PARALYTIC SHELLFISH TOXINS CHARACTERIZED FROM LYNGBYA WOLLEI DOMINATED MATS COLLECTED FROM TWO FLORIDA SPRINGS

By

AMANDA FOSS

A THESIS PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE

UNIVERSITY OF FLORIDA

2011

1

© 2011 AMANDA FOSS

2

To my wonderful parents

3

ACKNOWLEDGMENTS

I appreciate the support provided me by GreenWater Laboratories, Mark Aubel for his wonderful direction and aid with toxin analysis and Andrew Chapman for his expertise in phycology and sampling. I thank the chair, Ed Phlips, and members of my committee, Nancy Szabo and Karl Havens, for their priceless advice and keen research assistance. I thank Mete Yilmaz for his work on the molecular portion of this study, as well as expertise. I thank Andrew Reich with the Florida Department of Health for collaboration with GreenWater Labs and the research that kick started this project, and, of course a big thank you to Alicia Plakotaris and Johnny May for their help in the field.

4

TABLE OF CONTENTS

page

ACKNOWLEDGMENTS ...... 4

LIST OF TABLES ...... 7

LIST OF FIGURES ...... 8

ABSTRACT ...... 10

CHAPTER

1 INTRODUCTION ...... 12

Objectives ...... 12 Background ...... 13 Lyngbya Growth in Florida Springs ...... 13 Lyngbya Toxin Production ...... 14 Paralytic Shellfish Toxins...... 15 Properties and toxicity ...... 15 Potential risks and trophic transfer ...... 16 Toxin Extraction ...... 17 Toxin Analysis Techniques ...... 18 Molecular Biology of Lyngbya and Saxitoxin gene ...... 19

2 MATERIALS AND METHODS ...... 21

Sampling ...... 21 Site Description ...... 21 Collection ...... 22 Standards and Reagents ...... 24 Standards for Instrument Calibration and Quantification ...... 24 Reagents ...... 24 Sample Preparation ...... 25 Qualitative Analysis of Filamentous Macroalgae Samples ...... 25 DNA Isolation, Polymerase Chain Reaction, and Sequencing ...... 25 Toxin Samples ...... 27 Paralytic Shellfish Toxin Extraction ...... 28 Extraction Assessment ...... 28 SPE and Lyophilization Assessment ...... 28 Final Extraction Protocol...... 29 Paralytic Shellfish Toxin Analysis ...... 29 HPLC/ Fluorescence ...... 29 Peroxide Oxidation ...... 29 Periodate Oxidation ...... 30 Chromatographic and Fluorescence Conditions ...... 31

5

Detection Limits ...... 32 Sample Analysis ...... 32 HPLC/MS/(MS) ...... 32 Chromatographic and Mass Spectrometry Conditions ...... 32 Detection Limits ...... 33 Sample Analysis ...... 34

3 RESULTS ...... 48

Field Data ...... 48 Water Quality ...... 48 Field Observations ...... 50 Algal Analysis ...... 50 Molecular Biology ...... 51 HPLC/Fluorescence ...... 52 Peroxide Oxidation Products ...... 52 Extraction Assessment ...... 53 SPE and Lyophilization Assessment ...... 55 Sample Analysis ...... 56 HPLC/MS ...... 56 Extraction Assessment ...... 56 SPE and Lyophilization Assessment ...... 57 Sample Analysis ...... 57

4 DISCUSSION ...... 87

Toxin Extraction ...... 87 Analytical Methods ...... 88 Toxin Risk Assessment ...... 89

LIST OF REFERENCES ...... 94

BIOGRAPHICAL SKETCH ...... 97

6

LIST OF TABLES

Table page

2-1 List of sample collections ...... 42

2-2 Saxitoxin structural variants ...... 43

2-3 LC/FL gradient ...... 45

2-4 MDLs for PSTs via fluorescence in elution order ...... 45

2-5 LC/MS gradient ...... 45

2-6 MDLs and retention times of PSTs analyzed via LC/MS/(MS) ...... 46

2-7 Molecular and fragmentation ions monitored LC/MS ...... 47

3-1 Water quality and sample collection data for sample sites ...... 61

3-2 Water quality data for Blue Hole Spring and Silver Glen Springs from FDEP Initiative Report 2011 ...... 62

3-3 Mat thickness measured at Silver Glen Springs ...... 63

3-4 Qualitative algae list from all samples ...... 64

3-5 pH of each extraction solution before and after boiling ...... 69

3-6 Decarbamoylgonyautoxin 2&3 (dcGTX2&3) concentrations (µg g-1) for all sites and sampling months ...... 76

3-7 Decarbamoylsaxitoxin (dcSTX) concentrations (µg g-1) for all sites and sampling months ...... 77

4-1 Toxin content of L.wollei collected from Alabama and Florida ...... 93

4-2 Toxicity of PST variants reported in MU/µmol and relative toxicity to STX (STX-eq) ...... 93

7

LIST OF FIGURES

Figure page

2-1 Silver Glen Springs Recreational Park photo ...... 36

2-2 Series of Silver Glen Main Stem photos ...... 37

2-3 Series of Silver Glen Natural Well (Southeastern Vent) photos ...... 39

2-4 Series of Blue Hole (Jug) Spring photos ...... 41

2-5 Conversion of saxitoxin to fluorescent derivative for HPLC/FL detection ...... 44

3-1 Series of micrographs of Lyngbya wollei ...... 66

3-2 Series of micrographs of Lyngbya wollei epiphytes, all at 400x ...... 67

3-3 Fluorescence chromatogram of standard solution mixture of dcGTX &3, dcSTX, GTX 2&3, GTX 5, and STX ...... 69

3-4 Fluorescence chromatogram of oxidized extraction (0.1 M acetic acid) with peaks appearing at retention times the same as dcGTX2&3 and dcSTX/dcNEO ...... 70

3-5 Extraction technique data (HPLC/FL). Data is represented by normalizing peaks to standards of dcGTX2&3 and dcSTX ...... 71

3-6 Fluorescence chromatogram of oxidized extraction (0.1 M acetic acid) with peaks appearing at retention times the same as dcGTX2&3 and dcSTX and with 10µg STX spiked pre-extraction (Blue Hole collected 8/27/09) ...... 72

3-7 Extraction technique data (HPLC/MS). Includes L. wollei toxins, dcGTX2, dcGTX3 and dcSTX relative MS responses ...... 73

3-8 HPLC/MS chromatograms of dcGTX2 & dcGTX3 SIM ions from Blue Hole Spring extracted with two different acids ...... 74

3-9 HPLC/MS/MS chromatograms of dcSTX from Blue Hole Spring extracted with two different acids...... 75

3-10 Sum of dcGTX2 and dcGTX3 concentrations for all sites over sampling period...... 76

3-11 Concentrations of dcSTX for all sites over sampling period...... 77

3-12 HPLC/MS SCAN of Silver Glen Springs Main Stem with SIM ions for L. wollei Toxins 1-6 ...... 78

8

3-13 HPLC/MS SCAN of Silver Glen Springs Natural Well with SIM ions for L. wollei Toxins 1-6 ...... 79

3-14 HPLC/MS SCAN of Blue Hole with SIM ions for L. wollei Toxins 1-6 ...... 80

3-15 Averaged MS responses over period of study for L. wollei Toxins 1-6 at each site ...... 81

3-16 MS Responses to Lyngbya wollei Toxin #1 detected from all sites over the sampling period...... 82

3-17 MS Responses to Lyngbya wollei Toxin #2&3 detected from all sites over the sampling period...... 83

3-18 MS Responses to Lyngbya wollei Toxin #4 detected from all sites over the sampling period...... 84

3-19 MS Responses to Lyngbya wollei Toxin #5 detected from all sites over the sampling period...... 85

3-20 MS Responses to Lyngbya wollei Toxin #6 detected from all sites over the sampling period...... 86

9

Abstract of Thesis Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Requirements for the Degree of Master of Science

PARALYTIC SHELLFISH TOXINS CHARACTERIZED FROM LYNGBYA WOLLEI DOMINATED MATS COLLECTED FROM TWO FLORIDA SPRINGS

By

Amanda Foss

May 2011

Chair: Edward Phlips Major: Fisheries and Aquatic Sciences

Lyngbya wollei is a commonly observed cyanobacterium in Florida’s spring fed systems and is considered a nuisance organism by forming large benthic and floating mats. Two Florida springs, Silver Glen Springs, and Blue Hole Spring (Ichetucknee) with standing crops of Lyngbya were sampled and mats characterized via microscopy.

Molecular data was acquired by the amplification and sequencing of sxtA and sxtG genes and a section of 16S rRNA from a filament collected from Silver Glen Natural

Well. Paralytic shellfish toxins (PSTs) were characterized utilizing High Performance

Liquid Chromatography (HPLC) coupled with Fluorescence (FL) and HPLC coupled with

Mass Spectrometry (MS). Extraction techniques may convert less toxic L. wollei toxins to more toxic decarbamoylgonyautoxins (dcGTX2&3) and decarbamoylsaxitoxin

(dcSTX). Anything stronger than 0.01 M HCl should not be utilized if preservation of the original toxin profile is desired. It was also determined that the use of Solid Phase

Extraction did not provide additional support when analyzing the toxins with LC/MS or

LC/FL. Pre-column oxidation and LC fluorescence was not a sufficient method of determining PST variants extracted from L. wollei mats since oxidation products are either converted or co-elute with other known PSTs and do not allow for

10

characterization. LC/MS is the only method explored in this study with the ability to do so, but without suitable standards for LC/MS/MS, quanititation and full confirmation could not be made for all the L. wollei Toxins. Analysis with LC/MS/(MS) of extracted algal material did show that dcGTX23 and dcSTX were present in L. wollei mats collected in Florida springs, with evidence pointing to the presence of all L. wollei toxins.

In future work with L. wollei derived PSTs, standards of the L. wollei toxins 1-6 should be utilized for LC/MS/(MS) or post column LC/FL quantification to accurately determine total PST content.

11

CHAPTER 1 INTRODUCTION

Objectives

Cyanobacteria (a.k.a blue green algae) produce a myriad of active metabolites which are known to induce toxic responses, with a long history of directly harming human and animal populations around the world (Sivonen and Jones, 1999). In

Florida’s spring fed rivers, one of the dominant cyanobacteria is Lyngbya wollei, which forms large benthic and floating mats. L. wollei has long been suspected of producing a myriad of toxins (Seifert et al. 2007, Berry et al. 2004, Teneva et al. 2002, Carmichale et al. 1997). Due to its widespread presence in many of Florida’s systems used for human recreation, it has become a management concern (Cowell and Botts 1994, Stevenson et al. 2007). Toxins of interest include dermatoxins (toxins that damage the skin), hepatotoxins (toxins that damage the liver), and neurotoxins (toxins that damage nerve cells). Anecdotal reports of adverse skin reactions (rashes, hives, and blisters), gastrointestinal disorders, respiratory illness, and even temporary loss of consciousness following potential exposure to cyanobacteria in Florida waterways has been documented by the Florida Department of Health (FDOH), (personal communication), but specific studies relating to toxins are limited. The FDOH initiated an evaluation of freshwater Lyngbya and its toxins in Florida springs in 2004 (Miller 2006). The study reported the presence of “saxitoxin-like” compounds, which were designated unknown paralytic shellfish toxins (PSTs). The presence of saxitoxin was not confirmed with the use of LC/MS/MS and other PSTs were not analyzed. Since the morphologic and molecular identification of L. wollei is under some scrutiny (Joyner et al. 2008), and toxin

12

production cannot be determined from simply determining the presence of Lyngbya, characterizing PST toxins present in Florida’s L. wollei mats is an important goal.

Based on initial assessment of data, it is hypothesized that L. wollei mats in

Florida springs produce Paralytic Shellfish Toxins (PSTs) which are not accurately characterized by frequently utilized methods of detection. The goal of this study was to test for the presence of PSTs in L. wollei dominated mat samples from two Florida spring fed pools. Currently, many methods utilized for PST detection remain largely nonspecific and present problems when analyzing a complex toxin profile such as the one for PSTs extracted from L. wollei. Additionally, improper extraction of the PSTs can inadvertently change the original toxin profile, resulting in over- or under- estimation of actual toxicity, so a careful examination of extraction protocols must be conducted to prevent toxin profile shifts (Oshima 1995, Indresena & Gill 2000, Vale et al. 2008).

Background

Lyngbya Growth in Florida Springs

Macroalgae growth in Florida’s springs has been on the increase over the last 50 years (Stevenson et al. 2007). Blooms of filamentous cyanobacteria, such as Lyngbya wollei (Farlow ex Gomont) Speziale and Dyck, have become increasingly responsible for declining habitat and water quality (Cowell and Botts 1994). Recently, a survey submitted to the Florida Department of Environmental Protection (FDEP) evaluating algal growth and nutrients in 21 different springs, found that Lyngbya wollei and

Vaucheria spp. were the most common macroalgae present (Stevenson et al. 2007).

Although Lyngbya abundance was not related to spring water nutrient concentrations, it was positively correlated to human activities around sampling sites and to other indicators of nutrient supply (i.e. sediment phosphorus) (Stevenson et al. 2007). It is

13

even proposed that these increases in macroalgae abundance are in response to other factors, such as lower dissolved oxygen levels that effectively reduce grazer interactions

(Heffernamn et al. 2010).

Lyngbya Toxin Production

Lyngbya (Family Oscillatoriales) is a known toxin producer, with more research conducted on marine species, such as Lyngbya majuscula. Although toxin production has not been as thoroughly investigated in freshwater Lyngbya as in its marine counterpart, some information is available in the scientific literature. In Australia,

Lyngbya wollei has been shown to produce cylindrospermopsin (CYN) and deoxy- cylindrospermopsin, a known hepatotoxin (Seifert et al. 2007). Cylindrospermopsin is toxic to both the liver and kidneys and has the ability to cause severe gastroenteritis. A freshwater species of Lyngbya isolated from the Florida Everglades was shown to produce pahayokolide A (Miccosukee for “Everglades”), a compound that inhibits microbial and green algal growth (Berry et al. 2004). Berry et al. (2004) suggested that although extracellular concentrations of this metabolite were low, pahayokolide A may function as an allelochemical, inhibiting the growth of other cyanobacteria or algae that may potentially compete for nutrients or space. Lyngbya aerugineo-coerulea, another freshwater species, is common in southern Europe and is known to be toxic to mammals and fish cells (Teneva et al. 2002). Because the chemicals from this Lyngbya species were crudely extracted, it is unknown which chemical isolate(s) produce the cytotoxic response.

Samples of L. wollei collected from Lake Guntersville, Alabama have yielded PSTs that resulted in positive neurotoxic responses in mouse bioassays. The presence of the

PSTs decarbamoylgonyautoxins 2&3, decarbamoylsaxitoxin, as well as 6 novel PSTs

14

(L. wollei Toxins 1-6) were detected utilizing post column fluorescence derivatization

HPLC techniques, and confirmed with NMR (Carmichael et al. 1997, Yin et al. 1997,

Onodera et al. 1997). This, however, has not been documented in L. wollei samples collected from any other source. The PSTs produced by the Alabama Lyngbya wollei are less toxic derivatives of PSTs as determined by mouse bioassay. Yin et al. (1997) found that PST production and dry weight increased with calcium concentrations in

Lyngbya mats, a phenomenon which has been observed in previous growth studies

(Cowell and Botts, 1994).

Paralytic Shellfish Toxins

Properties and toxicity

Paralytic Shellfish Toxins (PSTs) are commonly referred to as saxitoxins (or paralytic shellfish poisons). PSTs are a group of neurotoxic alkaloids produced primarily by freshwater cyanobacteria and marine dinoflagellates, with over 57 variants reported to date (Wiese et al. 2010). Exposure to these toxins may result in illness, paralysis, and even death. The PSTs reversibly bind to voltage gated sodium channels, inducing neurological symptoms. The degree of the effect varies with the amount of toxin exposed to, the exposure route and which of the variants are present. The different PSTs are grouped by hydrophilicity and by the nature of the side chains

(carbamate, acetate, sulfate, etc.). Not all variants are of comparable toxicity and identification of specific PSTs is of importance, not only when analyzing for the toxins, but also when calculating potential risk. When analyzing for PSTs in shellfish for human consumption, a limit of 80 µg STX-equivalents per 100g of shellfish meat is set as the maximum acceptable limit in most countries, with few countries having lower established limits (Campas et al. 2007). For drinking water, there are even fewer

15

specific guidelines. A level of three µg STX-equivalents L-1 is recommended as the health alert level for drinking water in Australia based on documented human toxicity data (Fitzgerald et al. 1999). “STX equivalents” are determined from bioassay response, rather than analytical variants detected. Current methods are moving away from the mouse bioassay approach and toward analytical techniques, so the European

Commission of Marine Biotoxins in Shellfish rate the toxicity of some of the PSTs with toxicity equivalency factors (TEF) based on acute toxicity in mice. With saxitoxin representing the most toxic and rated as 1, the following toxins are rated as: NEO =1,

GTX1 = 1, GTX2 = 0.4, GTX3 = 0.6, GTX4 = 0.7, GTX5 = 0.1, GTX6 = 0.1, C2 = 0.1,

C4= 0.1, dcSTX = 1, dcNEO= 0.4, dcGTX2 = 0.2, dcGTX3 = 0.4, and 11-hydroxy-STX

=0.3 (Scientific Opinion of the Panel on Contaminants in the Food Chain on a request from the European Commission on Marine Biotoxins in Shellfish – Saxitoxin Group

2009). Unfortunately, this rating effort does not include all the potential variants and scientific literature is not always in agreement with regard to toxicity.

Potential risks and trophic transfer

Risk to animals and humans associated with Lyngbya PST production is currently unclear as reports of exposure are not be always linked to toxin production. Direct contact with the mats may cause mild neurotoxic symptoms in mammals (e.g., tingling, numbness, etc.), but ingestion (direct or indirect) may be required for strong adverse toxic effects. Direct ingestion has not been reported, but indirect ingestion of organisms exposed to PSTs may occur. Currently, it is unknown to what extent organisms feed on

Lyngbya, or if trophic transfers take place, but some work has been conducted on herbivory. Camacho and Thacker (2006) investigated amphipod herbivory on PST- producing Lyngbya wollei and found that the presence of PSTs stimulated the feeding of

16

Hyalella azteca (an omnivorous amphipod), even though the sheath prevented feeding.

Their results suggest that Hyalella can tolerate or detoxify the PSTs produced by

Lyngbya. Whether or not trophic transfer of PSTs produced by Lyngbya occurs is not yet known.

Toxin Extraction

Currently, there is not an effective published method for the extraction of PSTs from algal mats such as the ones produced by Lyngbya wollei. Extraction methods have been published for PSTs from phytoplankton (Ravn et al. 1997), but certain characteristics of Lyngbya (i.e., the presence of accessory pigments and a tough sheath) make extraction difficult. A reproducible extraction method for algal mat material such as Lyngbya is necessary for accurate qualitative and quantitative analysis of PST toxins. Sample collection, homogenization and preparation techniques, extraction of lyophilized algal material and additional clean-up techniques such as Solid

Phase Extraction (SPE) must be considered to improve reproducibility and to reduce interferences and false peaks.

Traditional extraction methods (e.g. AOAC Official Method 2005.06), based on human consumption of shellfish, utilize an acidic solution as the extraction media. PSTs are readily soluble in water and are considered heat sTable, but they are reported to degrade at higher pH (Botana 2008). Actually, both chemical and biological interconversion of PST variants is a common occurrence, making qualitative and quantitative determinations difficult. For instance, it has been demonstrated that extraction with 0.1 M hydrochloric acid (via boiling) may convert less toxic variants of

PSTs, such as GTX5 to more toxic ones, such as saxitoxin (STX) (Oshima 1995, Vale et al. 2008, Indresena & Gill 2000). Keeping the boiling times short and reducing the

17

concentration of acid may reduce the risk of hydrolyzing the less-toxic variants to the more toxic ones, while keeping the sample at a sufficient low pH to prevent degradation and other conversions. In addition, separation by SPE may also aid in reducing/removing many of the chemically active components that could change the extracted PSTs post-extraction, thereby allowing for longer hold times of samples prior to analysis.

Toxin Analysis Techniques

Analysis of extracted algal material has been conducted via multiple methods, including Mouse Bioassay (MBA), Enzyme Linked Immunosorbent Assay (ELISA), receptor binding assays, and Liquid Chromatography coupled with either Mass

Spectrometry or Fluorescence (LC/MS or LC/FL). Because MBA is considered unethical, assay approaches for toxins have moved toward analytical/instrumental testing. While ELISA is reportedly highly specific and sensitive, complex biological matrices are known to confound and interfere with results. Additionally, ELISA kits for saxitoxins are designed for the assay of only saxitoxin. Because other variants are not reported to have a 1:1 cross reactivity, STX-specific ELISAs are not useful quantitatively and may, depending on what toxin variants are present, severely underestimate the actual concentrations of STX-related toxins that are present. Although receptor binding assays are a better measure of how the extract may impact sodium channel receptors, these assays do not identify the variants present and, unfortunately, the cross reactivity of other similarly acting chemicals may cause interference. Methods employing chromatography include post- and pre-column oxidation techniques since PSTs have only a weak natural chromophore for spectrophotometric detection and requires oxidation to detect the compound. Post-column oxidation requires a complex setup and

18

daily maintenance requirements. The Association for Analytical Chemist Official method for shellfish toxins adopted the pre-column oxidation with hydrogen peroxide and periodic acid solutions which is considered by regulators to be the accepted method for the analysis of PSTs. Prechromatographic oxidation was utilized for detection of

PSTs present in Lyngbya dominated mats in this study. Confirmation of PST toxins and scans for multiple PSTs was conducted using Liquid Chromatography coupled with

Mass Spectrometry (LC/MS).

Molecular Biology of Lyngbya and Saxitoxin gene

Distantly related organisms, the dinoflagellates and cyanobacteria, both utilize the same biosynthesis pathway for PST production (Kellman et al. 2008). The saxitoxin biosynthetic gene cluster (sxt) was recently identified in several cyanobacteria species including Cylindrospermopsis raciborskii T3 (Kellmann et al. 2008), Anabaena circinalis

AWQC131C, Aphanizomenon sp. NH-5 (Mihali et al. 2009) and Raphidiopsis brookii D9

(Stucken et al. 2010). While C. raciborksii sxt biosynthesis gene cluster covers a 35 kb

(kilobase pair) region and contains 31 Open Reading Frames (ORFs), A. circinalis has

22 ORFs spanning a 29 kb region, Aphanizomenon sp. has 24 ORFs spanning a 27.5 kb region, and R. brookii has 24 ORFs in a 25.7 kb genomic region. Kellmann et al.

(2008) proposed that 11 to 14 of these genes are necessary for the production of the parent saxitoxin molecule and other genes in the biosynthetic cluster encodes for enzymes for the modification of the parent molecule. Therefore differences in the number of encoded enzymes in the sxt biosynthesis cluster from different organisms probably reflect the types of PST variants produced by these species. Interestingly, sxtX, which is present in C. raciborskii T3, Aphanizomenon flos-aquae NH-5, and

Lyngbya wollei , but absent in A. circinalis AWQC131C, encodes an enzyme similar to

19

cephalosporin hydroxylase which is proposed to perform N-1 hydroxlation of saxitoxin.

Indeed while the former two species form N-1 hydroxylated PSTs (e.g. neosaxsitoxin),

A. circinalis does not produce it and it has yet to be detected in L. wollei extracted samples (Kellmann et al. 2008). SxtI, a gene proposed to encode for a carbamoyltransferase, was naturally mutated in L. wollei previously analyzed from

Alabama and encoded an inactive enzyme (Kellmann et al. 2008). According to

Onodera et al. (1997), L. wollei species do not produce carbamoylated PSTs and it is suggested that this is due to the mutated sxtI gene (Kellmann et al. 2008).

PSTs produced by L. wollei may play an important role in the proliferation of the cyanobacterium by influencing grazer interactions. A recent paper characterizing the L. wollei genetics found genes coding for Multi Antimicrobial Extrusion (MATE) proteins of the NorM family and contains a gene, sxtPER, which is similar to permeases of the

Drug and Metabolite Transporter (DMT) family (Mihali et al. 2011). These proteins may increase extracellular transport of PSTs produced by L. wollei which could change our understanding of the role PST production of L. wollei plays , but further studies to determine if active transport is occurring would be required.

20

CHAPTER 2 MATERIALS AND METHODS

Sampling

Site Description

Three sites were chosen for Lyngbya collection, one in Blue Hole Spring

(Ichetucknee State Park) and two from Silver Glen Springs Recreation Area (Ocala

National forest). These springs were chosen as they have standing Lyngbya wollei dominated mats observable year round.

Silver Glen Springs is located in the Ocala National Forest in Marion County

Florida. It is utilized as a recreation area maintained by the United States Department of Agriculture (USDA) Forest Service and has a large combined spring pool emanating from 2 vents (east and southwest vents), flowing ¾ mile into Lake George, Florida’s second largest freshwater lake. Lake George flows into the St. Johns River out towards the Atlantic. Both freshwater and saltwater species of fish are prominent in both vents and throughout the spring run as it is influenced by natural sources of saline water.

The main pool is sectioned by three ropes, two encompass a large Lyngbya mat, meant to keep boaters from entering the spring pool, and the third restricts access to swimmers from the southwest vent. The eastern spring vent is a 1st magnitude (> 100 cubic feet per second (cfs) spring shaped as a conical depression, approximately 20 feet (ft) deep and is frequented by swimmers. Very little Lyngbya was observed in this open section, with pristine waters and a sandy bottom. The southwest vent (also known as the “Natural Well”) is a 2nd magnitude spring (>10-100 cfs). The spring is a vertical vent, approximately 40 ft deep and surrounded by Lyngbya mats. Both sectioned off areas of the spring were sampled over the period of this study.

21

Blue Hole Spring (aka Jug Spring) is a 1st magnitude spring in the Ichetucknee state park located in Columbia County FL. The spring directly contributes to the flow of the Ichetucknee River, which flows six miles until emptying into the Santa Fe River.

Although recreational activities such as tubing and canoeing are allowed down the main spring run, the activities in Blue Hole are limited to swimming, cave diving and snorkeling. Blue Hole Spring is one of many that feed into the Ichetucknee system, the second downstream of seven other named springs, but with the greatest contributing flow at over 100 cfs. Lyngbya mats in this spring are located along the bottom as well as vertically distributed along the northern face of the spring hole, loosely attached to the roots of the Bald Cypress (Taxodium distichum) trees pictured in Figure 2-4. The vertically distributed mat was the one sampled for this study.

Collection

The goal of this study was to collect representative L. wollei samples from the top of the mats for toxin analysis. A single line transect was utilized at each site to assure all collections were made from an area there was L. wollei material. A transect was established by running survey tape from one end of the mat to the other, approximately down the middle. Transects were re-established each collection period in the same direction and approximate area. The largest mat, at Silver Glen Main Stem area, required the longest transect of thirty-seven meters. The tape was run along the bottom across the length of the mat from east to west. The Natural Well transect was nine meters in length, running along the bottom across the mat from north to south. A 5 meter transect was utilized at Blue Hole spring, set up at 1 meter in depth. The mat at

Blue Hole Spring was not horizontally distributed like the mats in Silver Glen Springs.

22

The Lyngbya grows vertically, loosely attached to the roots of a cypress tree (Figure 2-

4; D, E, & F), therefore collections were taken at half the max depth of one meter.

Algal mat samples were collected as grab samples along the single line transect at each site. The five grab collection points were chosen based on random numbers generated in Microsoft® Excel. Each grab was approximately 20g wet and composited in 10”x10” bags for transport and analysis. Large invertebrates and debris were rinsed from the composited mat lightly with spring water. The samples were maintained below

10°C after collection and for transport. All sample preparation for lyophilization, DNA isolation, and algal analysis was conducted within 4 hours of collection.

There were six collection events where all three sites were visited. One composited collection per site was taken during a sampling event. A duplicate field collection was collected from one of the sites during each sampling event. This resulted in 4 total samples collected each event. A total of 24 samples were collected for this study (Table 2-1). Data acquired from sampling sites included canopy coverage

(densiometer, Wildlife Supply Company, Yulee, FL), depth (max, secchi & sampling), pH, Dissolved Oxygen (DO), specific conductivity, salinity, temperature (YSI MP6600 sonde, YSI Inc., Yellow Springs, OH) and other observations regarding mat size/distribution. Densiometer readings for overstory density (canopy coverage) were taken at every sample grab collection (5 readings per composited collection) and averaged for each sample site. Overstory density was determined from raw data by the following calculation: 100% - (# unfilled squares x 4.17) = average overstory density.

23

Standards and Reagents

Standards for Instrument Calibration and Quantification

Standards for analysis and instrument calibration were purchased from two sources. The Certified Reference Materials Program of the National Research Council of Canada provided Saxitoxin (STX), Neosaxitoxin (NEO), Decarbamoylneosaxitoxin

(dcNEO), Gonyautoxin 1&4 (GTX1&4), Gonyautoxin 2&3 (GTX2&3), Gonyautoxin 5

(GTX5), Decabamoyl Gonyautoxin 2&3 (dcGTX2&3) and N-sulfocarbamoyl- gonyautoxin-2 and -3 (C1&2) (Institute for Marine Biosciences, Halifax, Canada) . An additional source of saxitoxin (STX) utilized was purchased from the National Institute of

Standards and Technology Standard Reference Materials (Gaithersberg, MD). All standards were diluted in the same solution they were supplied in.

Reagents

HPLC grade acetonitrile, sodium hydroxide, glacial acetic acid, periodic acid, disodium hydrogen phosphate, hydrogen peroxide (30% wt. sol), hydrochloric acid

(certified ACS), (>99%), formic acid, and ammonium formate were all purchased from

Thermo Fisher Scientific (Waltham MA). Glutaraldehyde (25% in water) for algal preservation was also purchased from Thermo Fisher Scientific. All mobile phases for

HPLC were filtered through 0.45µm PVDF Millipore filters (Thermo Fisher Scientific,

Waltham MA) prior to utilization. Deionized water at 18 MΩ cm was provided in house by a Pure Lab Ultra Filtration System (Siemens Water Technologies Corp. Warrendale

PA).

24

Sample Preparation

Qualitative Analysis of Filamentous Macroalgae Samples

Algal mats were placed in a 17”x11”x1.5” tray within 4 hours of collection and mixed by hand, while rinsing with DI water to remove non-algal debris. Once thoroughly mixed, five subsamples (approximately 1 gram wet weight each) were removed from each field collected filamentous macroalgae sample and preserved with gluteraldehyde

(approximately 0.25%). These specimens were stored at 4°C until full algal identification/characterization was accomplished.

Microscopy was conducted by Andrew Chapman at GreenWater

Laboratories/CyanoLab in Palatka, FL. In preparation for microscopy, subsamples

(approximately one gram wet weight) of preserved samples were placed in plastic petri dishes with sample water or distilled water. A minimum of two wet mounts were made from the subsamples. The entire area of each wet mount was viewed at 100x on a

Nikon Eclipse TE200 inverted microscope equipped with phase contrast optics and epi- fluorescence. Higher magnifications were used as necessary for species identifications.

Identifications were taken to species level when possible. A list of observed species was constructed of species organized from most to least abundant based on empirical judgment of the analyst. Epi-fluorescence was utilized to aid in characterization of epiphytes attached to Lyngbya filaments.

DNA Isolation, Polymerase Chain Reaction, and Sequencing

A single filament of Lyngbya wollei was isolated within 4 hours of collection via microscopy from the sample collected in Silver Glen Springs Natural Well (9/29/09). An effort to cut a clean filament absent of epiphytic algae was made by observing the

25

filament at 100x and 400x with phase contrast optics prior to extraction for DNA isolation. The filament was stored at -4°C until DNA isolation could be conducted.

DNA analysis was conducted by Mete Yilmaz at the University of Florida,

Gainesville from a single filament according to Tillet and Neilan (2009). The filament was placed in 500 µL of XS buffer containing potassium ethyl xanthogenate (1%); Tris-

HCl, pH 7.4 (100 mM); EDTA, pH 8 (20 mM); ammonium acetate (800 mM); and SDS

(1%). The tube was incubated at 70 °C for 2 h, vortexed and incubated on ice for 30 min, followed by centrifugation at 15,000 g for 10 min. The supernatant was transferred to a clean microcentrifuge tube and 0.7 volumes of isopropanol was added. DNA was pelleted by centrifugation at 15,000 g for 10 min, washed with 70% ethanol, air-dried and resuspended in 50 µL of double distilled water.

Amplification and sequencing of 16sRNA gene was performed as described in

Yilmaz et al. (2008). The presence of saxitoxin biosynthesis genes was checked by partial amplification of sxtA and sxtG genes with primer pairs designed in this work.

Primer pair sxtAF (5’-AGCTGGACTCGGCTTGTTGCTTC) and sxtAR

(CACTTGCCAAACTCGCAACAGG) was used to amplify an approximately 657 bp fragment within polyketide synthase sxtA4 region. Primer pair sxtGF (5’-

ATTGAAGCACCAATGGCAGATCG) and sxtGR (5’AGAGTTCCGCGTCGGCGAAC) was utilized to amplify an approximate 700 bp fragment within the amidinotransferase gene, sxtG. Both PCR reactions contained 5 µL of genomic DNA, 20 pmol of each primer (Eurofins, MWG operon, Huntsville, AL), 200 µM of each deoxynucleoside triphosphate (Thermo Fisher Scientific, Waltham MA), 1.5 mM MgCl2, 10 µL of 5X green buffer, and 2 units of GoTaq® Flexi DNA polymerase (Promega, Madison WI) in a total

26

volume of 50 µL. Amplification was initiated with denaturation of the genomic DNA at

95°C for 3 min, followed by 33 cycles of 95°C for 30 sec, 58°C for 30 sec, 72°C for 1 min, and ended with an extension step at 72°C for 5 min. PCR products were purified from agarose gels (1.5% w/v) with the QIAquick Gel extraction kit (Qiagen, Germantown

MD). Sequencing of 16s rRNA was performed with sequencing primers reported in

Yilmaz et al. 2008 and sxt genes were sequenced on both strands using the same primers used in amplifications at the University of Florida’s Interdisciplinary Center for

Biotechnology Research core sequencing facility. Sequences were manually checked and corrected using Mega version 4.1 (Tamura et al. 2007).

Toxin Samples

Samples were prepared for toxin extraction after qualitative algal and DNA collections were made. Excess water was lightly squeezed from the mats and filaments were cut wet in 1 centimeter (cm) lengths to maximize homognization and decrease lyophilization time. Samples were frozen (-20oC) in freeze flasks and lyophilized at -

50oC (Thermo Savant Modulyo Freeze Dryer System, Thermo Fisher Scientific,

Waltham MA). Lyophilized one gram subsets were utilized for extraction experiments and analysis.

Sonication and mechanical homogenization presented difficulties by fouling equipment and/or not sufficiently homogenizing the material. Manually cutting wet filaments prior to lyophilization allowed for the most direct contact of extraction solvent with the inner cells, as determined from microscopy.

27

Paralytic Shellfish Toxin Extraction

Extraction Assessment

Methods for extraction of the dried algal material included the AOAC Official

Method 2005.06 for PST extraction and by varying the concentration and type of acid used. The difficulty when working with algal material such as Lyngbya is effectively extracting the toxin from the thick sheath without extracting accessory pigments that may interfere with analyses and avoiding conversions of the toxins during extraction.

The two acids assessed were hydrochloric acid (0.1, 0.01, 0.001 M concentrations) and acetic acid (0.1, 0.01, and 0.001 M concentrations). One sample collected from Blue

Hole (collected 4/2/10) was weighed out in 1 gram subsets and run in triplicate for each extraction. Samples were extracted in 50 mL of extractant solution at 100˚C ± 5˚C with contant stirring for 5 minutes and constant monitoring of temperature. Once cooled to room temperature, the samples were brought back to original volume with the addition of the original extractant solution. All samples were filtered with Whatman Puradisc™

0.45 µm PVDF syringe filters (Whatman Inc. Piscataway, NJ) and stored at -4 oC prior to oxidation, further preparation and analysis.

SPE and Lyophilization Assessment

A sample set collected in August 2009 was utilized, with Blue Hole, Silver Glen

Main stem, Silver Glen Natural Well, and duplicate collection of Silver Glen Natural

Well. Five milliliter aliquots of extractant solutions were passed through conditioned

200mg Alltec® C18 Solid Phase extraction cartridges (Grace Analytical, Deerfield IL) and rinsed with 0.003M HCl (3 mL). The load and wash fractions were collected and frozen in 15 mL glass centrifuge tubes. The fractions were lyophilized to dryness and reconstituted in 500 µL 0.003 M HCl for analysis with LC/MS at a concentration of 0.2

28

grams dried material per mL solution. This solution was diluted to the original extract concentration (0.02 g mL-1) for comparison with original LC/FL and LC/MS Scans.

Final Extraction Protocol

All samples from all collections were extracted in the same way with acetic acid.

One gram dry weight was added to 50 mL of 0.1 M acetic acid and the solution boiled for 5 minutes while temperature was being monitored closely. Once cooled, the volumes were brought back to original volume with 0.1 M acetic acid, filtered through

Whatman Puradisc™ 0.45 µm PVDF syringe filters, and stored at -4˚C until analysis.

Select samples were utilized for pre-column oxidation and LC/FL. All samples were analyzed via LC/MS/(MS).

Paralytic Shellfish Toxin Analysis

HPLC/ Fluorescence

Peroxide Oxidation

The basic structure of PSTs is a 3,4-propinoperhydropurine tricyclic system (Table

2-2). Oxidation with H2O2 cleaves the propino ring in the presence of an alkaline solution, forming a purine. This purine becomes fluorescent in acidic solution and can be monitored at an excitation of 340 nm and emission of 395 nm (Figure 2-7) (Botana

2008). With separation via HPLC, qualitative and quantitative determinations can be made. The Lawrence, Niedzwiadek, Menard method (2005) via prechromatographic oxidation and LC-fluorescence is the accepted AOAC International Official First Action

Method due to the successful quantification of multiple variants of PSTs. This method was used to monitor a wide range of PSTs known to occur, including STX, GTX 5,

C1&C2, GTX 2&3, dcGTX2&3, dcNEO & dcSTX.

29

The peroxide oxidation technique employed was as follows; 25 µL 10% (w/v) aqueous H2O2 was added to 250 µL 1 M NaOH and vortexed. One hundred microliters of sample and standards were added to the oxidant solution and allowed to react for 2 minutes at room temperature. Concentrated glacial acetic acid (25µL) was added to acidify the purine group and allow for fluorescence determination. Standards used to calibrate this method included STX, dcSTX, dcNEO, GTX2&3, C1&2, dcGTX2&3 &

GTX5. Oxidation products do eventually degrade, so all oxidized samples were analyzed within 8 hours of their oxidation.

Periodate Oxidation

Periodate oxidant was prepared fresh daily by mixing 5 mL each of 0.03 M

Periodic acid, 0.30 M ammonium formate and 0.3 M Sodium hydrogen phosphate and by adjusting the pH to 8.2 with 175 µL 0.2 M Sodium Hydroxide. All standards and solutions were oxidized in 100 µL aliquots with the addition of 500 µL oxidant solution, vortexing and one minute reaction time at room temperature. Five microliters of acetic acid were added to finalize the reaction prior to analysis via LC/FL. Periodate oxidation is utilized to monitor for GTX1&4 & NEO, so standard solutions of GTX1&4 & NEO were prepared via this method. Due to the significant amount of noise produced by L. wollei extracted samples, non resolved peaks occurring from oxidation of the PSTs present in the material, and the longer run times required to clear the system of retained fluorescent material, this method was abandoned for PST characterization. LC/MS was utilized to scan for PSTs that would normally be observed in the periodate oxidation procedure (NEO and GTX 1&4).

30

Chromatographic and Fluorescence Conditions

Most PSTs are water-soluble non-volatile compounds, only existing in ionic forms; therefore HPLC is the preferred method of separation. A Thermo Separations Product

(TSP) SpectraSystem® High Performance Liquid Chromatography System equipped with a P4000 Pump, AS3000 autosampler controlled by a SN4000 Interface and coupled with a FL3000 for fluorescence detection (Thermo Fisher Scientific, Waltham

MA) was used for HPLC/FL analysis. An Agilent SS420X interface (Agilent

Technologies, Inc., Pleasanton CA) was utilized for digitizing analog detector signals from the fluorescence detector and Xcalibur™ Software (Thermo Fisher Scientific,

Waltham MA) was utilized for acquisitioning and quantification of PSTs. The LC parameters were modified from the AOAC Official Method 2005.06 to accept a lower flow of 1mL/min and to provide sufficient resolution of PST oxidation products. A

150mm x 4.6 mm, 5 µm particle size reversed phase Aquasil C18 column (Thermo

Fisher Scientific, Waltham MA) was used for separation of oxidation products. Two mobile phases where utilized, Mobile Phase A: 0.1 M ammonium formate and Mobile

Phase B: 0.1 M ammonium formate in 5% acetonitrile. Both were adjusted to pH 6 w/ 6 mL 0.1 Acetic Acid. At a flow rate of 1mL/min, runs were 20 minutes with the following linear gradient: 100%-90% A over 5 minutes, hold at 90% A for 5 minutes, from 90%-

30% A over 3 minutes, 30% back to 100% over 7 minute (Table 2-3).

A 20µL injection was used for all LC/FL runs and all samples were monitored both with and without oxidation to verify there were no interfering naturally fluorescent peaks.

Fluorescence was monitored with an excitation of 340nm and an emission of 396 nm.

31

Detection Limits

Method Detection Limits (MDLs) were determined with a signal to noise ratio of 3:1 in the matrix for all but dcNEO, dcSTX and dcGTX2&3. Standard addition and five point standard curves were used to estimate conservative detection limits for dcNEO, dcSTX and dcGTX2&3. The method detection limits are reported in Table 2-4 analyzed with fluorescence with 20µL injections. Standards of dcGTX2&3 were linear to extract concentrations of 15 ppm (µg L-1), dcNEO to 5 ppm (µg L-1), and dcSTX to 10 ppm (µg

L-1).

Sample Analysis

Paralytic shellfish toxins in samples were evaluated by comparing peak areas to standards curves and standard addition via precolumn oxidation and HPLC/FL. A minimum of three point standard curves for dcGTX2&3 and dcSTX were prepared to bracket sample concentrations, with a standard check run at the end of each sample set and verified within 10% of initial standard and sample. All samples and standards to be analyzed together were oxidized fresh daily with replicate runs made on standards and on at least one sample per set to assure reproducibility of equipment. Fluorescence data was interpreted by normalizing peak areas of samples to standard curves of dcGTX2&3 and dcSTX, which were run daily. Resultant values achieved this way were reported in dcGTX3&3 and dcSTX equivalents.

HPLC/MS/(MS)

Chromatographic and Mass Spectrometry Conditions

Liquid chromatography/Mass spectrometry/Mass spectrometry (LC/MS/MS) and

LC/MS were utilized for the characterization of PST variants in all samples. A Thermo

Finnigan™ Surveyor HPLC system coupled with a Thermo Finnigan™ LCQ Advantage

32

MSn ion trap tandem mass spectrometer was utilized for validation of PSTs in samples.

A TSKgel® Amide 80 250mm x 2 mm HPLC Column (5µm particle size) was employed for chromatographic separation (Tosoh Bioscience LLC, Grove City, OH). Two mobile phases were used, C: 100% DI with 3.6 mM formic acid and 2 mM ammonium formate and solvent D: 95% (v/v) acetonitrile with 3.6 mM formic acid and 2 mM ammonium formate. The elution gradient was as follows; hold at 35% C for two minutes, 35%-70% over ten minutes, 70% held for 5 minutes, return to 35% in 10 minutes and hold for ten additional minutes for re-equilibration (Table 2-5).

Paralytic Shellfish Toxin standards utilized in all MS/MS analyses were directly infused on the mass spectrometer and autotuned with Xcalibur™ software. Once

MS/MS parameters and collision energies were optimized, Scans (150-500 m/z), Single

Ion Monitoring (SIM) and Selective Reaction Monitoring (SRM) were conducted on standards and samples.

Detection Limits

Limits of detection could only be calculated for toxins with available standards.

Detection limits were determined by standard addition techniques and utilizing a signal to noise ratio of 3:1 with 20µL injections for non detectable toxins. For toxins positively identified in the matrix, a conservative approach and standard addition was utilized to estimate method detection limits. The elution order of standards was not the same in the matrix as they were in standard solution due to shifts in retention time as verified via spikes. MDLs and retention times of PSTs for LC/MS and LC/MS/MS are reported in

Table 2-6.

33

Sample Analysis

Extraction Assessment Samples. All extracts from the extraction assessment portion of this study were analyzed with the matrix removed under N2 gas at 60˚C and reconstituted in 0.003M HCl to eliminate/minimize the matrix influence on ionization.

Samples were first analyzed for dcGTX2&3 and dcSTX via LC/MS/MS since florescence data exhibited response for only those two toxins. There was a quantitative disparity between LC/FL data and LC/MS data, so a Total Ion Current (TIC) scan was conducted from 150-500 m/z. Table 2-7 shows ions monitored for with LC/MS/(MS)

SCANs and SIM.

MS responses from TIC SCAN data was utilized for comparisons between samples for LWT 1-6 since standards were not available for L. wollei toxins. Data utilized for comparison of dcGTX2 and dcGTX3 was also achieved from SCANs. SIM data was utilized for dcSTX comparison. Matrix effects were assessed with the use of pre- and post- extraction spikes for STX, dcSTX and dcGTX2&3. Only post extraction spikes were utilized for dcNEO due to the limited availability of standard.

All collections. All samples extracted in 0.1 M acetic acid were analyzed with the following approach: TIC scans (150-500 m/z) for dcGTX2, dcGTX3, and L. wollei toxins

1-6. Standard addition with dcGTX2&3 and TIC SCANs were utilized for the quantitative data achieved for dcGTX2 and dcGTX3. TIC SCAN data resulted in high noise at the dcSTX and dcNEO retention times and could not be used for detection.

Instead, SIM was used to monitor for dcSTX (m/z 257) and dcNEO (m/z 273). Standard addition dcSTX with SIM data was utilized for quantification of dcSTX.

Pre-extraction spikes (STX) and post extraction spikes (STX, dcGTX2&3, dcSTX, dcNEO, GTX1&4 and NEO) were utilized with representative samples from each

34

collection site to evaluate retention time shifts and provide additional qualitative data.

One sample per collection site was also chosen to monitor (SIM) for ions of toxins with known standards (GTX 1&4 and NEO).

To determine reproducibility and handle variability of equipment, analysis was conducted by sample set (sample set = one collection month) and then by sample site.

A full sample set was analyzed in a single day with replicates to assess intraday variability. Samples analyzed by site were then handled in the same manner with all samples, including field replicates, analyzed in a single day. A sample analyzed at the beginning of a day was re-analyzed (SCAN and SIM) at the end of the day. Data for the entire day was accepted if the rerun analysis was <10% different for all toxins (by comparing MS responses). Field replicate data was averaged for final concentration calculations and comparison.

35

Figure 2-1. Silver Glen Springs Recreational Park photo; white lines represent ropes restricting general access; Natural Well and Main Stem were sampled in this study. Photo Courtesy of St. Johns River Water Management District

36

A B

C D

E F

Figure 2-2. Series of Silver Glen Main Stem photos A) Overlooking East Vent area B) Red buoys mark the 2-lines separating boat traffic from main spring area, where large Lyngbya mat is found (Main Stem) C) Underwater photo of Main Stem mat D) Lyngbya loosely attached to separating rope E) Florida Gar (Lepisosteus platyrhincus) in Main Stem F) Lyngbya mat October 2009 G) Lyngbya mat in August 2009 H) Lyngbya mat November 2009

37

G H

Figure 2-2. Continued

38

A B

C D

E F

Figure 2-3. Series of Silver Glen Natural Well (Southeastern Vent) photos A) Looking into sectioned off area of Natural well from swim area B) Looking out from Natural Well sample area C) Looking down at Natural Well vent D) Standing Lyngbya mat August 2009 E) Lyngbya and Vallisneria americana F) Lyngbya mat January 2010 with Tilapia and nests G) Lyngbya over natural vent August 2009 H) Lyngbya surrounding vent October 2009

39

G H

Figure 2-3. Continued

40

A B

C D

E F

Figure 2-4. Series of Blue Hole (Jug) Spring photos A) View of spring from wooden dock B) Cypress Trees with Lyngbya attached loosely to submerged roots C) Looking down Blue Hole Spring Vent D) View of vertically distributed mat loosely attached to tree roots E) & F) View of Lyngbya mat October 2009 G) & H) Lyngbya mat February 2010

41

G H

Figure 2-4. Continued

Table 2-1. List of sample collections Site Sampling Dates Silver Glen Main Stem 8/25/09* 9/29/09 10/23/09 11/24/09* 1/27/10 4/5/10 Silver Glen Natural Well 8/25/09 9/29/09 10/23/09* 11/24/09 1/27/10* 4/5/10 Blue Hole Spring 8/27/09 9/24/09* 10/26/09 12/4/09 2/3/10 4/2/10* * Duplicate field replicate collected at this time

42

Table 2-2. Saxitoxin structural variants

M.W. Toxin R1 R2 R3 R4 R5 (Free Base) STX H H H OCONH2 OH 299.3 ¯ GTX1 OH H OSO3 OCONH2 OH 411.4 ¯ GTX4 H OSO3 H OCONH2 OH 411.4 ¯ GTX2 H H OSO3 OCONH2 OH 395.4 ¯ GTX3 H OSO3 H OCONH2 OH 395.4 ¯ GTX5 H H H OCONHSO3 OH 379.4

NEO OH H H OCONH2 OH 315.1 ¯ ¯ C1 H H OSO3 CONHOSO3 OH 475.4 ¯ ¯ C2 H OSO3 H CONHOSO3 OH 475.4 dcSTX OH H H OCONH2 OH 256.3 ¯ ¯ dcNEO H H OSO3 CONHOSO3 OH 272.1 ¯ dcGTX2 H H OSO3 OH OH 352.3 ¯ dcGTX3 H OSO3 H OH OH 352.3 ¯ *LWT 1 H H OSO3 OCOCH3 H 378.3 ¯ *LWT 2 H H OSO3 OCOCH3 OH 394.3 ¯ *LWT 3 H OSO3 H OCOCH3 OH 394.3 *LWT 4 H H H H H 240.3

*LWT 5 H H H OCOCH3 OH 298.3

*LWT 6 H H H OCOCH3 H 282.3 *Structures acquired from Onodera et al. 1997, there are no known standards for these variants

43

Figure 2-5. Conversion of saxitoxin to fluorescent derivative for HPLC/FL detection

44

Table 2-3. LC/FL gradient Time (min) Solvent A Solvent B 0 100% 0% 5 90% 10% 10 90% 10% 13 30% 70% 20 100% 0%

Table 2-4. MDLs for PSTs via fluorescence in elution order (µg g-1 dry weight) Toxin MDL dcGTX 2&3 0.5 C1&C2 0.5 dcSTX 0.5 dcNEO 5.0 GTX 2&3 2.5 GTX 5 5.0 STX 1.0

Table 2-5. LC/MS gradient Time (min) Solvent C Solvent D 0 35% 65% 2 35% 65% 12 70% 30% 17 70% 30% 27 35% 60% 37 35% 65%

45

Table 2-6. MDLs and retention times of PSTs analyzed via LC/MS/(MS) SIM SRM Retention time Retention Time MDL MDL (min) (min) Toxin (µg g-1) (µg g-1) Matrix Standard STX 5 10 13.5 16.5 dcSTX 5 13 13.0 16.7 dcNEO 20 25 13.1 18.8 dcGTX2 19 100 8.0 7.6 dcGTX3 38 100 8.3 7.9 GTX1 10 ― 8.1 3.6 GTX4 12 ― 8.5 3.6 NEO 9 ― 16.9 13.1 LWT 1 ― ― ― 6.2 LWT 2&3 ― ― ― 6.6 LWT 4 ― ― ― 3.0 LWT 5 ― ― ― 12.0 LWT 6 ― ― ― 11.8

46

Table 2-7. Molecular and fragmentation ions monitored LC/MS SIM Toxin [M+H]+ ion Product Ions STX 300 300 282 266 221 204 dcSTX 257 257 239 222 dcNEO 273 273 255 225 dcGTX2 353 273 273 255 dcGTX3 353 353 335 273 255 GTX1 412 332 GTX4 412 412 NEO 316 316 *LWT 1 379 379 *LWT 2 395 395 *LWT 3 395 395 *LWT 4 241 241 *LWT 5 299 299 *LWT 6 283 283 * No known standards available, data from Onodera et al. 1997 Note: All other data adopted from Dell'Aversano et al. 2005 except product ions (determined from autotune, Xcalibur™ software)

47

CHAPTER 3 RESULTS

Field Data

Water Quality

Water quality data collected during this study (August 2009-April 2010) is outlined in Table 3-1. Temperatures remained constant throughout the study period, ranging no more than 0.2˚C from the reported average values at each site. Blue Hole Spring averaged 21.37˚C while Silver Glen Springs averaged 23.27˚C. In the Silver Glen Main

Stem, pH ranged from 7.85-8.02 while in the Natural Well Vent, pH was slightly lower ranging from 7.72-7.86. Blue Hole Spring pH was 7.48-7.59. Silver Glen Springs is naturally influenced by saline ground water inputs so salinity and Specific Conductivity

(S.C.) measurements are higher than that of Blue Hole. Salinity was highest from the

Natural Well Vent (1.016 psu), just slightly lower in the Main Stem (0.938 psu) and much lower in Blue Hole (0.142 psu). Specific conductance measured at the Natural

Well vent of Silver Glen Springs was the highest at 1982 µS cm-1, slightly lower at the main stem (1858 µS cm-1) and low at Blue Hole (298 µS cm-1). The data collected at the time of study are comparable to data from Florida Department of Environmental

Protection (FDEP).

Dissolved oxygen levels are low in all the spring systems near vents. They were higher during this study at Silver Glen Springs, both Main Stem (3.5-4.06 mg L-1) and

Natural Well areas (3.1-3.5 mg L-1), than that of Blue Hole Spring (1.8-2.2 mg L-1). This represents an increase from DO levels reported by the FDEP from 2001-2006 where the average in Silver Glen Springs was 2.78 mg L-1 and 1.83 in Blue Hole Spring. The overstory densities were quite different from site to site, potentially impacting the

48

Lyngbya mat growth and health. In Silver Glen, the Main Stem collections were free from canopy coverage, representing an overstory density of 0%; trees lining the embankment did not have any influence over the large mat in the middle of the spring.

Conversely, the mat sampled at Natural Well was surrounded by an embankment with overstory density varying from 7-21%, and an average of 14%. Blue Hole represented the highest canopy coverage with a range from 69-89%, an average of 81% coverage.

Samples collected at the top of the mats in Silver Glen Springs averaged 1.1 meters in depth at the Main Stem and 0.97 meters in the Natural Well. The collections from Blue

Hole were obtained at ½ the max depth of 1 meter, so all collections for all spring samples were roughly at the same depth of 1 meter.

The Florida Department of Environmental Protection (FDEP) produced an Initiative

Report in 2010 summarizing water quality data for over 49 spring vents (Harrington et al. 2010). The data released in that report is utilized here. Water quality for the Main

Vent in Silver Glen Springs was sampled 21 times over a period from 2001-2006. Blue

Hole Spring was sampled 18 times from 2001-2002 and 2004-2006. Data collected for the FDEP study (2001-2006) are reported in Table 3-2.

The FDEP Initiative Report states that Silver Glen Springs has nitrate levels near or at the range that would be considered true background levels, with Nitrate + Nitrite levels at 0.05 mg L-1. Conversely, nitrate levels in Blue Hole are averaged to be 0.695 mg L-1, which is higher than the FDEP proposed surface water standard for nitrogen in spring vents of 0.35 mg L-1. Sulfate, calcium, potassium, and Total Dissolved Solid levels are higher in Silver Glen Springs than that of Blue Hole Spring, but turbidity,

Color, and pH values are similar. Fecal coliforms were not detected at either spring.

49

Field Observations

Silver Glen Springs Lyngbya wollei appeared to have the most observed changes during this study with regard to the mat size and color. The average thickness of the mat (from top of the mat to max depth) was 12 cm (0.12m) in August 2009 (Table 3-3).

The mat remained between 12 cm and 19 cm until January 2010 where it gained over

10 cm for average thickness to 30 cm. One last sample period in April 2010 resulted in a measured average mat thickness of 46 cm. Other observations included the color of the mat, which seemed to change from a burnt brown/orange color in the summer to a darker black color in the winter months. Tilapia created deep beds within the mats for breeding purposes in January 2010 and Atlantic Stingray were abundant during the summer months. Mullet were abundant year round and were frequently observed picking up the Lyngbya filaments and either ingesting it or just the epiphytes covering the Lyngbya. It is unknown at this time if mullet actually fully ingest Lyngbya, a study examining the stomach contents of the fish would be required to determine that.

Silver Glen Natural Well Lyngbya appeared the darkest in color of the three sites.

The mat was thickest in November of 2009, an average of 32 centimeters (cm), although it appeared to decline in growth when sampling was conducted in April 2010, with an average thickness of 16 cm (Table 3-3). The thickness of the Lyngbya mat at

Blue Hole Spring was not measured since it was vertically distributed and the mat was intertwined with the roots of a cypress tree, although Lyngbya was present along each point of the transect.

Algal Analysis

All 24 samples collected were dominated by the filamentous cyanobacterium

Lyngbya wollei Farlow ex Gomont (syn. Plectonema wollei) as determined via

50

microscopy. Lyngbya wollei has been disputed in the literature for some time now, as molecular and morphological evidence is sometimes confounding (Joyner et al. 2008).

The mats collected here were determined to be L. wollei by meeting criteria published in

Speziale and Dyck (1992). In some samples collected, other Lyngbya species were identified (Lyngbya cf. aestuarii & Lyngbya cf. major), but only made up a small fraction of the entire sample. All collections were comprised of more than one alga, but the dominant algal species in every sample was L. wollei (Figure 3-1).

Silver Glen Main Stem yielded the highest algal species list, followed by Blue Hole

Spring and then Silver Glen Natural Well. Table 3-4 corresponds to qualitative data acquired from all samples. Fine oscillatorialean filaments reported in Table 3-4 were primarily Heteroleibleinia sp. with some Leibleinia sp. (epiphytic) and Leptolyngbya sp.

(metaphytic). The primary epiphytes identified were the previously mentioned oscillatorialean filaments, the pennate diatom Cocconeis sp. and the chroococcalean cyanophyte Chamaesiphon sp. Micrographs of commonly found epiphytes are in Figure

3-2. Changes in the amount and types of epiphytes on the L. wollei did occur from month to month, but were not quantitated.

Molecular Biology

A 1366 bp near full-length16S rRNA gene fragment was recovered. A blast search with other Lyngbya sequences in the National Center for Biotechnology

Information (http://www.ncbi.nlm.nih.gov/) database revealed highest identities to two L. wollei 16S rRNA sequences. While the sequence was 95% identical to L. wollei strain

Carmichael/Alabama (Accession: EU439567) over 1353 nucleotides, it was 96% identical to another L. wollei sequence (Accession: EU603708) over 726 nucleotides, both of which are saxitoxin producers (Kellman et al. 2008, Mihali et al. 2009) . Identities

51

to other Lyngbya sequences in the database ranged from 86% to 92%. A blast search including other species showed a Blennothrix sp. strain having 98% identity to our sequence over 1107 bp (Accession: EU586735). However the identity of 16S rRNA sequence of this Blennothrix strain to another Blennothrix sequence (Accession:

EU253968) in the NCBI database was around 89%. Although the 16S rRNA sequence of this filament shows a high identity (98%) to a sequence presumably obtained from

Blennothrix, we suggest that this is probably a misidentified species. Moreover we could not find a published article on this sequence and there are no available morphological descriptions for this species from which the sequence was obtained.

The partial sxtA gene sequence obtained from the single L. wollei filament showed

97% identity to the corresponding region in L. wollei strain Carmichael/Alabama

(Accession: EU629174), while identities to other sxtA genes was lower from 90% to

92% from saxitoxin producing species such as C. raciborskii, Aphanizomenon sp.,

Aphanizomenon flos-aquae, Aphanizomenon gracile, Aphanizomenon issatschenkoi,

Anabaena circinalis, Anabaena planktonica, and Anabaenopsis elenkinii.

A 672 bp sxtG sequence was recovered from the L. wollei filament which showed

98% identity to that of L. wollei strain Carmichael/Alabama (Accession EU629180).

Identities to other sxtG sequences from other species were lower from 93% to 96%.

HPLC/Fluorescence

Peroxide Oxidation Products

Precolumn peroxide oxidation provided good sensitivity to PST standards and reproducible oxidation products for fluorescence. Sample extractions of samples from

Blue Hole Spring with varying concentrations of acid were initially characterized with this

52

method. Figure 3-3 represents a typical chromatogram acquired from pre-column peroxide oxidation of a mixed standard solution.

Extraction Assessment

Initial assessments of Lyngbya extracts were made with regard to differences in

Colored Dissolved Organic Matter (CDOM) and pH. Since pigments can interfere with analyses it was important to utilize an extraction protocol that provided a representative toxin profile but did not result in high concentrations of pigment. Extractions with 0.1 M hydrochloric acid resulted in the highest concentration of extracted pigments, a very dark purple extract, followed by 0.001 M HCl. All other extractions, 0.01 M HCl and all acetic acid extractions, resulted in clear solutions with very little pigment co-extracted.

This was also observed in fluorescence chromatograms, where all non-oxidized samples exhibited multiple peaks, but there was a noticeable increase in peak size at a retention time of 16.8 minutes in non oxidized samples extracted with 0.1 M HCl and

0.001 M HCl. Although these peaks do not elute at retention times that would interfere with known toxin standards, samples with higher concentrations of pigment presented problems when attempting to concentrate the extract (lyophilization, SPE). pH, an important factor with regard to conversions and degradation of PSTs, was measured both before and after boiling the samples in the extraction solution to verify they remained acidic (Table 3-5).

Samples extracted with varying concentrations of acids were assessed using

LC/FL and later by LC/MS. Only two major peaks were observed in LC/FL chromatograms of peroxide oxidized samples in all extractions, retention times matched those of dcGTX2&3 and dcSTX/dcNEO. Peaks corresponding to standards of STX,

C1&2, GTX 2&3 and GTX 5 were not observed. Figure 3-4 represents common

53

chromatograms achieved through peroxide oxidation LC/FL experiments conducted on

Lyngbya samples. All extractions exhibited acceptable reproducibility with respect to replicates (<10% difference) and spike responses.

Variations in acid type and concentration resulted in changes to peak size and FL detector response. Peak areas of the triplicate extractions and replicate runs were averaged and compared to standard curves of dcGTX23 and dcSTX. Figure 3-5 summarizes the differences in extractions. As acid concentration went down, so did the peaks corresponding to both dcGXT2&3 and dcSTX. This could be due to many reasons, including inefficiency of sample extraction as the acid becomes more dilute, degradation of toxins in higher pH (less likely) and changes in the profile which provide a different response once oxidized and analyzed.

Even though the resultant peaks may have originated from multiple toxins (e.g. L. wollei PST toxins), it is theorized that the oxidation products are the same. However, quantification could not be achieved utilizing this method since standards of L.wollei toxins are not available, their optimal oxidation conditions are unknown, and the resultant response via fluorescence may not be equal. In addition, some known PST toxins also produce similar, if not the same oxidation products, and cannot be distinguished with this method. For instance, separation of dcNEO and dcSTX is not possible with precolumn oxidation and peak detection at retention times corresponding to oxidation products of either toxin cannot be fully characterized or quantitated since they exhibit a different fluorescent response at the same concentrations. Other toxins known to react this way include dcGTX2 and dcGTX3, which are typically combined in

54

pre-column oxidation since the oxidation products cannot be fully resolved with this method.

It should be noted that samples were re-analyzed two weeks after being stored at -

4˚C, and although responses to samples extracted with acetic acid and more dilute hydrochloric acid remained unchanged; the 0.1 M HCl acid extraction represented a

40% increase in the dcGTX2&3-eq and a 16% increase in dcSTX-eq peaks. This could be due to the continued conversions or changes in toxin profile due to hydrolysis as observed in other PST matrices in other studies.

Pre-column peroxide oxidation does not allow for specificity with regard to many toxins, including L. wollei toxins. This is an inherent problem with pre-column oxidation of PSTs and would require further work to improve upon the approach. Further characterization of extraction products was conducted by LC/MS.

SPE and Lyophilization Assessment

Solid Phase Extraction with C18 resulted in lowered fluorescence response.

There was an average recovery of 88% (81%-92%) for the dcGTX2&3 peak and 83%

(71%-90%) for the dcSTX peak. The spiked samples of STX resulted in an average recovery of 117% when compared to standard curves.

Initial lyophilization efforts resulted in significant losses, up to 80% for some samples (LC/FL). The removal of pigments and other compounds co-extracted was necessary to increase recoveries and allow for reconstitution of samples in less volume.

Post C18 extracted material was lyophilized and final recoveries via LC/FL were averaged 67% (62%-75%) for the dcGTX2&3 peak and an increase in response was observed for the dcSTX peak, with an average of 104% (98%-107%) from the original extract. The saxitoxin spike recovery was 75% post SPE and lyophilization.

55

Sample Analysis

Samples analyzed via pre-chromatographic peroxide oxidation LC/FL included collections from all spring sites in the month of August 2009 and April 2010. Other samples were not analyzed with this method since qualitative data from LC/FL is limited and chromatographic separation of L. wollei toxins could not be achieved. As a screening method for toxins with good chromatographic separation (i.e. STX, C1&C2,

GTX2&3, GTX 5), it provided a useful tool for additional confirmation. It also provided supporting extraction efficiency data when used with pre-extraction STX spikes, as exhibited in Figure 3-6.

HPLC/MS

Extraction Assessment

Samples extracted with varying acids analyzed by LC/MS exhibited similar responses with regard to L. wollei toxins and decarbamoyl toxins, with the exception to the 0.1 M HCl extracted samples. Figure 3-7 summarizes the MS responses to toxins detected in the varying acid extractions. Data from both LC/FL and LC/MS extractions supported the relationship between acid concentrations and total toxins extracted, with exception to the 0.1 M Hydrochloric acid extractions. The stronger HCl extraction resulted in only the detection of dcGTX2, dcGTX3, and dcSTX. The L. wollei toxins were not detected. This is potentially due to the conversion of the L. wollei toxins to dcGTX2&3 and dcSTX via hydrolysis. This was further supported by an increase in dcGTX2 response (812-1122%, average 977%), dcGTX3 response (459-3473%, average=1654%) and dcSTX MS/MS response (154-2689%, average 929%) in the 0.1

M HCl extracted material versus all other acid extractions (examples are in Figure 3-8 and 3-9).

56

The use of more dilute acids resulted in the detection of all the L. wollei toxin parent ions, dcGTX2&3 and dcSTX. Due to potential conversions, hydrochloric acid was avoided for final extractions on all samples and a 0.1 M acetic acid solution was chosen instead. The acetic acid extractions resulted in acceptable recoveries of pre- extraction and post extraction spikes, exhibited good reproducibility between extractions

(<10% difference) and allowed for the detection of all L.wollei toxins, dcSTX and dcGTX2&3.

SPE and Lyophilization Assessment

The use of SPE alone in an attempt to clean the sample matrix did not provide an increase in signal to noise and did not prevent shifts in retention time. The use of SPE in conjunction with lyophilization for samples analyzed via LC/MS resulted in significant levels of noise for both scans and MS/MS. Although LC/FL data showed that there was

>62% recovery of oxidized PST products, lyophilization of the extracts post SPE to achieve a higher extract concentration resulted in complete ion suppression and toxins

(L.wollei toxins, STX, dcSTX, dcGTX2&3) were not detectable.

Sample Analysis

All samples extracted with 0.1 M Acetic acid and analyzed with LC/MS (SCANs and SIM) contained decarbamoylgonyautoxin(s), decarbamoylsaxitoxin, and LWTs 1-6.

All other toxins were not present or below the detection limits established in this study, including dcNEO, GTX1, GTX4 and NEO. Some samples did indicate a presence of dcNEO via SIM, but the levels were below the MDL and could not be verified. Analysis with SIM indicated the presence of GTX1 as well, but LC/MS/MS was not conducted at this time to confirm the presence. Further method development and analysis must be conducted to establish the presence/absence of GTX1 and dcNEO.

57

Matrix effects were apparent for all standard spikes and included retention time shifts and reduced MS response to standards. Retention times for dcGTX2&3 shifted the least (<1 min) but dcSTX, dcNEO, NEO, GTX1&4 peaks were detected up to 5.7 minutes prior to standard retention times without the matrix. LC/MS/MS was utilized to verify retention times of standard spikes.

Field replicate data represented significant variability for some toxins and samples.

Differences greater than 20% were observed for some samples and even a 61% difference for LWT 6 collected 1/27/10 from SGS Natural Well. The highest variability in all duplicate analyses was observed with LWT1 and LWT 6 data (both non toxic LWTs).

This level of variability was not observed in replicate runs of the same sample (<10% difference) or in triplicate extractions of the same sample so it may be concluded that variability was inherent in the collections themselves and the L. wollei mat was not homogeneous in nature. Field replicate data was averaged to represent data from any given site.

Decarbamoylgonyautoxin 2&3 concentrations for all sites and sampling months are reported in Table 3-6. The Main Stem of Silver Glen Springs represented the lowest concentrations of dcGTX2&3 while SGS Natural Well had the highest reported concentrations for dcGTX2 and dcGTX3. However, Blue Hole had the highest average concentrations for both toxins with the lowest variability between sample dates

(standard deviation= 7; n=6). Concentrations of dcGTX2 ranged from not detectable

(<5 µg g-1) to 43 µg g-1 in Silver Glen Main Stem and dcGTX 3 concentration ranged from 10 µg g-1 to 19 µg g-1. Silver Glen Natural Well dcGTX2 concentrations ranged from 15 to 62 µg g-1 and dcGTX3 ranged from 12 to 29 µg g-1. Blue Hole Spring

58

concentrations of dcGTX2 ranged from 31 to 54 µg g-1 and dcGTX3 ranged from 16 to

29 µg g-1. The lowest concentrations of dcGTX2&3 for all sites were in the month of

September. The sum of both dcGTX2 and dcGTX3 are shown in Figure 3-10 over the entire sampling period.

Decarbamoylsaxitoxin (dcSTX) concentrations for all sites ranged from 16-33 µg g-

1 with the highest concentration reported in the month of January collected at SGS Main

Stem. The lowest concentrations of dcSTX for both Silver Glen samples were in the month of September, while it was lowest in February for Blue Hole. Conversely, the highest concentrations for Silver Glen were in January collections. Overall the concentration of dcSTX did not vary greatly for any sampling site over the entire collection period with an average of 24 µg g-1 (stdev=4.9, %CV=20%). Values for dcSTX concentration are reported in Table 3-7. A summary of results can be seen in

Figure 3-11.

Lyngbya wollei toxins were compared by MS responses only since standards were not available for quantification. Elution order of LWT toxins was not the same as in previous studies, instead, the order of elution in this study of L. wollei toxins was: LWT

4, LWT 1, LWT 2&3, LWT 6 followed by LWT 5. Decarbamoyl toxins dcGTX3 and dcGTX2 eluted after LWT 2&3 while dcSTX eluted between dcGTX2&3 and LWT 6. No other toxins were detected above detection limits in samples. Lyngbya wollei toxins 2 and 3 are isomers and have the same molecular weight of 396 ([M+H+] = 395).

Resolution of the two toxins was not apparent in this study, so they were combined as

LWT 2&3. SCANs did exhibit a peak at the same RT of LWT 2&3 with m/z 315,

59

representing the loss of a sulfate group and providing additional confirmation of LWT

2/3 presence.

Samples spiked pre-extraction with saxitoxin (10µg) were analyzed via MS/MS and it was discovered that LWT 5 fragments in the same way as saxitoxin even though its parent m/z is 299 (saxitoxin m/z is 300). In addition, the matrix shifted saxitoxin spikes to the same retention time as LWT 5, not allowing for resolution or the ability to differentiate between the two. Even when MS parameters were set to the strictest possible settings, interference was still apparent, so STX extracted spikes were not utilized as a measure of extraction efficiency for LC/MS. However, the fragmentation pattern observed for LWT 5 provided additional confirmation of its presence and structure as well as an MS/MS program for analysis of the toxin in L. wollei samples.

Representative SCAN data showing SIM ions for L. wollei toxins from each site can be viewed in Figures 3-12 to 3-14. L. wollei toxin 4, a non-toxic form of PST, resulted in the highest responses for all sites and sampling dates. Data was averaged from each sampling month and represented in Figure 3-15 to show that the toxin profile was similar for each site. Figures 3-16 to 3-20 represent MS response for all Lyngbya wollei toxins detected at each site from month to month as determined form TIC chromatograms. The data is derived from MS responses only since standards for quantification were not available. This limits the ability to make comparisons between toxins or infer which PSTs may dominate the profile. Although it appears that LWT 4 dominated the profile in all extractions, the response was not calibrated and actual toxin levels could not be determined.

60

Table 3-1. Water quality and sample collection data for sample sites (August 2009-April 2010) (averaged data, n=6) Blue Hole Silver Glen Main Stem Silver Glen Natural Well Parameter Average S.D. Average S.D. Average S.D. Temperature ˚C 21.6 0.1 23.2 0.1 23.0 0.0 pH (s.u.) 7.5 0.0 7.9 0.1 7.8 0.1 Salinity (psu) 0.14 0.00 0.94 0.02 1.02 0.00 Specific Conductance (µS cm-1) 298 9 1858 26 1982 22 Dissolved Oxygen (mg L-1) 2.1 0.2 3.8 0.2 3.2 0.2 Average Canopy Coverage % 81% 7% 0% 0% 14% 5% Average Collection Depth (m) 1.00 0.00 1.07 0.20 1.00 0.19 Note: S.D. = Standard Deviation

61

Table 3-2. Water quality data for Blue Hole Spring and Silver Glen Springs from FDEP Initiative Report 2011, collections from Silver Glen Springs only made at one site, the main pool area Blue Hole Spring Silver Glen Springs Parameter Average Range Average Range Nitrite & Nitrate (mg L-1) 0.695 0.510-0.790 0.049 0.046-0.055 Ammonia (mg L-1) 0.005 0.005-0.005 0.005 0.005-0.005 Orthophosphate (mg L-1) 0.047 0.039-0.047 0.027 0.024-0.028 Potassium (mg L-1) 0.36 0.34-0.41 8.70 7.20-9.40 Sodium (mg L-1) 3.0 2.7-3.0 252.0 216.0-257.0 Chloride (mg L-1) 4.9 4.3-5.5 450 380-480 Specific Conductance (µS cm-1) 298 287-309 1905 1540-2050 Sulfate (mg L-1) 4.7 4.2-28.0 170 150-180 Dissolved Oxygen (mg L-1) 1.83 1.26-2.76 2.78 2.03-4.10 Calcium (mg L-1) 54.0 47.9-56.8 73.0 62.7-79.1 Total Dissolved Solids (mg L-1) 168 150-164 1020 854-1060 Turbidity (mg L-1) 0.15 0.03-1.40 0.10 0.03-0.30 TOC (mg L-1) 0.5 0.5-1.3 0.5 0.5-1.3 Color (Pt-co) 2.5 2.5-5.0 2.5 2.5-5.0 pH (su) 7.50 7.14-7.72 7.74 7.42-7.92 Fecal Coliforms (col 100mL-1) 0 0 0 0

62

Table 3-3. Mat thickness measured at Silver Glen Springs, 5 collections per sample date SITE DATE Mat Thickness (cm) Average Silver Glen 8/25/09 20 10 10 10 10 12 Springs Main 8/25/09* 20 10 10 10 10 12 Stem 9/29/09 15 20 20 20 20 19 10/23/09 10 10 15 20 20 15 11/24/09 30 20 10 10 10 16 11/24/09* 30 35 10 10 10 19 1/27/10 40 10 40 40 20 30 4/5/10 40 50 50 40 50 46 AVERAGE 21 Silver Glen 8/25/09 10 10 5 25 5 20 Springs 9/29/09 20 10 5 5 40 16 Natural Well 10/23/09 10 30 10 40 10 20 10/23/09* 10 10 10 10 10 10 11/24/09 20 40 30 40 30 32 1/27/09* 30 30 10 10 10 18 1/27/10 35 35 30 20 10 26 4/5/10 5 30 30 10 5 16 AVERAGE 20 *Denotes field duplicate collection

63

Table 3-4. Qualitative algae list from all samples DATE Silver Glen Main Stem Silver Glen Natural Well Blue Hole Spring August Lyngbya wollei Lyngbya wollei Lyngbya wollei 2009 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Spirogyra sp. 1 Phormidium sp. 2 Microchaete sp. 1 Phormidium sp. 3 Oedogonium sp. 1 Stigeoclonium sp. 1 Oedogonium sp. 2 Ulva flexuosa Cladophora glomerata Anabaena sp. 1 oscillatorialean filament sp. 1 Pseudanabaena sp. 1 September Lyngbya wollei Lyngbya wollei Lyngbya wollei 2009 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Stigeoclonium sp. 1 Spirogyra sp. 1 Microchaete sp. 1 Phormidium sp. 3 Oedogonium sp. 1 Phormidium sp. 2 Anabaena sp. 1 Batrachospermum sp. 1 Microchaete sp. 2 Mougeotia sp. 1 October Lyngbya wollei Lyngbya wollei Lyngbya wollei 2009 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Phormidium sp. 3 Anabaena sp. 1 Homoeothrix sp. fine oscillatorialean filaments spp. Spirogyra sp. 1 Phormidium sp. 2 Oedogonium sp. 2 Microchaete sp. 1 Stigeoclonium sp. 1 Oedogonium sp. 1

64

Table 3-4. Continued DATE Silver Glen Main Stem Silver Glen Natural Well Blue Hole Spring Nov./Dec. Lyngbya wollei Lyngbya wollei Lyngbya wollei 2009 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Phormidium sp. 2 *Lyngbya cf. aestuarii fine oscillatorialean filaments spp. Lyngbya cf. major Homoeothrix sp. 1 Phormidium sp. 1 Phormidium sp. 2 Anabaena sp. 1 Phormidium sp. 4 Phormidium sp. 2 Microchaete sp. 1 Rhizoclonium hieroglyphicum Stigeoclonium sp. 1 Oedogonium sp. 1 Spirogyra sp. 1 Jan./Feb. Lyngbya wollei Lyngbya wollei Lyngbya wollei 2010 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Lyngbya cf. aestuarii Phormidium sp. 2 oscillatorialean filament sp. 2 chlorophyte filament sp. 1 Microchaete sp. 1 Mougeotia sp. 1 April Lyngbya wollei Lyngbya wollei Lyngbya wollei 2010 fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. fine oscillatorialean filaments spp. Stigeoclonium sp. 1 Phormidium sp. 3 * L. aestuarii (Mertens) Liebmann sensu Prescott 1962; unrevised in Komarek and Anagnostidis

65

A B

C D

E Figure 3-1. Series of micrographs of Lyngbya wollei A) Exhibiting false branching from Blue Hole Spring with phase contrast optics 400x B) False branching from Silver Glen Springs Natural Well 400x C) Silver Glen Main Stem filament, bright field 400x D) Filament from Blue Hole E) Multiple filaments at 100x from Silver Glen Natural Well

66

A B

C D Figure 3-2. Series of micrographs of Lyngbya wollei epiphytes, all at 400x A) Leibleinia sp. on Lyngbya viewed with epifluorescence from Silver Glen Springs Main B) Epifluorescent image of empty L. wollei sheath with attached cyanobacterial epiphytes collected Silver Glen Main C) Heteroleiblenia sp. attached to Lyngbya from Silver Glen Natural Well D) Cocconeis sp. attached to Lyngbya filament from Silver Glen Natural Well E) Chamaesiphon sp. attached to filament from Silver Glen Main Stem F) Cyanophyte epiphytes from Blue Hole G) Heteroleibleinia sp. from Silver Glen Main Stem H) Cocconeis on Lyngbya from Blue Hole

67

E F

G H Figure 3-2 continued

68

Figure 3-3. Fluorescence chromatogram of standard solution mixture of dcGTX &3, dcSTX, GTX 2&3, GTX 5, and STX (dcNEO not shown as it co-elutes with dcSTX)

Table 3-5. pH of each extraction solution before and after boiling Extraction pH before pH after 0.1 M HCl <1.5 4.0 0.01 M HCl 2.5 4.5 0.001 M HCl 5.0 6.0 0.1 M Acetic Acid 3.5 3.5 0.01 M Acetic Acid 4.5 5.5 0.001 M Acetic Acid 5.0 6.0

69

Figure 3-4. Fluorescence chromatogram of oxidized extraction (0.1 M acetic acid) with peaks appearing at retention times the same as dcGTX2&3 and dcSTX/dcNEO (Silver Glen Main Stem collected 8/25/09)

70

dcSTX

dcGTX2&3 Fl Response Fl

0.1 M 0.01 M 0.001 M 0.1 M 0.01 M 0.001 M HCl HCl HCl Acetic Acetic Acetic

Figure 3-5. Extraction technique data (HPLC/FL). Data is represented by normalizing peaks to standards of dcGTX2&3 and dcSTX

71

Figure 3-6. Fluorescence chromatogram of oxidized extraction (0.1 M acetic acid) with peaks appearing at retention times the same as dcGTX2&3 and dcSTX and with 10µg STX spiked pre-extraction (Blue Hole collected 8/27/09)

72

dcSTX dcGTX2 dcGTX3 LWT 1 LWT 2&3

LWT 4 MS Response MS LWT 5 LWT 6

0.1 M 0.01 M 0.001 M 0.1 M 0.01 M 0.001 M HCl HCl HCl Acetic Acetic Acetic

Figure 3-7. Extraction technique data (HPLC/MS). Includes L. wollei toxins, dcGTX2, dcGTX3 and dcSTX relative MS responses

73

A

B Figure 3-8. HPLC/MS chromatograms of dcGTX2 & dcGTX3 SIM ions from Blue Hole Spring extracted with two different acids A) 0.1 M hydrochloric acid. B) 0.1 M acetic acid

74

Figure 3-9. HPLC/MS/MS chromatograms of dcSTX from Blue Hole Spring extracted with two different acids (0.1 M hydrochloric acid and 0.1 M acetic acid).

75

Table 3-6. Decarbamoylgonyautoxin 2&3 (dcGTX2&3) concentrations (µg g-1) for all sites and sampling months Month SGS Main Stem SGS Natural Well Blue Hole Collected dcGTX2 dcGTX3 dcGTX2 dcGTX3 dcGTX2 dcGTX3 August ND 11 40 26 30 25 September ND 9 ND 17 29 23 October ND 16 44 26 38 16 Nov/Dec ND 17 60 28 49 21 Jan/Feb 43 17 26 17 51 28 April 21 8 22 11 41 24 Note: ND = not detectable above detection limit

120

100 µg/g 80 Silver Glen Main Stem

60 Silver Glen Natural Well

40

Blue Hole SUM dcGTX2 & dcGTX 3 dcGTX & dcGTX2 SUM Spring 20

0 Aug Sept Oct Nov/Dec Jan/Feb Apr

Figure 3-10. Sum of dcGTX2 and dcGTX3 concentrations for all sites over sampling period (reported in µg g-1 dry weight of lyophilized L. wollei). Error bars represent standard error.

76

Table 3-7. Decarbamoylsaxitoxin (dcSTX) concentrations (µg g-1) for all sites and sampling months Month Collected SGS Main SGS Natural Well Blue Hole August 22 25 25 September 17 16 28 October 22 24 30 Nov/Dec 31 28 25 Jan/Feb 33 29 17 April 25 23 19

40

35

Silver Glen 30 Main Stem

Silver Glen 25 Natural

Well dcSTX (µg/g) dcSTX

20 Blue Hole Spring

15

10 August September October Nov/Dec Jan/Feb April

Figure 3-11. Concentrations of dcSTX for all sites over sampling period (µg g-1 dry weight of lyophilized L. wollei). Error bars represent standard error.

77

Figure 3-12. HPLC/MS SCAN of Silver Glen Springs Main Stem (collected 1/27/10) with SIM ions for L. wollei Toxins 1-6

78

Figure 3-13. HPLC/MS SCAN of Silver Glen Springs Natural Well (collected 11/24/10) with SIM ions for L. wollei Toxins 1-6

79

Figure 3-14. HPLC/MS SCAN of Blue Hole (collected 4/2/10) with SIM ions for L. wollei Toxins 1-6

80

Silver Glen Main Stem

Silver Glen Natural Well

Blue Hole Spring MS Response MS

LWT 1 LWT 2&3 LWT 4 LWT 5 LWT 6

Figure 3-15. Averaged MS responses over period of study for L. wollei Toxins 1-6 at each site (error bars represent sample standard deviation)

81

Figure 3-16. MS Responses to Lyngbya wollei Toxin #1 detected from all sites over the sampling period. Error bars represent sample standard deviation.

82

Figure 3-17. MS Responses to Lyngbya wollei Toxin #2&3 detected from all sites over the sampling period. Error bars represent sample standard deviation.

83

Figure 3-18. MS Responses to Lyngbya wollei Toxin #4 detected from all sites over the sampling period. Error bars represent sample standard deviation.

84

Figure 3-19. MS Responses to Lyngbya wollei Toxin #5 detected from all sites over the sampling period. Error bars represent sample standard deviation.

85

Figure 3-20. MS Responses to Lyngbya wollei Toxin #6 detected from all sites over the sampling period. Error bars represent sample standard deviation.

86

CHAPTER 4 DISCUSSION

The results of this study show that Lyngbya dominated mats in two spring fed rivers in Florida are a potential source of Paralytic Shellfish Toxins (PSTs) with toxin profiles similar to Lyngbya mat samples collected and described from Alabama

(Carmichael et al. 1997). Morphological examination indicates that the Lyngbya in the mats is L. wollei. The 16S rRNA sequence obtained from an isolated filament of

Lyngbya shows 95% identity to the L. wollei strain Alabama, which also produces PSTs, and DNA isolation from a single filament confirmed the presence of genes associated with toxin production, including sxtA and sxtG genes. The latter observation provides additional evidence that L. wollei is the source of PSTs observed in Silver Glen Springs

Lyngbya mats. Nucleotide sequences of sxtA and sxtG genes from isolates showed a high level of identity with the L. wollei Alabama strain, in contrast to lower identities for sequences from other species.

Toxin Extraction

Methods utilized to extract PSTs from mat material can influence the identification and quantification of toxins. Working with lyophilized material that has been cut or mechanically homogenized allows for more contact of inner cells with the extract solution. The choice of extraction solution should focus on avoiding interconversions of

PSTs, maintain an acidic pH, and allow for extraction of as much of the toxins as possible. The use of acetic acid at 0.1 M concentration is suggested for any future work extracting PSTs from L. wollei. The use of Solid Phase Extraction (SPE) requires more investigation as C18 SPE did not aid in reducing noise or provide improved response in fluorescence or spectrometry analyses. An SPE method to concentrate the extract and

87

provide added sensitivity would be helpful for confirmation of suspect toxins, such as dcNEO and GTX1. The use of lyophilization alone and in conjunction with C18 did not result in improved sensitivity but instead resulted in complete ion suppression via Liquid

Chromatography/Mass Spectrometry (LC/MS).

Analytical Methods

Analytical methods used to detect PSTs extracted from Lyngbya are not all equivalent and may lead to an overestimation of toxicity. The results in this study show significant differences between methods. Pre-chromatographic oxidation Liquid

Chromatography/Fluorescence (LC/FL) techniques only resulted in peaks corresponding to decarbamoyl toxins dcGTX23 and dcSTX/dcNEO. Samples of Blue

Hole Spring (collected 4/2/10) extracted with 0.1 M Acetic acid, quantitated via LC/FL, resulted in toxin concentrations of 490 µg dcGTX2&3 g(dry weight)-1 and 26 µg dcSTX/dcNEO g(dry weight)-1. This approach led to an overestimation of actual toxicity since less toxic variants of PSTs added to the total area of the oxidation peaks.

The use of LC/MS provided a more accurate detection of PSTs in samples.

LC/MS data of Blue Hole Spring (collected 4/2/10) resulted in concentrations of decarbamoyl toxins of 41 µg dcGTX2 g(dry weight)-1, 24 µg dcGTX3 g(dry weight)-1

,and 19 µg dcSTX g(dry weight)-1. Taking into account relative toxicity factors, LC/FL analysis of Lyngbya samples yielded 6.4 times higher STX-equivalents than obtained using LC/MS analyses. Decarbamoyl toxins dcGTX and dcSTX were present in every sample, but made up less than 13% of the total PST constituents detected (7 % average) when based on MS response values. This is similar to percent composition determined for L. wollei mats collected in Alabama (Table 4-1).

88

Toxins detected via LC/MS included dcGTX2&3, dcSTX and L. wollei toxins

(LWTs) 1-6. These are the same toxins reported for mats from the Guntersville

Reservoir on the Tennessee River in Alabama (Carmichael et al. 1997, Onodera et al.

1997). Average concentrations of toxins (nmol g-1-dry weight) and percent composition in this study and that of Onodera et al. (1997) are shown in Table 4-1. The levels of dcSTX and dcGTX2&3 were comparable in both studies, but comparisons with L. wollei

Toxins cannot be made with the limited quantitative data achieved in this study. The highest percent composition for PSTs reported in Alabama Lyngbya samples was LWT

2, a toxic derivative of PST. It is unknown which LWT was dominant in this study without the use of standards, but mass spectrometer response was greatest with LWT

4, a non-toxic variant.

Toxin Risk Assessment

The important question with regard to PSTs is toxicity and potential health threats posed by L. wollei mats. Standards of L. wollei toxins 1-6 are needed to facilitate routine analyses of toxins to eliminate the need to conduct mouse bioassay (MBA) or some other toxicity screening. With regard to current measures of toxicity, only approximations can be made from data acquired utilizing assumptions about relative toxicity from previous studies. Table 4-2 shows toxicity data for PSTs detected in L. wollei mats as well as STX. The data reported by Oshima et al. (1995) and Onodera et al. (1997) is reported in Mouse Units (MU) µmol-1 of toxin using the AOAC Method for

Paralytic Shellfish Poisons, i.e. 20 g ddY strain male mice. One Mouse Unit (MU) is the dose of toxin required to kill a male mouse in 15 minutes via intraperitoneal injection

(Onodera et al 1997). Saxitoxin exhibits the highest toxicity at 2483 MU µmol-1 of toxin, based on research conducted by Oshima et al. (1995). Although heavily cited, this

89

reference is not in agreement with other research, which applies different toxicities to

PSTs. However, Oshima’s toxicity data is reported here in conjunction with Onodera’s

LWT 1-6 toxicity data, since both were derived using similar techniques. Lethal doses of toxic samples of L. wollei collected in Alabama ranged from 150 mg of lyohphilized L. wollei to nontoxic quantities of 1500 mg lyophilized L. wollei per kg mouse, with reported saxitoxin equivalents from 0 to 58 µg STX-eq g(dry weight)-1. In this study, total STX- eqs cannot be determined with the limited quantitative data achieved with regard to L. wollei toxins 1-6, but STX-eqs based on dcSTX & dcGTX2&3 concentrations represent minimum toxicity ranges from 34 to 61- µg STX-eq/g(dry weight)-1. This indicates L. wollei mats collected from Florida springs are comparable in toxicity, or more toxic, than

L. wollei mats from Alabama.

Care must be taken when analyzing PSTs originating from L. wollei dominated mats in the future. Traditional methods of detection, such as ELISA and pre-column oxidation HPLC/Fluorescence may over- or under- estimate toxicity values. Future studies assessing PST production in L. wollei should incorporate the use of postcolumn oxidation LC/FL and LC/MS, to determine what toxins are actually present. The use of standards is necessary for the final quantitation of L. wollei toxins and to relate levels to potential toxicity. The utilization of other techniques, such as tissue culture bioassays, may also provide additional information on actual toxin content of L. wollei in Florida without the need to conduct MBA.

Lyngbya wollei is prevalent in many of the Florida springs and is not limited to the springs sampled in this study. PSTs have been detected in Lyngbya wollei dominated mats collected from multiple Florida springs in conjunction with work conducted by the

90

FDEP (Florida Department of Environmental Protection) and the FDOH (Florida

Department of Health), including Wakulla Springs, Fanning Springs, Homosassa

Springs, Alexander Springs, Juniper Springs and multiple springs in the Ichetucknee

Spring System. PSTs are known to bioaccumulate in bivalves through filter feeding or exposure to extracellular toxins, but the thick sheath associated with Lyngbya filaments inhibit filter feeding and reduces the leaching of toxins into the surrounding water.

When Lyngbya senesces or when toxins are released from the Lyngbya sheath the

PSTs are thought to degrade to non-toxic derivatives or dilute to a non-toxic concentration (Carmichael et al. 1997). PSTs do not, however, degrade quickly in a buffered freshwater system, such as the freshwater springs of Florida, and may remain intact until photolytic or bacterial degradation acts on the toxins (Jones and Negri,

1997). On the other hand, the high flow of the spring systems does allow for rapid dilution of any toxins released.

There is considerable uncertainty regarding the actual risk of human exposure to the PSTs produced by L. wollei. Ingestion of Lyngbya directly is unlikely and the bioaccumulation or biomagnification of the toxins to herbivores or predators is unknown.

As levels of accessible drinking water in Florida and other regions decline, sources of surface water with L. wollei presence may be tapped to support the growing need. This may increase the risk of human exposure to PSTs and provide the incentive to procure workable standards for routine analysis of the PSTs associated with L. wollei.

Management decisions may be made to remove or treat L. wollei mats as awareness grows, and care must be taken to reduce occupational hazards related to handling, treating or removing the mats. Recreational exposure is of concern as well

91

since direct contact with L. wollei dominated mats in spring runs is likely to occur. PSTs may be limited in their ability to cross the dermal layer, but individuals with lesions or cuts in the skin may be of greater risk. To date, there are no known cases directly linking PSTs from L. wollei dominated mats to symptomatic responses. It is unknown if this is due to actual risks or if this is because epidemiological data is limited. Currently, data collected by the FDOH in areas dominated by L. wollei is focused on dermatoxic and gastrointestinal responses, with a lack of focus on neurotoxic responses, which can be mild. Questionnaires and reports would have to be extended to include neurotoxic responses associated with PST exposure, such as numbness, tingling sensations, headaches, nausea, blurred vision, and muscle incoordination. Epidemiological data of this nature might help to provide a better understanding of risks related to PSTs derived from L. wollei in the future.

92

Table 4-1. Toxin content of L.wollei collected from Alabama and Florida Alabama Blue Hole Silver Glen Lyngbya Percent Spring Springs Toxin nmol/g Composition nmol/g nmol/g dcSTX 40 1.4% 89 100 dcGTX2 110 3.9% 113 98 dcGTX3 41 1.5% 64 48 aLWT 1 690 24.6% ─ ─ LWT 2 390 46.7% ─ ─ LWT 3 920 0.7% ─ ─ aLWT 4 20 0.7% ─ ─ LWT 5 500 17.8% ─ ─ aLWT 6 97 3.5% ─ ─ ─ Could not quantify aNon-toxic derivatives

Table 4-2. Toxicity of PST variants reported in MU/µmol and relative toxicity to STX (STX-eq) aMouse Units Toxin (MU)/µmol Relative Toxicity aSTX 2483 1.00 adcSTX 1274 0.51 adcGTX2 1617 0.65 adcGTX3 1872 0.75 bLWT 1 <10 ― bLWT 2 178 0.07 bLWT 3 52 0.02 bLWT 4 <10 ― bLWT 5 326 0.13 bLWT 6 <10 ― aOshima et al. 1995 data bOnodera et al. 1997 data

93

LIST OF REFERENCES

Berry, J.P., Gantar, M., Gawley, R.E., Wang, M., and Rein, K.S., 2004. Pharmacology and toxicology of pahayokolide A, a bioactive metabolite from a freshwater species of Lyngbya isolated from the Florida Everglades. Comparative Biochemistry & Physiology Part C: Toxicology & Pharmacology 139, 231-238.

Botana L., 2008., Seafood and freshwater toxins. CRC Press, Taylor and Francis group,Boca Raton Fl. 2, 163-197.

Camacho, F.A. and Thacker, R.W., 2006. Amphipod herbivory on the freshwater cyanobacterium Lyngbya wollei: Chemical stimulants and morphological defenses. Limnology & Oceanography 51, 29.

Campas, M., Prieto-Simon, B., and Marty, J.L., 2007. Biosensors to detect marine toxins: Assessing seafood safety. Talanta 72, 884-895.

Carmichael, W.W., Evans, W.R., Yin, Q.Q., Bell, P., and Moczydlowski, E., 1997. Evidence for Paralytic Shellfish Poisons in the Freshwater Cyanobacterium Lyngbya wollei (Farlow ex Gomont) comb. nov. Appl.Environ.Microbiol. 63, 3104.

Cowell, B.C., Botts, P.S., 1994. Factors influencing the distribution, abundance and growth of Lyngbya wollei in central Florida. Aquat. Bot. 49, 1–17.

Dell'Aversano, C., Hess, P., and Quilliam, M.A., 2005. Hydrophilic interaction liquid chromatography-mass spectrometry for the analysis of paralytic shellfish poisoning (PSP) toxins. Journal of Chromatography A 1081, 190-201.

Fitzgerald D.J., Cunliffe D.A., Burch M.D., 1999. Development of Health Alerts for Cyanobacteria and Related toxins in Drinking water in South Australia. Environ Toxicol. 14. 203-09.

Harrington D., Maddox G., Hicks R.,2010. Florida springs initiative monitoring network report and recognized sources of nitrate. Florida Department of Environmental Protection; Division of Environmental Assessment and Restoration Bureau of Watershed Restoration.

Heffernan, J.B., Liebowitz, D.M., Frazer, T.K., Evans, J.M., and Cohen, M.J., 2010. Algal blooms and the nitrogen-enrichment hypothesis in Florida springs: Evidence, alternatives, and adaptive management. Ecological applications : a publication of the Ecological Society of America 20, 816.

Indrasena, W.M. and Gill, T.A., 2000. Food Chemistry and Toxicology - Thermal Degradation of Partially Purified Paralytic Shellfish Poison Toxins at Different Times, Temperatures, and pH. Journal of food science : an official publication of the Institute of Food Technologists 65, 948.

94

Jones, G.J. and Negri, A.P., 1997, Persistence and degradation of cyanobacterial paralytic shellfish poisons (PSPs) in freshwaters. Water Research 31, 525-533.

Joyner, J.J., Litaker, R.W., and Paerl, H.W., 2008. Morphological and Genetic Evidence that the Cyanobacterium Lyngbya wollei (Farlow ex Gomont) Speziale and Dyck Encompasses at Least Two Species. Applied & Environmental Microbiology 74, 3710- 3717.

Kellmann, R., Michali, T., and Neilan, B., 2008. Identification of a Saxitoxin Biosynthesis Gene with a History of Frequent Horizontal Gene Transfers. Journal of Molecular Evolution 67, 526-538.

Lawrence, J.F., Niedzwiadek, B., and Menard, C., 2005. Quantitative Determination of Paralytic Shellfish Poisoning Toxins in Shellfish Using Prechromatographic Oxidation and Liquid Chromatography with Fluorescence Detection: Collaborative Study. Journal of AOAC International 88, 1714-1732.

Mihali, K. T., Carmichael W. W., Neilan B., 2011. A putative gene cluster from a Lyngbya wollei bloomthat encodes paralytic shellfish toxin biosynthesis. PLoS ONE 6(2): e14657. doi:10.1371/journal.pone.0014657,

Mihali, K. T., R. Kellmann, and B. Neilan. 2009. Characterization of the paralytic shellfish toxin biosynthesis gene clusters in Anabaena circinalis AWQC131C and Aphanizomenon sp. NH-5. BMC Biochemistry 10:8.

Miller, C., 2006. Ecological and Toxicological Assessment of Lyngbya in Florida Springs. Final Report to Florida Department of Health Aquatic Toxins Program.

Onodera H, Satake M, Oshima Y, Yasimoto T, Carmichael WW, 1997. New saxitoxin analogues from the freshwater filamentous. cyanobacterium Lyngbya wollei. Nat. Toxins. 5(4). 146-51.

Oshima Y, 1995, Post-column derivatization liquid chromatographic method for paralytic shellfish toxins. Journal of AOAC International. 78. 528-532.

Ravn H, Anthoni U, Christophersen C, Nielsen P, Oshima Y. 1995. Standardized extraction method for paralytic shellfish toxins in phytoplankton. Journal of Applied Phycology. 7. 589-94.

Scientific Opinion of the Panel on Contaminants in the Food Chain on a request from the European. 2009. Commission on Marine Biotoxins in Shellfish – Saxitoxin Group. The EFSA Journal. 1019. 1-76.

Seifert, M., McGregor, G., Eaglesham, G., Wickramasinghe, W., and Shaw, G., 2007. First evidence for the production of cylindrospermopsin and deoxy-cylindrospermopsin by the freshwater benthic cyanobacterium, Lyngbya wollei (Farlow ex Gomont) Speziale and Dyck. Harmful Algae 6, 73-80.

95

Sivonen K. and Jones G., 1999. Cyanobacterial toxins. Toxic Cyanobacteria in Water: A Guide to Public Health, Significance, and Monitoring and Management. I. Chorus and J. Bertram (Edn). The World Health Organization. ISBN 0-419-23930-8. E & FN Spon. London, UK, 41-111.

Speziale, B.J. and Dyck, L.A., 1992. Lyngbya Infestations: Comparitive Taxonomy of Lyngbya wollei Comb. Nov. (cyanobacteria). Journal of Phycology 28, 693-706.

Stevenson, R.J., Pinowska A., Albertin A., Sickman J. 2007. Ecological condition of algae and nutrients in Florida Springs : The synthesis report. Environmental Assessment Section, Florida Department of Environmental Protection.

Stucken K., John U., Cembella A., Murillo A.A., Soto-Liebe K., 2010 The smallest known genomes of multicellular and toxic cyanobacteria: comparison, minimal gene sets for linked traits and the evolutionary implications. PLoS ONE 5(2): e9235. doi:10.1371/journal.pone. 0009235.

Tamura, K., Dudley, J., Nei, M., and Kumar, S., 2007. MEGA4: Molecular Evolutionary Genetics Analysis (MEGA) Software Version 4.0. Mol.Biol.Evol. 24, 1596-1599.

Teneva, I., Asparuhova, D., Dzhambazov, B., Mladenov, R., and Schirmer, K., 2003. The freshwater cyanobacterium Lyngbya aerugineo-coerulea produces compounds toxic to mice and to mammalian and fish cells. Environ.Toxicol. 18, 9.

Thacker, R.W., McLeod, A.M., and McLeod, S.W., 2005. Herbivore-induced saxitoxin production in the freshwater cyanobacterium Lyngbya wollei. Algological studies 159, 415-425.

Tillett, D. and Neilan, B.A., 2000. Xanthogenate nucleic acid isolation from cultured and environmental cyanobacteria. Journal of Phycology 36, 251-258.

Vale, C., Alfonso, A., Vieytes, M.R., Romaris, X.M., Arevalo, F., Botana, A.M., and Botana, L.M., 2008. In Vitro and in Vivo Evaluation of Paralytic Shellfish Poisoning Toxin Potency and the Influence of the pH of Extraction. Analytical Chemistry 80, 1770- 1776.

Wiese M., D'Agostino P.M., Mihali T.K., Moffitt M.C., Neilan B.A., 2010. Neurotoxic Alkaloids: Saxitoxin and Its Analogs. Marine Drugs. 8(7). 2185-2211.

Yin, Q., Carmichael, W.W., and Evans, W.R., 1997. Factors influencing growth and toxin production by cultures of the freshwater cyanobacterium Lyngbya wollei Farlow ex Gomont. J.Appl.Phycol. 9, 55.

Yilmaz, M., Phlips, E.J., Szabo, N.J., and Badylak, S., 2008. A comparative study of Florida strains of Cylindrospermopsis and Aphanizomenon for cylindrospermopsin production. Toxicon 51, 130-139.

96

BIOGRAPHICAL SKETCH

Amanda Foss was born and raised in Grand Rapids MI. She earned a B.S. in

Environmental Chemistry and Biology from Lake Superior State University in 2006. She was recruited to work for the private company GreenWater Laboratories in Palatka,

Florida analyzing toxins produced by cyanobacteria. While working at GreenWater, she received her M.S. in Fisheries and Aquatic Sciences from the University of Florida in the spring of 2011. Upon completion of the program, Amanda continued to work for

GreenWater Laboratories isolating and detecting naturally derived toxins.

97