K, Dittmann E, Börner T. Toxic and non-toxic strains of the cyanobacterium Microcystis aeruginosa contain sequences homologous to peptide synthetase genes. FEMS Microbiol Lett. 1996;135(2-3):295-303. http://www.ncbi.nlm.nih.gov/pubmed/8595871. Accessed May 2, 2018. 36. Neilan BA, Dittmann E, Rouhiainen L, et al. Nonribosomal Peptide Synthesis and Toxigenicity of Cyanobacteria. J Bacteriol. 1999;181(13):4089-4097. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC93901/pdf/jb004089.pdf. Accessed May 2, 2018. 37. Christiansen G, Fastner J, Erhard M, Börner T, Dittmann E. Microcystin biosynthesis in Planktothrix: genes, evolution and manipulation. Microbiology. 2003;185(2):564-572. doi:10.1128/JB.185.2.564 38. Tillett D, Dittmann E, Erhard M, von Döhren H, Börner T, Neilan BA. Structural organization of microcystin biosynthesis in Microcystis aeruginosa PCC7806: an integrated peptide–polyketide synthetase system. Chem Biol. 2000;7(10):753-764. doi:10.1016/S1074-5521(00)00021-1 39. Kaebernick M, Neilan BA, Rner TB, Dittmann E. Light and the Transcriptional Response of the Microcystin Biosynthesis Gene Cluster. Appl Environ Microbiol. 2000;66(8):3387- 3392. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC92160/pdf/am003387.pdf. Accessed May 2, 2018. 40. Boopathi T, Ki J-S. Impact of environmental factors on the regulation of cyanotoxin production. Toxins (Basel). 2014;6(7):1951-1978. doi:10.3390/toxins6071951 41. Sivonen K. Effects of Light, Temperature, Nitrate, Orthophosphate, and Bacteria on Growth of and Hepatotoxin Production by Oscillatoria agardhii Strains. Appl Environ Microbiol. 1990;56(9):2658-2666.

147 https://www.ncbi.nlm.nih.gov/pmc/articles/PMC184825/pdf/aem00090-0078.pdf. Accessed May 3, 2018. 42. Wang Q, Niu Y, Xie P, et al. Factors affecting temporal and spatial variations of microcystins in Gonghu Bay of Lake Taihu, with potential risk of microcystin contamination to human health. ScientificWorldJournal. 2010;10:1795-1809. doi:10.1100/tsw.2010.172 43. Paerl HW, Scott JT, McCarthy MJ, et al. It Takes Two to Tango: When and Where Dual Nutrient (N & P) Reductions Are Needed to Protect Lakes and Downstream Ecosystems. Environ Sci Technol. 2016;50(20):10805-10813. doi:10.1021/acs.est.6b02575 44. Chaffin JD, Davis TW, Smith DJ, Baer MM, Dick GJ. Interactions between nitrogen form, loading rate, and light intensity on Microcystis and Planktothrix growth and microcystin production. Harmful Algae. 2018;73:84-97. doi:10.1016/j.hal.2018.02.001 45. Monchamp M-E, Pick FR, Beisner BE, Maranger R. Nitrogen Forms Influence Microcystin Concentration and Composition via Changes in Cyanobacterial Community Structure. Neilan B, ed. PLoS One. 2014;9(1):e85573. doi:10.1371/journal.pone.0085573 46. Fishcer WJ, Altheimer S, Meier PJ, Dietrich DR, Hagenbuch B. Organic anion transporting polypeptides expressed in liver and brain mediate uptake of microcystin. Toxicol Appl Pharmacol. 2005;203:257-263. 47. Komatsu M, Furukawa T, Ikeda R, et al. Involvement of Mitogen-Activated Protein Kinase Signaling Pathways in Microcystin-LR-Induced Apoptosis after its Selective Uptake Mediated by OATP1B1 and OATP1B3. Toxicol Sci. 2007;97(2):407-416. doi:10.1093/toxsci/kfm054 48. Hagenbuch B, Gao B, Meier PJ. Transport of xenobiotics across the blood-brain barrier. News Physiol Sci. 2002;17:231-234. http://www.ncbi.nlm.nih.gov/pubmed/12433976. Accessed May 3, 2018. 49. MacKintosh RW, Dalby KN, Campbell DG, Cohen PTW, Cohen P, MacKintosh C. The cyanobacterial toxin microcystin binds covalently to cysteine-273 on protein phosphatase 1. FEBS Lett. 1995;371(3):236-240. doi:10.1016/0014-5793(95)00888-G 50. Bagu JR, Sykes BD, Craig MM, Holmes CFB. A molecular basis for different interactions of marine toxins with protein phosphatase-1: Molecular models for bound motuporin, microcystins, okadaic acid, and calyculin A. J Biol Chem. 1997;272(8):5087-5097. doi:10.1074/jbc.272.8.5087 51. MacKintosh C, Beattie KA, Klumpp S, Cohen P, Codd GA. Cyanobacterial microcystin- LR is a potent and specific inhibitor of protein phosphatases 1 and 2A from both mammals and higher plants. FEBS Lett. 1990;264(2):187-192. doi:10.1016/0014- 5793(90)80245-E 52. Falconer IR, Yeung DSK. Cytoskeletal changes in hepatocytes induced by Microcystis toxins and their relation to hyperphosphorylation of cell proteins. Chem Biol Interact. 1992;81(1-2):181-196. doi:10.1016/0009-2797(92)90033-H 53. Ding WX, Shen HM, Zhu HG, Ong CN. Studies on oxidative damage induced by cyanobacteria extract in primary cultured rat hepatocytes. Environ Res. 1998;78(1):12-18. doi:10.1006/enrs.1998.3843 54. Campos A, Vasconcelos V. Molecular mechanisms of microcystin toxicity in animal cells. Int J Mol Sci. 2010;11(1):268-287. doi:10.3390/ijms11010268

148 55. Chen L, Zhang X, Zhou W, et al. The Interactive Effects of Cytoskeleton Disruption and Mitochondria Dysfunction Lead to Reproductive Toxicity Induced by Microcystin-LR. PLoS One. 2013;8(1). doi:10.1371/journal.pone.0053949 56. Li X, Zhao Q, Zhou W, Xu L, Wang Y. Effects of chronic exposure to microcystin-LR on hepatocyte mitochondrial DNA replication in mice. Environ Sci Technol. 2015;49(7):4665-4672. doi:10.1021/es5059132 57. Chen T, Wang Q, Cui J, et al. Induction of apoptosis in mouse liver by microcystin-LR: a combined transcriptomic, proteomic, and simulation strategy. Mol Cell Proteomics. 2005;4(7):958-974. doi:10.1074/mcp.M400185-MCP200 58. Clark SP, Ryan TP, Searfoss GH, Davis M a, Hooser SB. Chronic microcystin exposure induces hepatocyte proliferation with increased expression of mitotic and cyclin- associated genes in P53-deficient mice. Toxicol Pathol. 2008;36(2):190-203. doi:10.1177/0192623307311406 59. Svircev Z, Baltić V, Gantar M, Juković M, Stojanović D, Baltić M. Molecular aspects of microcystin-induced hepatotoxicity and hepatocarcinogenesis. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev. 2010;28(1):39-59. doi:10.1080/10590500903585382 60. US EPA. Health Effects Support Document for the Cyanobacterial Toxin Microcystins. EPA-820R15102.Washington, DC; 2015. doi:820R15102 61. Deng X, Ito T, Carr B, Mumby M, May WS. Reversible phosphorylation of Bcl2 following interleukin 3 or bryostatin 1 is mediated by direct interaction with protein phosphatase 2A. J Biol Chem. 1998;273(51):34157-34163. doi:10.1074/JBC.273.51.34157 62. Guzman-Guillen R, Puerto M, Gutierrez-Praena D, et al. Potential use of chemoprotectants against the toxic effects of cyanotoxins: A review. Toxins (Basel). 2017;9(6). 63. He J, Li G, Chen J, et al. Prolonged exposure to low-dose microcystin induces nonalcoholic steatohepatitis in mice: a systems toxicology study. Arch Toxicol. 2017;91(1):465-480. doi:10.1007/s00204-016-1681-3 64. IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to Humans, Vol 94, Ingested Nitrate and Nitrite, and Cyanobacterial Peptide Toxins. Vol 94.; 2010. doi:10.1002/food.19940380335 65. Chen Y, Zhou Y, Wang J, et al. Microcystin-Leucine Arginine Causes Cytotoxic Effects in Sertoli Cells Resulting in Reproductive Dysfunction in Male Mice. Sci Rep. 2016;6(December):39238. doi:10.1038/srep39238 66. Zhou Y, Yuan J, Wu J, Han X. The toxic effects of microcystin-LR on rat spermatogonia in vitro. Toxicol Lett. 2012;212(1):48-56. doi:10.1016/j.toxlet.2012.05.001 67. Wang X, Ding J, Xiang Z, Jiang P, Du J, Han X. Microcystin-LR causes sexual hormone disturbance in male rat by targeting gonadotropin-releasing hormone neurons. Toxicon. 2016;123:45-55. doi:10.1016/j.toxicon.2016.10.011 68. Carvalho GMC, Oliveira VR, Casquilho NV, et al. Pulmonary and hepatic injury after sub-chronic exposure to sublethal doses of microcystin-LR. Toxicon. 2016;112:51-58. doi:10.1016/j.toxicon.2016.01.066 69. Xie L, Rediske RR, Gillett ND, O’Keefe JP, Scull B, Xue Q. The impact of environmental parameters on microcystin production in dialysis bag experiments. Sci Rep.

149 2016;6(November):1-10. doi:10.1038/srep38722 70. Cerasino L, Salmaso N. Diversity and distribution of cyanobacterial toxins in the Italian subalpine lacustrine district. Oceanol Hydrobiol Stud. 2012;41(3). doi:10.2478/s13545- 012-0028-9 71. Mankiewicz J, Walter Z, Tarczynska M, Palyvoda O, Wojtysiak-Staniaszczyk M, Zalewski M. Genotoxicity of cyanobacterial extracts containing microcystins from polish water reservoirs as determined by SOS chromotest and comet assay. Environ Toxicol. 2002;17(4):341-350. doi:10.1002/tox.10061 72. Kelker MS, Page R, Peti W. Crystal Structures of Protein Phosphatase-1 Bound to Nodularin-R and Tautomycin: A Novel Scaffold for Structure-based Drug Design of Serine/Threonine Phosphatase Inhibitors. J Mol Biol. 2009;385(1):11-21. doi:10.1016/J.JMB.2008.10.053 73. Jungblut A-D, Neilan BA. Molecular identification and evolution of the cyclic peptide hepatotoxins, microcystin and nodularin, synthetase genes in three orders of cyanobacteria. doi:10.1007/s00203-005-0073-5 74. Kuiper-Goodman T, Balconer I, Fitzgerald J. Human Health Aspects. In: Chorus I, Bartram J, eds. Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring and Management Determination of Organic Compounds in Natural and Treated Waters. London, UK: E & FN Spon Publishers; 1999:36. http://www.who.int/water_sanitation_health/resourcesquality/toxcyanbegin.pdf. 75. Moffitt MC, Neilan BA. Characterization of the nodularin synthetase gene cluster and proposed theory of the evolution of cyanobacterial hepatotoxins. Appl Environ Microbiol. 2004;70(11):6353-6362. doi:10.1128/AEM.70.11.6353-6362.2004 76. Chiswell RK, Shaw GR, Eaglesham G, et al. Stability of cylindrospermopsin, the toxin from the cyanobacterium,Cylindrospermopsis raciborskii: Effect of pH, temperature, and sunlight on decomposition. Environ Toxicol. 1999;14(1):155-161. doi:10.1002/(SICI)1522-7278(199902)14:1<155::AID-TOX20>3.0.CO;2-Z 77. Pearson L, Mihali T, Moffitt M, Kellmann R, Neilan B. On the Chemistry, Toxicology and Genetics of the Cyanobacterial Toxins, Microcystin, Nodularin, Saxitoxin and Cylindrospermopsin. Mar Drugs. 2010;8(5):1650-1680. doi:10.3390/md8051650 78. Wormer L, Cirés S, Carrasco D, Quesada A. Cylindrospermopsin is not degraded by co- occurring natural bacterial communities during a 40-day study. Harmful Algae. 2008;7(2):206-213. doi:10.1016/J.HAL.2007.07.004 79. de la Cruz AA, Hiskia A, Kaloudis T, et al. A review on cylindrospermopsin: the global occurrence, detection, toxicity and degradation of a potent cyanotoxin. Environ Sci Process Impacts. 2013;15(11):1979. doi:10.1039/c3em00353a 80. Runnegar MT, Xie C, Snider BB, Wallace GA, Weinreb SM, Kuhlenkamp J. In vitro hepatotoxicity of the cyanobacterial alkaloid cylindrospermopsin and related synthetic analogues. Toxicol Sci. 2002;67(1):81-87. http://www.ncbi.nlm.nih.gov/pubmed/11961219. Accessed May 8, 2018. 81. Humpage AR, Fontaine F, Froscio S, Burcham P, Falconer IR. Cylindrospermopsin Genotoxicity and Cytotoxicity: Role Of Cytochrome P-450 and Oxidative Stress. J Toxicol Environ Heal Part A. 2005;68(9):739-753. doi:10.1080/15287390590925465 82. Rogers EH, Zehr RD, Gage MI, et al. The cyanobacterial toxin, cylindrospermopsin,

150 induces fetal toxicity in the mouse after exposure late in gestation. Toxicon. 2007;49(6):855-864. doi:10.1016/J.TOXICON.2006.12.009 83. Aráoz R, Molgó J, Tandeau de Marsac N. Neurotoxic cyanobacterial toxins. Toxicon. 2010;56(5):813-828. doi:10.1016/J.TOXICON.2009.07.036 84. Rantala-Ylinen A, Känä S, Wang H, et al. Anatoxin-a synthetase gene cluster of the cyanobacterium Anabaena sp. strain 37 and molecular methods to detect potential producers. Appl Environ Microbiol. 2011;77(20):7271-7278. doi:10.1128/AEM.06022-11 85. Kellmann R, Mihali TK, Jeon YJ, Pickford R, Pomati F, Neilan BA. Biosynthetic intermediate analysis and functional homology reveal a saxitoxin gene cluster in cyanobacteria. Appl Environ Microbiol. 2008;74(13):4044-4053. doi:10.1128/AEM.00353-08 86. Soto-Liebe K, Murillo AA, Krock B, et al. Reassessment of the toxin profile of Cylindrospermopsis raciborskii T3 and function of putative sulfotransferases in synthesis of sulfated and sulfonated PSP toxins. Toxicon. 2010;56(8):1350-1361. doi:10.1016/J.TOXICON.2010.07.022 87. Jochimsen EM, Carmichael WW, An J, et al. Liver Failure and Death After Exposure To Microcystins. 1998:873-878. 88. Fawell J, Mitchell R, Everett D, Hill R. The toxicity of cyanobacterial toxins in the mouse: I Microcystin-LR. Hum &Experimental Toxicol. 1999;18(September 1998):162- 167. doi:10.1191/096032799678839842 89. Hereman TC, Bittencourt-Oliveira M do C. Bioaccumulation of Microcystins in Lettuce. J Phycol. 2012;48(6):1535-1537. doi:10.1111/jpy.12006 90. Lee S, Jiang X, Manubolu M, et al. Fresh produce and their soils accumulate cyanotoxins from irrigation water: Implications for public health and food security. Food Res Int. 2017;102:234-245. doi:10.1016/J.FOODRES.2017.09.079 91. WHO. Cyanobacterial toxins: Microcystin-LR in Drinking-water (Background document for development of WHO Guidelines for Drinking-water Quality). WHO/SDE/WSH/03.04/57; 2003;2:14. 92. Ingrid Chorus, Ian R. Falconer, Hen. Health Risks Caused By Freshwater Cyanobacteria in Recreational Waters. J Toxicol Environ Heal Part B. 2000;3(4):323-347. doi:10.1080/109374000436364 93. Li Y, Chen J an, Zhao Q, et al. A cross-sectional investigation of chronic exposure to microcystin in relationship to childhood liver damage in the three gorges reservoir region, China. Environ Health Perspect. 2011;119(10):1483-1488. doi:10.1289/ehp.1002412 94. Zhao Y, Xue Q, Su X, et al. First Identification of the Toxicity of Microcystins on Pancreatic Islet Function in Humans and the Involved Potential Biomarkers. Environ Sci Technol. 2016;50(6):3137-3144. doi:10.1021/acs.est.5b03369 95. Svircev Z, Krstic S, Miladinov-Mikov M, Baltic V, Vidovic M. Freshwater cyanobacterial blooms and primary liver cancer epidemiological studies in Serbia. J Environ Sci Health C Environ Carcinog Ecotoxicol Rev. 2009;27(1):36-55. doi:10.1080/10590500802668016 96. Teixeira M da G, Costa M da C, de Carvalho VL, Pereira M dos S, Hage E. Gastroenteritis epidemic in the area of the Itaparica Dam, Bahia, Brazil. Bull Pan Am Health Organ. 1993;27(3):244-253. http://www.ncbi.nlm.nih.gov/pubmed/8220519. Accessed May 9, 2018.

151 97. Carmichael WW, Azevedo SMFO, An JS, et al. Human fatalities form cyanobacteria: Chemical and biological evidence for cyanotoxins. Environ Health Perspect. 2001;109(7):663-668. doi:10.1289/ehp.01109663 98. Pouria S, de Andrade A, Barbosa J, et al. Fatal microcystin intoxication in haemodialysis unit in Caruaru, Brazil. Lancet. 1998;352(9121):21-26. doi:10.1016/S0140- 6736(97)12285-1 99. Soares RM, Yuan M, Servaites JC, et al. Sublethal exposure from microcystins to renal insufficiency patients in Rio de Janeiro, Brazil. Environ Toxicol. 2006;21(2):95-103. doi:10.1002/tox.20160 100. Tisdale ES. Epidemic of intestinal disorders in Charleston, W. Va., occurring simultaneously with unprecedented water supply conditions. Am J Public Health. 1931;21:198- 200. https://books.google.com/books?id=mjM7AAAAIAAJ&pg=PA198&lpg=PA198&dq=epi demic+of+intestinal+disorders+in+charleston,+wv,+occurring+simultaneously+with&sou rce=bl&ots=c2hEB- KQXQ&sig=NQxeLIOTx5KryxAAeeVQdHyyE6I&hl=en&sa=X&ved=0ahUKEwjzr9G SlfnaAhWwTt8KHTeIDpM. Accessed May 9, 2018. 101. Hindman SH, Carson LA, Favero MS, Petersen NJ, Schonberger LB, Solano JT. Pyrogenic Reactions During Haemodialysis Caused by Extramural Endotoxin. Lancet. 1975;306(7938):732-734. doi:10.1016/S0140-6736(75)90721-7 102. Williams C, Burns J, Chapman A, Flewlling L, Pawlowicz M, Carmichael W. Assessment of Cyanotoxins. Palatka, FL; 2001. 103. Preece EP, Moore BC, Swanson ME, Hardy FJ. Identifying best methods for routine ELISA detection of microcystin in seafood. Environ Monit Assess. 2015;187(2):12. doi:10.1007/s10661-014-4255-y 104. Manubolu M, Lee J, Riedl KM, Kua ZX, Collart LP, Ludsin SA. Optimization of extraction methods for quantification of microcystin-LR and microcystin-RR in fish, vegetable, and soil matrices using UPLC–MS/MS. Harmful Algae. 2018;76:47-57. doi:10.1016/j.hal.2018.04.009 105. Anderson DM, Glibert PM, Burkholder JM. Harmful algal blooms and eutrophication: Nutrient sources, composition, and consequences. Estuaries. 2002;25(4):704-726. doi:10.1007/BF02804901 106. Glibert P, Anderson D, Gentien P, Granéli E, Sellner K. The Global, Complex Phenomena of Harmful Algal Blooms. Oceanography. 2005;18(2):136-147. doi:10.5670/oceanog.2005.49 107. Hallegraeff GM. A review of harmful algal blooms and their apparent global increase. Phycologia. 1993;32(2):79-99. doi:10.2216/i0031-8884-32-2-79.1 108. Smayda TJ. Novel and nuisance phytoplankton blooms in the sea: Evidence for a global epidemic. In: Graneli E, Sundstrom B, Edler L, Anderson D, eds. New York City, NY: Elsevier; 1990:29-40. 109. Solomon S, Quin D, Manning M, et al. Climate Change 2007: The Physical Science Basis. In: Contribution of Wroking Group I to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. New York City: Cambridge University Press; 2007.

152 110. Pomati F, Matthews B, Jokela J, Schildknecht A, Ibelings BW. Effects of re- oligotrophication and climate warming on plankton richness and community stability in a deep mesotrophic lake. Oikos. 2012;121(8):1317-1327. doi:10.1111/j.1600- 0706.2011.20055.x 111. Parry M, CAnzaini O, Paultikof J, van der Linden P. Climate Change 2007: Impacts, Adaption and Vulnerability. Contribution of Working Group II to the Fourth Assessment Report of the Intergovernmental Panel on Climate Change. In: Cambridge, EN: Cambridge University Press; 2007. 112. Jeppesen E, Kronvang B, Meerhoff M, et al. Climate Change Effects on Runoff, Catchment Phosphorus Loading and Lake Ecological State, and Potential Adaptations. J Environ Qual. 2009;38(5):1930. doi:10.2134/jeq2008.0113 113. Paerl HW, Hall NS, Calandrino ES. Controlling harmful cyanobacterial blooms in a world experiencing anthropogenic and climatic-induced change. Sci Total Environ. 2011;409(10):1739-1745. doi:10.1016/J.SCITOTENV.2011.02.001 114. Paerl HW, Huisman J. Climate change: a catalyst for global expansion of harmful cyanobacterial blooms. Environ Microbiol Rep. 2009;1(1):27-37. doi:10.1111/j.1758- 2229.2008.00004.x 115. Mrdjen I, Fennessy S, Slonzewski J, Schaal A, Dennis R, Lee J. Tile Drainage and Anthropogenic Land Use Contribute to Harmful Algal Blooms and Microbiota Shift in Inland Water Bodies. Environ Sci Technol. 2018. 116. Paerl HW, Gardner WS, Havens KE, et al. Mitigating cyanobacterial harmful algal blooms in aquatic ecosystems impacted by climate change and anthropogenic nutrients. Harmful Algae. 2016;54:213-222. doi:10.1016/j.hal.2015.09.009 117. Dodds WK, Bouska WW, Eitzmann JL, et al. Eutrophication of U. S. freshwaters: Analysis of potential economic damages. Environ Sci Technol. 2009;43(1):12-19. doi:10.1021/es801217q 118. EPA: United States Environmental Protection Agency. Human Health Recreational Ambient Water Quality Criteria or Swimming Advisories for Microcystins and Cylindrospermopsin Draft Human Health Recreational Ambient Water Quality Criteria or Swimming Advisories for Microcystins and Cylindrospermopsin. 2016;(December):185. https://www.epa.gov/sites/production/files/2016-12/documents/draft-hh-rec-ambient- water-swimming-document.pdf. 119. Žegura B, Štraser A, Filipič M. Genotoxicity and potential carcinogenicity of cyanobacterial toxins - a review. Mutat Res - Rev Mutat Res. 2011;727(1-2):16-41. doi:10.1016/j.mrrev.2011.01.002 120. Hu C, Rea C, Yu Z, Lee J. Relative importance of Microcystis abundance and diverstiy in determining microcystin dynamics in Lake Erie coastal wetlands and downstream beach water. J Appl Microbiol. 2016;120(1):138-151. 121. Dai R, Wang P, Jia P, Zhang Y, Chu X, Wang Y. A review on factors affecting microcystins production by algae in aquatic environments. World J Microbiol Biotechnol. 2016;32(3):1-7. doi:10.1007/s11274-015-2003-2 122. Buratti FM, Manganelli M, Vichi S, et al. Cyanotoxins: producing organisms, occurrence, toxicity, mechanism of action and human health toxicological risk evaluation. Arch Toxicol. 2017;91(3):1049-1130. doi:10.1007/s00204-016-1913-6

153 123. Altaner S, Puddick J, Wood SA. Adsorprtion of ten micorcystin congeners to common laboratory-ware is solvent and surface dependent. Toxins (Basel). 2017;9(4). 124. US EPA. Drinking Water Health Advisory for the Cyanobacterial Microcystin Toxins. EPA-820R15100. Washington, DC; 2015. 125. Zhang F, Hu C, Shum CK, Liang S, Lee J. Satellite Remote Sensing of Drinking Water Intakes in Lake Erie for Cyanobacteria Population Using Two MODIS-Based Indicators as a Potential Tool for Toxin Tracking. Front Mar Sci. 2017;4(124):1-11. doi:10.3389/fmars.2017.00124 126. Gorham T, Jia Y, Shum CK, Lee J. Ten-year survey of cyanobacterial blooms in Ohio’s waterbodies using satellite remote sensing. Harmful Algae2. 2017;66:11-19. 127. Downing JA. Emerging global role of small lakes and ponds : little things mean a lot. 2010;29(1). 128. Ohio Environmental Protection Agency. Inland Lakes Program. http://www.epa.ohio.gov/dsw/inland_lakes/index.aspx. Published 2017. 129. Dodson S. I., Arnott S. E. CKL. Concep Ts & Synthesis the Relationship in Lake Communities Between Primary. America (NY). 2000;81(10):2662-2679. 130. Yang J, Lv H, Yang J, Liu L, Yu X, Chen H. Decline in water level boosts cyanobacteria dominance in subtropical reservoirs. Sci Total Environ. 2016;557-558:445-452. doi:10.1016/j.scitotenv.2016.03.094 131. Van Esbroeck CJ, Macrae ML, Brunke RI, McKague K. Annual and seasonal phosphorus export in surface runoff and tile drainage from agricultural fields with cold temperate climates. J Great Lakes Res. 2016;42(6):1271-1280. doi:10.1016/j.jglr.2015.12.014 132. King KW, Williams MR, Macrae ML, et al. Phosphorus Transport in Agricultural Subsurface Drainage : A Review. J Environ Qual. 2015;44(2):467-485. doi:10.2134/jeq2014.04.0163 133. Drury CF, Tan CS, Gaynor JD, Oloya TO, Welacky TW. Influence of Controlled Drainage-Subirrigation on Surface and Tile Drainage Nitrate Loss. J Environ Qual. 1996;25(2):317. doi:10.2134/jeq1996.00472425002500020016x 134. Rockwell DC, Warren GJ, Bertram PE, Salisbury DK, Burns NM. The U.S. EPA Lake Erie indicators monitoring program 1983–2002: Trends in phosphorus, silica, and chlorophyll a in the central basin. J Great Lakes Res. 2005;31(January 2005):23-34. doi:10.1016/S0380-1330(05)70302-6 135. Kumwimba MN, Zhu B, Muyembe DK. Estimation of the removal efficiency of heavy metals and nutrients from ecological drainage ditches treating town sewage during dry and wet seasons. Environ Monit Assess. 2017;189(9):434. doi:10.1007/s10661-017-6136-7 136. Carpenter SR, Caraco NF, Correll DL, W.Howarth R, Sharpley AN, Smith VH. Nonpoint pollution of surface waters with phosphorus and nitrogen. Ecol Appl. 1998;8(1998):559- 568. doi:10.1890/1051-0761(1998)008[0559:NPOSWW]2.0.CO;2 137. Lee C, Marion JW, Cheung M, Lee CS, Lee J. Associations among human-associated fecal contamination, microcystis aeruginosa, and microcystin at lake erie beaches. Int J Environ Res Public Health. 2015;12(9):11466-11485. doi:10.3390/ijerph120911466 138. Odagiri M, Schriewer A, Daniels ME, et al. Human fecal and pathogen exposure pathways in rural Indian villages and the effect of increased latrine coverage. Water Res. 2016;100:232-244. doi:10.1016/j.watres.2016.05.015

154 139. Robilotti E, Deresinski S, Pinsky BA. Norovirus. Clin Microbiol Rev. 2015;28(1):134- 164. doi:10.1128/CMR.00075-14 140. Dieter RA, Dieter RS, Gulliver G. Zoonotic diseases: health aspects of canadian geese. Int J Circumpolar Health. 2001;60(4):676-684. 141. Homer BC, Dewitz J, Yang L, et al. Completion of the 2011 national land cover database for the conterminous United States – representing a decade of land cover change information. Photogramm Eng Remote Sensing. 2015;81(5):345-354. 142. Tillett D, Parker DL, Neilan BA. Detection of Toxigenicity by a Probe for the Microcystin Synthetase a Gene (mcyA) of the Cyanobacterial Genus Microcystis: Comparison of Toxicities with 16S rRNa and Phycocyanin Operon (Phycocyanin Intergenic Spacer) Phylogenies. Appl Environ Microbiol. 2001;67(6):2810-2818. doi:10.1128/AEM.67.6.2810-2818.2001 143. Lee CS, Kim M, Yu Z, Lee J. Microbiota of Recreational Freshwater and the Implications of Environmental and Public Health. Front Microbiol. 2016;7:1826. 144. Haugland RA, Siefring SC, Wymer LJ, Brenner KP, Durour AP. Comparison of enterococcus density meansurements by quanititative polymerase chain reaction and membrane filer analysis at two freshwater recreational bodies. Water Res. 2005;39:559- 568. 145. Sipari H, Rantala-Ylinen A, Jokela J, Oksanen I, Sivonen K. Development of a chip assay and quantitative PCR for detecting microcystin synthetase e gene expressions. Appl Environ Microbiol. 2010;76(12):3797-3805. doi:10.1128/AEM.00452-10 146. Rantala A, Rajaniemi-Wacklin P, Lyra C, et al. Detection of microcystin-producing cyanobacteria in Finnish lakes with genus-specific microcystin synthetase gene E (mcyE) PCR and associations with environmental factors. Appl Environ Microbiol. 2006;72(9):6101-6110. doi:10.1128/AEM.01058-06 147. Borherova Z, Park E, Holloran K, Lee J. Water quality changes shortly after low head dam removal examined with cultural and microbial source tracking methods. River Res Appl. 2016;33:113-122. 148. Green HC, White KM, Kelty CA, Shanks OC. Development of rapid canine fecal source identification PCR-based assays. Environ Sci Technol. 2014;48(19):11453-11461. doi:10.1021/es502637b 149. Gourmelon M, Caprais MP, Mieszkin S, et al. Development of microbial and chemical MST tools to identify the origin of the faecal pollution in bathing and shellfish harvesting waters in France. Water Res. 2010;44(16):4812-4824. doi:10.1016/j.watres.2010.07.061 150. Boehm AB, Van De Werfhorst LC, Griffith JF, et al. Performance of forty-one microbial source tracking methods: A twenty-seven lab evaluation study. Water Res. 2013;47(18):6812-6828. doi:10.1016/j.watres.2012.12.046 151. Garcia-Mazcorro J, Castillo-Carranza S, Guard B, Gomez-Vazquez J, Dowd S, Brightsmith D. Comprehensive molecular characterization of bacterial communities in feces of pet birds using 16S marker sequencing. Microb Ecol. 2017;73(1):224-235. doi:10.1007/s00248-016-0840-7 152. Capone K, Dowd S, Stamatas G, Nikolovski N. Diversity of the human skin microbiome early in life. J Invest Dermatol. 2011;131:2026-2032. doi:10.1038/jid.2011.168 153. Swanson K, Dowd S, Suchodolski J, et al. Phylogenetic and gene-centric metagenomics of

155 the canine intestinal microbiome reveals similarities with humans and mice. Multidiscip J Microb Ecol. 2011;5(4):639-649. 154. US EPA. Recommendations for Cyanobacteria and Cyanotoxin Monitoring in Recreational Waters. Office of Water EPA 820-R-17-001. Washington, D.C.; 2017;(June):1-15. 155. Carlson RE, Simpson J. A coordinator’s guide to volunteer monitoring methods. North Am Lake Manag Soc. 1996:96. 156. Ohio Department of Health. Household Sewage Treatment System Failures in Ohio. Columbus, OH; 2013. 157. Hubbard RK, Newton GL, Hill GM. Water quality and the grazing animal. J Anim Sci. 2004;82 E-Suppl:255-263. doi:82/13_suppl/E255 158. Pearce AR, Chambers LG, Hasenmueller EA. Science of the Total Environment Characterizing nutrient distributions and fl uxes in a eutrophic reservoir , Midwestern United States. Sci Total Environ. 2017;582:589-600. doi:10.1016/j.scitotenv.2016.12.168 159. Kane DD, Conroy JD, Peter Richards R, Baker DB, Culver DA. Re-eutrophication of Lake Erie: Correlations between tributary nutrient loads and phytoplankton biomass. J Great Lakes Res. 2014;40(3):496-501. doi:10.1016/j.jglr.2014.04.004 160. Park HD, Iwami C, Watanabe MF, Harada KI, Okino T, Hayashi H. Temporal variabilities of the concentrations of intra- and extracellular microcystin and toxic Microcystis species in a hypertrophic lake, Lake Suwa, Japan (1991-1994). Environ Toxicol Water Qual. 1998;13(1):61-72. doi:10.1002/(SICI)1098- 2256(1998)13:1<61::AID-TOX4>3.0.CO;2-5 161. Bourne DG, Boyett H V., Henderson ME, Muirhead A, Willis BL. Identification of a ciliate (Oligohymenophorea: Scuticociliatia) associated with brown band disease on corals of the great barrier reef. Appl Environ Microbiol. 2008;74(3):883-888. doi:10.1128/AEM.01124-07 162. Rocke E, Liu H. Respiration, growth and grazing rates of three ciliate species in hypoxic conditions. Mar Pollut Bull. 2014;85(2):410-417. doi:10.1016/j.marpolbul.2014.04.050 163. Puytorac P, Batisse A, Bohatier J, et al. Proposition d’une classification du phylu Ciliophora Doflein. Comptes Rendus I’Academie Sci. 1974;278:2799-2802. 164. States U, Agency P. Health Effects Support Document for the Cyanobacterial Toxin Microcystins. 2015;(June). 165. Backer LC, Landsberg JH, Miller M, Keel K, Taylor TK. Canine cyanotoxin poisonings in the United States (1920s-2012): Review of suspected and confirmed cases from three data sources. Toxins (Basel). 2013;5(9):1597-1628. doi:10.3390/toxins5091597 166. Bojcevska H, Jergil E. Removal of cyanobacterial toxins ( LPS endotoxin and microcystin ) in drinking-water using the BioSand household water filter. 2003;(May):44. 167. Dias E, Louro H, Pinto M, et al. Genotoxicity of Microcystin-LR in In Vitro and In Vivo Experimental Models. 2014;2014. doi:10.1155/2014/949521 168. Tachi M, Imanishi SY, Harada K-I. Phosphoprotein Analysis for Investigation of In Vivo Relationship Between Protein Phosphatase Inhibitory Activities and Acute Hepatotoxicity of Micorcystin-LR. Environ Toxicol. 2007;22:620-629. doi:10.1002/tox 169. Yoshida T, Makita Y, Nagata S, et al. Acute oral toxicity of microcystin-LR, a cyanobacterial hepatotoxin, in mice. Nat Toxins. 1997;5(3):91-95. doi:10.1002/nt.1

156 170. Xu C, Shu WQ, Qiu ZQ, Chen JA, Zhao Q, Cao J. Protective effects of green tea polyphenols against subacute hepatotoxicity induced by microcystin-LR in mice. Environ Toxicol Pharmacol. 2007;24(2):140-148. doi:10.1016/j.etap.2007.04.004 171. Gehringer MM, Shephard EG, Downing TG, Wiegand C, Neilan BA. An investigation into the detoxification of microcystin-LR by the glutathione pathway in Balb/c mice. Int J Biochem Cell Biol. 2004;36(5):931-941. doi:10.1016/j.biocel.2003.10.012 172. Wang L, Wang X, Geng Z, et al. Distribution of microcystin-LR to testis of male Sprague- Dawley rats. Ecotoxicology. 2013;22(10):1555-1563. doi:10.1007/s10646-013-1141-2 173. Wieckowska A, Zein NN, Yerian LM, Lopez AR, McCullough AJ, Feldstein AE. In vivo assessment of liver cell apoptosis as a novel biomarker of disease severity in nonalcoholic fatty liver disease. Hepatology. 2006;44(1):27-33. doi:10.1002/hep.21223 174. Solter P, Liu Z, Guzman R. Decreased hepatic ALT synthesis is an outcome of subchronic microcystin-LR toxicity. Toxicol App Pharmacol. 2000;164(2):216-220. doi:10.1006/taap.2000.8895 175. Singhal SS, Saxena M, Ahmad H, Awasthi YC. Glutathione S-transferases of mouse liver: sex-related differences in the expression of various isozymes. BBA - Gen Subj. 1992;1116(2):137-146. doi:10.1016/0304-4165(92)90110-G 176. Sekijima M, Tsutsumi T, Yoshida T, et al. Enhancement of glutathione S-transferase placental-form positive liver cell foci development by microcystin-LR in aflatoxin B1- initiated rats. Carcinogenesis. 1999;20(1):161-165. http://www.ncbi.nlm.nih.gov/pubmed/9934864. Accessed March 7, 2018. 177. Carvalho GMC, Oliveira VR, Soares RM, et al. Can LASSBio 596 and dexamethasone treat acute lung and liver inflammation induced by microcystin-LR? Toxicon. 2010;56(4):604-612. doi:10.1016/j.toxicon.2010.06.005 178. Runnegar M, Berndt N, Kaplowitz N. Microcystin uptake and inhibition of protein phosphatases: effects of chemoprotectants and self-inhibition in relation to known hepatic transporters. Toxicol Appl Pharmacol. 1995;134(2):264-272. 179. Jayaraj R, Anand T, Rao PVL. Activity and gene expression profile of certain antioxidant enzymes to microcystin-LR induced oxidative stress in mice. Toxicology. 2006;220(2- 3):136-146. doi:10.1016/j.tox.2005.12.007 180. Ruch RJ, Boucher PD, Gentry BG, Shewach DS. Gap Junctional Intercellular Communication Increases Cytotoxicity and Reduces Resistance to Hydroxyurea. 2014;(November):1190-1202. 181. Trosko JE, Ruch JR. Department of Pediatrics and Human Development, Michigan State University, East Lansing, Michigan 48824, 2 Department of Pathology, Medical College of Ohio, Toledo, Ohio. Front Biosci. 1998;(3):208-236. 182. Klauning JE, Ruch RJ. Role of Intercellular Communication in Nongenotoxic Carcinogenesis. Lab Investig. 1990;62:135-146. 183. Trosko JE, Chang CC. Nongenotoxic Mechanisms in Carcinogenicity: Role of Intercellular Communication. In: Hart RW, Hoeger FG, eds. Branbury Report 31. Cold Springs Harbor, NY: Cold Spring Harbor Laboratory Press; :139-170. 184. Ohta T, Nishiwaki R, Yatsunami J, Komori A, Suganuma M, Fujiki H. Hyperphosphorylation of cytokeratins 8 and 18 by microcystin-LR, a new tumor promoter, in primary cultured rat hepatocytes. Carcinogenesis1. 1992;13:2443-2447.

157 185. Vinken M. Gap junctions and non-neoplastic liver disease. J Hepatol. 2012;57(3):655- 662. doi:10.1016/j.jhep.2012.02.036 186. Ruch RJ. Intercellular communication, homeostasis, and toxicology. Toxicol Sci. 2002;68(2):265-266. doi:10.1093/toxsci/68.2.265 187. Andreassen PR, Lacroix FB, Villa-Moruzzi E, Margolis RL. Differential subcellular localization of protein phosphatase-1 alpha, gamma1, and delta isoforms during both interphase and mitosis in mammalian cells. J Cell Biol. 1998;141(5):1207-1215. doi:10.1083/jcb.141.5.1207 188. Ruch RJ, Klauning JE. Inhibition of mouse hepatocytes intercellular communication by paraquat-generated oxygen free radicals. Toxicol Appl Pharmacol. 1988;94:427-436. 189. Nováková K, Babica P, Adamovský O, Bláha L. Modulation of gap-junctional intercellular communication by a series of cyanobacterial samples from nature and laboratory cultures. Toxicon. 2011;58(1):76-84. doi:10.1016/J.TOXICON.2011.05.006 190. Bláha L, Babica P, Hilscherová K, Upham BL. Inhibition of gap-junctional intercellular communication and activation of mitogen-activated protein kinases by cyanobacterial extracts – Indications of novel tumor-promoting cyanotoxins? Toxicon. 2010;55(1):126- 134. doi:10.1016/J.TOXICON.2009.07.009 191. Egaas E, Falls JG, Dauterman WC. A study of gender, strain and age differences in mouse liver glutathione-S-transferase. Comp Biochem Physiol Part C Comp. 1995;110(1):35-40. doi:10.1016/0742-8413(94)00079-P 192. Sharma R, Ahmad H, Singhal SS, Saxena M, Srivastava SK, Awasthi YC. Comparative studies on the effect of butylated hydroxyanisole on glutathione and glutathione S- transferases in the tissues of male and female CD-1 mice. Comp Biochem Physiol C. 1993;105(1):31-37. http://www.ncbi.nlm.nih.gov/pubmed/8101791. 193. Eriksson JE, Toivola D, Meriluoto JAO, Karaki H, Han YG, Harthshorne D. Hepatocyte deforemation induced by cyanobacterial toxins reflects inhibition of protein phosphatases. Biochem Biophys Reserach Communcations. 1990;173:1347-1353. 194. Ruch RJ, Trosko JE. Gap-Junction Communication in Chemical Carcinogenesis. Drug Metab Rev. 2001;33(1):117-121. 195. Woolbright BL, Williams CD, Ni H, et al. Microcystin-LR induced liver injury in mice and in primary human hepatocytes is caused by oncotic necrosis. Toxicon. 2017;125:99- 109. doi:10.1016/J.TOXICON.2016.11.254 196. Ueno Y, Nagata S, Tsutsumi T, et al. Detection of microcystins, a blue-green algal hepatotoxin, in drinking water sampled in Haimen and Fusui, endemic areas of primary liver cancer in China, by highly sensitive immunoassay. Carcinogenesis. 1996;17(6):1317-1321. doi:10.1093/carcin/17.6.1317 197. Svirčev Z, Simeunović J, Subakov-Simić G, Krstić S, Pantelić D, Dulić T. Cyanobacterial blooms and their toxicity in Vojvodina Lakes, Serbia. Int J Environ Res. 2013;7(3):745- 758. 198. Zhang F, Lee J, Liang S, Shum C. Cyanobacteria blooms and non-alcoholic liver disease: evidence from a county level ecological study in the United States. Environ Heal. 2015;14(1):41. doi:10.1186/s12940-015-0026-7 199. Nishiwaki-Matsushima R, Ohta T, Nishiwaki S, et al. Liver tumor promotion by the cyanobacterial cyclic peptide toxin microcystin-LR. 1992:420-424.

158 200. Ito E, Kondo F, Terao K, Harada K. Neoplastic nodular formation in mouse liver induced by repeated intraperitoneal injectins of microcystin-LR. Toxicon. 1997;35(9):1453-1457. 201. Falconer I, Humpage A. Tumor promotion by cyanobacterial toxins. Phycologia. 1996;35(6S):74-79. 202. Labine M, Minuk GY. Long-term, low-dose exposure to microcystin toxin does not increase the risk of liver tumor development or growth in mice. Hepatol Res. 2015;45(6):683-692. doi:10.1111/hepr.12394 203. Bojcevska H, Jergil E. Removal of cyanobacterial toxins (LPS endotoxin and microcystin) in drinking-water using the BioSand household water filter. 2003;(May):44. 204. Watanabe M, Ichimura T. Fresh- and Salt-water forms of Spirulina platensis in axenic cultures. Bull Jpn Soc Phycol. 1997;25:371-377. 205. Dragani TA, Manenti G, Gariboldi M, De Gregorio L, Pierotti MA. Genetics of liver tumor susceptibility in mice. Toxicol Lett. 1995;82-83(C):613-619. doi:10.1016/0378- 4274(95)03505-2 206. Schneider C, Rashband W, Eliceiri K. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671-675. 207. StataCorp. Stata Statistical Software Release 10. 2007. 208. Qureshi ST, Larivière L, Leveque G, et al. Endotoxin-tolerant Mice Have Mutations in Toll-like Receptor 4 ( Tlr4 ). J Exp Med. 1999;189(4):615-625. doi:10.1084/jem.189.4.615 209. Doucette GJ. Interactions between bacteria and harmful algae: A review. Nat Toxins. 1995;3(2):65-74. doi:10.1002/nt.2620030202 210. Heisler J, Glibert PM, Burkholder JM, et al. Eutrophication and harmful algal blooms: A scientific consensus. Harmful Algae. 2008;8(1):3-13. doi:10.1016/j.hal.2008.08.006 211. Shumway SE, Allen SM, Boersma PD. Marine birds and harmful algal blooms: Sporadic victims or under-reported events? Harmful Algae. 2003;2(1):1-17. doi:10.1016/S1568- 9883(03)00002-7 212. Ramanan R, Kim BH, Cho DH, Oh HM, Kim HS. Algae-bacteria interactions: Evolution, ecology and emerging applications. Biotechnol Adv. 2016;34(1):14-29. doi:10.1016/j.biotechadv.2015.12.003 213. Hu C, Rea C, Yu Z, Lee J. Relative importance of Microcystis abundance and diversity in determining microcystin dynamics in Lake Erie coastal wetland and downstream beach water. J Appl Microbiol. 2016;120(1):138-151. doi:10.1111/jam.12983 214. Ahn YH, Shanmugam P, Ryu JH, Jeong JC. Satellite detection of harmful algal bloom occurrences in Korean waters. Harmful Algae. 2006;5(2):213-231. doi:10.1016/j.hal.2005.07.007 215. Wynne TT, Stumpf RP, Tomlinson MC, Dyble J. Characterizing a cyanobacterial bloom in western Lake Erie using satellite imagery and meteorological data. Limnol Oceanogr. 2010;55(5):2025-2036. doi:10.4319/lo.2010.55.5.2025 216. Clark JM, Schaeffer BA, Darling JA, et al. Satellite monitoring of cyanobacterial harmful algal bloom frequency in recreational waters and drinking water sources. Ecol Indic. 2017;80(November 2016):84-95. doi:10.1016/j.ecolind.2017.04.046 217. Kahru M, Savchuk OP, Elmgren E. Satellite measurements of cyanobacteria bloom frequency in the Baltic Sea: Interannual and spatial variability. Mar Ecol Prog Ser.

159 2007;343:15-23. 218. US General Accounting Office. Costs And Uses Of Remote Sensing. GAO/RCED-83- 11Washington, DC; 1983. doi:GAP/RCED-83-111 219. U.S. Geological Survey. Landsat Missions Timeline | Landsat Missions. https://landsat.usgs.gov/landsat-missions-timeline. Published 2017. Accessed February 12, 2018. 220. Shen L, Xu H, Guo X. Satellite remote sensing of harmful algal blooms (HABs) and a potential synthesized framework. Sensors (Switzerland). 2012;12(6):7778-7803. doi:10.3390/s120607778 221. Blondeau-Patissier D, Gower JFR, Dekker AG, Phinn SR, Brando VE. A review of ocean color remote sensing methods and statistical techniques for the detection, mapping and analysis of phytoplankton blooms in coastal and open oceans. Prog Oceanogr. 2014;123:23-144. doi:10.1016/j.pocean.2013.12.008 222. International Ocean Color Coordinating Group. MODIS - Terra - IOCCG. ioccg.org. http://ioccg.org/sensor/modis-terra/. Published 2017. 223. Feng L, Hu C. Land adjacency effects on MODIS Aqua top-of-atmosphere radiance in the shortwave infrared: Statistical assessment and correction. J Geophys Res Ocean. 2017;122(6):4802-4818. doi:10.1002/2017JC012874 224. Capolsini P, Andréfouët S, Rion C, Payri C. A comparison of Landsat ETM+, SPOT HRV, Ikonos, ASTER, and airborne MASTER data for coral reef habitat mapping in South Pacific islands. Can J Remote Sens. 2003;29(2):187-200. doi:10.5589/m02-088 225. MODIS: Moderate Resolution Imaging Spectroradiometer. NASA. https://modis.gsfc.nasa.gov/data/. Accessed December 2, 2018. 226. Stelzer E, Loftin K, Struffolino P. Relations between DNA-and RNA-based molecular methods for cyanobacteria and microcystin concentration at Maumee Bay State Park Lakeside Beach, Oregon, Ohio. US Geol Surv. 2012;(No. 2013-. 227. Buckeye Lake State Park. http://parks.ohiodnr.gov/buckeyelake. Accessed March 16, 2018.

160 Appendix A. Supplementary Information

161 Table 19. Water chemistry data of samples collected in various Knox County lakes over a 3- month period.

Tile Drains Sampling NO - NH -N PO Date 3 3 4 or Animals Location (mg/L) (mg/L) (mg/L) Present? BT 06-23 16.3 0.06 0 Y BT 07-30 12 0.07 0.25 Y BT 09-04 0.5 0.09 0.11 Y BJ 06-16 0.07 0.13 0.71 N BJ 07-21 0 0.1 0.75 N BJ 08-20 x x x N CO 07-23 0.1 0.51 0.58 N D1 07-07 0.51 0.14 0 Y D2 07-07 5.2 0.69 0.1 Y D2 08-11 6.7 0.3 0.7 Y FC 07-07 1.38 0 0.4 Y FC 06-18 0.33 30 21.7 Y FC 07-28 x x x Y FC 07-28 x x x Y FC 07-28 x x x Y FC 07-28 x x x Y F1 06-25 0.01 0 0.24 Y F1 08-04 0 0 0.16 Y F2 06-25 0.05 0.01 0.03 Y F2 08-04 0.09 0.06 0.09 Y F3 06-25 0.03 0 0 Y F3 08-04 0.1 0 0 Y G2 08-22 2.23 0.27 0.06 N GC 06-08 0.98 0 0.17 N GC 06-30 0.22 0.01 0.05 N JB 07-02 0.02 0.14 0.24 N JB 08-06 0.07 0.04 0 N L1 08-20 0.06 0 0.08 Y L2 08-20 0.07 0.42 0.44 N LH 06-18 0.04 0 0.24 N

Continued

162 Table 19. Continued

LH 07-30 0.09 0 0.19 N LE 07-02 0.1 0.51 0.58 Y LE 07-23 x x x Y LE 08-22 0.05 0.1 0.17 Y RN 06-30 0.09 0.28 0.32 Y RN 08-24 0.09 0.06 0.27 Y RC 06-23 6.9 0.01 0.3 N RC 07-30 1.39 0.21 0 N RV 07-16 2.2 0 0.09 N RV 08-13 0.13 0.13 0.37 N SA 06-16 1.23 0 1.1 Y SA 07-21 0.03 0.07 0.4 Y WR 06-08 0.07 0.18 0.83 N WR 07-10 0.05 0.44 0 N Overall n/a 1.6 0.9 0.8 n/a Mean

163