Investigations on the Impact of Toxic Cyanobacteria on Fish : As

Home , Antillatoxin, Biliprotein, Immunostaining, Microviridin

INVESTIGATIONS ON THE IMPACT OF TOXIC CYANOBACTERIA ON FISH

- AS EXEMPLIFIED BY THE COREGONIDS IN LAKE AMMERSEE -

DISSERTATION

Zur Erlangung des akademischen Grades des Doktors der Naturwissenschaften an der Universität Konstanz Fachbereich Biologie

Vorgelegt von BERNHARD ERNST

Tag der mündlichen Prüfung: 05. Nov. 2008

Referent: Prof. Dr. Daniel Dietrich Referent: Prof. Dr. Karl-Otto Rothhaupt Referent: Prof. Dr. Alexander Bürkle

»Erst seit gestern und nur für einen Tag auf diesem Planeten weilend, können wir nur hoffen, einen Blick auf das Wissen zu erhaschen, das wir vermutlich nie erlangen werden«

Horace-Bénédict de Saussure (1740-1799) Pionier der modernen Alpenforschung & Wegbereiter des Alpinismus

ZUSAMMENFASSUNG

Giftige Cyanobakterien beeinträchtigen Organismen verschiedenster Entwicklungsstufen und trophischer Ebenen. Besonders bedroht sind aquatische Organismen, weil sie von Cyanobakterien sehr vielfältig beeinflussbar sind und ihnen zudem oft nur sehr begrenzt ausweichen können. Zu den toxinreichsten Cyanobakterien gehören Arten der Gattung Planktothrix. Hierzu zählt auch die Burgunderblutalge Planktothrix rubescens, eine Cyanobakterienart die über die letzten Jahrzehnte im Besonderen in den Seen der Voralpenregionen zunehmend an Bedeutung gewonnen hat. An einigen dieser Voralpenseen treten seit dem Erstarken von P. rubescens existenzielle, fischereiwirtschaftliche Probleme auf, die wesentlich auf markante Wachstumseinbrüche bei den Coregonenbeständen (Coregonus sp.; i.e. Renken, Felchen, etc.) zurückzuführen sind. So auch am Ammersee, wo die beschriebenen Wachstumseinbrüche vermeintlich sogar regelmäßig zum vorzeitigen Verenden bestimmter Coregonenjahrgänge führen. Interessanterweise hatten die Coregonen im Ammersee wiederholt einen außergewöhnlich blau gefärbten Darminhalt. Diese auffällige Färbung wird vermutlich von cyanobakteriellen Farbpigmenten verursacht und deutet darauf hin, dass die Coregonen im Ammersee mit Cyanobakterien in Kontakt kommen. Es scheint daher grundsätzlich nicht abwegig, dass die Schwierigkeiten der Ammersee-Coregonen in kausalem Zusammenhang zum Auftreten giftiger P. rubescens Filamente stehen könnten. Ziel des Dissertationsprojektes war es daher, • das Aufkommen, die Verteilung und die Toxizität von P. rubescens im Ammersee über einen aussagekräftigen Zeitraum detailliert zu erfassen, • in an die natürlichen Verhältnisse angelehnten Laborexperimente zu untersuchen, ob die vorgefundenen P. rubescens-Dichten Coregonenpopulationen beeinträchtigen und gesundheitlich schädigen können und schließlich, • zu prüfen, ob es Hinweise auf P. rubescens Expositionen und dadurch verursachte Schädigungen wildlebender Coregonen im Ammersee gibt.

Zur Bestimmung der Planktothrix-Dichte im Ammersee war es zunächst notwendig, ein Verfahren zu etablieren, bei dem die in Wasserproben enthaltenen Planktothrix-Filamente auf Filtern mittels Fluoreszenzmikroskopie und digitaler Bildverarbeitung einfach, schnell und akkurat quantifiziert werden können. Dieses Verfahren ermöglichte eine aufwendige Beprobung und damit eine aussagekräftige Beschreibung des zeitlichen und räumlichen Verteilungsmusters von P. rubescens im See. Bei den von April 1999 bis September 2004 durchgeführten Beprobungen zeigte sich dann, dass P. rubescens-Filamente im Ammersee – wenn auch in unterschiedlicher Dichte – durchgehend vorhanden waren. Der Bereich maximaler Zelldichten korrelierte jeweils von Mai bis Oktober

4 ZUSAMMENFASSUNG ______signifikant mit der unteren Grenze der euphotischen Tiefe und dem Beginn des Metalimnion. P. rubescens erreichte regelmäßig im Sommer maximale Dichten (zeitweilig bis zu 75.000 Zellen/ml). Darüber hinaus konnte P. rubescens auch während der winterlichen Vollzirkulation über den Wasserkörper verteilt in Zelldichten von bis zu 15.000 Zellen/ml nachgewiesen werden. In welchem Ausmaß sich P. rubescens entwickeln kann, scheint wesentlich von der Illumination des Metalimnion und damit von der Transparenz des Wassers abzuhängen. Zudem scheint P. rubescens auch von regelmäßigen Phosphat-Auszehrung und den hohen Stickstoffkonzentrationen in dem re-oligotrophierten See zu profitieren. In 27 bzw. 38 von 54 Planktonproben aus verschiedenen Monaten konnten mittels HPLC und ELISA Toxinanalyse diverse fischgiftige Microcystine (in erster Linie Microcystin-RR Varianten) nachgewiesen werden. Nahezu konstante Microcystin/Phycoerythrin Verhältnisse verdeutlichten, dass die von P. rubescens produzierten Microcystinmengen weitgehend unveränderlich sind. Dies bedeutet, dass bei einem Aufkommen von P. rubescens auch von einem Auftreten messbarer Microcystin-Belastungen auszugehen ist.

Die Auswirkungen von P. rubescens auf Coregonen wurden in Laborexperimenten untersucht, wobei die im Ammersee erfassten P. rubescens Dichten und verschiedene Expositionsformen berücksichtigt wurden. Vorversuche verdeutlichten, dass aufgrund einer kovalenten Bindung des Microcystins im Gewebe eine aussagekräftige Quantifizierung geringer Microcystinmengen in Fischgeweben generell schwierig und via ELISA-, HPLC- und PPAssay-Toxinanalytik unmöglich ist. Microcystine konnten im Fischgewebe aber qualitativ, mittels immunhistochemischer Anfärbung mit anti-Microcystin Antikörpern lokalisiert werden. Die exponierten Renken reagierten mit auffälligem Verhalten, einer gesteigerten Atemfrequenz und wiesen erhöhte Serumglukose-Konzentrationen auf. Merkmale, die in ihrer Gesamtheit als eindeutige Stressindikatoren zu bewerten sind. Pathologische Auffälligkeiten in der Leber, Niere und im Gastrointestinaltrakt verdeutlichen zudem beträchtliche Organschäden die auf nachhaltige Auswirkungen auf Organfunktionen schließen lassen. Die Tatsache, dass die geschädigten Gewebebereiche gemäß immunhistologischer Anfärbung vielfach auch Microcystin enthielten, veranschaulicht einen kausalen Zusammenhang von Gewebeschäden und der offensichtlichen Aufnahme von Microcystin. Eine erhöhte Empfindlichkeit gegenüber Ektoparasiten und eine erhöhte Mortalitätsrate deuten weiter daraufhin, dass durch die Wirkung von P. rubescens letztendlich auch die Kondition der experimentell exponierten Coregonen beeinträchtigt wurde. Insgesamt betrachtet war die Symptomatik der Auswirkungen in den verschiedenen Expositionsansätzen vergleichbar – die Intensität der Wirkung war hingegen dosisabhängig. Dies verdeutlicht, dass sich mit zunehmender P. rubescens-Zelldichte entsprechende Effekte früher und deutlicher ausprägen. Nichtsdestoweniger zeigte sich auch bei vergleichsweise geringer P. rubescens Dichten (≈1500 Zellen/ml) eine fischgiftige Wirkung.

5 ZUSAMMENFASSUNG ______

P. rubescens Zelldichten von mindestens 1500 Zellen/ml waren im Ammersee in etwa zur Hälfte der 261 Wochen andauernden Beobachtungsperiode festzustellen, was verdeutlicht, dass die Coregonen im Ammersee tatsächlich regelmäßig mit schädlichen P. rubescens Zelldichten konfrontiert sind. Dies scheint vor allem dann problematisch, wenn die Coregonenpopulation einer P. rubescens-Exposition nicht aktiv ausweichen kann (z.B. bei P. rubescens Entwicklungen die den gesamten Wasserkörper umfassen). In der Tat, auch im See selbst ergaben sich Anhaltspunkte für eine P. rubescens Exposition der Coregonen. So konnte gezeigt werden, dass die Fische regelmäßig P. rubescens Filamente inkorporieren und verdauen. Dadurch werden im Darm der Fische die in den Filamenten enthaltenen Metabolite freigesetzt. Die Freisetzung von Phycocyanin bewirkt die auffällige Blaufärbung des Darminhalts und die Freisetzung der Microcystine verursacht eine Microcystin- Exposition der Coregonen. Da Microcystin stichprobenartig zudem auch in Leberhomogenaten von Ammersee Coregonen nachzuweisen war, ist wahrscheinlich, dass das im Darm freigesetzte Microcystin über das Darmepithel in den Organismus gelangen und sich entsprechend nachhaltig auf den Gesundheitszustand und die physiologische Kondition der Coregonen auswirken kann.

Man kann daher davon auszugehen, dass die in den Laborexperimenten aufgezeigten Microcystin-Schäden auch in den Fischen im See auftreten und sich dauerhafte P. rubescens Vorkommen entsprechend substanziell auf die Coregonen auswirken. Hinzu kommt, dass neben den experimentell aufgezeigten unmittelbaren Schädigungen weiter auch indirekte Einflüsse (z.B. P. rubescens bedingte Veränderungen in der Umwelt der Coregonen) eine entscheidende Rolle spielen können. Insgesamt betrachtet unterstützen die bisherigen Erkenntnisse damit die Vermutung, dass das anhaltende Aufkommen von P. rubescens eine wesentliche Ursache für den Wachstumseinbruch und die schlechte physiologische Kondition der Coregonen in Voralpenseen wie dem Ammersee ist.

SUMMARY

Toxic cyanobacteria affect organisms of almost all stages of development and trophic levels. Especially threatened are aquatic organisms such as fish, as they can be affected by toxic cyanobacteria via multiple routes, and their options for exposure avoidance in waters containing toxic cyanobacteria are limited. Among the most toxic cyanobacteria are species of the genus Planktothrix, including Planktothrix rubescens. During the last decades, P. rubescens has become one of the predominant species of the phytoplankton community in several lakes in the pre-alpine regions. In some of those lakes (e.g. Lake Ammersee) the rise of P. rubescens has been observed to coincide with pronounced slumps in fishery yields bringing the professional fishery into existential difficulties. These slumps have primarily been characterised by prominent growth reduction of coregonids (Coregonus sp.), resulting in reduced fish fitness which appears to be associated with the regular disappearance of specific age groups of coregonid. As Lake Ammersee coregonids have repeatedly displayed blue coloured gut contents indicating coregonid contact with cyanobacteria, it appeared plausible that the challenge to this coregonid population might be causally related to the occurrence of toxic P. rubescens. The aim of the study was therefore • to characterise the density, distribution and toxicity of P. rubescens in Lake Ammersee, • to investigate environmental observations in controlled experimental exposure studies with respect to possible detrimental effects on coregonids and finally, • to assess the evidence linking P. rubescens exposure and adverse effects on feral coregonids in Lake Ammersee

A prerequisite for the proposed P. rubescens observations was the need for a rapid and precise method for the quantification of P. rubescens cell densities. Thus, initial work focused on the validation of an image processing system, which automatically measures fluorescing P. rubescens filaments on illuminated filters. Subsequent field studies using this system demonstrated that P. rubescens was present during the entire observation period from 1999-2004, albeit at varying cell densities. Filaments were regularly distributed over the entire water column during winter and stratified in distinct metalimnic layers during summer, reaching cell densities of up to 15,000 cells/ml and 75,000 cells/ml, respectively. P. rubescens mass occurrence was demonstrated to be strongly influenced by water transparency, i.e. illumination in the metalimnion. Microcystins (predominantly MC-RR variants) were detectable in 27 and 38 of 54 monthly seston samples via HPLC and ELISA measurements, respectively. These analyses further suggest that microcystin production by P. rubescens is consistent and consequently, that the appearance of P. rubescens coincides with measurable microcystin levels. The impact of P. rubescens on coregonids was examined in experimental exposure studies, where the environmentally observed P. rubescens cell densities and various forms of application were considered. Preliminary tests however revealed, that due to covalent binding of microcystins

7 SUMMARY ______within tissues the quantification of low microcystin concentrations in fish tissue is difficult if not impossible via ELISA-, HPLC- and PPAssay analyses. Consequently, the presence and localisation of microcystin was determined immunohistochemically, using anti-microcystin antibodies. Coregonids exposed to P. rubescens showed abnormal behaviour, increased ventilation rates and elevated plasma glucose levels, presumably representing a behavioural and physiological stress response. Histopathological alterations in liver, gastrointestinal tract and kidney suggested substantial tissue damage and therefore sustained alteration in normal organ function. The fact that these alterations were also immunopositive for microcystin further indicated an uptake of microcystins and causality of tissue damage and the presence of microcystin. In addition, susceptibility to ectoparasitic infestation and increased mortality in exposed fish suggested these P. rubescens associated effects to impair fish fitness. The pathology and stress response of exposed coregonids was comparable across the different exposure experiments. Although, even low cell densities (≈1500 cells/ml) resulted in significant injury, the progression and severity of the observed adverse effects occurred in a dose-dependent manner, indicating that the higher the P. rubescens cell densities and hence microcystin concentrations, the more pronounced and earlier the onset of the adverse effects.

P. rubescens cell densities greater than 1500 cells/ml were demonstrated to occur in Lake Ammersee during 47% of the 261 weeks observed, thus suggesting that Lake Ammersee coregonids are indeed regularly confronted with detrimental P. rubescens exposure situations. This is corroborated by field observations demonstrating P. rubescens filaments in gut contents of Lake Ammersee coregonids. This additionally gives evidence that feral coregonids actually ingest P. rubescens. These field investigations further demonstrated this exposure to result in an accumulation of P. rubescens components within the coregonid intestine, as the investigated fish showed a significant accumulation of cyanobacterial biliproteins explaining the prominent blue colouration of gut contents and implying possible coregonid exposure to P. rubescens toxins. Indeed, from coregonids sampled during P. rubescens bloom episodes in 1998 and 1999, four out of ten fish contained significant microcystin accumulation in the gut content, unambiguously demonstrating microcystin exposure of feral coregonids in Lake Ammersee. The detection of covalently-bound microcystin in liver tissue of Lake Ammersee coregonids furthermore demonstrates microcystins to traverse the ileal membrane and to accumulate in the liver. As corroborated by the experimental exposure studies, this makes substantial detrimental effects on the coregonids appear inevitable.

In conclusion, prolonged occurrence of toxic P. rubescens can thus be expected to substantially affect feral coregonids. In addition to the direct detrimental effects outlined above also indirect effects, such as P. rubescens-induced environmental changes are likely. The current investigation hence substantiates the initial hypothesis of a causal relationship between mass occurrences of P. rubescens and challenged coregonid populations in pre-alpine lakes such as Lake Ammersee.

PUBLICATIONS & PRESENTATIONS

PEER REVIEWED ARTICLES:

First author: • Ernst, B., Hoeger, S.J, O´Brien, E. & Dietrich, D.R.: Abundance and toxicity of Planktothrix rubescens in the pre-alpine Lake Ammersee, Germany. Submitted for publication in Harmful Algae. • Ernst, B., Hoeger, S.J, O´Brien, E. & Dietrich, D.R. (2007): Physiological stress and pathology in European whitefish (Coregonus lavaretus) induced by subchronic exposure to environmentally relevant densities of Planktothrix rubescens. Aquatic Toxicology 82, 15-26. • Ernst, B., Hoeger, S.J, O´Brien, E. & Dietrich, D.R. (2006): Oral toxicity of the microcystin- containing cyanobacterium Planktothrix rubescens in European whitefish (Coregonus lavaretus). Aquatic Toxicology 79, 31-40. • Ernst, B., Neser, S., O´Brien, E., Hoeger, S.J. & Dietrich, D.R. (2006): Determination of filamentous cyanobacteria Planktothrix rubescens in environmental water samples using an image processing system. Harmful Algae 5, 181-189. • Ernst, B., Dietz, L., Hoeger, S.J. & Dietrich, D.R. (2005): Recovery of MC-LR in fish liver tissue. Environmental Toxicology 20, 449-458. • Ernst, B., Hitzfeld, B.C. & Dietrich, D.R. (2001): Presence of Planktothrix sp. and cyanobacterial toxins in Lake Ammersee, Germany and their impact on whitefish (Coregonus lavaretus L.) Environmental Toxicology 16, 483-388.

Co-author: • Hoeger, S.J., Schmid, D., Blom, J., Ernst, B. & Dietrich, D.R. (2007): Specifics of microcystin- RR variants: consequences for analytical procedures and risk assessment. Environmental Science and Technology 41, 2609-2616.

MAGAZINE ARTICLES: • Ernst, B. & Dietrich, D.R. (2001): Cyanobakterien auf dem Vormarsch. Geoskop, GEO, 12, 217-219.

9 PUBLICATIONS & PRESENTATIONS ______

ORAL PRESENTATIONS: • Ernst, B. & Dietrich, D.R.: Blaugrünes Alsterwasser, Roter Chiemsee und fragwürdige Nahrungsergänzungsmittel: zu den fatalen Gefahren von Cyanobakterien. 3. interdisziplinäres Forum der Arthur und Aenne Feindt Stiftung, 19. April 2006, Hamburg, Germany. • Ernst, B. & Dietrich, D.R.: Toxic Planktothrix rubescens and its subchronic impact on European whitefish (Coregonus lavaretus). 6th International Conference on Toxic Cyanobacteria, 21. - 26. Jun. 2004, Bergen, Norway. • Ernst, B.: Toxins of Planktothrix rubescens and their subchronic impact on European whitefish (Coregonus lavaretus L.). Second Late Summer Workshop der GDCh, 29. Sep. - 01. Oct. 2003, Maurach, Germany. • Ernst, B.: Auswirkungen toxischer Cyanobakterien auf Fische am Beispiel der Renken (Coregonus lavaretus L.) im Ammersee. 5th EESL Statuskolloquium 19. - 20. Nov. 2001, Konstanz.

ABSTRACT & POSTER PRESENTATIONS:

First author: • B. Ernst & Dietrich, D.R.: Regular exposure of coregonids (Coregonus sp.) to Planktothrix rubescens may be causal for reduced fish weight and fitness and hence recurrent slumps in fishery yields in pre-alpine lakes. 7th International Conference on Toxic Cyanobacteria, 05.-10. Aug. 2007, Rio das Pedras, Brazil. • Ernst, B., Neser, S., Hitzfeld, B.C. & Dietrich, D.R.: Determination of filamentous cyanobacteria in water samples using the image processing system Visiometrics IPS. 10th International Conference on Harmful Algal Blooms, 21. - 25. Oct. 2002, St. Petersburg, Florida, USA. • Ernst, B., Hitzfeld, B.C. & Dietrich, D.R.: Microcystin contamination of fish from a European pre-alpine lake via chronic exposition to Planktothrix sp.: A human health hazard? 10th International Conference on Harmful Algal Blooms, 21. - 25. Oct. 2002, St. Petersburg, Florida, USA. • Ernst, B., Neser, S. & Dietrich, D.R.: Bestimmung der Dichte fädiger Cyanobakterien in Wasserproben mit dem digitalen Bildverarbeitungssystem Visiometrics IPS 5th EESL Statuskolloquium 19. - 20. Nov. 2001, Konstanz, Germany. • Ernst, B., Hitzfeld, B.C. & Dietrich, D.R.: Presence of Planktothrix sp. and cyanobacterial toxins in Lake Ammersee, Germany and their impact on whitefish (Coregonus lavaretus L.) 5th International Conference on Toxic Cyanobacteria, 15. - 20. Jul. 2001, Noosa, Queensland, Australia.

10 PUBLICATIONS & PRESENTATIONS ______

• Ernst, B., Hitzfeld, B. & Dietrich, D.R.: Detection of cyanobacterial toxins in whitefish (Coregonus lavaretus L.) from lake Ammersee. SOT 39th Annual Meeting, 19. - 23. Mar. 2000, Philadelphia, PA, USA.

Co-author • D.R. Dietrich, B. Ernst & Day, B.W.: Human consumer death and algal supplement consumption: A post mortem assessment of potential microcystin-intoxication via microcystin immunohistochemical (MC-IHC) analyses. 7th International Conference on Toxic Cyanobacteria, 05.-10. Aug. 2007, Rio das Pedras, Brazil. • A. Fischer, S. J. Hoeger, D. Feurstein, B. Ernst & Dietrich, D. R.: Importance of organic anion transporting polypeptides (OATPs) for the toxicity of single microcystin congeners in vitro. 7th International Conference on Toxic Cyanobacteria, 05.-10. Aug. 2007, Rio das Pedras, Brazil. • Dietz, L., Ernst, B., Höger, S.J. & Dietrich, D.R.: Recovery of MC-LR in fish liver. 6th International Conference on Toxic Cyanobacteria, 21. - 26. Jun. 2004, Bergen, Norway. • Schmid, D., Ernst, B., Höger, S.J. & Dietrich, D.R.: Characterization and differentiation of two microcystins from Planktothrix spec. isolated from a pre-alpine lake in Europe. 6th International Conference on Toxic Cyanobacteria, 21. - 26. Jun. 2004, Bergen, Norway.

AWARDS & FUNDINGS: • Environment Award of the Environment and Living Foundation, University of Konstanz, Germany (2007). • Ph.D. scholarship by the Arthur & Aenne Feindt Foundation, Hamburg, Germany. • Congress grant by the Environment and Living Foundation, University of Konstanz (2004). • Congress grant by the German Research Foundation, DFG (2007).

CONTENTS

1. INTRODUCTION

1.1. CYANOBACTERIA - A GENERAL INTRODUCTION …………………………… 14 1.2. CYANOBACTERIAL TOXINS …………………………………………………… 19 OLIGOPEPTIDES ………………………………………………………………… 20 ALKALOIDS …………………………………………………………………… 24 OTHER CYANOBACTERIAL TOXINS …………………………………………… 27 CYANOBACTERIAL TOXINS – COMPARISON OF TOXIC POTENTIALS …………… 29 1.3. CYANOBACTERIA: EFFECTS ON FISH ………………………………………… 31 CYANOBACTERIA INDUCED FISH KILLS ………………………………………… 32 THE ICHTHYOTOXICITY OF MICROCYSTIN ……………………………………… 34 ICHTHYOTOXICITY OF OTHER CYANOBACTERIAL TOXINS ……………………… 49 1.4. THE RISE AND FALL OF P. RUBESCENS AND FISHERY YIELDS IN LAKE AMMERSEE, GERMANY – HISTORY AND GOAL OF THE STUDY ……… 55

2. METHODICAL INOVATIONS

2.1. DETERMINATION OF THE FILAMENTOUS CYANOBACTERIA P. RUBESCENS IN ENVIRONMENTAL WATER SAMPLES USING AN IMAGE PROCESSING SYSTEM …………………………………… 57 ABSTRACT ………………………………………………………………………… 57 INTRODUCTION ………………………………………………………………… 58 MATERIAL & METHODS ………………………………………………………… 59 RESULTS ………………………………………………………………………… 63 DISCUSSION ……………………………………………………………………… 65 ACKNOWLEDGEMENTS ………………………………………………………… 67 2.2. RECOVERY OF MC-LR IN FISH LIVER TISSUE ………………………………… 68 ABSTRACT ………………………………………………………………………… 68 INTRODUCTION ………………………………………………………………… 69 MATERIAL & METHODS ………………………………………………………… 70 RESULTS ………………………………………………………………………… 73 DISCUSSION ……………………………………………………………………… 76 ACKNOWLEDGEMENTS ………………………………………………………… 80

3. EXPOSURE EXPERIMENTS

3.1. ORAL TOXICITY OF THE MICROCYSTIN-CONTAINING CYANOBACTERIUM PLANKTOTHRIX RUBESCENS IN EUROPEAN WHITEFISH (COREGONUS LAVARETUS) …………………… 81 ABSTRACT …………………………………………………………………… 81 INTRODUCTION ………………………………………………………………… 82 MATERIAL & METHODS ………………………………………………………… 83 RESULTS ………………………………………………………………………… 87 DISCUSSION ……………………………………………………………………… 93 ACKNOWLEDGEMENTS ………………………………………………………… 95

3.2. PHYSIOLOGICAL STRESS AND PATHOLOGY IN EUROPEAN WHITEFISH (COREGONUS LAVARETUS) INDUCED BY SUBCHRONIC EXPOSURE TO ENVIRONMENTALLY RELEVANT DENSITIES OF P. RUBESCENS …………… 96 ABSTRACT …………………………………………………………………… 96 INTRODUCTION ………………………………………………………………… 97 MATERIAL & METHODS ………………………………………………………… 98 RESULTS ………………………………………………………………………… 102 DISCUSSION ……………………………………………………………………… 109 ACKNOWLEDGEMENTS ………………………………………………………… 112

4. FIELD-STUDIES

4.1. ABUNDANCE AND TOXICITY OF PLANKTOTHRIX RUBESCENS IN THE PRE ALPINE LAKE AMMERSEE, GERMANY ..……………………… 113 ABSTRACT …………………………………………………………………… 113 INTRODUCTION ………………………………………………………………… 114 MATERIAL & METHODS ………………………………………………………… 115 RESULTS ………………………………………………………………………… 120 DISCUSSION ……………………………………………………………………… 125 ACKNOWLEDGEMENTS ………………………………………………………… 133 4.2. THE ADVERSE EFFECTS OF PLANKTOTHRIX RUBESCENS ON COREGONIDS (COREGONUS LAVARETUS) IN LAKE AMMERSEE – FURTHER FIELD-OBSERVATIONS ……………………… 134 ABSTRACT …………………………………………………………………… 134 INTRODUCTION ………………………………………………………………… 135 MATERIAL & METHODS ………………………………………………………… 136 RESULTS ………………………………………………………………………… 137 DISCUSSION ……………………………………………………………………… 139

5. GENERAL DISCUSSION

5.1. ASSESSMENT ON THE IMPACT OF PLANKTOTHRIX RUBESCENS ON FERAL COREGONID IN LAKE AMMERSEE ……………………………… 142 INITIAL SITUATION ……………………………………………………………… 142 THE TOXICITY OF P. RUBESCENS IN LAKE AMMERSEE - BASIC CONSIDERATIONS 144 DIRECT IMPACT OF TOXIC P. RUBESCENS ON LAKE AMMERSEE COREGONIDS … 144 P. RUBESCENS-INDUCED ENVIRONMENTAL CHANGES: INDIRECT EFFECTS ON COREGONIDS ……………………………………… 147 CONSEQUENCES ………………………………………………………………… 149 CONCLUDING ASSESSMENT ……………………………………………………… 150 5.2. ASSESSMENT OF HUMAN HEALTH HAZARD, RISING FROM THE ABUNDANCE OF TOXIC P. RUBESCENS IN LAKE AMMERSEE ……………… 151 IRRITATION & ACCIDENTAL INTOXICATION DURING RECREATIONAL WATER ACTIVITIES ………………………………… 151 INTOXICATION VIA UPTAKE OF CONTAMINATED DRINKING WATER AND FOOD 152

6. ABBREVIATIONS ……………………………………………………………………… 154

7. REFERENCES ………………………………………………………………………… 155

8. APPENDIX ………...…………………………………………………………………… 174

1. INTRODUCTION

1.1. CYANOBACTERIA - A GENERAL INTRODUCTION

“LAKE ALBERT IN WAGGA WAGGA HAS BEEN CLOSED BECAUSE OF TOXIC LEVELS OF BLUE-GREEN

ALGAE …” (ABC News, Australia 25.01.07), “ALGAE OUTBREAK PROMPTS WATER WARNING IN

QUEBEC – HUNDREDS OF QUEBECERS ARE BEING WARNED NOT TO DRINK THEIR WATER BECAUSE OF

AN OUTBREAK OF BLUE-GREEN ALGAE …” (TheStar.com, Canada 06.07.07), “WATER SPORTS HAVE

BEEN BANNED AT A MANCHESTER LAKE FOLLOWING REVELATIONS THAT TOXIC BLUE-GREEN ALGAE

HAS BLOOMED IN THE WATER …” (BBC News, England 23.08.07). Headlines like these, regularly show people worldwide quite plainly that in a lot of waters a distinct problem is growing: Cyanobacteria (synonyms: blue-greens, blue-green algae, cyanoprokaryotes, cyanophyceans, cyanophytes, myxophyceans, etc.) – a few micrometer small organisms with a promising potential for the future, but also a source of considerable nuisance and hazard for animal and human health.

The variety of names highlights the important position that cyanobacteria occupied in the scientific past. Since their earliest characterisation by Linné (1753), they have been a matter of interest for many scientists of various disciplines, including botanists (Geitler, 1932; Vaucher, 1803), microbiologists (Forti, 1907), limnologists (Lampert, 1981; Skulberg, 1964), biochemists (Singh et al., 2005) and toxicologists (Falconer et al., 1981). Cyanobacteria provide an extraordinary contribution to human affairs in every day life, they are important primary producers (Whitton & Potts, 2000 and references therein) and they contribute globally to soil and water fertility in rice fields and other agricultural wetlands (Whitton, 2000). Cyanobacteria produce an incredible number of metabolites and extraordinary pigments, making them interesting to the pharmaceutical, cosmetics and colours & dyes industries (Singh et al., 2005) and even in the utilisation of alternative energies (Skulberg, 1994). Cyanobacterial mass occurrences however also present a considerable nuisance for the management of water bodies. Surface scum, water colouration, unpleasant odour and particularly the production and release of highly potent toxins affect aquaculture, drinking water treatment, crop irrigation and recreational water use (Dietrich & Hoeger, 2005; Hitzfeld et al., 2000). The water quality problems caused by dense cyanobacterial populations are intricate and have great health and economic impacts. Thus, public concern is rising and with it, the interest in cyanobacteria and cyanobacterial toxins.

What are cyanobacteria, where do they occur and where do they come from? Cyanobacteria are ancient gram-negative prokaryotes. Fossil occurrences of cyanobacteria are thought to date to approximately 3.5 billion years ago (Schopf & Packer, 1987). Thus, they belong to the very 14 1. INTRODUCTION ______earliest life forms. Fossil cyanobacteria are primarily isolated from stromatolites1. Despite their long evolutionary history, the fossil forms resemble very closely the currently occurring species (Schopf, 2000). Cyanobacterial cells are encased by a cell wall and lipopolysaccharides. A few species are additionally covered by exopolysaccharides, which protect against digestion by potential grazers (Friedland et al., 2005; Kolmakov & Gladyshev, 2003; Lewin et al., 2003). Cyanobacterial cells include no nucleus2 and no other membrane covered cellular organelles (e.g. golgi, mitochondria, endoplasmatic reticulum and chloroplasts). Their ribosomes, synthesising their proteins, are of the bacterial type. Thus, the only class (classis: cyanobacteria) of the Cyanophyceae is taxonomically categorised as bacteria (summarised in Adams & Duggan, 1999; Mur et al., 1999 and van den Hoek & Jahns, 2002). The morphology of cyanobacteria is very diverse comprising unicellular, colonial as well as multicellular filamentous forms. They include species with or without specialised cells (i.e. heterocysts and akinets) and possess many more characteristic specialities. Based on primarily morphological characteristics, cyanobacteria are systematically classified into five orders: Chroococcales (including the toxic genera Microcystis), Nostocales (including the toxic genera Anabaena, Aphanizomenon, Cylindrospermopsis & Nodularia), Oscillatoriales (including the toxic genera Planktothrix and Lyngbya), Pleurocapsales and Stigonematales (summarised in Whitton & Potts, 2000 and van den Hoek & Jahns, 2002). According to the International Code of Botanical Nomenclature there are 150 genera including about 2000 species (Mur et al., 1999; van den Hoek & Jahns, 2002). Not even 50 of these species have yet been shown to produce toxins however, those toxic species are very widespread, leading to the actuality that approximately 75 % of waters containing cyanobacteria also contain cyanobacterial toxins (Sivonen & Jones, 1999).

Their diversity enables cyanobacteria to colonise a tremendous variety of ecosystems, including both largely barren and infertile, as well as nutrient rich conditions. Cyanobacteria have been isolated from hot springs (e.g. Yellowstone National Park; see Ward & Castenholz, 2000 and references therein), from extremely halophilic (e.g. the Dead Sea, Israel; see Oren, 2000 and references therein) and alkaline waters (e.g. Lake Bogoria, Kenya; see Krienitz et al., 2003), as well as from cold glacial- and polar lakes (Vincent, 2000 and references therein; Jungblut et al., 2005). They can be found in terrestrial and arid habitats, where they grow epi- and endolithic on rocks and walls (summarised in Wynn-Williams, 2000 and Pentecost & Whitton, 2000). The majority however arise in aquatic environments, i.e. marshland and waters, including salt- brackish- and in particular freshwater (Mur et al., 1999; van den Hoek & Jahns, 2002). Altogether, cyanobacteria are ubiquitous and thus it is not surprising that also cyanobacterial toxins can be detected worldwide (Sivonen & Jones, 1999).

1 Stromatolites are sedimentary growth structures formed via trapping, binding, and cementation of sedimentary grains by microorganisms, especially cyanobacteria (Stal, 2000) 2 The DNA-containing nucleoplasm is without an envelope (Adams & Duggan, 1999) 15 1. INTRODUCTION ______

NUTRITIONAL

PRE-EXISTING CYANO- CONDITIONS POPULATIONS MORPHO- & HYDROLOGY OF THE WATER BODY

CYANOBACTERIAL WATER COLUMN TEMPERATURE, LIGHT BLOOMS STABILITY

GRAZING COMPETITION WITH WEATHER OTHER SPECIES

Fig. 1.1: Cyanobacterial bloom development is based on a complex interaction of various factors which are influenced by each other and moreover by daily, seasonal and long-ranging alterations (e.g. climatic changes).

The reproduction of cyanobacteria is asexual, via cell division, filament (i.e. trichome-) fragmentation or by the formation of special hormogonia3 (Mur et al., 1999; van den Hoek & Jahns, 2002 and references therein). A few species can additionally form akinetes, resting cells which develop from vegetative cells and can endure several years of unfavourable environment (Adams & Duggan, 1999). Whenever conditions are favourable, cyanobacteria can proliferate rapidly, resulting in the formation of cyanobacterial blooms4. Those blooms can reach densities of up to 109 cells/ml (Zohary & Madeira, 1990) and normally accumulate at the surface or in the near surface stratum of a lake (Oliver & Ganf, 2000). A few species (e.g. Planktothrix rubescens, Cylindrospermopsis raciborskii) however can also accumulate in distinct layers below the surface stratum, for example in the metalimnion of stratified lakes (Blikstad-Halstvedt et al., 2007; Falconer, 2005; Hoeger et al., 2004; Jacquet et al., 2005). Cyanobacterial blooms occur especially in regions with elevated nutrient input into water bodies (i.e. eutrophication), which are consequential from either natural circumstances or anthropogenic influences for example a lack of concomitant sewage treatment (Bartram et al., 1999). This resulted in the assumption that cyanobacterial mass occurrence serves as a unique indicator of eutrophication. However, this must be reviewed as recent scientific findings clearly demonstrate, that cyanobacterial bloom development is based on a complex interaction of numerous factors (summarised in Mur et al., 1999; Oliver & Ganf, 2000 and Falconer, 2005), including water body morphology, water column stability, the temperature and light regime and weather conditions (Fig. 1.1). Thus it has become evident that the reduction of external loading (i.e. re- oligotrophication) alone does not guarantee the disappearance of cyanobacteria from a lake (Jacquet et al., 2005; Morabito et al., 2002).

3 Filament fragments that detach by active gliding motion and gradually develop into new filaments (van den Hoek & Jahns, 2002) 4 A rapid increase in the population of algae/cyanobacteria to densities as to render it visible to the human eye (Vollenweider, 1976)

16 1. INTRODUCTION ______

Cyanobacteria are aerobic photoautotrophic organisms – meaning that their principal mode of energy metabolism is photosynthesis and their life processes require beside inorganic nutrients primarily water, carbon dioxide and light (Mur et al., 1999; van den Hoek & Jahns, 2002). It has been demonstrated that certain cyanobacterial species are able to survive long periods in complete darkness (Micheletti et al., 1998) and furthermore that approximately 50% of the existing species show a distinct ability for facultative heterotrophic nutrition5 (Adams & Duggan, 1999 and references therein). The ability to colonise those diverse and extreme habitats requires further particular qualities in addition to the basic physiological properties described above. Indeed, cyanobacteria possess a few remarkable specialisations allowing the occupation of ecological niches and a competitive advantage compared to other photoautotrophic organisms. The photosynthetic pigments of cyanobacteria include besides chlorophyll a and carotinoids, the accessory pigments phycocyanin (providing their characteristic blue-green colour), allophycocyanin and occasionally phycoerythrin (van den Hoek & Jahns, 2002). These biliproteins are combined in phycobilisomes and enable maximum light utilisation, from the spectrum between 400 and 700 nm wavelengths, also including the green light range (550-650 nm wavelength), which is largely inaccessible to the majority of green algae and plants (Oliver & Ganf, 2000; van den Hoek & Jahns, 2002). Most cyanobacterial species additionally perform chromatic adaptation (Mur et al., 1999; Oliver & Ganf, 2000). This means that their synthesis of photosynthetic pigments is particularly susceptible to light quality and environmental influences and in consequence that species are able to produce the accessory pigment needed to absorb light most efficiently. Those extraordinary pigments and abilities allow cyanobacteria to exist under low light conditions and thus in ecological niches which are unreachable for most eukaryotic algae, e.g. under ice cover or in the metalimnion of lakes (Blikstad-Halstvedt et al., 2007). Cyanobacteria furthermore have a remarkable ability to store essential nutrients and metabolites (e.g. carbohydrate granules, lipid globules, cyanophycin granules, polyphosphate bodies, carboxysomes, etc.) enabling them to endure temporary nutritional poverty (summarised in Mur et al., 1999; Oliver & Ganf, 2000 and van den Hoek & Jahns, 2002). Several species moreover have the capability for nitrogen fixation putting them at an advantage during conditions with limited availability of inorganic nitrogen (Adams & Duggan, 1999; Oliver & Ganf, 2000). Nitrogen fixing species use the enzyme nitrogenase to convert molecular nitrogen directly into utilisable ammonium. This usually occurs in heterocysts6. As another speciality, many cyanobacterial species possess gas vesicles - gas filled, cytoplasmatic inclusions enclosed by a semi-permeable membrane and acting as a buoy. In most cyanobacterial species the buoyancy resulting from those vesicles can be regulated, by cellular inclusion of photosynthetic products (corresponding to cellular ballast and changing the cellular turgor pressure which may end in the collapse of vesicles), or via the generation and degradation of gas vesicles (Oliver & Ganf, 2000 and references therein). This buoyancy regulation allows planktonic

5 Heterotrophic species consume and assimilate organic substrates (e.g. glycogen granules, lipid globules, etc.) in order to get carbon and energy for growth and development (Adams & Duggan, 1999) 6 Specialised, nitrogen fixing cells with a thickened cell wall and without photosynthetic activity, so that an anoxic cellular environment can be maintained, which is an essential for the reduction of nitrogen via nitrogenase (Adams & Duggan, 1999) 17 1. INTRODUCTION ______cyanobacteria to stratify and move effectively in the water column, adapting to the best position within vertical gradients of chemical and physical factors (Dokulil & Teubner, 2000; Mur et al., 1999; Oliver & Ganf, 2000). Finally, most cyanobacteria species attain maximum growth rates by temperatures above 25°C (Dokulil & Teubner, 2000 and references therein). Thus, at high temperatures cyanobacteria have an obvious growth advantage over green algae and diatoms, which prefer much lower temperatures.

These physiological properties however vary between different cyanobacterial species. Thus, not all species are adapted to the various environmental conditions in the same way. This allows the emergence of diverse ecostrategists7 (Mur et al., 1999). Scum-forming ecostrategists (e.g. Microcystis sp., Anabaena sp., Aphanizomenon sp., etc.) attain best growth conditions and simultaneously protection from detrimental high light intensities at the water surface through continuous movement throughout the near-surface layer (Falconer, 2005). When buoyancy regulation is disturbed (e.g. due to changing weather conditions), these species accumulate in unpleasant-smelling scums at the water surface. Homogenously distributed ecostrategists (e.g. Planktothrix agardhii, Limnothrix redekei, etc.) develop in circulating water bodies where they are distributed over the whole water body (Briand et al., 2002). In contrast, stratifying ecostrategists (e.g. Planktothrix rubescens, Cylindrospermopsis raciborskii) develop distinct layers primarily in the metalimnion of thermally stratified lakes (Blikstad-Halstvedt et al., 2007; Falconer, 2005). These species are adapted to low light conditions and benefit from the nutrient rich metalimnic situation. Benthic ecostrategists (e.g. Oscillatoria limnosa, Phormidium sp., etc.) form coherent mats on the bottom sediments of water bodies that are sufficiently clear to allow light penetration to the ground (Mez et al., 1997; Wood et al., 2007). Benthic ecostrategists can thus also occur in oligotrophic lakes. In contrast, the mass development of nitrogen fixing ecostrategists (e.g. Anabaena sp., Aphanizomenon sp., Cylindrospermopsis sp., Nodularia sp., etc.) can be mainly related to periodic nitrogen limitation occurring in both shallow and deep water bodies mostly exhibiting eutrophic and hypertrophic conditions (Mur et al., 1999; Oliver & Ganf, 2000).

Based on the described morphological and physiological fundaments, cyanobacteria have managed to survive millions of years. Moreover, they temporarily even developed burgeoning growth - during the Precambrian, also called » the age of cyanobacteria «, 2.5 to 0.5 million years ago. Due to oxygen release associated with their photosynthetic activity, this age culminated in the accumulation of oxygen in the biosphere (Adams & Duggan, 1999). Thus, cyanobacteria actually cleared the way for the predominantly aerobic life on earth as exists today – all the more astonishing that they now increasingly attract attention because of the risks they pose to human and animal life due to the production of highly toxic metabolites.

7 Classification of ecotypes, according to their behaviour in the water column and according to their adaptation for specific environmental conditions (Mur et al., 1999) 18 1. INTRODUCTION ______

1.2. CYANOBACTERIAL TOXINS

Cyanobacterial mass occurrences and detrimental effects on human and animal life have been associated at least since the early Middle Ages. As noticed by Bartram et al. (1999), there are reports from the Chinese Han dynasty of loosing troops from poisoning as soldiers drank from a river that was green in colour, approximately thousand years ago. In 1189, Gerald of Wales documented Lake Llangorse, to “turn bright green” and to “became scarlet” throughout his Journey through Wales (Belov et al., 1999). Although largely anecdotal and not documented on scientific analyses as we know it today, these and other reports suggest that not only cyanobacteria, but also cyanobacterial toxins might have accompanied life on earth for hundreds of years. The first observation giving scientific evidence for poisonings caused by cyanobacterial mass occurrence was given by George Francis (1878). He documented a bloom of Nodularia spumigena “… forming a thick scum like green oil paint … as thick and pasty as porridge” in Lake Alexandria, Australia, and demonstrated this bloom to act poisonously, and to cause rapidly death of sheep, horses, dogs and pigs subsequently to symptoms including stupor, unconsciousness, as well as convulsions and spasm. Since this time, the awareness of toxic cyanobacteria has increased continuously. At first, by a rising number of scientific documentations of fish kills and livestock mortalities of animals living in and drinking from lakes and ponds containing toxic cyanobacteria (summarised in Falconer, 2005 and Kuiper-Goodman et al., 1999), and since the nineteen-eighties, by the chemical and toxicological characterisation of numerous toxic metabolites which were isolated from cyanobacterial blooms (Dow & Swoboda, 2000; Sivonen & Jones, 1999).

Cyanobacteria produce a variety of unusual substances which are considered as secondary metabolites and thus appear not essential for the organism’s metabolic pathways by itself (Bentley, 1999). Most of them serve yet unclear functions, however, some have been associated with various bioactive, as well as toxic attributes (summarised in Dow & Swoboda, 2000; Kreitlow et al., 1999 and Singh et al., 2005). The latter belong to diverse groups of natural toxins regarding both, their chemical and toxicological characteristics, including alkaloids, lipopolysaccharids and multifarious oligopeptides that cause neurotoxic, hepatotoxic, cytotoxic, as well as dermatotoxic and irritant effects. Almost all cyanobacterial toxins are intracellular toxins. Therefore, they are primarily released into the water by natural bloom senescence or anthropogenically induced cell death connected with water treatment (e.g. chlorination, algaecide application) and less by a continuous toxin

19 1. INTRODUCTION ______excretion. Organisms may be exposed either from a direct uptake of cyanobacterial toxins (dissolved in water or accumulated in contaminated food) or from the uptake of toxin containing cyanobacterial cells and their subsequent digestion resulting in a toxin release within the digestive tract (Kuiper-Goodman et al., 1999; Landsberg, 2002).

OLIGOPEPTIDES Globally the most frequently found cyanobacterial toxins are oligopeptide toxins (Sivonen & Jones, 1999). They primarily occur in blooms in fresh and brackish waters and are produced by more than 30 genera belonging to all taxonomical classis (i.e. Oscillatoriales, Nostocales, Chroococcales, Stigonematales and Pleurocapsales). More than 600 peptides have been characterised including 80 structural archetypes of compounds (Welker & Döhren, 2006). While most of them (e.g. aeruginosins, microginins, anabaenopeptins, cyanopeptolins, microviridins, cyclamides) are considered to provide negligible or at most slight toxicity (e.g. inhibiting chymotrypsin, plasmin and proteases)(Chorus, 2006; Grach-Pogrebinsky et al., 2003; Ishida et al., 1997; Itou et al., 1999; Rohrlack et al., 2003; Sano & Kaya, 1995; Sano & Kaya, 1996, etc.), nodularins and especially microcystins have been associated with numerous severe animal and human intoxications all over the world (Falconer, 2005; Kuiper-Goodman et al., 1999).

Microcystin & Nodularin Microcystins and structural related nodularins are cyclic peptides named after the organism from which they were first isolated: Microcystis aeruginosa (Botes et al., 1982; Botes et al., 1985) and Nodularia spumigena (Rinehart et al., 1988). Microcystins have since also been isolated from various Anabaena sp., Planktothrix sp. and less frequently from Nostoc sp., Anabaenopsis sp. and Hapalosiphon sp. In contrast, nodularin production is restricted to Nodularia sp. yet (Sivonen & Jones, 1999). Both, microcystins and nodularins are produced non-ribosomally, by multi enzyme complexes and peptide synthetase genes (summarised in Falconer, 2005). They contain either seven (microcystins; Fig. 1.2), or five (nodularins; Fig. 1.3) amino acids, building a cyclic structure.

D-Glu Mdha

Adda D-Ala

X Z D-Measp Fig. 1.2: Chemical structure of microcystins (MW: 900-1,100 D), i.e. cyclo-(D-Alanine-X-D-MeAsp-Z-Adda-D- glutamate-Mdha), where X and Z can be occupied by various L-amino acids. 20 1. INTRODUCTION ______

D-Glu Adda

Mdhb

L-Arg D-Measp

Fig. 1.3: Chemical structure of nodularin (MW: 824 D), i.e. cyclo-(D-Alanine-X-D-MeAsp-Z-Adda-D- glutamate-Mdha)

Both are primarily composed of D-amino acids and collectively contain the unusual Adda8 as well as D-Glutamic and D-Methylaspartic acid (Measp). The microcystin molecule moreover contains D-Alanine, N-Methyldehydroalanine (Mdha) and two L-amino acids. Structural variations in microcystins have been characterised for all seven amino acids, but most frequently with substitution of the L-amino acids and demethylation of the Mdha- and Measp-residue. This variability results in more than 75 naturally occurring microcystin congeners (Spoof, 2004). The nodularin molecule is completed by N-Methyldehydrobutyrine (Mdhb) and D-Alanine. In contrast to microcystins, structural variations of nodularin are restricted to limited demethylations. Thus, only seven naturally occurring nodularin congeners have been characterised to date (summarised in Craig & Holmes, 2000 and Sivonen & Jones, 1999).

Microcystins and nodularins are extremely stable and persistent once released into the water. Rapid chemical hydrolysis has only been obtained under environmentally irrelevant conditions including exposure to 6 M hydrochloric acid and high temperatures (Harada, 1996). In natural environments microcystins and nodularins may persist for weeks until their photochemical breakdown and isomerisation. Photochemical degradation can be accelerated through photosensitisation due to the presence of photopigments and humic substances (summarised in Sivonen & Jones, 1999 and Welker et al., 2001). Despite their chemical stability, microcystins have been shown to be susceptible to breakdown by aquatic bacteria. After an initial lag phase, which may persist for several days, the biodegradation process commences and microcystin removal of up to 90 percent can be achieved within 2-10 days depending on environmental conditions, e.g. water temperature, water body hydrology, initial microcystin concentration, etc. (Jones et al., 1994; Lahti et al., 1997; Rapala et al., 2005; Welker et al., 2001).

8 Adda is 3-amino-9-methoxy-2,6,8,-trimethyl-10-phenyl-4,6,-decadienoic acid, an unusual D-amino acid 21 1. INTRODUCTION ______

Although the bacterial and photosensitised degradation is assumed to be much more efficient than non-catalysed breakdown, it appears most probable that high microcystin and nodularin concentrations - which mostly occur locally, within and around a senescent cyanobacterial bloom - initially decrease due to dilution within the water body rather than via chemical or biological degradation (Jones & Orr, 1994). Such dilution may prevent acute intoxication however it broadens contamination and thus may increase the number of exposed organisms.

The uptake of microcystin and nodularin by exposed organisms from either water or ingested cyanobacterial cells is dominated by the polar characteristic of the cyclic peptides. Except a few microcystin congeners containing hydrophobic amino acids (e.g. microcystin-LW and microcystin- LF), microcystins and nodularins are hydrophilic and therefore soluble in water and hence also in blood of exposed organisms (Botes et al., 1982; Harada, 1996). Due to their hydrophilic character, nodularins and most microcystins however are unable to penetrate lipid membranes via passive diffusion (Falconer, 2005; Vesterkvist & Meriluoto, 2003). Thus, the majority of those peptides ingested are unable to pass the epithelium of the ileum and consequently remain in the digestive tract, from where the toxins are most likely excreted via the faeces (Fujiki et al., 1996). However, ingested microcystins and nodularins have been shown to be at least fractionally transported across the ileum into the venous bloodstream and also from the portal vein into hepatocytes via bile acid membrane transporters (e.g. organic anion transporters (OATPs)) (Eriksson et al., 1990; Fischer et al., 2005; Runnegar et al., 1991, etc.). Presumably due to the first pass effect9 and the high density of transporting peptides in hepatocytes, the liver is the main target organ for both accumulation and detoxification of microcystins and nodularins. This categorises them as primarily hepatotoxic (Falconer et al., 1986; Kaya, 1996; Robinson et al., 1991, etc.). Microcystins are however also detectable in other OATP-expressing organs (i.e. intestine, kidney, brain, lung, heart, etc.), albeit to a considerably lesser extent (Ito et al., 2000; Robinson et al., 1989; Robinson et al., 1991; Spoof et al., 2003a, etc.). This indicates that toxicity in these organs is also likely.

The toxicity of microcystins and nodularins is mainly mediated via a strong inhibition of protein phosphatases (PP). This inhibition results from a non-covalent interaction of the peptide’s Adda- glutamate domain with the catalytic site of especially the serine/threonin phosphatases PP1 and PP2A (Honkanen et al., 1990; MacKintosh et al., 1990; Yoshizawa et al., 1990). As the Adda- glutamate domain is well conserved across various microcystins and nodularins, the toxicity of almost all structural variants shows low variation (LD50: ≥50 µg/kg bw; mouse toxicity (i.p.)10), except for a few congeners containing considerable modifications to the Adda-glutamate region and thus showing no or negligible PP-inhibition (summarised in Sivonen & Jones, 1999). The initial binding (formed within minutes) can be followed by the formation of an additional irreversible covalent bond between the Mdha-residue of microcystins and a cysteine residue of the

9 Molecules which were absorbed via the digestive system enter the hepatic system prior to other organs due to the flow direction of the bloodstream leaving the digestive system 10 Single-dose level that will cause death in 50 per cent of exposed animals within 7-14 days. The LD50 depends, besides the toxicity of the applied substance, on the species sensitivity and especially on the form of application 22 1. INTRODUCTION ______

PP-molecule (Craig et al., 1996; Goldberg et al., 1995). The formation of the covalent bond requires several hours and indeed strengthens the interactions between microcystins and PPs, however it does not strengthen PP-inhibition (Bagu et al., 1995; Maynes et al., 2006). A covalent bond cannot arise with nodularin due to a lack of Mdha. Correspondingly, it is also impeded in conjunction with microcystin congeners that include significant alterations at the Mdha-residue (Craig et al., 1996; Hastie et al., 2005). As protein phosphatases are key cellular enzymes which regulate and control a variety of cellular functions and processes via the dephosphorylation of proteins (Barford et al., 1998; Cohen, 1989; Janssens & Goris, 2001), the inhibition of PPs by microcystins and nodularins results in a hyperphosphorylation of phosphate regulated proteins (Falconer & Yeung, 1992; Ohta et al., 1992; Wickstrom et al., 1995; Yoshizawa et al., 1990). Such hyperphosphorylation causes a tremendous cascade of consequences in hepatocytes, including disintegration of cellular structure, loss of cell-cell adhesion, disruption of cellular metabolism, interference with signal transduction and disturbance of cell cycle control (summarised in Craig & Holmes, 2000; Falconer, 2005; Gehringer et al., 2004; Kuiper-Goodman et al., 1999). In addition to the described consequences of PP-inhibition, further damage may result from a microcystin induced generation of reactive oxygen species (ROS), which also affects cellular activities (summarised in Gehringer et al., 2004) and furthermore, from binding to the ATP synthetase which represents a trigger of mitochondrial apoptotic signalling (Mikhailov et al., 2003). Any or all of these alterations may result in severe liver damage (e.g. apoptosis, necrosis, loss of lobular architecture, intrahepatic haemorrhage, hepatic insufficiency, etc.) and subsequently death in the case of acute intoxications, as well as increased cellular proliferation and thus tumour promotion and/or initiation following chronic exposure. As PPs are present in various tissues and due to the organotropism of nodularin and microcystin, comparable cellular alterations and tissue damage also occur in organs other than liver, however these are usually less severe (Falconer et al., 1992; Khan et al., 1995; Nobre et al., 1999; Nobre et al., 2004).

For assessing the toxicity of a substance, not only are uptake, distribution and modes of action fundamental, but detoxification and degradation must also be considered. Microcystins and nodularins are highly resistant to eukaryotic peptidases. This is probably due their cyclic structure and their amino acid composition predominantly consisting of D-amino acids (Harada, 1996). The persistence of microcystins is additionally elevated by their covalent and thus irreversible binding to protein phosphatases (summarised in Falconer, 2005 and Kuiper-Goodman et al., 1999).

23 1. INTRODUCTION ______

ALKALOIDS Despite their chemical relationship, cyanobacterial alkaloid toxins differ considerably in their structure and stability, their mode of action and thus also in their toxicological consequences. Cyanobacterial alkaloids include dermatotoxic skin irritants (e.g. lyngbyatoxin and aplysiatoxin) with inflammatory and presumably even skin tumour promoting activity. They further include cytotoxic (e.g. cylindrospermopsin) and neurotoxic (e.g. PSPtoxins (saxitoxins) and anatoxins), metabolites that regularly cause animal and human poisonings.

Cylindrospermopsin Cylindrospermopsin came to attention as a result of severe gastroenteritis outbreak among children drinking from a water supply in Queensland, Australia, containing a bloom of Cylindrospermopsis raciborskii (Byth, 1980; Hawkins et al., 1985). Cylindrospermopsin is primarily detected in tropical and subtropical waters although increasing occurrence has also been observed in temperate climates, for example in New Zealand, Europe and South and North America (Falconer, 2005 and references therein). It is mainly produced by Cylindrospermopsis raciborskii. However, also other cyanobacterial species (e.g. Umezakia natans, Anabaena bergii, Anabaena lapponica, Raphidiopsis curvata and some Aphanizomenon sp.) have meanwhile been shown to produce cylindrospermopsin (Falconer, 2005 and references therein, Rucker et al., 2007; Spoof et al., 2006).

The toxin was first isolated and characterised by Ohtani et al. (1992). It comprises a tricyclic guanidine combined with hydroxymethyl uracil and is a highly water-soluble molecule (Fig. 1.4). Cylindrospermopsin is stable in darkness and shows slow breakdown in pure water under environmentally relevant conditions. Similar to microcystins, the photochemical breakdown of cylindrospermopsin is accelerated (≥90% degradation within 2-3 days) via photosensitisation in the presence of photosensitive pigments (Chriswell et al., 1999). Cylindrospermopsin has a flat, tricyclic molecule structure with rotational bonds within the two main components and can thus intercalate into the DNA double helix causing chromosome breaks and irreversible inhibition of protein synthesis (Falconer, 2005). Additional toxic effects have been assumed. Consequently, cylindrospermopsin causes a time- and dose-dependent cytotoxicity and presumably possesses mutagenic, clastogenic and even carcinogenic activity (Falconer, 2005).

Fig. 1.4: Chemical structure of cylindrospermopsin (MW: 415 D).

24 1. INTRODUCTION ______

In consequence of the inhibition of protein synthesis, cylindrospermopsin primarily injures tissues presenting a high synthesis/turnover of proteins, i.e. liver and kidney, intestine, and the immune system including spleen and thymus. Poisonings may lead to decreased functionality of these organs and subsequently even death, depending on the duration and dose of exposure (Briand et al., 2003; Falconer, 2005 and references therein).

PSPtoxins (Saxitoxins) Saxitoxins are primarily known in conjunction with paralytic shellfish poisoning (PSP). They were originally isolated from shellfish, where they may accumulate due to filter-feeding of toxic marine dinoflagellates (Anderson, 1994). As a consequence of the consumption of contaminated shellfish, PSPtoxins become available for higher trophic levels and have been associated with animal, livestock and human mortalities (Briand et al., 2003; Kuiper-Goodman et al., 1999; Landsberg, 2002). In cyanobacteria, PSPtoxins have been found in Aphanizomenon flos-aquae, various Anabaena sp. (predominantly A. circinalis), Cylindrospermopsis raciborskii (Sivonen & Jones, 1999; van Apeldoorn et al., 2007) as well as in Lyngbya wollei and Planktothrix sp. (Carmichael et al., 1997; Onodera et al., 1997b; Pomati et al., 2000). PSPtoxins are carbamate alkaloid toxins with a tricyclic structure of hydropurine rings (Fig. 1.5). There are at least 21 derivates, which differ among others predominantly in the substitution of five variable positions (Oshima, 1995; Sivonen & Jones, 1999; van Apeldoorn et al., 2007 and references therein). Similar to microcystin and cylindrospermopsin, PSPtoxins are also rather persistent toxins. Their chemical breakdown in darkness and room temperature is slow, often requiring more than three months for ≥90 % breakdown (Jones & Negri, 1997). However, no detailed studies have as yet been carried out on its breakdown in sunlight and it is unknown if this can be accelerated via photosensitisation. The severity of toxicity differs considerably among diverse PSPtoxin variants, with saxitoxin being the most toxic. The toxicity of all congeners is attributable to a generalised blockade of sodium channels in the excitable membranes in nerve axons (Briand et al., 2003; Kuiper- Goodman et al., 1999 and references therein). This results in a dose-dependent, partial or complete inhibition of action potential transmission in peripheral nerves and skeletal muscles. Thus, PSPtoxins cause general nerve dysfunction, culminating in paralysis, respiratory depression, and in the case of an acute dosage death due to respiratory failure.

Fig. 1.5: Chemical structure of PSPtoxin (e.g. saxitoxin; MW: 241-491 D). 25 1. INTRODUCTION ______

Anatoxin-a Presumably the most widespread cyanobacterial alkaloid is anatoxin-a. Anatoxin-a occurrences and episodes, including animal and human poisonings, have been reported from all over the world (Kuiper-Goodman et al., 1999 and references therein; Edwards et al., 1992; Gugger et al., 2004; Gunn et al., 1992; Wood et al., 2007). Anatoxin-a was first isolated from Anabaena flos-aquae (Devlin et al., 1977) and later from various other Anabaena species as well as from species of Aphanizomenon, Planktothrix, and Cylindrospermopsis (Sivonen & Jones, 1999; van Apeldoorn et al., 2007 and references therein). It is a low weight secondary amine (Fig. 1.6) and whilst the toxin appears relatively stable in darkness, in contrast to saxitoxin, it undergoes a rapid photochemical degradation in sunlight and even more in alkaline conditions. The toxin moreover appears to be readily degraded by bacteria that are associated with cyanobacterial filaments (summarised in Sivonen & Jones, 1999). Anatoxin-a is a post-synaptic nicotinic agonist and thus potent neurotoxin. It binds to neuronal nicotinic acetylcholine receptors at neuromuscular junctions, mimicking the effect of acetylcholine. This causes an influx of Na+ and subsequently, depolarisation, which results in the opening of voltage sensitive Ca++ and Na+ channels. In contrast to acetylcholine, anatoxin-a is not susceptible to enzymatic hydrolysis by acetylcholinesterases and consequently causes a prolonged stimulus blocking subsequent electrical transmission and potentially leading to muscular paralysis. Symptoms of anatoxin-a toxicity hence include muscle fasciculation, loss of coordination, staggering, gasping, convulsion and subsequent acute intoxication death by respiratory paralysis (summarised in Briand et al., 2003; Kuiper-Goodman et al., 1999 and Metcalf & Codd, 2005).

The anatoxin-a homologue homoanatoxin-a is produced by certain Phormidium, Anabaena and Raphidiopsis species (Furey et al., 2003; Namikoshi et al., 2003; Skulberg et al., 1992).

Homoanatoxin-a is a methyl derivate of anatoxin-a with slightly lower potency (based on the LD50 in mice following intraperitoneal application). It has been shown to block muscular contraction by neurostimulation in phrenic nerve-hemidiaphragm. Homoanatoxin-a toxicity is based on an enhanced Ca++ flux in the cholinergic nerve terminals. Comparable to anatoxin-a it acts very quickly and may cause death due to respiratory arrest within minutes (Lilleheil et al., 1997; van Apeldoorn et al., 2007 and references therein).

Fig. 1.6: Chemical structure of anatoxin-a (MW: 165 D). 26 1. INTRODUCTION ______

Fig. 1.7: Chemical structure of anatoxin-a(s) (MW: 252 D).

Anatoxin-a(s) Anatoxin-a(s) is a naturally occurring organophosphate. It is a unique phosphate ester of cyclic N- hydroxyguanine (Fig. 1.7) which decomposes rapidly in alkaline settings but is stable under neutral and acidic conditions (Matsunaga et al., 1989). Anatoxin-a(s) has been identified in blooms of Anabaena flos-aquae and A. lemmermannii (Henriksen et al., 1997; Onodera et al., 1997a). Symptoms of acute anatoxin-a(s) toxicology include muscle weakness, convulsion, respiratory distress and death due to respiratory failure. As these neurotoxic symptoms are very similar to those of anatoxin-a, anatoxin-a(s) was initially considered to be an anatoxin-a homologue. However, in contrast to anatoxin-a, anatoxin-a(s) is a acetylcholinesterase inhibitor (Cook et al., 1989), blocking the acetylcholine regenerating enzyme in a manner similar to organophosphate insecticides (e.g.malathion, parathion) and poison gases (i.e. tabun and sarin).

OTHER CYANOBACTERIAL TOXINS Only a few cyanobacterial toxins can neither be categorised to oligopeptides or alkaloids. The most prominent are Lipopolysaccharids (LPS), which are located at the outer cyanobacterial cell wall and thus are no intracellular toxins.

Lipopolysaccharides (LPS) Cyanobacterial LPS were first isolated from Anacystis nidulans (Weise et al., 1970). LPS are endotoxins and thus generally an integral component of the cell wall of Gram negative bacteria, which include cyanobacteria. They are condensed products of sugar and fatty acids, whose composition shows considerable variation among bacteria and also among cyanobacteria (Sivonen & Jones, 1999). LPS have pyrogenic activity and can act as irritants mainly due to their fatty acid component (Fig. 1.8). Exposure may also elicit an allergenic response, causing fever and induce gastroenteritis. However, cyanobacterial LPS are considerably less potent than LPS from pathogenic bacteria such as Salmonella (summarised in Briand et al., 2003; Kuiper-Goodman et al., 1999).

27 1. INTRODUCTION ______

Fig. 1.8: Chemical structure of the toxic component of lipopolysaccharides.

Beta-N-methylamino-L-alanine (BMAA) The unique non-protein amino acid BMAA is structurally similar to methylated alanine (Fig. 1.9). It is previously known from the seeds of the Guam cycad Cycas micronesica (Vega & Bell, 1967). BMAA has lately gained attention as it has been shown to biomagnify in the Guam ecosystem and to be associated with an increased occurrence of amytrophic lateral sclerosis/Parkinsonism dementia complex in the indigenous Chamorro people (Spencer et al., 1987). Recent investigations demonstrate the toxicity of the cycad seeds to result from cyanobacteria, living as endosymbionts in the coralloid roots (Cox et al., 2003). Meanwhile, BMAA has been shown to be produced by all known groups of cyanobacteria (i.e. Chroococcales, Nostocales, Oscillatoriales, Pleurocapsales and Stigonematales) in various environments (Cox et al., 2005). Once ingested, BMAA can be bound by proteins within the body, functioning as an endogenous neurotoxic reservoir that slowly releases toxin directly into the cerebral tissue through protein metabolism (Murch et al., 2004). The toxicity of BMAA is primarily based on the potent interaction as an agonist of glutamate AMPA/kainate receptors (excitotoxin) at glutamergic synapses in the brain and spinal cord (Weiss et al., 1989), whereas additional toxic effects have been assumed (Murch et al., 2004).

Fig. 1.9: Chemical structure of Beta-N-methylamino-L-alanine (BMAA).

28 1. INTRODUCTION ______

CYANOBACTERIAL TOXINS – COMPARISON OF TOXIC POTENTIALS The famous alchemist Paracelsus (1493-1541) raised the wise appraisal: "All things are poison and nothing is without poison, only the dose permits something not to be poisonous." In other words: the toxicity of a substance is determined by its dosage. When comparing the effective dosages of cyanobacterial toxins with other poisons of natural origin, cyanobacterial toxins range with curare and cobra toxin and thus display far higher toxicity than for example hydrocyanic acid (see also Tab. 1.1). This confirms the toxic potential and risk arising from poisonous cyanobacteria. Moreover, cyanobacterial blooms are often not dominated by a single species, but rather by a variety of species including non-toxic and toxic strains and species, a constellation that may generate various toxins concurrently. The toxin composition and toxicity of a cyanobacterial bloom is thus defined by its species composition as well as variations in the toxicity within one species. This results in considerable spatial and temporal variations in toxin compositions and concentrations within natural samples, often even within the same habitat. Thus, additive effects consequential of exposure to various cyanobacterial toxins simultaneously are likely to occur (Sivonen & Jones, 1999).

Tab. 1.1: Comparison of the mean acute lethal dose (LD50) of various cyanobacterial (bold) and other natural occurring toxins following intraperitoneal application to mice

ORGANISM OF ORIGIN TOXIN LD50 [µG/KG BW]

Clostridium botulinum botulinum toxin-a 0.00003 1 Clostridium tetani tetanus toxin 0.001 2 div. cyanobacteria saxitoxin 10 3 Naja naja cobra toxin 20 4 div. cyanobacteria anatoxin-a(s) 30 5 div. cyanobacteria microcystins ≥50 6 Nodularia spumigena nodularins 50 7 div. cyanobacteria anatoxin-a 200 8 Chondrodendron tomentosum tubocurarine (curare) 200 9 div. cyanobacteria cylindrospermopsin 2100 10 div. Rosacea, stone fruits, etc. hydrocyanic acid 10,000 9

REFERENCES: 1Hardegree & Tu, 1988; 2Gill, 1982; 3Briand et al., 2003 and references therein; 4Tu, 1991; 5Cook et al., 1988; 6Sivonen & Jones, 1999 and references therein; 7Eriksson et al., 1988; 8Devlin et al., 1977; 9Marquardt & Schäfer, 2004; 10Ohtani et al., 1992.

Considering the potential of cyanobacterial toxins and, in view of the ubiquitous occurrence of cyanobacteria it is not surprising, that toxic cyanobacteria affect organisms of almost all stages of development and trophic levels. Irrespective of this situation, the current knowledge on the

29 1. INTRODUCTION ______toxicity of cyanobacterial secondary metabolites is widely focused on their toxicity in terrestrial mammals especially rodents, which is most likely due to their relevance in human risk assessment. In natural environments, cyanobacterial toxins are hazardous to terrestrial mammals via the uptake of toxin containing water and contaminated food (Briand et al., 2003; Dietrich & Hoeger, 2005). In stark contrast, aquatic organisms are exposed to toxic cyanobacteria via multiple routes and the exposure duration is generally considerably longer.

30 1. INTRODUCTION ______

1.3.

CYANOBACTERIA: EFFECTS ON FISH

Organisms living in an environment displaying a mass occurrence of toxic cyanobacteria are surrounded by water containing either toxic cells and/or dissolved toxins. Those organisms are not only threatened by the ingestion of cyanobacteria and their toxins, toxic cells and contaminated water furthermore flow through their gills and they feed, breed and develop in that water. Hence aquatic organisms can be more profoundly influenced by cyanobacterial toxins than terrestrial animals. Regarding to this very interesting are fish. As they are located at a high trophic level within the aquatic food web, fish may be directly influenced by cyanobacterial toxins, via accumulation of cyanobacterial toxins from contaminated food (e.g. zooplankton, mussels, fish) (Engström-Öst et al., 2002; Kankaanpää et al., 2005b; Karjalainen et al., 2005; Smith & Haney, 2006), but also via indirect effects on their environment. Indeed, toxic cyanobacteria influence the zooplankton assemblage which may result in a qualitative and/or general loss of food organisms (DeMott et al., 1991; Laurén-Määttä et al., 1997; Sivonen & Jones, 1999). Moreover, cyanobacterial mass occurrences may modulate exogenous factors (e.g. pH-values, oxygen, ammonium, etc.) resulting in an insufficient sometimes even hostile environment (Wiegand & Pflugmacher, 2005 and references therein). Whether fish are affected by cyanobacteria depends on several factors: Which cyanobacterial species are involved? Do they produce toxins and if yes how much? Are the toxins released into the water and/or accumulated by potential food organisms? Additionally important is to what extent cyanobacteria and a given fish population come into contact with one another and thus whether it is a benthic, stratified, or scum forming bloom or if cyanobacteria are evenly distributed over the entire water body. Besides the sensitivity of the fish species in question it furthermore appears crucial, if the fish recognise and subsequently may avoid an exposure to cyanobacterial toxins. The course and dimension of various cyanobacteria induced fish kills however suggest the latter occurrences to be rather low.

31 1. INTRODUCTION ______

CYANOBACTERIA INDUCED FISH KILLS Whilst wildlife and predominantly livestock mortalities caused by toxic cyanobacteria regularly evoke great public concern (Edwards et al., 1992; Gugger et al., 2004; Gunn et al., 1992; Holschuh, 2001; Kuiper-Goodman et al., 1999; Mez et al., 1997; Negri et al., 1995; Wood et al., 2007, etc.), cyanobacteria induced fish mortalities mostly remain unnoticed. Nevertheless toxic cyanobacteria regularly cause considerable fish kills spanning the entire globe, involving almost all toxic cyanobacterial genera and affecting herbivorous, planktivorous, omnivorous and piscivorous fish species (Tab. 1.2). Fish kills most likely occur with the breakdown of a cyanobacterial bloom (Albay et al., 2003; Bürgi & Stadelmann, 2002; Nascimento & Azevedo, 1999; Rodger et al., 1994). Typically, environmental conditions change, disturbing cellular buoyancy and resulting in an accumulation of cyanobacterial cells at the water surface. Surface conditions, in particular elevated irradiance, high temperatures, wind, wave action, etc. cause bloom senescence, cyanobacteria cell lyses and toxin release. Fish remaining in those contaminated areas are then unavoidably exposed to cyanobacterial toxins. The decomposition of the senescent bloom material often simultaneously results in substantial oxygen depletion (Jewel et al., 2003; Nascimento & Azevedo, 1999), which, depending on the dimension of the cyanobacterial bloom, may cause insufficient oxygen conditions. In consequence cyanobacteria induced fish mortalities are caused by either poisoning due to toxin release, anoxia or a combination of both (Malbrouck & Kestemont, 2006 and references therein, Jewel et al., 2003; Pollux & Pollux, 2004; Toranzo et al., 1990). Hence dead fish appear after the disappearance of cyanobacteria cells. Thus, a lot of fish kills as induced by cyanobacterial toxins remain unrealised and are rather attributed to insufficient oxygen conditions (Gaete et al., 1994; Zambrano & Canelo, 1996). It is therefore most likely that the proportion of fish kills caused by cyanobacterial toxins is actually higher than thought. Even so, not every cyanobacterial bloom causes a fish kill. Released toxins should usually dissolve rapidly within the water body and toxin concentrations thus mostly remain beneath lethal levels. Moreover, fish may migrate within the water body and thus avoid an exposure to a toxic cyanobacterial bloom. It is also crucial to what extent exogenous (i.e. temperature, oxygen, alkalinity, etc.) and endogenous factors (i.e. mobility, physiological and nutritional condition, resistance to stress and diseases, etc.) modulate the consequences of a given toxin exposure (Hofer & Lackner, 1995). Nevertheless, also non-lethal and chronic exposure to toxic cyanobacteria may cause severe organ damage and impairment also resulting in an accumulation of cyanobacterial toxins within exposed fish (Andersen et al., 1993; Carbis et al., 1997; Sipiä et al., 2001a; Sipiä et al., 2002; Xie et al., 2005, etc.).

32 1. INTRODUCTION ______

Tab. 1.2: A summary of documented fish kills associated with cyanobacterial mass occurrence

CYANOBACTERIAL SP. AFFECTED FISH SPECIES YEAR LOCATION INVOLVED

Anabaena circinalis not specified 1880 div. Polish lakes 1 Anabaena sp. carp (Cyprinus carpio), perch (Perca sp.) 1913 div. German lakes 2 & roach (Rutilus rutilus) M. aeruginosa not specified 1913-1943 div. South-African lakes 3 Anabaena circinalis not specified 1914 div. Hungary lakes 1 Aphanizomenon sp. not specified 1931-1933 div. lakes in Iowa, USA 1 Anabaena sp. not specified 1940-1942 Lake Ymsen, Sweden 1 Aphanizomenon sp. not specified 1942 Zuiderzee, Holland 1 Aphanizomenon sp. carp, pike (Esox lucius) & perch 1946 div. rivers & lakes in Wisconsin, USA 4 Anabaena sp., buffalo fish (Ictiobus sp.) & carp 1948 Storm Lake, Iowa, USA 5 Microcystis sp. Nostoc sp. not specified 1956-1958 Waco, Texas, USA 6 M. aeruginosa, carp, roach, catfish (Ictalurus punctatus) 1956-1959 Wolga, Russia 1 Anabaena sp. & bream (Abramis sp.) Anabaena sp., perch 1961 div. lakes in Microcystis sp. Saskatchewan, Canada 7 Nodularia sp., not specified 1964,1982, Black Sea, Georgia 8 others 1992, 1997 Aphanizomenon sp. not specified 1967 Lake Winnisquam, New Hampshire, USA 9 Oscillatoria sp. roach 1982 Vesijarvi, Finland 10; 11 (red coloured) Anacystis cynaea pike & pikeperch (Stizostedion lucioperca) 1982-1987 div. lakes in Latvia 12 Microcystis sp. carp & silverside (Labidesthes sp.) 1984 Aculeo Lake, Chile 13; 14 Aphanizomenon sp. tench (Tinca tinca), bream, roach, perch, pike 1984 Sempacher See, Switzerland 15 & smelt (Osmerus sp.) Microcystis sp. not specified 1988 Forez, France 16 Anabaena sp. rainbow trout (Oncorhynchus mykiss) 1989 fish farm, Spain 17 Anabaena sp., not specified 1991 Darling river, Australia 18 Microcystis sp. Synechocystis sp. menhaden (Brevoortia tyrannus) 1991 Barra Lagoon, Brazil 19 Anabaena sp., brown trout (Salmo trutta) 1992 Loch Leven, Scotland 20 Microcystis sp., Aphanizomenon sp. Microcystis sp. catfish 1996 Patos Lagoon, Brazil 21 Microcystis sp smelt & ruffe (Gymnocephalus cernuus) 1994-1996 Lake Ijsselmeer30 Planktothrix sp. *Prymnesium sp., perch, roach & bleak (Alburnus alburnus) 1997 Aland, Finland 22 Oscillatoria sp. Aphanizomenon sp., small fish ( not specified) 1997 Lago Varese, Italy 23 Oscillatoria sp. Planktothrix rubescens not specified 1997 Lake Spanca, Turkey 24 not specified bream, roach, perch, pike div. Finnish lakes 25 & coregonids (Coregonus sp.) Microcystis sp. not specified div. ponds in southern USA 26 Microcystis sp. not specified 2000-2003 Brackish lake in Turkey 27 A. flos-aquae, carp, tilapia (Oreochromis sp.), catla (Catla catla) 2002 div. fish ponds M. aeruginosa & silvercarp (Hypophthalmichthys sp.) in Bangladesh 28 not specified perch, pike, tench, carp 2003 Romeinwerd, Netherlands 29 & topmouth gudeon (P. parva)

* haptophyte, no cyanobacterium REFERENCES: 1Schwimmer & Schwimmer, 1968; 2Seydel, 1913; 3Stephens, 1948; 4Mackenthum & Herman, 1948; 5Rose, 1953; 6Davidson, 1959; 7Gorham et al., 1964; 8Devidze, 1998; 9Sawyer et al., 1968; 10Persson et al., 1984; 11Berg et al., 1986; 12Druvietis, 1998; 13Gaete et al., 1994; 14Penaloza et al., 1990; 15Bürgi & Stadelmann, 2002; 16Sevrin-Reyssac & Pleticosic, 1990; 17Toranzo et al., 1990; 18Humpage et al., 1993; 19Nascimento & Azevedo, 1999; 20Rodger et al., 1994; 21Yunes et al., 1998; 22Lindholm et al., 1999; 23Giovannardi et al., 1999; 24Albay et al., 2003; 25Fischer & Dietrich, 2000; 26Zimba et al., 2001; 27Albay et al., 2004; 28Jewel et al., 2003; 29Pollux & Pollux, 2004; 30Ibelings et al., 2005.

33 1. INTRODUCTION ______

As fish is an important food source, detrimental effects such as those induced by toxic cyanobacteria may not only cause significant economic losses in aquaculture (Andersen et al., 1993; Kent, 1990; Toranzo et al., 1990; Zimba et al., 2001), but may also represent a possible route for human exposure (Deblois et al., Toxicon in press; Jewel et al., 2003; Magalhaes et al., 2003; Magalhaes et al., 2001; Soares et al., 2004; Xie et al., 2005) and result in sustained effects on the nutrition of regional populations. Thus detailed investigations on the various and multiple effects of cyanobacterial toxins on fish are essential.

THE ICHTHYOTOXICITY OF MICROCYSTIN

The median lethal doses (LD50) for microcystins (predominantly MC-LR) have been determined for diverse fish species by different forms of toxin application (Tab. 1.3). Fish species differ in their sensitivity to microcystin, e.g. cyprinids have been shown to be up to 50-times more sensitive than trout (Andersen et al., 1993; Fischer & Dietrich, 2000; Tencalla, 1995). Differences within the same species emphasise an influence of the nutritional and physiological condition on microcystin toxicity to exposed fish (Carbis et al., 1996a; Malbrouck et al., 2004b; Råbergh et al., 1991). Although less sensitive to microcystin via the intraperitoneal route, fish are more sensitive than mice to orally applied microcystin, which suggests fish to be particularly susceptible to the effects of oral ingestion of microcystin and microcystin containing cyanobacteria.

Uptake of Microcystin Although microcystin containing cyanobacterial blooms have unambiguously been associated with numerous fish kills, the microcystin uptake of fish has not been clarified in detail so far. Indeed, pathological alterations in moribund and dead fish resemble those of fish treated with microcystin (Andersen et al., 1993; Rodger et al., 1994; Toranzo et al., 1990; Zimba et al., 2001). However environmentally detected concentrations of microcystin and cyanobacterial crude extracts have not yet been demonstrated to cause acute fish mortality (Bury et al., 1995; Carbis et al., 1996a; Phillips et al., 1985; Tencalla et al., 1994).

Tab. 1.3: The median lethal dose (LD50) of microcystin-LR in various fish species and mouse

INTRAPERITONEAL [µG MC-LR/KG BW] ORAL APPLICATION APPLICATION

Trout 1700-6600 1 400-1000 2 (Oncorhynchus mykiss) Carp <1700 6 300-550 3 (Cyprinus carpio) 20-50 4 Loach *138 5 (Misgurnus mizolepis)

Mouse ≥5000 7, 8 50 9 (Mus sp.)

* application of MC-LRequiv. (primarily containing MC-RR)

REFERENCES: 1Tencalla et al., 1994; 2Kotak et al., 1996; 3Råbergh et al., 1991; 4Carbis et al., 1996a; 5Li et al., 2005; 6Tencalla, 1995; 7Yoshida et al., 1997; 8Fawell et al., 1999; 9Krishnamurthy et al., 1986.

34 1. INTRODUCTION ______

Principally, there are a variety of possible routes for the uptake of microcystins. Dissolved microcystins might be absorbed from water, via the gills, the intestine or the integument and fish might further be intoxicated due to the digestion of ingested cyanobacterial cells, contaminated food and subsequent microcystin release within the intestine. As deduced from various experimental microcystin exposure of trout, the most likely uptake route is the oral ingestion of either dissolved microcystin and/or microcystin containing cells with a consequent toxin uptake via the digestive system (Bury et al., 1995; Tencalla et al., 1994). Therefore, fish filtering water for food (primarily herbivorous and planktivorous species) appear to run a much higher risk of ingesting toxic cyanobacteria than species that take food selectively (e.g. most omnivorous and piscivorous species), resulting in a species-specific susceptibility to microcystin (Adamovsky et al., 2007; Chen et al., 2007). Xi et al. (2005) detected microcystins in tissue but not in gut content of fish naturally exposed to toxic M. aeruginosa, concluding, that microcystins might moreover be intoxicated via ingestion and digestion of food organisms containing accumulated toxin. However, these observations might also be attributed to far earlier ingestion of cyanobacteria, whereas cyanobacterial cells and toxins were no longer determinable in gut contents because they had already been digested and/or excreted (Soares et al., 2004). Digestion of cyanobacterial cells in the intestine of fish varies within different cyanobacterial species. Cells of various Aphanizomenon sp. for example have been shown to be almost totally decomposed during fish digestion while others, predominantly species comprising cell walls including additional exopolysaccharides (e.g. Microcystis sp.), were barely digested (Carbis et al., 1997; Gavel et al., 2004; Kamjunke et al., 2002a; Kamjunke et al., 2002b; Lewin et al., 2003). However, even cell walls of obviously intact cells have been shown to become permeable throughout the gut passage and may thus allow toxin release within the digestive tract (Moore & Scott, 1985).

As microcystins are highly resistant to acidic and enzymatic degeneration, once released into the fish intestine microcystin degradation is assumed to occur slowly, resulting in high microcystin availability. It is suggested, that microcystins are absorbed across the epithelium of the ileum and thus reach the venous bloodstream of fish via bile acid membrane transporters (Boaru et al., 2006; Fischer & Dietrich, 2000; Meier-Abt et al., 2007). Bury et al. (1998b) conclude from studies on the uptake of radiolabeled microcystin in trout, that microcystins may partly pass the ileal epithelium via passive transport mechanisms. Nevertheless, the ileal epithelium provides a barrier for the transfer of microcystin into the bloodstream, for which reason a substantial proportion of microcystin appears to remain in faeces (Tencalla & Dietrich, 1997; Xie et al., 2004). Indeed less than 5% of microcystin orally applied to trout (Oncorhynchus mykiss) has been shown to enter the bloodstream (Bury et al., 1998b; Tencalla & Dietrich, 1997), and also salmon (Salmo salar) intraperitoneally exposed to microcystins incorporated only 38-62 % of the applied microcystin into their tissues (Williams et al., 1997a; Williams et al., 1995). The proportion of

35 1. INTRODUCTION ______microcystin absorbed into blood has been shown to depend on time, dose, route of exposure and on the chemical properties of the specific microcystin congeners (Bury et al., 1998b; Fischer & Dietrich, 2000; Tencalla, 1995; Williams et al., 1997a; Williams et al., 1995; Xie et al., 2004). Ileal absorption of microcystin is assumed to be affected by the constitution and length of the species intestine. Trout possessing a short, thick-walled gut have been shown to be much less sensitive to microcystin than cyprinids which possess a comparably long intestine (Fischer & Dietrich, 2000; Tencalla, 1995). Differences in the microcystin sensitivity of diverse fish species can, besides the species foraging strategy, also be ascribed to differing organ morphologies, in particular the intestine.

Transport and Organotropism of Microcystins Having traversed the epithelial barrier, microcystin accumulates primarily in the liver of exposed fish following the portal venous bloodstream and distribution of OATPs (Cazenave et al., 2005; Fischer & Dietrich, 2000; Soares et al., 2004; Williams et al., 1997a; Williams et al., 1995). Microcystins have been shown to accumulate in liver of carp (Cyprinus carpio) and trout orally exposed to M. aeruginosa as of 1 h post application (Fischer & Dietrich, 2000, Tencalla et al., 1997). Maximum hepatic microcystin accumulations were usually determined within 3-8 h post application subsequent to intraperitoneal and oral exposure of both, cyprinids and trout (Fischer & Dietrich, 2000, Malbrouck et al., 2003; Tencalla & Dietrich, 1997; Williams et al., 1995, Williams et al., 1997a). Smaller amounts have also been shown to accumulate in kidney, blood, gill, bile, intestine and brain while only minimal amounts have been detected in muscle tissue (Cazenave et al., 2005; Fischer & Dietrich, 2000; Williams et al., 1995; Xie et al., 2005). It is yet not clear, whether microcystins reach those organs directly, via the bloodstream bypassing the hepatopancreas, or downstream of the hepatopancreas subsequent to an overload of presystemic hepatic elimination11 capacities (Carbis et al., 1996b; Fischer & Dietrich, 2000).

Biochemical Changes as Indicators of Microcystin Toxicity in Fish Upon reaching the metabolism of an exposed organism, the toxicity of a given contaminant is principally based on molecular alterations. Only the interaction(s) of a toxin with receptors or other target molecules, their inhibition or activation initialise physiological, pathological and other functional alterations observable and measurable in exposed organisms (Hofer & Lackner, 1995; Schlenk & Di Giulio, 2002). In this respect, the ichthyotoxicity of microcystin is mainly mediated by four basic molecular effects: inhibition of protein phosphatases (PPs), an inhibition of ATPases, an elevated formation of reactive oxygen species (i.e. oxidative stress) and effects on detoxification via glutathione conjugation.

11 First pass elimination of toxins in liver, thus impeding/restricting detrimental effects in downstream organs 36 1. INTRODUCTION ______

PP-INHIBITION: Also in fish, PPs appear the main target molecules of microcystin (Råbergh et al., 1991; Tencalla & Dietrich, 1997) and their inhibition causes substantial disturbance of various elementary metabolic processes (Mezhoud et al., Aquatic Toxicology in press; see also chapter 1.2). The timecourse and extent of PP-inhibition has been investigated in several studies including various fish species. Tencalla & Dietrich (1997) demonstrated that hepatic PP-activity in trout is reduced to 60% within 1 h and to 100% within 3 h post oral application of toxic

M. aeruginosa (equiv. to 5.7 mg MC-LRequiv./kg bw). Also in carp, the hepatic PP-activity totally disappeared within 3 h post gavage of M. aeruginosa (equiv. to 1.8 mg MC-LRequiv./kg bw) (Tencalla, 1995). The hepatic PP-activity in catfish (Ictalurus punctatus) decreased to 66-81% 24 h post intraperitoneal application of 0.5-2 mg MC-LR/kg bw and was also shown to decrease significantly following immersion in water containing microcystin (≥1 mg MC-LR/l) (Snyder et al., 2002).

The IC50 of MC-LR induced PP-inhibition in liver and kidney homogenates of diverse cyprinid species has been determined to be 0.1-0.3 nM (Tencalla, 1995; Xu et al., 2000), those in liver homogenates of trout to 0.05-0.2 nM (Sahin et al., 1995; Tencalla & Dietrich, 1997) thus indicating low inter-species variation. However, the extent of PP-inhibition has been shown to depend on the nutritional and physiological status of exposed fish (Malbrouck et al., 2004a; Malbrouck et al., 2004b) and also on the characteristics of specific congeners. For example, MC-RR has been identified as a weaker PP-inhibitor than MC-LR and MC-YR (Xu et al., 2000). Xu et al. (2000) moreover demonstrated differences within the inhibition of various PPs, showing the inhibition of PP1 (IC50 = 0.9-3.6 nM) isolated from grass carp (Ctenopharyngidin idellus) to be at least three-times less effective than inhibition of isolated PP2A (IC50 = 0.3-0.6 nM). This is in accordance with data obtained in mammals where IC50s for PP-inhibition of 1.7 and 0.04 nM were reported for PP1 and PP2A, respectively (Honkanen et al., 1990). Xu et al. (2000) further suggests that PP2A might be the principal microcystin sensitive enzyme in fish tissues, as the IC50 of MC-LR induced PP-inhibition in crude tissue homogenates was similar to those determined for isolated PP2A.

ATPASE-INHIBITION: Studies investigating the effect of M. aeruginosa extracts on gill fractions derived from carp and tilapia (Oreochromis sp.) suggest cyanobacterial-induced inhibitory effects on the activity of Na+/K+ ATPases in fish gill (Bury et al., 1998a; Bury et al., 1996b; Gaete et al., 1994; Zambrano & Canelo, 1996). ATPases are important enzymes, carrying various ions across cell membranes. Thus ATPase-inhibition in gill of fish exposed to cyanobacteria is assumed to disturb the transport of ions across the gill epithelium resulting in a sustained ionic imbalance and consequently gill dysfunction. While Gaete et al. (1994) and Zambrano & Canelo (1996) attribute the observed ATPase inhibition primarily to microcystin, Bury et al. (1998a & 1996b) regard the epithelial ATPase inhibition to result from bioactive metabolites other than microcystins (e.g. fatty acids). Malbrouck et al. (2003) determined neither ionic imbalance nor detrimental effects on the activity of gill Na+/K+ ATPases in goldfish (Carassius auratus)

37 1. INTRODUCTION ______intraperitoneally exposed to microcystin (125 µg MC-LR/kg bw), thus corroborating the latter point of view and revealing the need for further research.

ELEVATED ROS FORMATION (OXIDATIVE STRESS): The biotransformation of contaminants in fish often results in an elevated formation of reactive oxygen species (ROS) (Schlenk & Di Giulio, 2002). This also applies to microcystin (Bláha et al., 2004; Jos et al., 2005; Li et al., 2003; Li et al., 2005; Prieto et al., 2006) and is possibly promoted by a phosphorylation imbalance as a consequence of PP-inhibition (Li et al., 2003). ROS increase may overload the ROS- complementary antioxidative systems including both enzyme (e.g. superperoxiddismutase, glutathionperoxidase, catalase, etc.) and compound-based response (e.g. glutathione, vitamin E & C, etc.) and thus result in an ROS surplus (Hofer & Lackner, 1995; Schlenk & Di Giulio, 2002). As ROS can oxidise intracellular molecules (causing enzyme inhibition, lipid peroxidation, DNA damage, etc.), elevated ROS levels cause severe cellular damage (i.e. oxidative stress) and may finally culminate in cell death (Hofer & Lackner, 1995; Zhang et al., 2007). Microcystin-mediated oxidative stress in fish has been confirmed by three approaches: (i) via measurement of ROS levels, (ii) via alterations in components of the antioxidative systems, and (iii) via an increase of products resulting from lipidperoxidation. In vitro experiments on isolated hepatocytes and lymphocytes demonstrated an increase of ROS concentrations after 15 min and a doubling of ROS concentrations within 2 h of exposure to 10 µg MC-LR/l (Li et al., 2003; Zhang et al., 2007). This, despite a simultaneous activation of several ROS-eliminating enzymes (Li et al., 2003). A microcystin-mediated activation of antioxidative enzymes has also been determined in vivo (Jos et al., 2005; Li et al., 2005; Prieto et al., 2006), following intraperitoneal application of microcystins (500 µg MC-LRequiv./kg bw) as well as chronic gavage of microcystins and toxic

Microcystis to tilapia (equiv. to 1.2 mg MC-LRequiv./kg bw for up to 21 days) and loach (Misgurnus mizolepis; equiv. to 10 µg MC-LRequiv./kg bw for 28 days). Those studies suggest the antioxidative enzyme response to be time-, dose- and tissue-dependent. It is thus not surprising that antioxidant enzymes appear able to eliminate oxidative stress induced by low microcystin concentrations (Li et al., 2005), whereas very high microcystin concentrations (≥500 µg MC-

LRequiv./kg bw) have been demonstrated to induce oxidative stress that results in tissue damage, e.g. lipid peroxidation (Jos et al., 2005; Prieto et al., 2006).

GLUTATHIONE LEVEL AND GLUTATHIONE S-TRANSFERASE AKTIVITY: In contrast to the enzymatic ROS response, the antioxidative response of glutathione is not consistent. While Li et al. (2003) observed a microcystin-induced decrease of the glutathione concentration in carp hepatocytes exposed to microcystin (10 µg MC-LR/l ≈ 10 nM MC-LR), Adamovsky et al. (2007) and Bláha et al. (2004) noted an ambiguous response of glutathione levels in carp and silver carp (Hypophtalmichthys molitrix) exposed to microcystin containing cyanobacterial blooms. Glutathione acts as a co-substrate for glutathione peroxidase and glutathione S-transferase and is thus important for both the detoxification of ROS and of microcystin itself (Wiegand &

38 1. INTRODUCTION ______

Pflugmacher, 2005). Indeed, microcystin-glutathione conjugates have been demonstrated to form within 30 min, via covalent binding to Mdha, the moiety that also binds to PPs (Pflugmacher et al., 1998). However, if a cellular glutathione concentration in the millimolar range is assumed (≈ 5 mM; see Stryer, 1996), significant effects on the glutathione concentration due to binding of microcystin appear unlikely at intracellular microcystin concentrations in the nanomolar range. It rather appears plausible that the observed response in glutathione levels largely results from secondary effects, i.e. binding of degradation products due to microcystin-induced cell damage. As for the conjugation of other toxic metabolites, the formation of microcystin-glutathione conjugates in fish is catalysed by glutathione S-transferase (GST) (Pflugmacher et al., 1998). Increasing GST-activities thus usually indicate an adaptation of the organism to exposure to certain contaminants (Bláha et al., 2004; Hofer & Lackner, 1995). Indeed, Wang et al. (2006) demonstrated a significant rise in the expression level of GST mRNA from tilapia subsequent to a single intraperitoneal dose of 50 µg MC-LR/kg bw. Several other studies have also shown an increase in GST activity in fish treated with microcystin or microcystin containing cyanobacteria (Adamovsky et al., 2007; Cazenave et al., 2006b; Pietsch et al., 2001; Qiu et al., 2007; Wiegand et al., 1999). In contrast, others have demonstrated microcystin to suppress GST activity (Cazenave et al., 2006a; Malbrouck et al., 2003; Pflugmacher et al., 1998) or to have no effect on GST activity at all (Malbrouck et al., 2004b). Recently Cazenave et al. (Water Research, in press) observed a decrease in GST activity in the Rio de la Plata livebearer Jenynsia multidentata when fed with 0.1 µg MC-LR/kg bw and an increase in GST activity when fed with 1 µg MC-LR/kg bw thus indicating the microcystin- induced GST response to be dose-dependent. Moreover, Qiu et al. (2007) attribute species-specific differences in microcystin susceptibility to variances in detoxification capacity. Indeed, the microcystin-induced GST response has been shown to vary between different GST isoforms and even within the same GST isoform isolated from different fish species (Fu & Xie, 2006; Liang et al., 2007). Therefore, the variability in GST response may predominantly originate from variations in microcystin dosage and from fish size and species, as already characterised for other ichthyotoxic contaminants (Egaas et al., 1999). In addition, GST has been shown to respond to compounds other than microcystin (e.g. lipopolysaccharids, natural organic matter, etc.) further implying a possible modulation of the microcystin-induced GST response (Best et al., 2002; Cazenave et al., 2006b; Palikova et al., 2007; Pietsch et al., 2001, Wang et al., 2006).

Physiological and Condition-related Indicators of Microcystin Toxicity in Fish The described microcystin mediated biochemical responses result in distinct and measurable changes of physiological parameters up to and including fish growth and fitness, enabling an additional characterisation of the microcystin ichthyotoxicity.

CARDIOVASCULAR EFFECTS: Best et al. (2001) documented significant changes in the cardiac function of brown trout (Salmo trutta) larvae exposed to microcystin and aqueous extracts of

39 1. INTRODUCTION ______

M. aeruginosa (equiv. to 5-500 µg MC-LRequiv./l). The larvae showed elevated heart rates (chronotropism), increased stroke volumes (inotropism) and a significant increase in cardiac output. Those cardiovascular alterations are regarded to reflect a stress response (Best et al., 2001) which might result from a stress-mediated elevation of energy demands (Hofer & Lackner, 1995). However, not only pure microcystin, but also the cyanobacterial extracts (independent of microcystin content) resulted in significant cardiovascular alterations indicating that predominantly other cyanobacterial metabolites than microcystins (e.g. cyanobacterial LPS, fatty acids) affect cardiovascular functions (Best et al., 2001).

CORTISOL LEVELS: Immersion of brown trout in lysates of toxic M. aeruginosa (24-42 µg MC-LR/l) has been shown to cause a significant increase in plasma cortisol levels (Bury et al., 1996a). Cortisol is the main steroid hormone in fish with multiple functions in liver metabolism, osmoregulation and in the immunosystem. Cortisol release is influenced by several stressors and as such alterations in fish blood cortisol levels are considered as a sensitive indicator of stress (summarised in Barton et al., 2002). Hence, the elevation in blood cortisol levels of exposed trout, reveal microcystin as an unambiguous stressor in fish metabolism (Bury et al., 1996a).

GLYCOGEN LEVELS: Exposure of fish to either microcystin or microcystin-containing extract have further been shown to result in a decrease in hepatic glycogen levels (Huynh-Delerme et al., 2005; Lecoz et al., 2008; Malbrouck et al., 2004a; Malbrouck et al., 2004b; Råbergh et al., 1991) and a subsequent increase in blood glucose concentration (Bury et al., 1996a). These are either ascribed to a catecholamine- and cortisol-stimulated mobilisation of energy and thus stress mediated energy demand (Barton et al., 2002; Bury et al., 1996a), and/or consequential to a disrupted glycogen homeostasis resulting from PP-inhibition and thus hyperphosphorylation of phosphorylase-a (Råbergh et al., 1991). A prolonged utilisation of glycogen may result in a depletion of energy stores and consequently reduced fish growth and fitness (Bury et al., 1996a).

FISH FITNESS AND FISH GROWTH: Few investigations describe decreasing condition and growth in carp and trout chronically exposed to microcystins, either via repeated application of contaminated food or immersion in water containing microcystin (Bury et al., 1995; Li et al., 2004). Reduced fish condition and growth reveal both physiological stress and pathological alterations and can lead to a suppression of the immunosystem and thus to an elevated susceptibility to infections and toxins (Barton et al., 2002). This confirms investigations which suggest that fish may not only be directly affected via microcystin but also via an elevated vulnerability to other detrimental influences (Ibelings et al., 2005; Kent et al. 1988; Zhao et al., 2006).

ION HOMEOSTASIS: Microcystins applied intraperitoneally to carp and silver carp have been shown to cause a decrease in total plasma protein (TPP) levels (Navratil et al., 1998; Vajcová et al., 1998). A comparable TPP decrease was demonstrated by Kopp & Hetesa (2000), in common

40 1. INTRODUCTION ______carp following exposure to various cyanobacterial crude extracts containing microcystins. According to Barton et al. (2002), changes in the TPP concentration may indicate osmoregulatory dysfunction. Hence the observed decreases in TPP levels suggest microcystin-mediated effects on the ion homeostasis, which is typical for detrimental effects of various contaminants on freshwater fish (Hofer & Lackner, 1995). Indeed, investigations in carp and trout exposed to microcystin-containing extract of M. aeruginosa present significant alterations in serum and plasma ion concentrations, i.e. decreasing sodium and chloride levels (Bury et al., 1996a; Carbis et al., 1996b; Carbis et al., 1997), thus confirming a possible microcystin-mediated influence on ion homeostasis. The observed ionic imbalance has been ascribed to a microcystin-mediated catecholamine release increasing the permeability of gill and consequently raising diffusional loss of ions (Bury et al., 1996a). Observed effects have further been attributed to a microcystin mediated inhibition of Na+/K+-ATPase activity (ion pumps) and thus ion transport in gill (Carbis et al., 1996b; Gaete et al., 1994; Zambrano & Canelo, 1996; see also above). However, the observed ionic imbalance has to be mostly attributed to cyanobacterial components other than microcystin, as pure microcystin caused neither alterations in plasma osmolality nor an inhibition of Na+/K+-ATPase activity, regardless whether intraperitoneally applied to goldfish or in immersed tilapia (Bury et al., 1996b; Malbrouck et al., 2003). Definitive conclusions on the impact of microcystin on ion homeostasis require further studies. Nevertheless the ionic imbalance induced by cyanobacteria may disturb physiological balance and thus impact elementary functions in fish metabolism, e.g. maintenance of membrane potentials, transmission of nerve impulses, etc. (Penzlin, 2002).

BLOOD PARAMETERS: Carp and silver carp orally and intraperitoneally exposed to microcystin containing cyanobacteria and pure microcystin, respectively showed a marked decrease in leucocyte counts (Palikova et al., 1998). Palikova et al. (1998) primarily ascribe this to a decrease of lymphocytes and neutrophil myelocytes corresponding with a decrease of phagocyte activity and thus implicating microcystin mediated immunosuppressive effects in specific immune response. This is in accordance with Zhang et al. (2007 & 2006) who demonstrated time- and dose-dependent apoptotic cell death in isolated carp lymphocytes following exposure to 1 nM MC-LR. Also lymphocytes and phagocytes isolated from trout displayed a dose- and time- dependent decrease in viability when exposed to microcystin (≥1 µM MC-LR). Dose-dependent effects on lymphocytes proliferation have also been reported (Rymuszka et al., 2007; Sieroslawska et al., 2007). These immunosuppressive effects appear additionally influenced by cyanobacterial components other than microcystin, as they are far more pronounced in fish exposed to cyanobacterial lysates than to pure microcystin (Palikova et al., 1998). This is corroborated by the findings of Kopp & Hetesa (2000), who observed a significant decrease in the leucocrit of carp exposed to cyanobacteria containing no microcystin. Silver carp exposed intraperitoneally to pure microcystin further displayed decreasing haemoglobin and haematocrit values (Vajcová et al., 1998). According to Barton et al. (2002) declining haemoglobin concentrations and haematocrit values may indicate a stress response

41 1. INTRODUCTION ______possibly concluding in anaemia and thus decreased oxygen binding capacities. As implicated by a comparative study on carp investigating the impact of toxic cyanobacteria and pure microcystin, the decline of haemoglobin and haematocrit cannot with certainty be ascribed to microcystin mediated effects (Navratil et al., 1998). In contrast to this, fish treated with microcystin respond with an unambiguous dose-dependent increase in the plasma activity of intracellular enzymes including lactatedehydrogenase (LDH), alaninaminotransferase (ALT), aspartateaminotransferase (AST). This applies to both, experimentally exposure to either microcystin or microcystin containing cyanobacteria (Bury et al., 1997; Carbis et al., 1996b; Kopp & Hetesa, 2000; Li et al., 2004; Malbrouck et al., 2003; Navratil et al., 1998; Råbergh et al., 1991; Sahin et al., 1995; Tencalla et al., 1994; Vajcová et al., 1998) as well as environmental exposure to toxic M. aeruginosa (Carbis et al., 1997). The elevation of plasma LDH, AST and ALT activities is thought to result from an enzyme release subsequent to cellular injury, hence indicating severe tissue (especially liver) damage. Indeed, enzyme activities increased as of 2-3 h post application, are accompanied by incipient liver pathology (Råbergh et al., 1991; Tencalla et al., 1994). Increased serum bile and bilirubin concentrations as observed in carp exposed to toxic M. aeruginosa reveal this damage to result in a sustained restriction of liver function (Carbis et al., 1996b; Carbis et al., 1997).

Cytological and Pathological Indicators of Microcystin Toxicity in Fish The knowledge on the ichthyotoxicity of microcystins is largely based on pathological and cytological studies in carp, trout and other fish species (e.g. zebrafish, catfish, goldfish, loach, salmon and killifish) as well as in isolated cells. These studies include experimental immersion, intraperitoneal and oral application of either pure microcystin or microcystin containing extracts, cell lysates and intact cells and predominantly correspond to acutely toxic microcystin concentrations. Only a few studies have investigated subchronic or chronic pathology of microcystin in fish (Li et al., 2004; Qiu et al., 2007). The experimental findings are complemented by few investigations on environmentally exposed fish (Andersen et al., 1993; Carbis et al., 1997; Qiu et al., 2007). Allowing for species-specific differences in sensitivity as well as variations in dose, duration and route of exposure, these investigations reveal that the ichthyopathology of microcystin proceeds similarly in all cases.

CLINICAL SYMPTOMS AND PHYSICAL CHANGES: Physically, microcystin exposed fish respond primarily with an unspecific disturbance of pigmentation concluding in prominent darkening comparable to their physical response to various other toxins and pathogens (Carbis et al., 1996a; Kotak et al., 1996; Phillips et al., 1985). Microcystin exposed fish moreover present time- and dose-dependent behavioural changes initially including increased activity and hyperventilation (Baganz et al., 2004; Baganz et al., 1998; Carbis et al., 1996a; Cazenave et al., Water Research in press; Phillips et al., 1985) and switching to lethargy and a slowdown of ventilation and daytime motility with continuous and

42 1. INTRODUCTION ______progressive exposure (intraperitoneal and oral microcystin application) (Baganz et al., 2004; Baganz et al., 1998; Kotak et al., 1996; Phillips et al., 1985; Tencalla et al., 1994). Increasing severity has further been shown to provoke gaping at the water surface, disturbed swimming behaviour and loss of balance concluding in lethargy interrupted by pulses of frenetic swimming (Carbis et al., 1996a; Kotak et al., 1996; Phillips et al., 1985; Råbergh et al., 1991).

LIVER: In accordance with the organ distribution of microcystin and comparable to toxicity in mammals, severe pathology is particularly observed in the hepatopancreas of fish (Råbergh et al., 1991; Snyder et al., 2002; Sugaya et al., 1990; Tencalla et al., 1994). Morphological changes mostly include discolouration of the liver, to a bright yellow and sometimes even spongy appearance, as well as distinct liver enlargement (Fischer & Dietrich, 2000; Kotak et al., 1996; Råbergh et al., 1991; Sugaya et al., 1990; Tencalla et al., 1994; Vajcová et al., 1998). The latter may be associated with a significant increase of the hepatosomatic index (HSI) (Bury et al., 1997; Kotak et al., 1996). Histopathology reveals a loss of cell-cell contacts, isolation of hepatocytes, engorgement of hepatic sinusoids and vessel degeneration (Andersen et al., 1993; Bury et al., 1997; Carbis et al., 1996a; Fischer & Dietrich, 2000; Liu et al., 2002; Malbrouck et al., 2003; Phillips et al., 1985; Råbergh et al., 1991; Rodger et al., 1994; Snyder et al., 2002; Tencalla & Dietrich, 1997; Tencalla et al., 1994; Zimba et al., 2001). In contrast to mammals, only few studies have demonstrated microcystin- mediated haemorrhage in fish (Sugaya et al., 1990; Tencalla & Dietrich, 1997; Tencalla et al., 1994). It is thus assumed that the abnormal liver appearance and liver enlargement observed in microcystin exposed fish is primarily a result of massive hydropic degeneration of hepatocytes (Kotak et al., 1996). Indeed, almost all histopathological investigations on the ichthyotoxicity of microcystin reveal widespread hepatocyte pyknosis, vacuolisation, swelling (megalocystis) and sustained hepatocyte necrosis. These are often accompanied by an infiltration of leucocytes and macrophages, suggesting an associated inflammatory response (Andersen et al., 1993; Bury et al., 1997; Carbis et al., 1996a; Carbis et al., 1997; Fischer & Dietrich, 2000; Fournie & Courtney, 2002; Kotak et al., 1996; Li et al., 2001; Li et al., 2007; Malbrouck et al., 2003; Palikova et al., 2004; Phillips et al., 1985; Pichardo et al., 2007; Råbergh et al., 1991; Rodger et al., 1994; Sugaya et al., 1990; Tencalla & Dietrich, 1997; Tencalla et al., 1994; Vajcová et al., 1998; Zimba et al., 2001). Several pathological and cytological investigations have described microcystin-induced apoptosis, as indicated by hepatocyte shrinkage (atrophy), the concentration and margination of chromatin, karyorrhexis, karyolyses and membrane blebbing (Carbis et al., 1996a; Carbis et al., 1997; Fischer et al., 2000; Fischer & Dietrich, 2000; Fladmark et al., 1998; Li et al., 2007; Liu et al., 2002; Malbrouck et al., 2003; Palikova et al., 2004; Pichardo et al., 2005; Pichardo et al., 2007). Studies on isolated carp hepatocytes demonstrated significant signs of apoptosis and thus cytotoxicity at ≥10 µg MC-LR/l (Li et al., 2007). Cytological investigations suggest that the form of cell death depends on the duration and dose of microcystin exposure as hepatocytes incubated in

43 1. INTRODUCTION ______microcystin concentrations ≤100 µg MC-LRequiv./l primarily responded with apoptotic and hepatocytes incubated in microcystin concentrations ≥500 µg MC-LRequiv./l usually with necrotic cell death (Fischer et al., 2000; Li et al., 2001; Li et al., 2007; Pichardo et al., 2007). Ultrastructural investigations on hepatocytes and liver sections of microcystin exposed fish identify cytoskeletal breakdown, lysis of cell membrane, nuclear degeneration, as well as swelling, fragmentation and vesiculation of endomembrane systems (i.e. golgi body, mitochondria and the endoplasmic reticulum) to be the main cause of liver pathology, primarily attributable to the various microcystin mediated effects on cell functions resulting from PP-inhibition (Kotak et al., 1996; Li et al., 2001; Li et al., 2004; Qiu et al., 2007; Råbergh et al., 1991; Tencalla & Dietrich, 1997). The liver pathology described has been shown to occur in a time- and dose-dependent manner (Tab. 1.4). Carp intraperitoneally exposed to microcystin presented with ultrastructural liver alterations as of 30 min post application (Råbergh et al., 1991). Initial pathology is predominantly diffuse, progressing to increased perivascular accumulation of pathological alterations in larger areas of tissue and occasionally the entire organ (Fischer et al., 2000; Fischer & Dietrich, 2000; Kotak et al., 1996; Malbrouck et al., 2003; Tencalla & Dietrich, 1997). In conclusion the massive hepatocyte degeneration is thought to cause severe liver malfunction and even total collapse of liver functions. The root cause of microcystin induced fish death is thus liver failure rather than hypovolumic shock (Fournie & Courtney, 2002; Kotak et al., 1996; Li et al., 2004; Råbergh et al., 1991).

KIDNEY: In addition to the described liver pathology, some investigations on microcystin-exposed fish demonstrate a distinct impact on renal cells and tissue. The renal pathology as induced by microcystin appears generally restricted to the proxima in the posterior part of the kidney (Fischer & Dietrich, 2000; Kotak et al., 1996). The observed alterations include glomerullar alterations (e.g. mesangial swelling, expanding of capillaries and widening of Bowman’s space), as well as degeneration of the tubular lining (e.g. vacuolisation, pyknosis and necrosis in tubular epithelial cells) resulting in an exfoliation of cell fragments and thus protein deposition in the tubular lumina (Carbis et al., 1996a; Fischer & Dietrich, 2000; Kotak et al., 1996; Råbergh et al.,

Tab. 1.4: Lowest observed microcystin dosages causing significant pathology in liver and kidney of fish (i.e. carp1&3 and catfish2), following different form of exposure

Intraperitoneal application Oral application Immersion [µg MC-LR/kg bw] [µg MC-LR/kg bw] [mg MC-LR/l]

Liver 2.51 2501 12 Kidney 251 4003 1.71

REFERENCES: 1Carbis et al., 1996a; 2Snyder et al., 2002; 3Fischer & Dietrich, 2000.

44 1. INTRODUCTION ______

1991; Vajcová et al., 1998). Effects have also been shown on interstitial tissue including congestion, melanisation and the formation of oedema (Råbergh et al., 1991). Carp orally exposed to toxic M. aeruginosa extract (equiv. to 400 µg MC-LRequiv./kg bw) displayed renal alterations as of 1-3 h post gavage gradually extending to widespread pathology involving almost the entire organ within 24 h (Fischer & Dietrich, 2000). In contrast to this, Carbis et al. (1996a) estimated the proportion of damaged tubuli in affected renal tissue of intraperitoneally exposed carp

(25-50 µg MC-LRequiv./kg bw) to be 2-5%. This suggests that comparable to liver, also the severity and progress of microcystin mediated renal pathology occurs time- and dose-dependently and furthermore is dependent on the route of toxin application (Tab. 1.4).

GILL: Fish gills have been shown to be affected by microcystin containing cyanobacteria via two different ways: An irritation and mechanically-induced epithelial abrasion resulting from cyanobacterial cells and filaments ensnarling within gill lamellae (Rodger et al., 1994; Toranzo et al., 1990) and/or microcystin-induced toxicity. Fish gills are known to represent a main entry route for many contaminants and in light of this should also be considered as an exposure route for microcystins (Cazenave et al., 2005; Hofer & Lackner, 1995). Most authors however regard the outer gill epithelium as a largely insurmountable barrier for microcystins, primarily based on the size and hydrophilic character of most microcystin congeners. This links gill toxicity primarily to oral ingested microcystin (Bury et al., 1995; Cazenave et al., 2005; Tencalla et al., 1994). Indeed, fish intraperitoneally and orally exposed to microcystin presented with distinct gill pathology corroborating the entry of microcystin into gills via the arterial bloodstream (Carbis et al., 1996a; Sugaya et al., 1990). Fish exposed to either microcystin or microcystin-containing cyanobacteria have been shown to respond with enhanced mucus excretion of the gill epithelium and increased swelling and clubbing (i.e. distal lamella hyperplasia) of the secondary lamellae as a consequence of epithelial cell hypertrophy (Carbis et al., 1996a; Carbis et al., 1997; Rodger et al., 1994). Progressive effects result in epithelial pyknosis and comprehensive epithelial necrosis which may result in a separation of the epithelium from the lamellar capillaries (Carbis et al., 1996a; Carbis et al., 1997; Rodger et al., 1994; Toranzo et al., 1990). Sugaya et al. (1990) further recognised anaemia in gill of goldfish intraperitoneally exposed to microcystin. Studies on the time- and dose-dependence of microcystin-induced gill pathology are rare. Roger et al. (1994) described distinct gill lesions in dead trout from a fish kill associated with a bloom of microcystin containing A. flos-aquae, containing approximately 16-19 µg MC-LRequiv./l. The only experimental investigation on the pathological efficiency of various microcystin dosage on gills, was carried out in common carp and resulted in a lowest observed effective dose of 2.5 µg

MC-LRequiv./kg bw and 250 µg MC-LRequiv./kg bw for intraperitoneal and oral microcystin application, respectively (Carbis et al., 1996a). Carbis et al. (1996a) additionally observed distinct pathological alterations in gill tissue of carp exposed to 1.7 mg MC-LRequiv./l after just six days of exposure and slight changes within the first days of exposure. In summary gill anaemia and the

45 1. INTRODUCTION ______observed lesions are assumed to reduce oxygen uptake and thus to exacerbate or even cause anoxia in fish. This is particularly critical when oxygen depletion resulting from senescent cyanobacterial blooms is considered (Carbis et al., 1996a).

OTHER AFFECTED ORGANS: Besides liver-, kidney- and gill pathology, exposure of catfish, trout and goldfish to toxic M. aeruginosa and microcystin has sporadically been shown to cause swelling and congestion of the spleen (Phillips et al., 1985; Sugaya et al., 1990; Zimba et al., 2001). Phillips et al. (1985) furthermore observed trout, intraperitoneally exposed to 30 mg M. aeruginosa/kg bw, to respond with various neuronal effects, e.g. meningeale oedema, congestion of meningeale and perispinal capillaries as well as scattered neuronal necrosis in the cerebellar and visual area of brain. Fischer & Dietrich (2000) moreover observed considerable pathology in the gastrointestinal tract of carp orally exposed to M. aeruginosa (equiv. to. 400 µg

MC-LRequiv./kg bw), including increased intestinal pyknosis and apoptosis, as well as intestinal haemorrhages and exfoliation of epithelial cells into the gut lumen. However, severity and thus impact of those pathological alterations have generally been ranked behind the consequence of gill, kidney and liver pathology.

Tissue Regeneration and Biotransformation of Ingested Microcystin in Fish The recovery from non-lethal microcystin toxicoses implies that fish can detoxify and metabolise incorporated microcystins. Indeed, Williams et al. (1995) observed a continuous clearance of tritium labelled microcystin in salmon as of 5 h post intraperitoneal application. In agreement with this, Adamovsky et al. (2007) observed rapid (within days) microcystin clearance from liver and muscle tissue of carp and silver carp following immersion in toxic cyanobacteria for several weeks. The biotransformation of microcystin in fish has been suggested to begin similarly to the process in numerous other organisms via enzymatic catalysation of glutathione conjugation, i.e. phase-II- detoxification, which particularly increases the water solubility of the cyclic peptide and thus results in enhanced toxin excretion (Kondo et al., 1996; Pflugmacher et al., 1998; Takenaka, 2001; Wang et al., 2006). Analyses of bile extracts obtained from trout orally exposed to M. aeruginosa (equiv. to 5.6 mg

MC-LRequiv./kg bw) showed significant protein phosphatase inhibition as of 1 h post gavage (Sahin et al., 1996). This indicates microcystin contaminations in bile and hence suggests that microcystin is excreted from the liver via bile. Sahin et al. (1996) determined the highest concentration of PP-inhibition in bile extracts 3 h post gavage (corr. 3.5 µg MC-LRequiv./ml) before declining significantly (≤0.5 µg MC-LRequiv./ml as of 24 h post gavage). In a parallel investigation in the same trout, Tencalla & Dietrich (1997) found that microcystin amounts, extractable from the liver, also decrease after reaching a maximum concentration 3 h post application. These results are in agreement with Mohamed & Hussein (2006), who also observed decreasing microcystin amounts in liver and intestine and concurrent increasing microcystin concentrations

46 1. INTRODUCTION ______in bile of tilapia post exposure a toxic M. aeruginosa bloom. Similar findings have been reported in fish following intraperitoneal application of microcystin, where extractable microcystin amounts from liver were shown to decrease after reaching maximum concentration 5-8 h post application (Malbrouck et al., 2003; Williams et al., 1995; Williams et al., 1997). In contrast to this, progression of liver damage in microcystin treated fish usually continues considerably beyond 24 h (Bury et al., 1997; Fischer et al., 2000; Fischer & Dietrich, 2000; Fournie & Courtney, 2002; Kopp & Hetesa, 2000; Malbrouck et al., 2003; Råbergh et al., 1991; Tencalla & Dietrich, 1997; Tencalla et al., 1994; Vajcová et al., 1998), and thus considerably beyond the peak of microcystin excretion as observed by Sahin et al. (1996). This suggests that microcystins and/or microcystin conjugates excreted via bile within the first hours post application may not be the causal agent of certain liver pathology and in conclusion, that microcystins were partly not excreted and remain within affected tissue. Immunohistochemical staining of liver tissue from carp, orally exposed to toxic M. aeruginosa, demonstrated a considerable concurrence of pathological alterations which co-localised with microcystin as deduced from immunohistochemical staining for microcystin (Fischer & Dietrich, 2000). Highest immunopositivity and thus microcystin concentration was reached within 12 h post application. In contrast to the microcystin amounts extractable from the liver of microcystin treated fish, microcystin immunopositivity of those carp liver tissue did not decline after maximum microcystin concentration (Fischer & Dietrich, 2000). This corroborates the latter assumption that the microcystin causing liver pathology is not excreted but rather remains in the liver. Moreover it rather suggests that microcystins are covalently bound as most of the noncovalently bound microcystins (insomuch as they have not been excreted) would be expected to re- dissolve from tissue sections during the staining procedure (i.e. deparaffinisation, rehydration and repeated aqueous rinsing). There are no studies on the detoxification of microcystin covalently bound in fish tissue. However, as fish have been shown to recover from certain severe pathologies (Fournie & Courtney, 2002; Malbrouck et al., 2003), it appears that either covalently bound microcystins are metabolised in the course of time, or proteins and peptides with covalently bound microcystin are replaced by de novo protein synthesis.

Behavioural Alterations Resulting from Microcystin Exposure of Fish As already discussed in the context of clinical symptoms, microcystin exposed fish may change their general behaviour considerably. Those behavioural changes apply to both behaviour of single individuals and shoals and are interpreted as escape reaction and response to stress and elevated energy demands (Carbis et al., 1996a; Cazenave et al., Water Research in press). Baganz et al. (2004) estimated the lowest observed effective microcystin concentration causing behavioural aberrations in zebrafish (Danio rerio) to be ≤0.5 µg MC-LR/l. Cazenave et al. (Water Research in press) observed an elevation of fish velocity and movement in the Rio de la Plata livebearer Jenynsia multidentata orally exposed to 0.01 µg MC-RR/kg bw. Moreover, planktivorous silver carp and tilapia have been shown to decrease grazing rates in the presence of

47 1. INTRODUCTION ______microcystin containing cyanobacteria and thus to reduce food intake (Beveridge et al., 1993; Keshavanath et al., 1994). Baganz et al. (1998) finally observed that also spawning activity of zebrafish decreased subsequent to microcystin exposure (50 µg MC-LR/l). As the survival of a fish population depends not only on physiological fish fitness and the severity of induced organ damage, but also on appropriate fish behaviour (Hofer & Lackner, 1995), microcystin induced alterations on fish motility, velocity, feeding and spawning may detrimentally affect endurance of a fish population.

Effects of Microcystin on the Embryo and Larval Development of Fish Microcystin-induced effects on development, hatching, and survival of fish eggs, embryos and larvae have been investigated via exposure of various fish species either via immersion or injection of microcystin and microcystin containing cyanobacteria. Wiegand et al. (1999) demonstrated via immersion in radiolabeled microcystin that zebrafish may incorporate microcystins along all stages of both embryonic and larval development. The toxicity of microcystin to fish embryos and larvae has been shown to depend on the duration, dose and perhaps more critically on the developmental stage of exposure (Jacquet et al., 2004a; Lecoz et al., 2008; Liu et al., 2002, Oberemm et al., 1999; Oberemm & Becker, 1997; Wang et al., 2005). The lowest observed effective concentration causing a significant reduction of embryo survival was

5 µg MC-LRequiv./l (Oberemm & Becker, 1997). Highest toxicity has been observed during the initial stages of development (e.g. cleavage) (Liu et al., 2002; Ojaveer et al., 2003) and juvenile fish have been suggested to be far less sensitive than embryos and larvae (Liu et al., 2002). The differences in the development specific sensitivity are primarily attributed to changes in the amount of microcystin uptake, e.g. differences in membrane permeability etc. (Wang et al., 2005). As the chorionic membrane of fish eggs lacks mechanisms enabling active microcystin transport (e.g. OATPs), it is assumed to form a decisive barrier for the uptake of microcystin into fish eggs (Jacquet et al., 2004a; Lecoz et al., 2008; Wang et al., 2005). Wiegand et al. (1999) however propose that microcystin may penetrate the chorionic membrane via pores (app. diameter: 0.2µm) as zebrafish eggs accumulated approximately 0.5 ng MC-LR/egg following immersion in MC-LR at a concentration of 2.5 mg/l. In comparison to the latter, an injection of just 0.1 pg MC-LR/embryo has been shown to be sufficient to reduce survival of zebrafish embryos by up to 90% (Jacquet et al., 2004a). Exposure of fish eggs to microcystin or microcystin containing cyanobacteria either by injection or immersion has not only been shown to affect embryo survival and thus hatching rates but also on the time of hatching of surviving embryos, with both delayed as well as precocious or premature hatching being observed (Jacquet et al., 2004a; Keil et al., 2002; Liu et al., 2002; Oberemm et al., 1999; Ojaveer et al., 2003; Palíkova et al., 2003; Zhang et al., Toxicon in press). Those treatments further revealed severe posthatching effects including reduced growth (Oberemm et al., 1999; Oberemm et al., 1997; Zhang et al., Toxicon in press) and a significant increase in the incidence of malformations thus indicating a teratogenic potential for microcystin (Huynh-Delerme et al.,

48 1. INTRODUCTION ______

2005; Jacquet et al., 2004a; Keil et al., 2002; Liu et al., 2002; Oberemm & Becker, 1997; Oberemm et al., 1999; Ojaveer et al., 2003; Palíkova et al., 2003; Wang et al., 2005; Zhang et al., Toxicon in press). The malformations as observed in hatched larvae include eye, torso and tail deformation, the formation of oedema and haematoma in the heart area, histopathological changes on yolk sack, oesophagus, swimbladder, intestine and hepatopancreas as well as slowed blood circulation. Wang et al. (2005) ascribed the observed malformations to a disruption of organogenesis resulting from a microcystin-mediated interference with the distribution of ß-catenin and cadherin and consequent loss of blastula coherence. Several investigations on cyanobacterial extracts further suggest that cyanobacterial compounds other than microcystin and/or synergistic effects of the complex cyanobacterial biomass may enhance embryotoxicity (Keil et al., 2002; Lecoz et al., 2008; Oberemm & Becker, 1997; Oberemm et al., 1999; Palikova et al., 2007). In conclusion, the present data propose that exposure to microcystin containing cyanobacteria can not only directly reduce embryo survival, but also restrict survival of larvae due to premature or delayed hatching and malformations. Hence it may affect species recruitment and thus threaten survival of an entire population.

ICHTHYOTOXICITY OF OTHER CYANOBACTERIAL TOXINS

Nodularin The ichthyotoxicity of nodularins is less elucidated than that of microcystins. However, as the cyclic peptides are very similar both structurally and functionally, it is assumed that their ichthyotoxicity may also be comparable. Although not shown explicitly for fish, the ichthyotoxicity of nodularin is thus predominantly ascribed to inhibition of serin/threonine protein phosphatases (Kankaanpää et al., 2002a; Kankaanpää et al., 2002b). Nevertheless, there are structural differences that may determine the toxicological attributes of the peptides: nodularin is a pentapeptide and thus of lower mass and size than microcystins, and, in contrast to the majority of microcystin congeners, it includes methyldehydrobutyrin (Mdhb) instead of methyldehydroalanin (Mdha) preventing covalent binding to other molecules, e.g. protein phosphatases (Kankaanpää et al., 2002b). Field studies as well as controlled exposure experiments reveal that fish may ingest nodularin during blooms of N. spumigena (Kankaanpää et al., 2005b; Karlsson et al., 2003a; Karlsson et al., 2003b; Mazur-Marzec et al., 2007; Sipiä et al., 2001a; Sipiä et al., 2001b; Sipiä et al., 2006; van Buynder et al., 2001). Studies on the transfer of nodularin within the food web demonstrate that fish may accumulate nodularin also via the intake of contaminated food, e.g. zooplankton, which is primarily ascribed to the fact that nodularin forms no covalent bonds and may thus accumulate more easily along the aquatic food chain (Engström-Öst et al., 2002; Karjalainen et al., 2005; Sipiä et al., 2001a).

49 1. INTRODUCTION ______

Similarly to microcystin, the ileal epithelium is thought to be a decisive barrier against the uptake of nodularin (Karjalainen et al., 2005). Indeed, Kankaanpää et al. (2002a) and Karjalainen et al. (2005) estimate the proportion of nodularin uptake following oral exposure to be 0.03-0.5 %. Naturally exposed fish have been shown to accumulate nodularin primarily in the liver (Kankaanpää et al., 2002a; Karjalainen et al., 2007; Sipiä et al., 2001b; Sipiä et al., 2002). Only negligible amounts have been determined in muscle to date (van Buynder et al., 2001). Studies investigating biochemical, physiological and condition-related responses in nodularin- exposed fish are rare. Planktivorous fish have been shown to reduce feeding rates in the presence of nodularin, which may result in significant growth inhibition (Karjalainen et al., 2007; Karjalainen et al., 2005). The knowledge on the ichthyotoxicity of nodularin is mostly based on pathological and cytological observations. Kankaanpää et al. (2002a) found inflamed, bile-coloured, swollen intestines in trout orally exposed to 440 µg NOD/kg bw. Those trout however primarily presented with severe liver damage, including loss of liver architecture and degenerative hepatocyte damage (i.e. pyknosis, apoptosis and necrosis) hence confirming organotropism and thus hepatotoxicity of nodularin (Kankaanpää et al., 2002a). Investigating nodularin-induced apoptosis in isolated salmon hepatocytes, Fladmark et al. (1998) determined a mean effective concentration (EC50) of 400 nM. In contrast to the ichthyotoxicity of microcystins, Kankaanpää et al. (2002a) observed a rapid recovery of the livers of fish exposed to nodularin. This can primarily be attributed to the fact that nodularin may not bind covalently within tissues and thus can be rapidly detoxified/eliminated. The detoxification of nodularin in fish tissue is assumed to occur primarily via conjugation to thiol-containing proteins (e.g. glutathione and cysteine) (Kankaanpää et al., 2002a). Despite this, nodularin conjugates have not as yet been detected in fish tissue (Karlsson et al., 2003b). The ichthyotoxicity of nodularin has indeed primarily been investigated under laboratory conditions, however, bioaccumulation of nodularin in fish caught from natural environments and fish mortalities associated with nodularin-containing cyanobacterial blooms (Tab. 1.2) reveal an clear indication that the experimentally observed impact of toxic Nodularia sp. also applies in natural environments.

Cylindrospermopsin Cylindrospermopsin has been shown to accumulate in visceral tissue (1.2 µg/g tissue dw) of rainbow fish (Melanotaenia eachamensis) exposed to a natural bloom of Cylindrospermopsis raciborskii in aquaculture ponds (Saker & Eaglesham, 1999). It has further been shown to accumulate in snails, crayfish, mussel and frogs (Saker & Eaglesham, 1999; Saker et al., 2004; White et al., 2006; White et al., 2007) where intoxication is primarily attributed to the ingestion of toxic cells sometimes resulting in dramatic effects on mortality rates (Saker & Eaglesham, 1999; White et al., 2006; White et al., 2007). This might also apply to fish, however has not yet been explicitly proven. As world wide frequency and distribution of cylindrospermopsin is increasing, further investigations on the ichthyotoxicity of cylindrospermopsin appear necessary.

50 1. INTRODUCTION ______

PSPtoxins (Saxitoxins) It has long been assumed that poikilothermic vertebrates including fish were not affected by PSPtoxins (Prakash et al., 1971). During the last decades, however, experiments and field studies (primarily numerous marine observations where PSPtoxin containing ›› red tides12 ‹‹ have been shown to cause devastating fish kills) have demonstrated the lethality of PSPtoxins for fish (Landsberg, 2002). As verified PSPtoxin producers, some cyanobacteria (e.g. Aphanizomenon flos- aquae, Anabaena circinalis, Cylindrospermopsis raciborskii, etc.) have been regularly associated with fish kills in freshwaters (Tab. 1.2). Indeed, Aphanizomenon flos-aquae was shown to produce metabolites that Sawyer et al. (1968) described as a “very-fast-death-factor” in fish almost fifty years ago. Meanwhile, the mean lethal PSPtoxin dose (LD50) has been determined for a variety of fish species and range between 400-750 µg PSP/kg bw and 4-12 µg PSP/kg bw for oral and intraperitoneal application, respectively (White, 1984). Fish exposure to PSPtoxins can occur either via toxin uptake from water or ingestion of PSP containing phytoplankton, as well as through the food web by consumption of PSP contaminated zooplankton and shellfish (Sephton et al., 2007 and references therein). Direct PSP intoxication has been shown to result in higher mortality rates than intoxication via ingestion of PSP- contaminated food. Hence, in natural environments primarily herbivorous fish and those feeding at least temporarily on algae and cyanobacterial phytoplankton (e.g. most small fish species and first-feeding, juvenile fish) are threatened by PSPtoxin. Beside species-specific differences, PSP- toxicity in fish is thus apparently determined by the toxin content in dinoflagellate and cyanobacterial cells, the route of toxin uptake and in particular the food preference (Landsberg, 2002 and references therein). While PSP bioaccumulation in fish resulting from exposure to cyanobacteria has not been documented yet, dinoflagellate borne PSPtoxin has been detected in stomach contents, liver-, gut- and gill tissue of various fish species. Negligible PSP contamination has been observed in muscle. Thus the risk of human poisoning following consumption of PSP contaminated fish is assumed to be negligible (summarised in Landsberg, 2002). The mode of action responsible for PSP ichthyotoxicity has not been explicitly determined in fish. Nevertheless, based on rodent toxicity studies it can mainly be attributed to the blockade of sodium channels in nerve axons and a subsequent inhibition of impulse transmission (see also chapter 1.2). This is in accordance with Salierno et al. (2006), who recently demonstrated that exposure of killifish (Fundulus heteroclitus) to PSPtoxin causes a decrease in c-fos expression unambiguously revealing neuronal effects in the exposed organism. Indeed, PSPtoxins affect fish metabolism as already low amounts of intraperitoneal administered PSPtoxin have been shown to increase the activity of phase-I- and phase-II-detoxification enzymes in salmon. Interestingly, this also indicates that that fish may actually detoxify PSPtoxins (Gubbins et al., 2000). Despite detoxification, fish exposed to PSPtoxin respond with prominent symptoms, e.g. disoriented

12 "Red tide" is a common name for an event in which estuarine, marine, or fresh water phytoplankton accumulate rapidly and form dense, visible patches near the water's surface that turn the water colour to red or brown 51 1. INTRODUCTION ______swimming, disturbance of equilibrium, reduced ventilation and apparent paralysis. Fish die within 20-60 min of exposure to the lethal dose. In contrast to this, fish may recover when showing mild symptoms (Landsberg, 2002; Sephton et al., 2007 and references therein). PSP exposure of fish embryos (≥400 µg PSP/l) has moreover been shown to generate sensorimotoral changes (e.g. paralysis, reduced reaction, etc.) and morphological alterations (e.g. malformations and oedema, etc.), reduced growth and increased larval mortality in a dose-dependent manner (Lefebvre et al., 2004; Oberemm et al., 1999). Deducible from the described ichthyotoxicity of PSP-containing dinoflagellates, also cyanobacteria are thought to be capable of reducing larval recruitment and to affect fish behaviour, growth and survival when comprising acutely PSP-concentrations. Even so, more detailed studies appear essential.

Anatoxins The impact of anatoxins (i.e. primarily anatoxin-a, homoanatoxin and anatoxin-a(s)) on aquatic fauna has hardly begun to be assessed. Most cyanobacterial species containing anatoxins additionally produce other toxins and/or occur simultaneously with other toxic cyanobacterial species. As anatoxin-a is moreover comparably unstable and transient when dissolved in water (see chapter 1.2) and thus hardly detectable in natural environments, mortalities of fish and other aquatic organism are rather associated with microcystin and PSP than anatoxin-a intoxication. It appears likely, that fish incorporate anatoxins via the ingestion of cyanobacterial cells, ensuing digestion and consequent toxin release within the digestive tract. Indeed, Carmichael et al. (1975) reported death of goldfish after intraperitoneal injection and oral application of pure anatoxin-a, whereas no effects were observed when goldfish were exposed to an aqueous toxin extract, thus concluding that anatoxin is not readily absorbed across the gill membranes. In contrast to this, carp exposed to Anabaena suspension (≤107 cells/ml) via immersion accumulated anatoxin-a following the toxin concentration in solution suggesting diffusion of anatoxin-a from the ambient medium to fish metabolism via external surface, gills and/or digestive tract (Osswald et al., 2007). Those carp have further been shown to respond with behavioural changes, increased mortality and to contain up to 0.8 µg anatoxin-a/g dw suggesting that fish may effectively accumulate anatoxin-a. Extract of Anabaena spiroides and anatoxin-a(s) have moreover been shown to irreversibly inhibit acetylcholinesterase isolates from several fish species with kinetics comparable to those observed subsequent to organophosphate pesticide exposure. This indicates a potential for neurotoxicity for anatoxin-a(s) in fish (Mahmood & Carmichael, 1987; Monserat et al., 2001). Juvenile carp exposed to Anabaena suspension containing anatoxin-a have been shown to respond with rapid opercular movement and abnormal swimming behaviour and died within 30 h when exposed to 107 Anabaena cells/ml (Osswald et al., 2007). Oberemm et al. (1999) observed alterations in heart rates of zebrafish when exposed to pure anatoxin-a, however these effects were temporary and no

52 1. INTRODUCTION ______chronic effects or mortalities were observed. Although toxicology and pharmacological actions have not been elucidated in detail yet, it thus appears presumable that cyanobacterial anatoxins may cause neurotoxicity in fish in a manner similar to mammals - by acting either as a postsynaptic cholinergic agonist or by inhibition of acetylcholine esterase - thus apparently affecting individual fish and fish populations especially during senescence of cyanobacterial blooms.

Other Cyanobacterial Metabolites Toxic to Fish The fact that numerous investigations demonstrate cyanobacterial crude extracts to be more toxic than included concentrations of pure toxins implies the existence of more ichthyotoxic compounds and that those may modulate the effectiveness of known cyanobacterial toxins. Indeed, Wright et al. (2006) reported the cyanobacterium Fischerella ambigua, although lacking known cyanobacterial toxins, to release compounds capable of inhibiting the early development of zebrafish embryos into the surrounding environment. Papendorf et al. (1997) moreover described lupenyl acetat and mueggelone (compounds isolated from Aphanizomenon flos-aquae) to affect embryo larval development of zebrafish. Further cyanobacterial metabolites (e.g. antillatoxins, hermitamides, jamaicamides, kalkippyrone, malyngamide H and pahayokolide A) have been isolated from Lyngbya sp. and demonstrated to be toxic to fish (Edwards et al., 2004; Graber & Gerwick, 1998; Nogle et al., 2001;

Orjala & Gerwick, 1996; Orjala et al., 1995a; Tan et al., 2000). Antillatoxin (LD50 = 0.05 µg/ml) is counted as the most ichthyotoxic metabolite isolated from marine “plants” (Orjala et al., 1995b). Teneva et al. (2003) demonstrate Lyngbya aerugieo-coerulea to produce compounds causing cytotoxic effects in a piscine liver cell line. All of these Lyngbya-produced compounds have been verified through experimental studies, however, it remains unclear to what extent they affect aquatic organisms in the environment. Nevertheless, mass occurrence of L. majuscula has been demonstrated to play a role in the decline of live mass, diversity and species composition of fish in Moreton Bay, Australia, thus confirming Lyngbya sp. to affect fish populations also in nature (Pittman & Pittman, 2005).

Another group of apparently ichthyotoxic cyanobacterial compounds are lipopolysaccharides (LPS). Immersion of halibut larvae (Hippoglossus hippoglossus L.) in LPS isolated from Aeromonas bacteria demonstrated that LPS can be incorporated into fish via the integument and/or gut (Dalmo et al., 2000). Carp exposed to bacterial LPS responded with profound and differential effects on the detoxification metabolism (e.g. inhibition of cytochrome P450 induction and decreasing cytochrome P450 levels) in the liver and immune organs (Marionnet et al., 1998). Similarly to bacterial LPS, cyanobacterial LPS have also been shown to be potentially capable of reducing GST activity in zebrafish embryos (Best et al., 2002). This indicates a reduction of the detoxification capacity and thus may explain the observed modulation of the toxicity of various contaminants including toxic cyanobacterial peptides and alkaloids (Best et al., 2001; Best et al., 2002; Pietsch et al., 2001).

53 1. INTRODUCTION ______

Beside the cyanobacterial toxins described here, there are further compounds comprising toxic potential that has not been investigated in regard to ichthyotoxicity explicitly yet (e.g. BMAA). It thus appears likely, that the number of isolated, ichthyotoxic cyanobacteria borne compounds will increase with additional scientific investigations in future.

In summary, scientific achievements clearly demonstrate that cyanobacterial toxins may impact fish via various and multiple influences. The actual hazard arising from toxic cyanobacteria for fish is primarily governed by the cyanobacterial cell/filament density. Hence, cyanobacterial abundance may change in the course of time especially when waters are subjected to anthropogenic influences (e.g. eutrophication and re-oligotrophication). However, as various cyanobacterial ecostrategists display effective adaptation to almost all environmental conditions the hazard of cyanobacterial induced ichthyotoxicity unlikely to vanish.

54 1. INTRODUCTION ______

1.4. THE RISE AND FALL OF P. RUBESCENS AND FISHERY YIELDS IN LAKE AMMERSEE, GERMANY – HISTORY AND GOAL OF THE STUDY

Lake Ammersee is a typical pre-alpine lake, located in the south of Germany at 553 m altitude. The lake is dimictic with a surface area of 46.6 km2, a total volume of 1750 x106 m3 and a maximum and average depth of 81.1m and 37.5m, respectively (Grimminger 1982). Water residence time is 2.7 years, whereby the lake’s principal water source is the river Ammer – with a mean flow rate of 16.6 m3/s. Due to the large catchment area of the river Ammer the lake collects water from an area of 993 km2, including widely natural and agricultural, but also urban and industrially influenced regions. Studies on the phytoplankton assemblage in Bavarian pre-alpine lakes describe Lake Ammersee to lack relevant abundance of blue-green algae (i.e. cyanobacteria) in 1942 (Gessner, 1950). As of this time, the lake underwent a distinct phase of nutritional enrichment (i.e. eutrophication), primarily as a result of increased urbanisation, detergent use and intensification of agriculture in the catchment area. Total phosphorous reached mean yearly concentrations of up to 60 µg/l until the 1970’s (Steinberg, 1980). This eutrophication obviously entailed an increase of P. rubescens abundance as appreciable P. rubescens densities were observed for the first time in 1955, albeit quantitatively still of secondary importance (Reimann, 1955 cited in Steinberg, 1980). During investigations on the effects of eutrophication on Lake Ammersee phytoplankton from 1975 to 1977, Steinberg (1980) then determined P. rubescens as a constant and dominant part of phytoplankton reaching localised accumulations of up to 1.6 Million filaments/l. In June 1972, the lake Ammersee fishery cooperative chronicled that “the burgundy-blood algae (i.e. P. rubescens) changed the colour of the entire lake to red”. Later, they reported “numerous dead coregonids floating at the water surface” and “dead fish lying at the bottom of the lake”. Concurrently, coregonids (Coregonus lavaretus) caught, representing the existential basis for the professional fishery, slumped from a mean yield of 60 t/year up until 1972 to an average of 15 t/year from 1973 to 1982 (Rauch et al., 1961-1999). Due to a reduction of anthropogenic influences, the continued eutrophication of Lake Ammersee was halted and reversed, i.e. a re-oligotrophication process was initiated. In consequence, the phosphorous input continuously decreased and Planktothrix sp. widely disappeared during the 1980’s (Steinberg & Lenhart, 1991), and coregonid catches recovered, reaching a maximum of 146 t in 1987 (Rauch et al., 1961-1999). As the yearly mean total phosphorous concentrations declined to approximately 10 µg/l during the 1990’s, P. rubescens astonishingly experienced a renaissance, with biomasses in excess of 200 µg/l from 1990 to 1993, in 1996 and 1998 (Kucklentz et al., 2001; Teubner et al., 2004). Again, coregonids caught slumped to yields of less than

55 1. INTRODUCTION ______

15 t/year in 1991-1993 and 2001-2005 (Rauch et al., 1961-1999), primarily attributed to a massive decrease in coregonids growth and fitness (Wißmath, 2004). Twofold annuli, as regularly observable on the scales of Lake Ammersee coregonids, demonstrated that the determined growth retardation resulted from an additional period of starvation, beyond the one normally occurring during the winter months (Wißmath et al., 1992). The observed phenomena have been discussed heatedly in a context of lacking zooplanktic food-species, high fish densities and recurrent metalimnic oxygen deficiencies (Mayr, 2001; Morscheid & Morscheid, 2001; Wißmath, 2004). The apparent coincidences of the appearance of P. rubescens mass development and decreased coregonid yields further imply a possible link between the P. rubescens occurrence and the observed changes in growth and population dynamics. Indeed, also Braun (1953) documented the sudden rise of massive Burgundy-blood algae development in Lake Hallwil, Switzerland in 1898, to coincide with the initiation of a rapid decrease in coregonid yields. Braun (1953) also ascribed the disappearance of coregonids in Lake Murten due to increased P. rubescens abundance. In 1998, coregonids caught in Lake Ammersee had conspicuous blue-coloured gut contents, which may most likely be attributed to the digestion of ingested cyanobacteria (i.e. most probably P. rubescens) and a consequential release of cyanobacterial biliproteins within the coregonid intestine. This suggests that coregonids actually ingest P. rubescens. As preliminary studies revealed those P. rubescens to include cyanobacterial toxins (i.e. microcystin) it was proposed that the observed disturbance in growth and population dynamics of coregonids results at least partially from P. rubescens ichthyotoxicity.

The aims of this thesis were therefore • to investigate the effects of an environmentally relevant dose of P. rubescens orally applied to coregonids and to assess the coregonids specific sensitivity in the context of microcystin toxicity previously reported in carp and trout (chapter 3.1) • to investigate the physiological stress response and organ pathology in coregonids sub- chronically exposed to ambient water containing P. rubescens cell densities, known to be typical of Lake Ammersee and other pre-alpine lakes (chapter 3.2) • to characterise the temporal and spatial abundance and toxicity13 of P. rubescens in Lake Ammersee in conjunction with the consideration of potential detrimental effects on the indigenous coregonid population (chapter 4.1) • and finally, to look for manifestations of P. rubescens induced adverse effects in wild caught coregonids, comparable to those previously observed experimentally (chapter 4.2). A prerequisite for the proposed investigations however was the opportunity for a rapid and precise observation of P. rubescens in Lake Ammersee and a reliable judgment on potential microcystin accumulations in fish tissue. Thus initial work aimed to validate an image processing system for the quantification of P. rubescens cell densities (chapter 2.1) and to quantify the microcystin recovery from tissue of MC-LR -exposed trout (chapter 2.2).

13 Except for one studies characterising P. rubescens to contain anatoxin-a (Viaggiu et al., 2004), P. rubescens toxicity is associated to the production of microcystins. Thus, toxin analyses as carried out in this study were limited to the detection of microcystins 56

2. METHODICAL INOVATIONS

2.1. DETERMINATION OF THE FILAMENTOUS CYANOBACTERIA PLANKTOTHRIX RUBESCENS IN ENVIRONMENTAL WATER SAMPLES USING AN IMAGE PROCESSING SYSTEM

Bernhard Ernst1, Stephan Neser2, Evelyn O’Brien1, Stefan J. Hoeger1, Daniel R. Dietrich1

1 Environmental Toxicology, University of Konstanz, P.O. Box X918, 78457 Konstanz, Germany 2 Dept. of Mathematics and Natural Sciences, University of Applied Sciences Darmstadt, Schöfferstrasse 3, 64295 Darmstadt, Germany

Published in Harmful Algae 5 (2006) 281–289

ABSTRACT Cyanobacteria occur in surface waters worldwide. Many of these produce peptides and/or alkaloids, which can present a risk for animal and human health. Effective risk assessment and management requires continuous and precise observation and quantification of cyanobacterial cell densities. In this respect, quantification of filamentous Planktothrix sp. is problematic. The aim of this study was to develop an automated system to count filamentous Planktothrix rubescens using image processing. Furthermore, this study aimed to assess optimum sample volumes and filament density for measurement precision and to validate image processing measurement of P. rubescens for an effective risk assessment. Three environmental samples and one cultured sample of P. rubescens were collected by filtration onto nitrocellulose filters. Filament lengths were determined using fluorescence microscopy combined with an image processor. Cell density could be calculated from the resulting images. Cyanobacteria could easily be discriminated from algae via their fluorescence properties. The results were found to be independent of the mode of image acquisition. The precision of total filament length determination was dependent on the total filament length on the filter, i.e. analyses of highest precision could be expected for filters containing 2000 to 20,000 µm filaments per mm2. When using suitable filtration volumes, the detection limits of the described method are sufficient for an effective risk assessment. To summarise, this procedure is a fast, easy and accurate method to determine cell densities of filamentous P. rubescens in water samples without costly and tedious manual handling.

KEYWORDS: Cyanobacteria; Filament; Image processing; Planktothrix; Cell quantification; Risk assessment

57 2. METHODICAL INOVATIONS ______

INTRODUCTION Cyanobacteria occur worldwide in coastal and surface waters. Surveys in various countries have demonstrated that about 75 percent of samples containing cyanobacteria are toxic. Due to nutritional enrichment (eutrophication), occurrences of toxic cyanobacterial blooms in surface waters, e.g. species of the genera Microcystis, Anabaena, Planktothrix and Aphanizomenon, are becoming a growing problem (Bartram et al., 1999). In addition, albeit in contrast to this situation, the intentional nutritional re-depletion of eutrophic surface waters (re- oligotrophication) resulted in regular blooms of Planktothrix rubescens in several European pre- alpine lakes (Ernst et al., 2001; Jacquet et al., 2005; Mez, 1998; Morabito et al., 2002). P. rubescens is a low light adapted, filamentous cyanobacterium, made up of cells, which contain gas vesicles enabling the filaments to adjust their buoyancy in the water column in order to achieve optimal use of the ambient environment (Walsby et al., 1998). Consequently, P. rubescens builds blooms distributed over the whole vertical water column during winter circulation and in metalimnic layers during summer stratification. Furthermore, buoyancy disturbances can result in P. rubescens blooms at the lake surface (Ernst et al., 2001; Jacquet et al., 2005). P. rubescens blooms and layers can attain densities of up to 150,000 cells/ml (Hoeger et al., 2005). At least 46 cyanobacterial species are able to produce neurotoxins, e.g. anatoxin-a, anatoxin-a(s), and saxitoxin, a range of dermatoxins and/or predominantly potent protein phosphatase inhibitors, such as microcystins and nodularins (Chorus et al., 2000; Sivonen & Jones, 1999). In addition to producing a range of other metabolites with unknown toxicological potential, including anabaenopeptins, microviridins and cyanopeptolins (Blom et al., 2003), species of the genera Planktothrix have been shown to contain the highest amounts of microcystin (<5.6 mg/g dw) (Fastner et al., 1999b). The release of these cyanobacterial toxins may present a serious risk for wild and domestic animals as well as for human health, as recently reviewed by Dietrich & Hoeger (2005). As a result of incidents attributed to toxic cyanobacteria the world health organisation (WHO) and several national authorities world-wide have recommended risk assessment plans and safety levels to include cyanobacteria as a parameter, which must be monitored for water quality control (Azevedo, 2001; Chorus et al., 2000; Codd et al., 2005; Falconer, 2001). Effective risk assessment and management requires continuous and precise observations of cyanobacterial biomass and/or cell densities (Chorus & Bartram, 1999). Quantification of Planktothrix species is difficult as the individual cells arranged to form a filament are hardly distinguishable. Furthermore, filament counts cannot be automatically correlated to biomass or cell densities because Planktothrix sp. exhibit large variations in filament length and filaments overlay one another and are often curved in a given sample when observed on a slide or filter, making measurement of filament length difficult and inaccurate. Quantification of cell volumes is difficult because centrifugation is laborious due to the gas vesicles incorporated in Planktothrix cells for buoyancy. Furthermore, quantification via determination of photopigments,

58 2. METHODICAL INOVATIONS ______e.g. chlorophyll and/or biliproteins is not reliable due to regulation of pigments with various growth conditions (Feuillade, 1994) and false positive results due to pigments of eukaryotic algae and zooplankton coexisting within the same environment. Gjolme et al. (2004) demonstrated protein concentrations to best reflect cyanobacterial biomass. However, as pigment and biomass parameters, protein measurement may easily be overestimated in environmental seston samples due to the coexistence of eukaryotic algae and zooplankton within the same environment. As many of the lakes containing Planktothrix sp. are used for recreational purposes and several even as drinking water reservoirs (Hitzfeld et al., 2000; Hoeger et al., 2005), a rapid and precise procedure for quantification of Planktothrix sp. is essential. For cell quantification of filamentous cyanobacteria most methods of choice are based on microscopic identification and counting (Bailey-Watts & Kirka, 1981; Hoogveld & Moed, 1993; Olson, 1950). This approach has the caveat of increased demand on both manpower and skill of the personnel as well as limitations in the speed with which filament densities can be determined. Cyanobacterial species use the biliproteins phycocyanin and allophycocyanin to harvest light for photosynthesis. Some species, including Planktothrix sp., additionally contain the biliprotein phycoerythrin (Anagnostides & Komárek, 1988; Glazer, 1985). When examined under blue light excitation, phycoerythrin and phycocyanin fluoresce orange and red, respectively. Therefore, cyanobacteria can be enumerated by visualising the autofluorescence of phycoerythrin and/or phycocyanin using epifluorescence microscopy (Sieracki & Wah Wong, 1999; Walsby & Avery, 1996). Walsby & Avery (1996) designed and described a semi-automated procedure to count Planktothrix cell densities. This method involves the transfer of epifluorescent microscope images of filter to a computer, followed by determination of filament length via computer image analysis. This is a fast and accurate method, which measures the length of several filaments simultaneously. The aim of our study was to develop automation in counting filamentous P. rubescens using arrays of filament images, i.e. to improve and expand on the method of Walsby & Avery, reducing the manual interactions required for measurement and thus reducing overall time per sample. Furthermore, this study aimed to asses optimum sample volumes and filament density for measurement precision and to validate image processing measurement of P. rubescens for an effective risk assessment.

MATERIAL & METHODS

Samples P. rubescens samples 1-3 were environmental seston samples of various P. rubescens densities collected from Lake Ammersee, Germany in July 2001. They were taken from the metalimnion (10-12 m depth) using a Ruttner flask sampler. The taxonomy of each sample was determined via

59 2. METHODICAL INOVATIONS ______light microscopy and classification was according to Anagnostidis & Komárek (1988). Samples were fixed with Lugol’s solution (iodine-potassium iodide solution) and stored in darkness at room temperature for 24 h until filtration. Samples 1-3 were used for method validation. Sample 4 was a culture sample of P. rubescens, isolated from a Lake Ammersee seston sample in autumn 2002 and cultivated in BG11 medium according to Rippka et al. (1979). Sample 4 was used to investigate the robustness of the method and to compare cell counts and chlorophyll a measurements.

Filtration, Epifluorescence microscopy, Video transfer and Analysis Mode Defined volumes of the samples were filtered onto nitrocellulose filters (pore size 8 µm, diameter 25 mm, Schleicher & Schuell, Germany). Samples were filtered using a standard filtration apparatus (Millipore, Germany) with an absolute surface area (Asurface) of 283.5 mm2 for each filter. Filters were air dried and stored in 6-well cell culture plates (Sarsted, Germany), darkness and at room temperature until analysis. Cyanobacterial filaments were observed with an epifluorescence microscope (ECLIPSE TS100; Nikon, Germany) using 100-, 200- and 400-fold magnification. Image analysis of the filters was performed using an epifluorescence microscope (Zeiss Standard 25 including an HBO 50/AC-lamp, Germany) with a x10 objective (Zeiss A-Plan x10/0.25). Samples were illuminated through a filter block allowing blue-light excitation (λ = 450-490 nm). Several fields-of-view from each membrane-filter were transferred to a Pentium II PC (450 MHz, 384 MB RAM, NVIDIA 128/128ZX graphics card, Windows 2000) with a monochrome CCD- camera (KAM02E, EHD, Germany, resolution 752 x 582 pixel, ½ in. CCD) in combination with an IDS Falcon frame-grabber and were digitalised using the image processing system Visiometrics IPS 1.119 (Visiometrics GbR, Germany). Brightness and contrast were adjusted until filaments appeared as white lines on a dark background (Fig. 2.1). Length scales were calibrated using the gridlines of a Neubauer haemocytometer (Brand, Germany) as reference (the calibration factor was px = 0.86 µm/pixel and py = 0.87 µm/pixel; resulting in an area per field-of-view (Aview) of 0.32 mm2 with a x10 objective). Filters containing the filaments were focussed and image analysis was started via seven automatic steps using the image processing system as follows: 1. For each field-of-view 50 digitised video frames were averaged in order to achieve an image with high signal-to-noise ratio. 2. A 5 x 5 median filter was applied, to remove small bright features in the background while preserving the outline of the filaments. 3. Spatial intensity gradients in the image background arising from the inhomogeneity of the illumination and due to the inherent and spatially variable fluorescence of the membrane were corrected using the local adaptive background correction of Visiometrics IPS. 4. The image was automatically segmented into filaments and background by binarisation with a single intensity threshold.

60 2. METHODICAL INOVATIONS ______

a b

100 µm 100 µm

Fig. 2.1: (a) Image of a filter membrane containing P. rubescens filaments enumerated by fluorescence microscopy and (b) skeletons of the filaments calculated using the image processing system Visiometrics IPS.

5. The resulting image was thinned using the skeleton routine of Visiometrics IPS (Fig. 2.1), which employs the skeletonising algorithm published by Arcelli et al. (1975). 6. Skeletons, with a length <30 pixels (user-defined threshold) were omitted from the analysis to minimise unspecific contributions to the total filament length (Fig. 2.1). 7. To determine the total length of the remaining filaments, the region of each filament pixel was analysed: For each of the left, right, top and bottom neighbouring pixels, one interpixel distance was added to the total filament length. For the top-left, top-right, bottom-left and bottom-right neighbours, √2 times the interpixel distance was added. Since during the analysis each distance is counted twice, the final result Lraw is divided by two. The lengths of the remaining filaments were determined by calculation from the number of skeleton pixels and summed to yield the total length of all filaments per field-of-view. Measuring length by means of interpixel distances has the disadvantage of overestimating the actual length in certain directions (Walsby & Avery,

1996). This error can be corrected by application of a statistical correction factor of 0.948 to Lraw. Hence, the total corrected filament length in the view is given as:

Lview = Lraw x 0.948

The total length of filaments per filter (Lfilter) was calculated using the following equation:

Lfilter = Asurface x Lview x Aview –1 for the x10 objective Lfilter = 886.6 x Lview

To obtain an approximation for the total cell count (Ctotal) per filter the total length of filaments per filter (Lfilter) was divided by the average cell length (Laverage) of P. rubescens:

Ctotal = Lfilter x Laverage–1

Laverage was determined by measuring the cell length of 29 randomly selected cells of different filaments. However, cells alined into a filament were only barely distinguishable (Anagnostides & Komárek, 1988). Average cell length of P. rubescens isolated from Lake Ammersee was determined to be 2.8 ±0.44 µm. This is in accordance to the description of Geitler (1932), who described cell lengths to range from 2 to 4 µm. For calculation of total cell counts, average cell length was assumed to be 3 µm/cell for P. rubescens. Finally, cell densities were calculated by dividing the determined cell count Ctotal through the filtered sample volume. 61 2. METHODICAL INOVATIONS ______

Method Validation To investigate if filaments were homogeneously distributed on the filters, the results of differently oriented picture grids were compared. For this purpose, 20 ml of the environmental samples (samples 1-3) were collected by filtration onto a membrane as described above. Each filter was counted four times, analysing 20 different fields-of-view per filter selected along varying grids (Fig. 2.2). To determine how many fields-of-view per filter must be analysed in order to obtain a representative cell count, 10, 20, 40, 60 and 80 randomly selected fields-of-view were analysed for each filter of the environmental Planktothrix-samples (samples 1-3). Analyses were then compared to determine, if the results were dependent on the number of fields of view analysed per filter. The filament-capacity of the method was determined by analysing filters with various sample volumes. Thus, 0.1-100 ml of a culture sample (sample 4) was filtered onto different membrane- filters. Each filter was analysed by counting ten fields-of-view. Analyses were then compared to investigate, whether or not the resulting counts per field-of-view correlated to the filtered sample volumes.

For comparison of cell counts and chlorophyll a measurements, eight different volumes of the culture sample (sample 4) were filled with tap water to give a final volume of one litre. Various volumes (10-30 ml) of these samples were filtered onto nitrocellulose filters and stored for cell count analyses as described above. Between 250 and 750 ml of the remaining volumes were filtered onto GF/C filters (Whatmann, UK) and chlorophyll a concentrations were determined according to the standard protocol DIN 38412 L16. Chlorophyll a measurements and cell count analyses were finally compared to determine correlation of the data.

Statistics Data analyses were carried out using JMP® 4 (USA) Software. Values represent the mean ± standard error of the mean (SEM). Results were analysed for statistical differences using analysis of variance (ANOVA) and the Tukey-Kramer Multiple Comparisons Test (p ≥0.05). Regression analyses were performed using Microsoft Excel.

Fig. 2.2: Scheme of picture grids analysed in order to screen the filament distribution on the membrane- filters (ten fields-of-view were analysed along each direction; r means analysis of 20 randomly distributed fields-of-view per filter).

62 2. METHODICAL INOVATIONS ______

a b

Fig. 2.3: (a) Light microscopy of a Lake Ammersee seston sample containing diatoms, other eukaryotic algae and cyanobacteria predominantly filamentous P. rubescens and (b) differentiation of cyanobacteria in the same field-of-view from eukaryotic algae using epifluorescence microscopy (magnification x100).

RESULTS In Lake Ammersee field-samples, cyanobacterial filaments could be automatically distinguished from other planktonic organisms by both their structure and fluorescence properties. Using blue light excitation, the P. rubescens filaments fluoresced orange while eukaryotic algae did not fluoresce at all (Fig. 2.3). After filtration and video transfer, filaments appeared as white lines on a dark background (Fig. 2.1). Analysis of the filtered environmental samples investigated for method validation yielded filament densities of 25,601 µm filament per mm2 filter (sample 1), 12,052 µm filament per mm2 filter (sample 2) and 3345 µm filament per mm2 filter (sample 3) corresponding to approximately 60,490, 28,475 and 7904 cells/ml, respectively. Neither significant differences nor tendencies of differences could be observed analysing the three filters along diverse picture grids thus a systematic error due to gradients in filament distribution on the filters can be excluded (Fig. 2.4).

10000 9000 8000 7000 6000 5000

[µm] L view 4000 3000 r 2000 1000 r r 0 sample 1 sample 2 sample 3 Fig. 2.4: Comparison of P. rubescens cell counts analysed by varying picture grids to test for homogeneous filament distribution on the filter membrane. Three environmental samples of various filament densities were filtered. There were no significant differences (n = 20; error bars = SEM).

63 2. METHODICAL INOVATIONS ______

10000 10 fields / filter 20 fields / filter 9000 40 fields / filter 8000 60 fields / filter 80 fields / filter 7000 6000 5000

L view [µm] L view 4000

3000 2000 1000

0 sample 1 sample 2 sample 3 Fig. 2.5: Cell counts of three environmental P. rubescens samples of various filament densities. Each filter was counted analysing 10, 20, 40, 60 and 80 fields-of-view per filter. There were no significant differences between the cell counts resulting from variable number of fields-of-view per filter (error bars = SEM).

Additionally, there were no significant differences between the cell counts resulting from variable number of fields-of-view per filter. Although, the standard error of the mean decreased with an increasing number of fields-of-view analysed per filter (Fig. 2.5). In order to determine the measurement precision and the ideal range of filament density on the filter, several filters from an identical culture sample, but with various sample volumes were analysed. For most samples double volume corresponded to double counts. There was a significant correlation between filtered volumes and the analysed counts per field-of-view (R2 = 0.98). Analyses of filters containing a filament density of 80 to 60,000 µm filament per mm2 filter resulted in a mean cell count of 81,790 with a standard error of the mean ±5,132 cells/ml (Tab. 2.1). Measurement precision increased (85,701 ±2629 cells/ml), if cell counts corresponded to filament densities ranging between 2000 to 20,000 µm filament/mm2 filter (Tab. 2.1), indicating the error of measurements to depend on the filament density on the filter. Finally, there was a significant correlation between cell counts and chlorophyll a measurements resulting in a ratio µg chlorophyll a/l:cells/ml of 1/2000 (Fig. 2.6).

Tab. 2.1: Capacity of the method: filters holding various volumes of a P. rubescens culture sample were analysed for µm filament per field-of-view (Lview), µm filament per mm2 filter and P. rubescens cell density

Filtered volume [ml] 0.1 0.2 0.4 0.6 0.8 1 2 4 6 8 10 20 40 60 80 100

Lview [µl] 25 35 170 220 251 383 606 1,190 1,641 1,977 3,055 6,020 9,114 12,739 16,970 18,256

[µm filament -2 mm filter] 78 110 530 687 785 1,197 1,895 3,723 5,131 6,182 9,552 18,826 28,499 39,835 53,064 57,087

3 -1 [cells x 10 ml ] 73 52 125 108 93 113 90 88 81 73 90 89 67 63 63 54

Analyses of filters containing between 2000 and 20,000 µm filaments/mm2 filter resulted in comparable cell counts (highlighted). In comparison, analyses of filters with <2000 µm/mm2 tended to result in an overestimation of cell counts while analyses of filters with >20,000 µm/mm2 resulted in lower cell counts.

64 2. METHODICAL INOVATIONS ______

150 125 y = 0.0005x 100 R2 = 0.9433 75

[µg/l] Chl a 50 25

0 Fig. 2.6: Comparison of chlorophyll a 0 50000 100000 150000 200000 250000 measurements and cell counts analysing cell count [cells/ml] a P. rubescens culture sample (sample 4).

DISCUSSION The analyses of the field samples confirm former findings that cyanobacteria can be distinguished from other algae using epifluorescence microscopy as these other planktonic species (diatoms, green algae, etc.) have a much weaker fluorescence than cyanobacteria (Sieracki & Wah Wong, 1999; Walsby & Avery, 1996). Discrimination can be achieved by setting the brightness level of the Visiometrics IPS system to a value at which cyanobacterial filaments can be recognised, but no organisms with weaker fluorescence. Thus, the determination of cyanobacterial cell densities by image analysis can be applied to cultured as well as natural samples. This conclusion also corroborates the earlier findings of Leboulanger et al. (2002), who determined vertical P. rubescens distribution in Lake Bourget, France, via the fluorescence properties using a submersible spectrofluorometer.

The counting of samples on membrane-filters has several advantages: samples must only be focussed in one focal plane, the filaments do not move on the microscope slide, fluorescence properties increase if filaments are dried, samples can be stored on the filters for several months without loss of quality and storage space requirements are reduced. Thus, the samples may be reanalysed if necessary (Walsby & Avery, 1996). The data obtained with the image processing system presented here demonstrate, that fully automated cell quantification of filamentous cyanobacteria can be carried out in a robust and highly reproducible manner. This is largely due to the following improvements in image processing: 1. Filaments were separated from background automatically, by binarisation with a single intensity threshold, followed by thinning of the resulting image using the skeleton routine of Visiometrics IPS, which employs the skeletonising algorithm published by Arcelli et al. (1975). A prerequisite for this automatic measurement is a clear, sharp image and low noise. 2. These images can be obtained by generating an average image of 50 individual video frames. 3. Image quality was additionally improved using a median filter and the automatic locally adaptive background correction of Visiometrics IPS. Thus, the image processing described here, represents a valuable improvement over the initial automation protocol suggested by Walsby & Avery (1996). 65 2. METHODICAL INOVATIONS ______

To obtain significant cell counts by analysis of the lowest possible number of randomly distributed fields-of-view, filaments must be distributed on the filter homogeneously. This condition was proven by examination of different modes of image acquisition. Comparison of cell counts determined along varying picture grids showed no differences demonstrating cell counts to be independent of the mode of image acquisition. This indicates that, indeed, filament distribution is homogeneous.

Measurement precision increased with an increasing number of fields-of-view used for filter analysis, as shown by a decreasing standard error of the mean for measurements resulting from variable number of fields-of-view per filter. However, for practical purposes, cell counts were shown, to be independent of the number of fields-of-view used for analysis. Consequently, the analysis of ten randomly taken fields-of-view is sufficient to obtain representative cell counts per filter. Walsby & Avery (1996) even suggested that the analysis of five randomly taken fields-of- view would suffice in representative cell counts. In contrast to our investigations, Walsby & Avery used a x4 objective for their image analyses. Consequently, these authors analysed larger fields- of-view. Whereas, despite that in the present study 10 fields-of-view are considered necessary, using a x10 objective is an improvement, as it results in better filament resolution, pronounced differences in fluorescence properties. In addition, higher magnification reduces the number of filament intersections and filament touching per field-of-view. This simplifies Planktothrix filament discrimination from other structures (eukaryotic algae, zooplankton) and increases the image processing accuracy.

The measurement precision depended on the total filament length on the filter. This is a function of the filament concentration, the filtered volume and the filter area analysed, which in turn depends on the number of fields-of-view analysed and their magnification. In order to determine the measurement precision of the method and the ideal range of filament density on the filter, filters with various sample volumes of an identical culture sample were analysed. Cell counts correlated significantly with the filtered volume analysing the culture sample. However, cell counts tended to be higher when analysing filters containing <2000 µm filaments per mm2 filter and tended to be lower when analysing filters containing >20,000 µm filaments per mm2 filter demonstrating discrepancies for filters containing high and low filament densities, respectively. This observation is also supported by earlier studies (Embleton et al., 2003; Walsby & Avery, 1996). The underestimation resulting from the analysis of filters with high filament densities is most likely due to superimposition of filaments on the filter, while inaccuracies analysing filters of low filament densities are probably caused by a statistical error due to increasing standard errors with decreasing filament densities. This is confirmed by the findings of Walsby & Avery (1996) who demonstrated the standard errors tend to decrease as a hyperbolic function of filament concentration. Using the described method, analyses of highest

66 2. METHODICAL INOVATIONS ______precision could be expected for analyses of filters in a range of 2000-20,000 µm filaments per mm2 filter. This could easiest be achieved by sample dilution or an increase of the volume filtered.

The WHO has provided recommendations for a framework of risk assessment starting at a first alert level of 20,000 cells/ml (Chorus & Bartram, 1999; Falconer, 2001). The Brazilian regulation for drinking water envisions an expansion in cyanobacterial monitoring at a density of 10,000 cells/ml (Azevedo, 2001). Filtration of 20 ml of a sample containing 10,000 cells/ml would result in a cell density of 200,000 cells/filter corresponding to approximately 2000 µm filament per mm2 filter. A sample containing 150,000 cells/ml would correspond to the highest documented P. rubescens cell density demonstrated for Lake Zürich, Switzerland (Hoeger et al., 2005). Ten ml of this corresponds to 16,000 µm/mm2. Therefore using suitable filtration volumes, the detection limits of the described method are sufficient for an effective risk assessment as recommended by the above named authorities.

Furthermore, the results demonstrate a significant correlation between cell counts and chlorophyll a measurements resulting in a ratio of 1 µg Chlorophyll a/l per 2000 cells/ml. This is in agreement with Chorus et al. (2000), who assumed the same chlorophyll a to cell ratio relating to alert levels for risk assessment regarding cyanobacterial contamination. This highlights the accuracy of the described method, which is additionally, in contrast to pigment, biomass and protein analyses, not susceptible to disturbances, e.g. effects caused by sample turbidity and/or the presence of other organisms (algae and zooplankton) in the samples. This method thus provides more accuracy than commonly used methods, especially in the analysis of environmental samples. In addition, this method, requiring only a few minutes per filter, is much faster than other methods normally used for quantification of P. rubescens, e.g. most notably microscopic counting. Therefore, a higher number of samples can be counted in a given time frame allowing for continuous and precise observations of Planktothrix cell densities, an essential prerequisite for effective risk assessment and management on toxic cyanobacteria. If an appropriate epifluorescence microscope and sufficient computer capacity are present, costs for the described method are limited to camera equipment, frame grabber and the image processing software. The method described here is therefore a reasonable alternative to the relatively expensive acquirement of a submersible spectrofluorometer and/or other technical laborious measurement procedures.

ACKNOWLEDGEMENTS We would like to thank the Arthur and Aenne Feindt Foundation (Germany) as well as the European Union (PEPCY QLRT-2001-02634) for kindly funding parts of this study.

67 2. METHODICAL INOVATIONS ______

2.2. RECOVERY OF MC-LR IN FISH LIVER TISSUE

Bernhard Ernst, Lisa Dietz, Stefan J. Hoeger, Daniel R. Dietrich

Environmental Toxicology, University of Konstanz, P.O. Box X918, 78457 Konstanz, Germany

Published in Environmental Toxicology 20 (2005) 449–458

ABSTRACT Cyanotoxins, particularly microcystins, have been shown to be a hazard to human health. Microcystins accumulate in aquatic organisms probably as a result of irreversible binding to liver protein phosphatases. The aim of this study was to describe the recovery of microcystin from fish liver using various detection methods, with MC-LR as the representative congener. These findings are discussed in conjunction with the current procedures and limit values used for human risk assessment. Following incubation of liver homogenates with various MC-LR concentrations, the homogenates were extracted by a water/methanol/butanol mixture via different treatments and subsequently analysed via the colorimetric protein phosphatase inhibition assay (cPPA), HPLC, and anti-Adda ELISA. Detection via cPPA appeared to yield the highest recovery of MC-LR, although the presence of unspecific background may have resulted in overestimation of the true recovery. The recoveries determined via HPLC and anti-Adda ELISA were comparable to each other. The limits of detection were 0.01-2.4 µg MC-LR/g liver tissue, depending on the method used. Maximum MC-LR recovery from samples incubated with 10 and 100 µg MC-LR/g ranged between 44% and 101%. Recovery from samples incubated with 1 µg MC- LR/g liver tissue was below 3%. Lower recovery is assumed to result from irreversible, covalent microcystin protein binding, as confirmed by Western blotting of liver homogenates with anti- Adda immunoprobing. The results demonstrate that further investigation of and improvement in routinely applied microcystin methods for fish tissue and/or food analyses are needed for a reliable risk assessment.

KEYWORDS: Microcystin; Recovery; Fish; Tissue; Risk assessment; Cyanobacteria; Liver

68 2. METHODICAL INOVATIONS ______

INTRODUCTION Cyanobacteria occur worldwide in coastal and surface waters. To date, at least 46 cyanobacterial species have been shown to produce potent hepato- and/or neurotoxins. Approximately 75% of water samples containing cyanobacteria also contain toxic cyanobacterial metabolites (Sivonen & Jones, 1999). The most widespread cyanobacterial toxins are the microcystins and the related nodularins. These cyclic peptides are produced mainly by the cyanobacterial genera Anabaena, Anabaenopsis, Microcystis, Oscillatoria, and Nostoc. So far, nearly 80 variants of microcystins have been identified (Dietrich & Hoeger, 2005), which are responsible for the deaths of terrestrial wildlife, livestock (Briand et al., 2003), and fish (Landsberg, 2002) all over the world. Human injury, that is, liver necrosis and acute diarrhoea/gastroenteritis after acute exposure to microcystins (Annadotter et al., 2001; Byth, 1980; Pouria et al., 1998; Teixeira et al., 1993; Turner et al., 1990) and an increased incidence of primary liver or colorectal cancer after chronic exposure (Yu, 1995; Zhou et al., 2002) have been reported. This highlights the need to consider the acute and chronic effects of microcystin exposure via nutritional intake, including water, especially as human deaths have been associated with, but not proven to result from, the consumption of drinking water or food contaminated with cyanobacterial toxins (Chorus et al., 2000; Dietrich & Hoeger, 2005; Falconer, 2001). Indeed, the World Health Organisation (WHO) has recommended a provisional guideline of a maximum microcystin concentration of 1.0 µg MC-LR/l final drinking water. In this regard, it was assumed that most of the microcystin ingested daily came from contaminated drinking water (80%). However, other sources of oral microcystin exposure exist, for example, contaminated food, uptake during recreational activity, and self-inflicted exposure via cyanobacterial food supplements (Dietrich & Hoeger, 2005). Microcystins have been shown to accumulate in various aquatic organisms including mussels (Karlsson et al., 2003a; Williams et al., 1997c), crustaceans (Kankaanpää et al., 2005a; Liras et al., 1998), and fish (Kankaanpää et al., 2002a; Karlsson et al., 2003b; Sipiä et al., 2001a; Soares et al., 2004; Williams et al., 1997a). The microcystin concentrations detected in field samples of aquatic organisms varied between 0.01 and 100 µg/g tissue (Magalhaes et al., 2003; Magalhaes et al., 2001; Mohamed et al., 2003; Williams et al., 1997b; Williams et al., 1997c). Most microcystins accumulate in the liver because of the first-pass effect; however, sufficient microcystins can pass via the liver to other organs including muscle, kidney, and brain (Fischer & Dietrich, 2000; Williams et al., 1997a). For detection, microcystins are routinely extracted from animal tissue by freezing/thawing or sonication using methanol or a mixture of water/methanol/butanol (water/MeOH/BuOH) as extraction solvent (Amorim & Vasconcelos, 1999; Eriksson et al., 1989; Kankaanpää et al., 2002b; Mohamed et al., 2003; Prepas et al., 1997; Sipiä et al., 2001b). Microcystins are routinely detected in the resulting tissue extracts via HPLC-UV (Andersen et al., 1993; Eriksson et al., 1989; Lawrence & Menard, 2001; Magalhaes et al., 2001), colorimetric (cPPA), or radioactive protein

69 2. METHODICAL INOVATIONS ______phosphatase inhibition assay (Andersen et al., 1993; Malbrouck et al., 2003; Malbrouck et al., 2004b; Prepas et al., 1997; Tencalla & Dietrich, 1997; Williams et al., 1997c; Williams et al., 1995), or ELISA (Amorim & Vasconcelos, 1999; Magalhaes et al., 2003; Magalhaes et al., 2001; Sipiä et al., 2001b). However, discussion on the applicability and quality of these analytical methods is ongoing, as microcystins are at least partly covalently bound to PPs, and therefore the microcystin concentrations reported in tissue samples may reflect only freely available microcystins (Dietrich & Hoeger, 2005; Meriluoto, 1997). The aim of this study was to compare different microcystin detection methods, sample pretreatments, and extraction steps in order to determine the most reliable method or methods for routine determination of microcystin concentrations in fish tissue. These findings are discussed in conjunction with the currently employed human risk assessment procedures and limit values for contaminated foodstuffs (e.g. fish).

MATERIALS & METHODS

Sample Preparation and MC-LR Incubation Double-distilled water was purified to 18.2 MΩcm using a Milli-Q system (Millipore, Germany). All other chemicals were of the highest analytical grade commercially available. Microcystin-LR was obtained from Alexis (Switzerland). Rainbow trout (Oncorhynchus mykiss) were obtained from a local fish hatchery (250-300 g/fish). The fish were killed with a blow to the head, and the livers (2-4 g/fish) were removed, weighed, pooled, and placed in a sample buffer (10 ml of buffer/g tissue) containing 10 mM Tris HCl, 140 mM NaCl, 5 mM EDTA, Triton X-100 (1%), 1 mM PMSF, and 1 mM DTT. Tissue was minced using an Ultra Turrax T25 (Janke & Kunkel, Germany) and homogenised using a Dounce Homogeniser 3431-E20 (Thomas Technological Service, USA). Each homogenate was divided into aliquots, one of which served as a control. Homogenates were incubated with 1, 10, and 100 µg MC-LR/g tissue (Tab. 2.2). MC-LR incubation took place in continuously rotating glass vials at 30°C for 20 h in order to achieve a representative amount of covalently bound microcystin complexes (Craig et al., 1996). Four or five different homogenates of pooled liver samples were used for each MC-LR concentration (Tab. 2.2). Controls and MC-LR samples were handled identically.

Tab. 2.2: Experimental setup: Incubated fish liver weights and MC-LR quantities applied

Incubation Quantity of MC-LR Volume of Theoret. conc. on Number of [µg MC-LR/g incubated liver added resuspension column/assay incubations/ tissue] [g] [µg] [ml] [µg MC-LR/ml] controls

control 0.2-1 0 1 0 5 1 1; 2.5 1; 2.5 1; 0.5 1; 5 4 10 0.5 5 1 5 4 100 0.2 20 1 20 5

70 2. METHODICAL INOVATIONS ______

Liver MC-LR homogenat standard

Incubation

Subsample 1 Subsample 2 Subsample 3

Centrifugation Extraction

Supernatant Pellet

Extraction

C -clean up C -clean up C -clean up 18 18 18 Fig. 2.7: Experimental setup: SDS PAGE Scheme of sample aliquots, sub- immunoprobing Treatment Treatment Treatment S P E sample extraction and pre- MC analyte MC analyte MC analyte treatment.

Sample Splitting and Extraction All liver homogenate aliquots (samples), incubated for 20 h with or without MC-LR, were divided into 3 subsamples after incubation. The first subsample was stored at –20°C until use for SDS- PAGE/Western blotting. The second subsample was centrifuged (15,000 x g) for 20 min at 4°C. The supernatant (treatment S) obtained was stored at –20°C until further cleanup steps prior to microcystin analyses. The remaining pellet (treatment P) and subsample 3 (treatment E) were subsequently subjected individually to a single MC-LR extraction (Fig. 2.7). Extraction was performed using a 75:20:5 (v/v/v) mixture of water/methanol/butanol, which has been demonstrated to yield the best possible extraction (Kankaanpää et al., 2002b). The extraction involved alternate shaking and ultrasonication at 35 kHz at hourly intervals over an 8 h period. Extracts were centrifuged (20 min at 15,000 x g), and the resulting supernatants were stored at –20°C for further cleanup steps prior to microcystin analysis (Fig. 2.7).

Analytical Sub-sample Pre-treatment and Microcystin-Analysis Prior to microcystin analysis, all three subsample pre-treatment types (S, P, and E) were purified and concentrated using C18 end-capped solid-phase extraction (SPE) cartridges (Chromabond C18ec, 500 mg; Macherey-Nagel, Germany). For SPE, samples were diluted with water to give methanol concentrations of <5%. The cartridges were preconditioned using 9 ml of 100% methanol, followed by 9 ml of water. Samples were applied to the cartridges slowly, followed by two washing steps using 9 ml of MQ and 9 ml of 10% methanol. Samples were then eluted from the solid phase in the cartridge using 12 ml of 100% methanol, and the eluents were dried under a nitrogen stream and finally resuspended in 20% methanol (Tab. 2.2) to give the final treatment type-specific microcystin analytes (MC analytes; Fig. 2.7), which were stored at –20°C until microcystin analysis.

71 2. METHODICAL INOVATIONS ______

The colorimetric protein phosphatase assay (cPPA) with 4-nitrophenylphosphate (Acros Organics, Belgium) as substrate was performed as described by Heresztyn & Nicholson (2001), using recombinant protein phosphatase 1 from E. coli (New England BioLabs Inc., UK) at an end concentration of 0.375 units/ml with MC-LR as a standard. The PP-inhibiting capacities of the respective MC analytes were compared to an MC-LR standard curve in the same assay. The detection range (20%-80% inhibition of PP1) of the colorimetric PPA used was 1.5-15 µg MC-LR/l, with a derived IC50 of 4.8 µg MC-LR/l. Enzyme solution (20 µl/well) was added to 20 µl of sample in 96-well plates and incubated at 37°C for 5 min. Substrate solution (200 µl/well) was added and incubated at 37°C for 2 h, and the absorption was measured at 405 nm using an SLT Reader. The absorption was measured before and after incubation, and substrate conversion was determined as the difference between the first and second measurements. Microcystin concentrations were calculated via comparison with substrate conversion of the MC-LR standards. Each MC analyte was analysed 3 times in duplicate. The anti-Adda ELISA Kit (Abraxis LLC, USA) employed in the tests is based on an antiserum raised against the unique C20 amino acid 3-amino-9-methoxy-2,6,8-trimethyl-10-phenyl-4,6- decadieonic acid (Adda; see Fischer et al., 2001). The ELISA was performed according to the manufacturer’s instructions. Each MC analyte was analysed 3 times in duplicate. HPLC was performed using Beckman (Germany) HPLC equipment (Autosampler 507e, Solvent Module 125) with an analytical C18 column (Grom-Sil 120 ODS-4 HE, 5 µm, 250 x 4 mm). A gradient with water (0.05% TFA) and acetonitrile (0.05% TFA) as the mobile phase was used according to the method described by Lawton et al. (1994). MC-LR was detected using a photodiode array SPD-M10A VP (Shimadzu, Germany) and identified via retention time and typical spectrum in comparison with internal MC-LR standards. MC-LR concentrations were calculated using peak area and peak height. HPLC MC-LR analysis was carried out once for each MC-LR analyte. For qualitative detection of covalently bound microcystin adducts, subsamples 1 (Fig. 2.7) were separated via 10% SDS PAGE in accordance with Laemmli (1970). The protein content of each treatment was determined according to the method of Bradford (1976) and adjusted to give a protein load of 60 µg protein/lane. Separated proteins were transferred onto a nitrocellulose membrane via Western blot. The membranes were blocked using TTBS + 1% BSA for 30 min, and MC-LR adducts were detected via incubation with polyclonal sheep anti-Adda serum (diluted 1:1000 in blocking buffer) at room temperature for 1 h according to Fischer & Dietrich (2000). Membranes were washed using TTBS (3 x 5 min) and incubated with secondary antibody (anti- sheep IgG-AP, diluted 1:5000 in TTBS; Sigma-Aldrich, Germany) at room temperature for 1 h. After washing with TTBS (3 x 5 min) and TBS (1 x 15 min), specific bands were finally stained using Sigma Fast Red® (Sigma-Aldrich, Germany) according to the manufacturer’s instructions. The molecular weights of detected adducts were estimated by comparison with full-range rainbow marker proteins RPN 800 (Amersham, UK).

72 2. METHODICAL INOVATIONS ______

Statistics Data analyses were carried out using JMP® (USA) software. Values represent the mean ± SD of at least three separate experiments. Results of the HPLC, ELISA, and PPA analyses, as well as results of different treatments, were analysed for statistical differences using analysis of variance (ANOVA) and the Tukey-Kramer multiple comparisons test (p ≥0.05). The sum of MC-LR concentrations analysed in analyte samples of treatments S and P (S + P) were tested for statistical differences (p ≥0.05) to concentrations analysed via treatment E using the Student’s t test.

RESULTS All MC analytes were analysed by cPPA, HPLC, and anti-Adda ELISA. Detected MC-LR concentrations in the respective analytes and their corresponding deduced tissue concentrations and recoveries are shown in Tables 2.3-2.5. Analyte samples of treatment S were analysed for quantification of soluble MC-LR. Samples of treatment P were analysed to recover the remaining, extractable, and not readily soluble MC-LR. Analyte samples of treatment E were analysed to quantify the overall extractable MC-LR in the incubated tissue homogenates. In all treatment approaches, the sum of MC-LR concentrations analysed in the S + P analyte samples were not significantly different from recovery from whole tissue homogenate extraction (treatment E), regardless of the MC-LR concentration and detection methods used. The highest MC-LR recovery generally was achieved using whole-tissue homogenate extraction (treatment E) without previous centrifugation and separation (Tab. 2.3-2.5). However, the overall extractable MC-LR analysed by treatment E did not always differ significantly from the quantities of soluble MC-LR in treatment S.

Tab. 2.3: MC-LR concentrations detected, corresponding tissue concentrations and calculated recovery after 20 h incubation of fish liver tissue with various MC-LR concentrations using different sample treatments (Fig. 2.7) with colorimetric protein phosphatase inhibition assay (cPPA) as the analytical method

cPPA

Detected Detected Number of Incubation concentration tissue conc. incubations [µg MC-LR/g] [µg/ml] [µg/g] Recovery [%] analysed

S Control 0.27 ± 0.212 4 1 i.d. i.d. i.d. 3 10 3.51 ± 0.511 7.02 ± 1.022 70 ± 10.2 3 100 14.19 ± 2.793 70.95 ± 13.96 71 ± 13.9 4 P Control 0.38 ± 0.415 4 1 i.d. i.d. i.d. 3 10 1.21 ± 0.298 2.42 ± 0.596 24 ± 05.9 3 100 1.89 ± 0.964 9.45 ± 4.820 9 ± 04.8 3 E Control 0.39 ± 0.219 4 1 i.d. i.d. i.d. 3 10 5.05 ± 1.288 10.10 ± 2.576 101 ± 25.8 4 100 18.59 ± 5.593 92.95 ± 27.96 93 ± 28.0 3

i.d. = indistinguishable from controls

73 2. METHODICAL INOVATIONS ______

Tab. 2.4: MC-LR concentrations detected, corresponding tissue concentrations and calculated recovery after 20 h incubation of fish liver tissue with various MC-LR concentrations using different sample treatments (Fig. 2.7) with high performance liquid chromatography (HPLC) as the analytical method

HPLC

Detected Detected Number of Incubation concentration tissue conc. incubations [µg MC-LR/g] [µg/ml] [µg/g] Recovery [%] analysed

S Control n.d. 3 1 n.d. n.d. n.d. 3 10 2.25 ± 0.31 4.5 ± 0.60 45 ± 06.2 3 100 12.63 ± 1.66 63.2 ± 8.30 63 ± 08.3 4 P Control n.d. 3 1 n.d. n.d. n.d. 3 10 0.21 ± 0.11 0.4 ± 0.22 4 ± 02.2 3 100 1.03 ± 0.50 5.2 ± 2.50 5 ± 02.5 4 E Control n.d. 3 1 n.d. n.d. n.d. 3 10 2.19 ± 0.50 4.4 ± 1.00 44 ± 10.0 4 100 16.10 ± 1.37 80.5 ± 6.85 81 ± 06.8 4

n.d. = not detectable

When comparing the different detection methods, determinations by HPLC and ELISA yielded comparable MC-LR recoveries in all MC-LR concentrations and sample treatments applied. Analyses with cPPA generally resulted in higher recovery than detection via either HPLC or ELISA (Tab. 2.3-2.5), although this was not significant at all MC-LR concentration levels (1, 10, and 100 µg/g) used. Analyses of controls via cPPA resulted in an average background noise equivalent to 0.3-0.4 µg MC-LR/ml analyte (Tab. 2.3). Nonspecific positive signals were also observed with the anti-Adda ELISA and control sample analytes (Tab. 2.5). This background noise, however, corresponded to ≤0.01 µg MC-LR/ml analyte and therefore was at least 30 times lower than the background

Tab. 2.5: MC-LR concentrations detected, corresponding tissue concentrations and calculated recovery after 20 h incubation of fish liver tissue with various MC-LR concentrations using different sample treatments (Fig. 2.7) with anti-Adda-ELISA as the analytical method

ELISA

Detected Detected Number of Incubation concentration tissue conc. incubations [µg MC-LR/g] [µg/ml] [µg/g] Recovery [%] analysed

S Control <0.01 3 1 0.01 ± 0.003 0.01 ± 0.003 1 ± 00.3 3 10 2.33 ± 0.116 4.66 ± 0.232 47 ± 02.3 3 100 10.42 ± 2.009 52.10 ± 10.05 52 ± 10.1 3 P Control <0.01 3 1 0.02 ± 0.006 0.02 ± 0.006 2 ± 00.6 3 10 0.37 ± 0.135 0.74 ± 0.270 7 ± 02.7 3 100 0.99 ± 0.247 4.95 ± 1.235 5 ± 01.2 3 E Control 0.01 ± 0.005 3 1 0.03 ± 0.013 0.03 ± 0.013 3 ± 01.3 3 10 2.89 ± 0.536 5.78 ± 1.072 58 ± 10.7 3 100 13.49 ± 0.352 67.45 ± 1.760 68 ± 01.8 3

74 2. METHODICAL INOVATIONS ______observed in the cPPA analyses. Conversely, no background noise relevant to the MC-LR peak retention time was observed in the HPLC analyses. To compare the results obtained with the three methods of microcystin analysis, concentrations detected via cPPA and ELISA were corrected for nonspecific background noise (subtraction of the background noise from the raw value in the analysis). Limits of detection for MC-LR in the liver homogenate samples used in this study were 1.2 µg MC-LR/ml (cPPA, Tab. 2.3), 0.2 µg MC-LR/ml (HPLC, Tab. 2.4), and 0.01 µg MC-LR/ml (anti-Adda ELISA, Tab. 2.5). These detection limits translate to minimum MC-LR tissue concentrations of 2.4 µg MC-LR/g for cPPA (Tab. 2.3), 0.4 µg MC-LR/g for HPLC (Tab. 2.4), and 0.01 µg MC-LR/g for anti-Adda ELISA (Tab. 2.5). Maximum MC-LR recovery from liver homogenate samples incubated with 10 and 100 µg MC-LR/g, for example, for treatment E (whole homogenate extract), ranged between 44% and 101%, depending on the initial MC-LR concentration and detection method used (Tab. 2.3- 2.5). The reliability of the recovery data (the variance) largely depended, however, on the analytical method used. The analytical methods ranked in order from worst to best are: cPPA < HPLC < anti-Adda ELISA. In contrast, MC-LR recovery from samples incubated with 1 µg MC-LR/g liver tissue was extremely low (1-3%), even after water/MeOH/BuOH extraction. Indeed, anti-Adda ELISA analyses provided a detectable MC-LR concentration of only 0.01-0.03 µg MC-LR/g liver tissue, depending on the sample pretreatment used (Tab. 2.5). As a result of having a higher limit of detection, neither the cPPA (Tab. 2.3) nor the HPLC-PDA (Tab. 2.4) analysis was able to detect MC-LR in the 1 µg/g liver homogenate samples. Western blot analysis showed that anti-Adda immunostaining of positive protein adducts (30-35 kD) could be detected in all liver homogenate samples incubated with MC-LR (1, 10, and 100 µg/g liver; Fig. 2.8).

75 kD MC-positive adducts 50 kD

35 kD

30 kD

Control 100 µg/g 10 µg/g 1 µg/g Marker protein

Fig. 2.8: Immunostaining of MC-LR-adducts in fish liver homogenates following 20 h incubation with 1, 10 and 100 µg MC-LR/g tissue. Molecular weights were estimated via comparison with marker proteins.

75 2. METHODICAL INOVATIONS ______

DISCUSSION The results demonstrated varying levels of MC-LR recovery from fish liver homogenates, depending on the treatment of the homogenate subsamples and the detection method used. Comparing recovery resulting from different subsample treatments, whole-tissue homogenate extraction provided comparable MC-LR recovery to the sum of recoveries of treatments S and P (E = S + P). This suggests that water/MeOH/BuOH extraction increases MC-LR recovery from incubated liver tissue. However, as the overall extractable MC-LR analysed in treatment E samples did not always differ significantly from those of soluble MC-LR in treatment S, the additional MC-LR recovery achieved by homogenate water/MeOH/BuOH extraction appears to be limited. For example, water/MeOH/BuOH extraction is not useful for extracting covalent bond microcystin. Indeed, repetition of the extraction procedure resulted in no additional MC-LR recovery (data not shown). These findings are in agreement with those of Kankaanpää et al. (2002b), who demonstrated that extraction of nodularin from animal tissue yielded the best extraction results (highest recovery) using water/MeOH/BuOH 75:20:5 (v/v/v) with an 8 h extraction time and that extraction repetition did not improve recovery. On the contrary, repetitious extraction increased the amount of matrix compounds interfering with HPLC analysis of nodularin.

MC-LR recovery detected by cPPA was generally higher than that determined by anti-Adda ELISA or HPLC, independent of subsample treatment and the amount of MC-LR used for incubation of the homogenate. This is most likely a result of the high background of non-MC-LR- related PP inhibition, as strongly suggested by the PP inhibition observed in the controls. It is assumed that this high background of non-MC-LR-related PP inhibition may be the result of (i) specific endogenous PP1 inhibitors (Oliver & Shenolikar, 1998) in the liver homogenates liberated during the liver homogenisation process and/or (ii) unspecific influences arising from matrix effects. These observations were corroborated by the findings of Sipiä et al. (2001a), who demonstrated that nodularin could not be detected in the muscle of Atlantic salmon via MC-LR ELISA, whereas analysis by cPPA resulted in 55-65 ng NOD/g. Sipiä et al. (2001a) also concluded that the disparate findings between the analyses by MC-LR ELISA and by cPPA were the result of turbidity and colour as well as of matrix-interfering compounds in the tissue homogenates and the resulting extracts. Consequently, this unspecific PP inhibition causes an overestimation of the microcystin and nodularin contamination in tissues in general and of MC-LR contamination in fish liver homogenates specifically, as presented in this study. Therefore, cPPA with PP1 appears inappropriate for routine microcystin and nodularin detection in tissue samples. In contrast to cPPA detection, overall MC-LR recovery from liver homogenates determined by HPLC was comparable to that determined by anti-Adda ELISA. However, no MC-LR was detectable by HPLC in homogenates incubated with 1 µg MC-LR/g, whereas the analysis with

76 2. METHODICAL INOVATIONS ______anti-Adda ELISA resulted in detectable MC-LR concentrations of up to 0.03 µg/ml analyte. Similar observations were made by Kankaanpää et al. (2002a), who demonstrated that no nodularin was detectable via HPLC analysis in liver tissue from sea-trout, orally dosed with nodularin; in contrast MC-LR ELISA analysis resulted in nodularin concentrations of up to 1.2 µg/g tissue. These concentrations exceeded the detection limit for nodularin quantification in liver tissue (0.15 µg NOD/g) of the chromatographic system employed and thus should actually have been detectable via HPLC. Kankaanpää and co-authors concluded, in accord with Metcalf et al. (2000) that the discrepancy between HPLC and ELISA analyses is most likely a result of the detection of additional nodularin conjugates via the ELISA assay, whereas these conjugates would not be readily detectable via HPLC. The detection limit of the HPLC system employed in this study was approximately 10 ng per injection. Thus, the injection volume of 50 µl of analyte resulted in an absolute detection limit of 0.2 µg MC-LR/ml analyte. As the MC-LR concentrations (anti-Adda ELISA) that could be determined in the homogenates incubated with 1 µg MC-LR/g were far below the HPLC detection limit, it is not surprising that no MC-LR was detectable in these samples using HPLC. Karlsson et al. (2003b) reported that matrix effects hindered HPLC-UV detection of nodularin in liver tissue samples of flounder, and Sipiä et al. (2001a) described HPLC as an inappropriate method for nodularin analysis in liver tissue samples using mobile and stationary phases with UV detection, as small concentrations (≤0.3 µg NOD/10 µl injection) of nodularin easily escaped detection. Contrary to these findings, the present results suggest that HPLC-UV is an accurate analytical method for quantification of microcystin in tissue samples, as long as microcystin tissue contamination is greater than 0.2 µg MC-LR/ml analyte, as specified by the sample treatment and extraction method used in this study. Unspecific competitive background binding also was observed in the anti-Adda ELISA with control samples, representing less than 0.01 µg MC-LR/ml approximately. Sipiä et al. (2001a) described nodularin concentrations of less than 0.01 µg/g liver tissue to be below the level of quantification in MC-LR ELISA because of matrix effects. Kankaanpää et al. (2002a) described a level of non-specific binding to antibodies of approximately 0.02 µg MC/g liver tissue of sea trout analysed for nodularin with MC-LR ELISA. Kankaanpää et al. (2005a) also suggested that the theoretical detection limits of MC-LR ELISA for nodularin analysis in hepatopancreas and muscle tissue of prawns are affected by low-level matrix effects because of unspecific binding to and/or denaturing of the antibodies. Matrix effects in the analyses of nodularin in liver tissue samples of flounder also have been reported by Karlsson et al. (2003b), suggesting that analysis via MC-LR ELISA is not optimal if it is the only means of toxin analysis in tissue. In contrast, the results of the study reported here demonstrated that background/matrix-associated effects and thus the limit of detection in the anti-Adda ELISA employed were at least 30 times lower than in the cPPA and HPLC methods used. Moreover, in conjunction with the MC-LR amounts recovered from incubated homogenates in this study, overestimation of MC-LR contamination of tissue because of unspecific measurement by the anti-Adda ELISA appeared negligible.

77 2. METHODICAL INOVATIONS ______

In summary, when comparing the microcystin detection methods employed in this study and the results obtained with those previously reported, anti-Adda ELISA appears to be the most appropriate method for the detection of microcystin in tissue samples. In tissue samples contaminated with relatively high microcystin concentrations (>0.4 µg MC-LR/g), both anti-Adda ELISA and HPLC-UV appear to be suitable methods for reliable microcystin detection in tissue samples. However, generally, a prerequisite for acceptable analyses is triplicate analyses and relevant standards and controls. The complementary use of different detection methods, that is, simultaneously using two or more analytical methods for the same analytes, is highly recommended, in agreement with other authors (Meriluoto, 2004; Metcalf et al., 2000), in order to achieve reliable detection of microcystin contamination in tissue samples.

Kankaanpää et al. (2002b) specified three possible reasons for incomplete recovery of nodularin and microcystin, as also observed in the study reported here: (i) loss during the analytical procedure, (ii) metabolism (conjugation) in tissue and, (iii) covalent binding of microcystin to macromolecules (proteins and peptides). One possible way to verify the abundance of microcystin- protein adducts is through immunoprobing using antibodies raised against microcystin or microcystin fragments, for example, the Adda moiety (Fischer & Dietrich, 2000; Hitzfeld et al., 1999; Mikhailov et al., 2003). As noncovalently bound MC-LR is expected to elute from a SDS- PAGE denaturing gel, Adda-positive bands observed in the Western blots most likely represent MC-LR protein adducts in the liver homogenates analysed. That cyanobacterial peptides such as nodularin, which do not appear to covalently bind to proteins, could not be detected in Western blots using the appropriate antibodies corroborates this (Mikhailov et al., 2003; Schmid et al., 2004). The presence of bands visible by anti-Adda immunostaining is therefore a distinct indication of the presence of covalently bound MC-LR adducts in incubated homogenates. This is in agreement with former studies, showing adducts in the 28-38 kD range that most likely represent PP-microcystin adducts to the liver endogenous protein phosphatases (Ernst et al., 2001; Fischer & Dietrich, 2000; Hitzfeld et al., 1999; Mikhailov et al., 2003). In agreement with the findings of previous investigations (Amorim & Vasconcelos, 1999; Kankaanpää et al., 2002b; Meriluoto, 1997; Williams et al., 1997a; Williams et al., 1997b), we observed loss in recovery that was probably attributable to irreversible covalent microcystin binding. According to MacKintosh et al. (1990), the microcystin-binding capacity in mice liver is expected to be approximately 1 µg/g. Similarly, Yoshida et al. (1998) estimated the amount of irreversibly bound microcystin in mice liver as 0.7 µg MC/g liver. Assuming a similar microcystin-binding capacity for fish, binding capacities may be expected also to be saturated for homogenates incubated with 1 µg MC/g. Consequently, microcystin recovery was only 1-3% in homogenates incubated with 1 µg MC/g. To extrapolate the above findings and conclusions to the routine situation of laboratory analyses of food samples and their safety assessment according to regulatory recommendations, the WHO recommendations and guidance values were employed in the analysis of the tissue levels used in

78 2. METHODICAL INOVATIONS ______this study in order to provide a reasonable example of risk calculation and extrapolation. The

WHO suggests a tolerable daily intake (TDI) of 0.04 µg MC-LRequiv./kg food a day, where

MC-LRequiv. is the sum of all microcystin congener concentrations likely to be in the food as contaminants. To determine an interim maximum acceptable concentration (IMAC) in fish used for consumption, the following equation was applied in accordance with the publication by Falconer (2001): IMAC = TDI x BW x POT/AFC BW: body weight POT: proportion of toxin consumed in form of contaminated fish AFC: average fish consumption

Average human body weight is assumed to be 60 kg. Based on Egyptian and Brazilian eating habits, the AFC ranges from 100 to 300 g per day per person (Magalhaes et al., 2001; Mohamed et al., 2003). According to the European Commission (2004), worldwide fish consumption is calculated as 43 g per person per day (including fish from marine water, brackish water, and freshwater). People in Europe on average consume 67 g per person per day, ranging from 31 g in Austria to 167 g in Portugal. Average German fish consumption is 35 g/day. However, only 8 g of that is freshwater fish or fish from mildly saline waters, which are more likely to be affected by cyanobacterial blooms (http://europa.eu.int/comm/fisheries). Performing a universal risk assessment is difficult because of large differences in consumption and exposure conditions (high variation in AFC and POT). Consequently, guideline values for microcystin contamination of fish must be based on local customs, conditions, and circumstances. Using various assumed AFC and POT levels, the interim maximum acceptable contamination for fish was calculated and demonstrated to vary between 0.002 and 0.18 µg/g tissue (Tab. 2.6). Assuming detection limits in liver tissue (as demonstrated here) to be similar to that in muscle tissue, these IMACs were then compared to the three analytical methods used for MC-LR analysis in this study, from which it was determined that cPPA is generally inappropriate for microcystin tissue contamination analysis. Considering the relevant limits of detection, anti-Adda ELISA and HPLC may be used for certain IMACs (Tab. 2.6). However, it also was clearly demonstrated that for providing safe and healthy food for consumers neither anti-Adda ELISA nor HPLC, depending on the AFC, is sufficiently sensitive to allow reliable detection and thus

Tab. 2.6: Estimation of an interim maximum AFC [g/d]/ POT [%] 0.25 0.5 0.75 acceptable contamination (IMAC) for fish

IMAC [µg/g] based on a tolerable daily intake of 0.04 µg MC/kg bw. The IMAC was calculated based 10 0.060 E; H 0.120 E; H 0.180 E; H on 60 kg average body weight, a diverse

E E E 50 0.012 0.024 0.036 average fish consumption (AFC) and various percentages of uptake of toxin (POT) via fish 300 0.002 0.004 0.006 consumption. E detectable using anti-Adda-ELISA H detectable using HPLC

79 2. METHODICAL INOVATIONS ______regulation of fish contaminated with microcystin (Tab. 2.6). In addition, on the basis of the findings of MacKintosh et al. (1990), Yoshida et al. (1998), and the results reported here, it can be assumed that in tissue contaminated with concentrations of ≤1 µg MC/g, most microcystin is bound covalently. In this respect, and given the unsatisfactory detection limits as mentioned above, further investigation and improvement of routinely applicable microcystin methods for fish tissue and/or food analyses are essential requirements for an effective risk assessment. Current investigations have demonstrated that such recent developments and improvements might include immunoaffinity chromatography, LCMS, and MALDI-TOF analyses (Hormazabal et al., 2000; Karlsson et al., 2003a; Karlsson et al., 2003b; Lawrence & Menard, 2001).

ACKNOWLEDGEMENT We thank the Arthur and Aenne Feindt Foundation (Germany) as well as the European Union (PEPCY QLRT-2001-02634) for kindly funding parts of this study.

3. EXPOSURE EXPERIMENTS

3.1. ORAL TOXICITY OF THE MICROCYSTIN-CONTAINING CYANOBACTERIUM PLANKTOTHRIX RUBESCENS IN EUROPEAN WHITEFISH (COREGONUS LAVARETUS)

Bernhard Ernst, Stefan J. Hoeger, Evelyn O’Brien, Daniel R. Dietrich

Environmental Toxicology, University of Konstanz, P.O. Box X918, 78457 Konstanz, Germany

Published in Aquatic Toxicology 79 (2006) 31–40

ABSTRACT The microcystin-producing cyanobacterium Planktothrix is one of the most widespread genera amongst toxin producing cyanobacteria in European lakes. In particular, the metalimnic blooms of Planktothrix rubescens have been associated with growing problems in the professional freshwater fishery as a decrease in yearly yields in the important coregonids fishery often coincides with the appearance of P. rubescens. P. rubescens is a cyanobacterial species known to produce toxic compounds, e.g. microcystins. Although microcystins have been reported to affect fish health, behaviour, development and growth and have also been associated with feral fish kills, there is currently no specific information on the effects of toxic Planktothrix filaments in fish and especially coregonids. Therefore, the aim of this study was to investigate the effects of an environmentally-relevant dose of P. rubescens filaments orally applied to coregonids and to discuss the findings in the context of microcystin toxicity previously reported in carp and trout. A single dose of P. rubescens culture, at a density of 80,000 cells per 120 µl, was applied to coregonids thus corresponding to 0.6-0.9 µg microcystin-LRequiv./kg body weight. Behavioural changes and opercular beat rates, growth, hepatosomatic index, condition and plasma glucose were determined. Liver, kidney, gill and the gastrointestinal tract were assessed histopathologically and immunhistologically. Exposed fish showed behavioural changes, increased beat rates and elevated plasma glucose levels, possibly representing a physiological stress response. Histopathological alterations in liver, gastrointestinal tract and kidney, also immunopositive for microcystin suggested causality of tissue damage and the in-situ presence of microcystins. The observed combination of stress and organ damage may explain the frequently reduced weight and thus the fitness noted in coregonids subjected to regular occurrences of stratified and dispersed P. rubescens blooms, e.g. in lake Ammersee, Bavaria, Germany.

KEYWORDS: Fish; Microcystin; Planktothrix; Coregonids; Cyanobacteria; Whitefish

81 3. EXPOSURE EXPERIMENTS ______

INTRODUCTION Toxic cyanobacteria occur world wide in fresh and coastal waters. Due to their ability to produce highly toxic metabolites, i.e. the neurotoxins anatoxin, saxitoxin and the potent protein phosphatase inhibitors microcystins and nodularins, mass occurrences of cyanobacteria have been associated with human intoxications (mild to lethal) and mortality of wild and domestic animals (Briand et al., 2003; Dietrich et al., 2008). To date, at least 46 cyanobacterial species are known to produce toxins, thus it is not surprising that approximately 75% of cyanobacteria samples taken in surface waters have been shown to contain toxins (Sivonen & Jones, 1999). The microcystin-producing Planktothrix is one of the most important genera amongst the widespread toxin producing cyanobacteria in temperate climates. Planktothrix species are predominant in several monomictic and dimictic European lakes, especially in the pre-alpine regions (Davis et al., 2003; Henriksen, 2001; Krupa & Czernas, 2003; Lindholm et al., 2002; Mez, 1998; Morabito et al., 2002; Utkilen et al., 2001). Planktothrix sp. generally occur in eutrophic waters, building blooms during winter circulation and metalimnic layers during lake stratification in summer. Planktothrix rubescens can also occur in mesotrophic and even oligotrophic lakes that have recently undergone an anthropogenically induced phase of nutritional re-depletion (re-oligotrophication) (Ernst et al., 2001; Jacquet et al., 2005). Drinking water management (Hitzfeld et al., 2000; Hoeger et al., 2005) and professional freshwater fishery (Ernst et al., 2001) have been faced with growing problems due to metalimnic blooms of P. rubescens. Indeed, a decrease in yearly yields in the important coregonids fishery was observed to coincide with the appearance of P. rubescens blooms in Swiss and German lakes as of the beginning of the last century (Braun, 1953; Ernst et al., 2001). Mass occurrence of toxic cyanobacteria have been associated with feral fish kills (Jewel et al., 2003; Rodger et al., 1994; Toranzo et al., 1990). Toxic effects have been shown for various fish species, including salmoniformes, siluriformes, cypriniformes as well as perciformes and clear differences in fish species sensitivity to toxic cyanobacteria in general (as cell suspensions or bloom material) and microcystins (primarily microcystin-LR (MC-LR)) have been demonstrated

(Malbrouck & Kestemont, 2006). The LD50 value for orally applied MC-LR to carp was reported to be <1.7 mg/kg body weight (Tencalla, 1995), while the LD50 for orally applied MC-LR to trout was found to range between 1.7 and 6.6 mg/kg body weight (Tencalla et al., 1994). Similar species differences were demonstrated for intraperitoneally applied MC-LR (Carbis et al., 1996a; Råbergh et al., 1991; Tencalla et al., 1994). The observed species-specific sensitivities to microcystins have been interpreted as resulting from anatomical, physiological and behavioural differences amongst the various fish orders (Fischer & Dietrich, 2000; Tencalla, 1995). Uptake of cyanobacterial toxins by fish results primarily following oral ingestion of toxins or toxic cyanobacterial cells and to a negligible extent from toxin uptake via the gill epithelium (Bury et al., 1998b; Tencalla et al., 1994). Fish exposed to cyanobacterial bloom material or acutely toxic concentrations of microcystins presented with liver, kidney and gill pathology, specific inhibition

82 3. EXPOSURE EXPERIMENTS ______of protein phosphatases and other downstream effects, e.g. increased liver enzyme values in the serum. In addition behavioural changes and decreased development of juvenile fish have been observed (Malbrouck & Kestemont, 2006). Most fish exposure experiments have been conducted using either pure MC-LR or MC-LR containing Microcystis aeruginosa. Compared to other cyanobacteria, Planktothrix sp. have been shown to contain the highest amounts of microcystin per gram dry weight (Fastner et al., 1999b). In contrast to M. aeruginosa, Planktothrix sp. contain various demethylated variants of MC-RR (Blom et al., 2001; Keil et al., 2002; Luukkainen et al., 1993). Apart from some reports on Planktothrix associated fish kills (Berg et al., 1986), there is no information on the toxicity of Planktothrix sp. filaments and demethylated MC-RR variants in fish and especially coregonids. In addition, Planktothrix sp. produce a range of other metabolites, e.g. anabaenopeptins, microviridins and cyanopeptolins (Blom et al., 2003) with yet unknown toxicological properties. Therefore, the toxic potential of Planktothrix blooms to freshwater fish, and especially coregonids populations is presently difficult to assess. Coregonids (Coregonus sp.), one of the most important species for professional fishery, are among the dominant fish species in most of the stratified European lakes. Ernst et al. (2001) proposed a possible link between the occurrence of toxic Planktothrix blooms and changes in growth and population dynamics of coregonids, thus providing a possible explanation for the observed decreases in fishery yields. However, despite the existence of a plausible explanation by association and the serendipity of observed events, there is at present no information available on the specific sensitivity of coregonids to cyanobacterial blooms and/or cyanobacterial toxins. With respect to the anatomical organisation, the digestive tract of coregonids includes a stomach, pyloric caeca and a short ileum. Thus coregonids are anatomically very similar to trout, both taxonomically belonging to the order of salmoniformes. However, in contrast to trout, coregonids feed exclusively on plankton. The feeding habits of the adult coregonids therefore resemble more those of the planktivorous cyprinid species rather than those of the predominantly piscivorous salmonids. Consequently, neither carp (cyprinids) nor trout (salmonids) serve as reliable surrogate species to assess the toxicity of Planktothrix blooms in coregonids. The aim of this study was therefore to investigate the effects of an environmental relevant dose of P. rubescens filaments orally applied to coregonids (Coregonus lavaretus L.) and to discuss the findings in the context of toxicity studies with M. aeruginosa bloom of freeze-dried material and/or MC-LR in carp and trout.

MATERIAL & METHODS

Chemicals and Reagents All chemicals were of the highest analytical grade commercially available. Microcystin-LR (MC- LR) was obtained from Alexis (Switzerland). [D-Asp3-(E)-Dhb7]-microcystin-RR (Asp3-Dhb7-MC- RR) was kindly provided by Judith Blom, University of Zurich, Switzerland and [D-Asp3]- microcystin-RR (Asp3-MC-RR) by Jussi Meriluoto, Abo Akademi University Turku, Finland.

83 3. EXPOSURE EXPERIMENTS ______

Cultivation, Toxin Extraction and Characterisation of P. rubescens P. rubescens was originally isolated from a Lake Ammersee seston sample in autumn 2002 and cultivated in BG11 medium according to the method described by Rippka et al. (1979). P. rubescens cell density was determined via image processing as described by Ernst et al. (2006b). Microcystin-LR equivalent content and microcystin congener composition of P. rubescens were characterised via anti-Adda MC-ELISA (Abraxis, USA; see Fischer et al., 2001), colorimetric phosphatase-inhibition assay (cPPA; see Heresztyn & Nicholson, 2001), and HPLC using extract samples (70% methanolic- followed by solid phase extraction; see Ernst et al., 2005). Adda-ELISA and cPPA analyses were carried out on three independent replicates of duplicate samples. MC-LR equivalent content (MC-LRequiv.) was calculated using an MC-LR standard as reference. HPLC was performed as described by Ernst et al. (2005). Microcystin congeners were characterised using retention time and known typical spectra in comparison with Asp3-Dhb7-MC-RR and Asp3-MC-RR standards. Quantification of the Asp3-MC-RR variant(s) was achieved using the peak area which was then transformed to express MC-LR equivalent content

(MC-LRequiv.).

Fish Exposure One-year-old coregonids (Coregonus lavaretus), with an average weight and average length of 9.9 ±1.9g and 11.4 ±0.7cm, respectively, were obtained from the fisheries administration department Uri, Altdorf, Switzerland. Fish were fed (15% of body weight daily) with frozen chironomid larvae (Honka, Germany). Coregonids were acclimatised for one week prior to P. rubescens gavage in 100 litre flow-through tanks. Tanks were supplied with tap water and additionally aerated using commercial 5 W aeration pumps (Tetra, Germany). Previously described P. rubescens cell densities determined in lakes attained up to 150,000 cells/ml (Hoeger et al., 2005). Maximum cell densities determined in Lake Ammersee range up to 80,000 cells/ml (Ernst et al., 2001). Fuentes & Eddy (1997) determined for trout fry an average water uptake of up to 0.5 ml kg-1 h-1. Assuming a comparable water uptake for coregonids and considering the above mentioned P. rubescens cell densities, a fish weighing 10 g approximately requires 4 and 8 days, respectively, in order to ingest a dose of 80,000 cells. As the described P. rubescens blooms/ layers usually persist for weeks and months (Ernst et al., 2001; Jacquet et al., 2005), this dose appeared to be environmental relevant and was subsequently applied in the gavage experiment. After acclimatisation, fish were anesthetised by 100 mg/l ethyl 3-aminobenzoate methanesulfonate (MS-222; Fluka, Germany) for gavaging. A single 120 µl dose of P. rubescens culture, at a density of 80,000 cells per 120 µl, was applied to 24 coregonids using blunt-tip gavage syringes (Roth, Germany). Six coregonids served as sham-control and received a single dose of 120 µl 0.9% of NaCl while an additional six coregonids were used as corresponding control (no gavage). Following gavage, P. rubescens gavaged coregonids were placed in two separate tanks (12 fish per tank), control and sham-control fish were placed in separate tanks (6 fish per

84 3. EXPOSURE EXPERIMENTS ______

Tab. 3.1 A: Number of coregonids sampled post application (p.a.) from two groups gavaged with a single P. rubescens dose and the respective control- and sham-control groups for histological and analytical assessment

Tank n Treatment Before 9 h p.a. 24 h p.a. 48 h p.a. 72 h p.a.

I 6 Control 3 3 II 6 Sham-control 3 3 III 12 Exposed 3 3 3 3

IV 12 Exposed 3 3 3 3

tank) (Tab. 3.1 A). Water temperature was measured twice daily. The 95% confidence interval of the tank temperature was [14-15 °C] and [12-15 °C] in the tanks stocked with exposed and sham- control fish, respectively.

Behavioural observations and opercular beat rate Fish were observed for behavioural changes daily. The opercular beat rate of sham-control and exposed fish was determined one hour before (basal rate) and 1, 3, 19, 25, 29, 46, 54 and 70 h post gavage (Tab. 3.1 B) by counting the opercular movement of four individuals per tank for 15 s each. The determined opercular counts were adjusted for temperature effect via addition of a correction factor c (c = (15 °C – actual temperature) x 0.5) derived from a normalised opercular rate in the sham-control tank. Adjusted counts were multiplied by four to give an opercular rate per min.

Experimental Parameters Six P. rubescens gavaged coregonids were sampled 9, 24, 48 and 72 h post-application. Three fish of the control group were sampled before and 48 h post-application, three fish of the sham-control 24 h and 72 h post-application as depicted in Tab. 3.1 A. The experimental parameters determined during and post-exposure are shown in Tab. 3.1 B and described below:

WEIGHT: Wet weight of fish was determined on a fine balance (PB3002; Mettler, Germany).

LENGTH: Length was determined from nose-tip to tail using a centimetre ruler and determined to the nearest mm.

Tab. 3.1 B: Time-points of parameter determination during the experiment

Time post application [h]

-1 +1 3 9 19 2425 29 46 48 54 70 72

Beat rate x x x xxxx x x Weight x x x x HSI x x x x Condition x x x x Plasma glucose x x x x Histopathology x x x x Immunhistology x x x x

85 3. EXPOSURE EXPERIMENTS ______

CONDITION FACTOR: Condition was determined using the condition factor: CF = weight [g] x (length [mm])-3 x 105, (Barton et al., 2002). HSI: The hepatosomatic index (HSI) was calculated as HSI = liver weight [g] x 100 x body weight [g] -1).

PLASMA GLUCOSE: Plasma glucose was determined in blood taken from the caudal vein, using a 1 ml syringe and a 0.40 x 20 mm needle (Braun, Germany). Whole blood samples were immediately frozen and stored at –20 °C. Before measurement, samples were thawed and centrifuged at 16,000 x g for 15 min. Subsequently, plasma glucose was determined by analysing the remaining supernatant using an automatic sensor (Accu–Check; Roche, USA) according to the manufacturer’s instructions. Samples were analysed in triplicate.

HISTOPATHOLOGY: Liver, kidney, gill, pylori and hindgut were dissected from each fish, a representative tissue sample placed into a labelled tissue cassette and briefly fixed in 4% PBS- buffered formalin. Tissues were routinely processed by RCC Ltd. (Itingen, Switzerland), i.e. paraffin embedded, sectioned to 3-5 µm, mounted on microscope slides and stained with haematoxylin and eosin (H&E). Histopathological assessment was carried out by light microscopy at 40 to 400-fold magnification. Pathological changes were classified as none (0), mild (1), moderate (2), strong (3) and severe (4), including intermediate classes, e.g. 0.5, 1.5 etc.

IMMUNOHISTOCHEMICAL DETERMINATION OF MICROCYSTIN: Liver, kidney, gill and intestine tissue on polysin-coated glass slides were deparaffinised in 100% xylol, rehydrated in descending ethanol concentrations (100%, 95% and 70%) and then incubated with 1 mg/ml type XIV bacterial protease (Sigma-Aldrich, Germany) in PBS for antigen-demasking at 37°C for 10 min.

Endogenous peroxidase was blocked with 3% H2O2 at room temperature for 15 min. Endogenous biotin was blocked using a commercial blocking kit (Avidin/Biotin Blocking Kit; BioGenex, USA). Slides were further blocked with normal goat serum (1:500 in PBS; Vector Laboratories, U.K.) for 20 min and a casein solution (Power BlockTM; BioGenex, USA) for 10 min at room temperature. Adda antiserum (see Fischer et al. 2001) was diluted 1:5000 in Power BlockTM and applied to the tissue section in a humidified atmosphere for 16 hours at 4 °C. Antigen-antiserum complexes were visualised using a HRP-labelled, biotin-streptavidin amplified detection system and 3- amino-9-ethylcarbazole (AEC) chromogen (Super SensitiveTM; BioGenex, USA). Sections were counterstained at room temperature for 6 min with Mayer’s haematoxylin (Sigma-Aldrich, Germany), rinsed with tap water and mounted using Chrystal/MountTM (Biomeda, USA) and Shandon HistomountTM (Thermo Electron Corporation, Germany). An organ was classified to be microcystin-positive when microcystin-positive areas were above background chromogen staining and the positive staining areas were congruently observable in two independently stained serial sections from the same organ sample.

Statistical Analyses As fish were gavaged individually, individuals were considered to represent replicates for statistical purposes. The potential influence of group holding of the respective fish per treatment/

86 3. EXPOSURE EXPERIMENTS ______sham-control/ control group was assumed to be negligible as the influence of group holding should be comparable across groups. As there were no statistical differences between data at the two time-points, data of the control fish were combined and represented as the control cohort (six individuals). Individual data from sham-control fish were combined as already described for the controls. Statistical analyses were carried out using GraphPad Prism 4® (USA) Software. Values were given as the mean ± standard deviation (SD) of at least three individuals for beat rate determinations and of at least five individuals for weight, HSI, condition factor and serum glucose analyses.

OPERCULAR BEAT RATE: Statistical differences in opercular beat rates were analysed using a one- way ANOVA followed by Bonferroni’s Multiple Comparison Test to compare beat rates of the multiple time-points of the sham-control fish (Tab. 3.2) and the rates of the 12 individuals (4 fish per tank) assessed before gavage (basal rate; see Tab. 3.1 B and 3.2). An F-Test for determination of homogeneity of variances followed by an unpaired t-test (two-tailed) with Welch’s correction was employed to compare the opercular beat rates observed in the sham-control with those of the P. rubescens gavaged fish at each time-point of the experiment.

CONDITION FACTOR: Condition factors were tested for statistical differences using a F-test to determine homogeneity of variances followed by an unpaired t-test (two-tailed) to compare the condition factor determined in sham-control with those of the control fish. A Bartlett’s test and one-way ANOVA followed by a Dunnett’s post-test were employed to compare the multiple time- points of the P. rubescens gavaged fish with the sham-control fish cohort.

PLASMA GLUCOSE: Glucose levels were tested for statistical differences using a F-test to determine homogeneity of variances followed by an unpaired t-test (two-tailed) to compare the plasma glucose concentration determined in sham-control with those determined in the control fish. A Bartlett’s test and one-way ANOVA followed by a Dunnett’s post-test were employed to compare the multiple time-points of the P. rubescens gavaged fish with the combined sham- control fish.

HISTOPATHOLOGY: Pathological changes in the respective tissues are given as median ± mean absolute deviation (MAD) of the individual ranks of at least four examined individuals. Ranking of pathological changes was tested for statistical differences using the non-parametric Mann- Whitney U-Test. Significant differences were determined at the **p <0.01 and *p <0.05 level for all statistical analyses.

RESULTS

P. rubescens Characterisation

The administered P. rubescens culture was shown to contain 2.1 ±0.03 µg MC-LRequiv./mg dw and

3.3 ±0.26 µg MC-LRequiv./mg via HPLC and Adda-ELISA analysis, respectively. Toxin quantification via the colorimetric PP-inhibition assay (cPPA) suggested that the P. rubescens

87 3. EXPOSURE EXPERIMENTS ______

Tab. 3.2: The opercular beat rate determined in coregonids before the gavage procedure (basal rate, before application (b.a.)) and comparison of beat rates determined for sham-control coregonids and coregonids, exposed to a single dose of P. rubescens at various time-points post application (p.a.)

Time to application Sham-control [beats/min] Exposed ([h] p.a.) [beats/min]

1 109 ±7 (4) 134 ±12 (8) ** 3 103 ±2 (4) 117 ±11 (8) ** 19 94 ±4 (4) 115 ±12 (8) ** 25 107 ±6 (3) 108 ±7 (8) 29 117 ±8 (3) 117 ±6 (8) 46 95 ±4 (3) 103 ±6 (8) 54 115 ±5 (3) 116 ±15 (6) 70 109 ±12 (3) 114 ±9 (6)

When time to application is 1 h b.a. the basal rate is 107 ±10 beat/min (12). Values are given as mean ± SD; the number of fish observed is given in parenthesis; significant differences between sham-control and exposed fish are indicated (** for p <0.01).

culture extract contained 0.03 ±0.02 µg MC-LRequiv./mg and thus less than that determined via either HPLC or Adda-ELISA. More in-depth HPLC analysis demonstrated one broad main peak with a shoulder and retention time and spectrum comparable to an overlay of the Asp3-Dhb7-MC-RR and Asp3-MC-RR peaks. In addition, further analysis of retention times and spectra of supplementary peaks observed, suggested the presence of anabaenopeptines and cyanpeptolines in the P. rubescens culture extract (data not shown).

Fish Exposure (Behavioural Observations and Opercular Beat Rate) Following gavage and subsequent reanimation from MS-222 a very few of the gavaged fish were observed to regurgitate a small amount of the P. rubescens suspension through the gill. Three treated fish died during anaesthesia and were replaced immediately with additional fish. With the exception of the anaesthesia-related mortality, no further mortalities were observed subsequent to anaesthesia and reanimation. Three hours post application, the exposed fish appeared more susceptible to startling and swimming behaviour was more hectic than observed in the corresponding sham-control fish. These behavioural differences, however, decreased and became indiscernible from the behavioural patterns of the sham-control fish with increasing duration of the experiment. A significantly elevated opercular beat rate was observed at the 1 h post application (p.a.) time- point in the P. rubescens culture exposed fish when compared to the corresponding sham-control fish. This increased opercular beat rate was also significantly elevated at 3 and 19 h p.a. (Tab. 3.2), however, the large variation of the individual data at the respective time-points suggests that the significant effects at the 3 and 19 h time-points p.a., although mathematically correct, may not be of biological relevance as the data lie within the range of the normal opercular beat rates of the sham-control fish. No differences were observed between the opercular beat rate of the sham-control fish and the basal rate determined prior to the start of the experiment.

88 3. EXPOSURE EXPERIMENTS ______

Tab. 3.3: Development of weight, hepatosomatic index (HSI), condition and plasma glucose of coregonids, exposed to a single dose of P. rubescens in comparison to control fish for each time-point post application (p.a.)

Treatment Control Sham-control Exposed

9h p.a. 24h p.a. 48h p.a. 72h p.a. Mean weight [g] 9.1 ±3.45 (6) 9.6 ±1.33 (5) 10.1 ±2.10 (6) 10.0 ±1.77 (6) 9.0 ±3.01 (6) 9.4 ±1.26 (5) HSI [mg/g] 1.17 ±0.205 (5) 1.09 ±0.178 (6) 1.11 ±0.156 (5) 1.11 ±0.081 (6) 1.06 ±0.104 (6) 1.19 ±0.153 (5) Condition factor 0.73 ±0.030 (6) 0.73 ±0.065 (5) 0.69 ±0.055 (6) 0.70 ±0.071 (6) 0.67 ±0.133 (6) 0.65 ±0.070 (5) Plasma glucose 15.6 ±05.95 (6) 31.2 ±08.40 (5) 54.1 ±27.08 (6) 43.3 ±48.62 (5) 17.4 ±04.35 (6) 30.5 ±14.38 (6) [mg/100ml]

Values are given as mean ± SD; the number of fish assessed is given in parenthesis; no significant differences were observed between sham-control and exposed fish at the p <0.05 level.

Weight, HSI, Condition Factor and Plasma Glucose No significant differences in weight or HSI were observed between treatment and control groups (Tab. 3.3). The condition factors determined in the P. rubescens culture exposed fish appeared lower than those of the corresponding controls and also appeared to decrease with increasing experiment time. This trend was, however, not statistically significant. Plasma glucose levels were not significantly different between treatment and the sham-control group (Tab. 3.3). P. rubescens exposure, however, appeared to increase the variability of the glucose levels determined. This was confirmed by statistically different variances as demonstrated by Bartlett’s test analysis. This high variability was probably due to the low number of individuals used, and may thus have prevented detection of statistically significant differences between exposed and sham-control fish. Handling of the fish clearly increased stress and consequently plasma glucose levels (Tab. 3.3), as evidenced by the differences between the control and the sham-control fish.

Histopathological Changes

LIVER: Neither control nor sham-control fish presented with histopathological changes in the livers beyond the normal range observed in coregonids of this age group. In contrast, P. rubescens gavaged fish presented with an time-dependent increase in liver pathology which consisted of focal hepatocytes with granulated cytosol, focal disintegration of the parenchymal liver architecture, cell dissociation, chromatin margination, diffuse focal necrosis and apoptosis, necrosis peripheral to central veins, ruptured vessels, dilated sinusoids, infiltrations of mononuclear cells and coagulative necrosis peripheral to central veins (Fig. 3.1). The degree of pathological change observed in P. rubescens gavaged fish was significantly different from the corresponding controls at 24, 48 and 72 h p.a. (Tab. 3.4).

KIDNEY: Sporadic changes in the renal tubules of control and exposed fish were characterised by the occurrence of low frequencies of apoptotic cells, regenerating cells and epithelial cell exfoliation. In distinction to corresponding controls, kidneys of P. rubescens gavaged fish presented with enhanced tubular degeneration which included vacuolisation and cell shedding, proteinaceous casts and calcium-phosphate-precipitates in the tubular lumina (Fig. 3.2). In 89 3. EXPOSURE EXPERIMENTS ______

A B

CN CN

P P

GC P CN

25 µm 25 µm

C CN BI LA

IL P CN P MC +

25 µm 25 µm

Fig. 3.1: Liver tissue of control (A) and exposed coregonids 24 h (B) and 48 h (C) post application. Sections (A–C) stained with H&E or microcystin antibodies (BI). Exposed fish presented with dilated sinusoids (white arrows), a partial loss of liver architecture (LA), infiltration of leucocytes (IL), granulation of hepatocyte cytosol (GC), pyknosis (P) and coagulative necrosis (CN). Comparison of H&E and immunostained sections (B and BI) demonstrates the presence of microcystin (MC+) in histologically changed tissue sections (circle).

addition, coagulative necrosis was observed in the interstitium. However, the degree of pathological change observed in P. rubescens gavaged fish was not significantly different from the corresponding controls (Tab. 3.4).

Tab. 3.4: Histopathological changes observed in various organs of P. rubescens exposed fish compared to control- and sham-control fish

Time Liver Kidney Gill Pylori Hindgut

Control 0h 0.5 ±0.2 (3) 1 ±0.3 (3) 0.5 ±0.0 (3) 0 ±0.3 (3) 0.8 ±0.3 (2) 48h 1 ±0.7 (3) 0.5 ±0.0 (3) 0.5 ±0.0 (3) 0 ±0.5 (3) 0.5 ±0.0 (2)

Sham-control 24h 0.5 ±0.0 (3) 1 ±0.0 (3) 0.5 ±0.2 (3) 0 ±0.0 (3) 1 ±0.0 (3) 72h 0.5 ±0.3 (3) 1 ±0.2 (3) 0.5 ±0.0 (3) 0 ±0.0 (3) 1 ±0.2 (3)

Exposed 9h 1 ±0.3 (6) 1.5 ±0.2 (6) 0.5 ±0.1 (6) 0.5 ±0.5 (6) 1.8 ±0.9 (6) 24h 1.5 ±0.5 (6) * 1 ±0.1 (6) 1 ±0.2 (5) 0.5 ±0.2 (6) * 2.3 ±0.5 (6) ** 48h 1.5 ±0.3 (6) ** 0.5 ±0.4 (6) 0.5 ±0.2 (5) 0 ±0.1 (6) 1.5 ±0.6 (6) 72h 1 ±0.3 (6) * 1.3 ±0.5 (6) 1 ±0.1 (5) 0 ±0.3 (6) 1 ±0.3 (6)

Histopathological changes were ranked from none (0) to severe (4) including intermediate classes, e.g. 0.5, 1.5 etc. Values are presented as median ± mean absolute deviation for each time point; the number of fish assessed is given in parenthesis. For statistical analysis individuals of the control groups were summed to one control- and one sham-control cohort. Significant differences between sham-control and exposed fish are indicated (*p <0.05 and ** p <0.01).

90 3. EXPOSURE EXPERIMENTS ______

A B TD CN CP

TD FN TD CN

CP 50 µm 50 µm

C BI

V FN

50 µm 50 µm

Fig. 3.2: Kidney tissue of control (A) and exposed coregonids 9 h (B) and 72 h (C) post application. Sections (A–C) stained with H&E or microcystin antibodies (BI). Exposed fish presented with focal (FN) and coagulative (CN) interstitial necrosis in kidney, as well as a degeneration of the tubular lining (TD), including vacuolation (V) and exfoliation (CE) of tubular epithelial cells into the tubular lumen, proteinaceous casts (white arrows) and calcium phosphate precipitates (CP). Comparison of H&E and immunostained sections (B and BI) demonstrates the presence of microcystin (MC+) in histologically changed tissue sections (circles).

PYLORI AND HINDGUT: Pylori and hindgut of control and sham-control fish were unremarkable. In contrast, distinct histopathological changes were observed in exposed coregonids. Characteristic were mild epithelial degeneration in the pylori, and loss of the mucosa structure, frayed gut villi, exfoliation of epithelial cells, widespread cell lysis, infiltration of leucocytes and intraluminal protein cast deposition in the hindgut (Fig. 3.3). The pathological changes appeared to be most predominant at the 24 h p.a. time-point and then decreased to background at 72 h p.a. This observation was also apparent from the degree of pathological change classifications, with the 24 h p.a. time-point pathological changes being significantly higher than those of the sham- control for both the pylori and the hindgut (Tab. 3.4).

GILLS: Pathological changes attributable to P. rubescens gavage were not unequivocal discernible from changes (vacuolisation, sporadic lamellar tip clubbing) observed in the control fish. This is also corroborated by the observation that no significant differences were observed in overall pathological scores between gills of P. rubescens gavaged and control fish (Tab. 3.4). 91 3. EXPOSURE EXPERIMENTS ______

A B CE

L CE 50 µm 50 µm

C LA CE BI

MC+ MC+

MC+ MC+ LA

MC+

50 µm 50 µm

Fig. 3.3: Pylori (A) and hindgut tissue (B & C) of exposed coregonids 24 h post application. Sections (A–C) stained with H&E or microcystin antibodies (BI). Exposed fish presented with epithelial degeneration and cell lysis (white arrows), exfoliation of epithelial cells lumen (CE), presence of leucocytes (L) and loss of the gut architecture (LA). Comparison of H&E and immunostained sections (B and BI) demonstrates the presence of microcystin (MC+) in affected gut regions.

Microcystin -Immunohistochemistry Microcystin-positive staining was most distinct in the liver of the exposed fish, with both the number of positive fish and the area of positive staining being highest at the 24 h p.a. time-point. Microcystin-positive staining could also be demonstrated in foci with coagulative necrosis and disrupted vessels (Fig. 3.1). Kidney tissues were variable with regard to positive microcystin-staining, which occurred most frequently in fish of the 9 and 72 h time-points and less frequently in fish of the 24 and 48 h p.a. time-point. However, immunopositive staining was primarily observable in kidneys with obvious pathology, i.e. overt interstitial and epithelial necrosis and proteinaceous casts (Fig. 3.2). Immunopositive staining was also detectable in the pylori and hindgut of P. rubescens gavaged fish at 9 h and 24 h p.a. Microcystin-staining was primarily observed in the mucosal epithelium, while in the pylori this was restricted to the villi tips (Fig. 3.3).

92 3. EXPOSURE EXPERIMENTS ______

DISCUSSION

P. rubescens Characterisation The administered P. rubescens culture was shown to contain microcystin corresponding to an amount of 2.1 and 3.3 µg MC-LRequiv./mg dw using HPLC- and ELISA-analysis, respectively. Differences between ELISA and HPLC assessment are in accordance to those published in other studies (Hawkins et al., 2005; Rapala et al., 2002). The differences between the quantities determined by ELISA and HPLC and those determined via the cPPA are in accordance to observations previously described by Rapala et al. (2002), who demonstrated that cPPA data achieved with extracts of microcystin-producing Planktothrix strains returned only 5% of the values obtained by ELISA and HPLC. The quantities of microcystin congeners determined via ELISA and HPLC were therefore considered more reliable than those determined via the cPPA assay. HPLC results suggested the main microcystin congener(s) present in the P. rubescens extract to be Asp3-MC-RR variant(s). This observation concurs with previous studies, describing demethylated variants of MC-RR to be the predominant microcystin-congeners in cyanobacteria of the P. agardhii/rubescens group in European lakes (Blom et al., 2001; Ernst et al., 2001; Fastner et al., 1999a; Jacquet et al., 2005; Kurmayer et al., 2005; Luukkainen et al., 1993; Spoof et al., 2003b). Microcystin-quantities detected in P. agardhii/rubescens range from 0.3 to 6 µg/mg dw (Ernst et al., 2001; Fastner et al., 1999b; Kurmayer et al., 2005). The microcystin-composition and quantities found for the lake Ammersee P. rubescens used for coregonid gavaging are comparable to those typically described for P. agardhii/rubescens in European lakes. Assuming an average filament-length of 231.9 cells/filament (derived from filament measurements in Lake Ammersee over a one year period; see Ernst, 2000), an average weight of 8 µg/1000 filaments (Gammeter et al., 1997) and based on microcystin-quantities determined by HPLC and ELISA, the 80.000 cells applied per fish (9.9 ±1.9 g) corresponded to a single dose of

0.6-0.9 µg MC-LRequiv./kg bw (0.3 mg P. rubescens dw/kg bw).

Behavioural Observations and Opercular Beat Rate As already observed in earlier studies with other fish species (Snyder et al., 2002; Tencalla & Dietrich, 1997), a few of the coregonids were able to regurgitate and expel parts of the gavaged cyanobacteria through the gills. Since the regurgitated amounts appeared to be minimal, gavage was considered successful. The observed regurgitation, however, suggests that coregonids aim to actively avoid uptake of P. rubescens. That P. rubescens represents a stressor to coregonids can be inferred from the behavioural changes, including hectic swimming and an increased ventilation as indicated by the significantly elevated opercular beat rate, observed in coregonids 1 h post gavage. Increased ventilation rates were also reported by Carbis et al. (1996a) and Kotak et al. (1996) irrespective of whether the microcystin-exposure was via the oral or intraperitoneal

93 3. EXPOSURE EXPERIMENTS ______route. While in the study described by Carbis et al. (1996a) the higher ventilation rate in carp treated with MC-LR was possibly attributable to simultaneously occurring gill damage, no gill pathology was observed in the current coregonid study. Therefore, the elevated ventilation observed in the gavaged coregonids appear more likely to result from stress induced by gastrointestinal P. rubescens than by gill damage.

Weight, HSI, Condition Factor and Plasma Glucose Although no significant effects of P. rubescens exposure were observed on weight, HSI or condition factor, stress related effects were observable using plasma glucose levels as the end- point. Indeed, handling of fish per se increased plasma glucose levels (mobilisation of glycogen stores) as already demonstrated in carp by Svobodová et al. (1999), i.e. levels in sham-control fish were approximately double the basal levels of non-handled control fish. P. rubescens treatment increased this basal plasma glucose level by approximately three- to four-fold in the period 9-24 h p.a., supporting the above observation that ingested P. rubescens provides for an increased stress to coregonids. Whether the increased plasma glucose levels are solely related to a catecholamine- and cortisol stimulated mobilisation of energy and thus stress-mediated higher energy demand (Barton et al., 2002; Bury et al., 1996a) or also as a consequence of a disruption glycogen homeostasis resulting from protein phosphatase inhibition as suggested by Malbrouck et al., (2004a) and Råbergh et al. (1991), cannot be clearly deduced from the data presented here. However, the influence of the toxin-mediated protein phosphatase inhibition appears secondary, as plasma glucose levels appear to return to at least control levels within 48 h p.a., while an extended exposure to toxins (P. rubescens) via the gastrointestinal would be expected to result in an appreciable and continued elevation of plasma glucose levels.

Histopathology and Immunohistochemistry The above interpretation is further corroborated by the observation that the earliest time-point with significantly enhanced liver pathology and immunohistochemically defined presence of microcystin was between 24 and 48 h p.a., a time-frame where plasma glucose levels appeared to be declining back to control levels. Consequently the stress, as depicted by plasma glucose levels, does not appear to be associated with, or a consequence of liver pathology. Evidence for a prolonged exposure to toxins is provided by the observation that increased gastrointestinal pathology was observed up to 48 h p.a. and this then gradually declined to control levels, while a slightly delayed onset of pathology was observed in the liver. The described pathological changes in the gastrointestinal tract, liver and kidney are characteristic of pathological lesions described earlier for microcystin intoxications (Carbis et al., 1996a; Fischer et al., 2000; Fischer & Dietrich, 2000; Fournie & Courtney, 2002; Kotak et al., 1996; Malbrouck et al., 2003; Råbergh et al., 1991; Rodger et al., 1994; Tencalla & Dietrich, 1997; Tencalla et al., 1994). The combined data of pathology and microcystin-immunohistochemistry suggest microcystin uptake and accumulation to be highest 24 h p.a., which appears slightly slower than what was observed for trout and carp

94 3. EXPOSURE EXPERIMENTS ______

(Fischer et al., 2000; Fischer & Dietrich, 2000). However, due to the use of non-acutely lethal doses of P. rubescens material in this coregonid study, the direct comparison with previous trout and carp studies (Carbis et al., 1996a; Fischer et al., 2000; Fischer & Dietrich, 2000; Råbergh et al., 1991; Tencalla et al., 1994) is tentative at best. Indeed, while Fischer et al. (2000) and Fischer & Dietrich (2000) demonstrated a continuum of increasing pathological alteration and microcystin-immunostaining in the liver of exposed trout and carp, the dose of P. rubescens employed in this study resulted in a maximal effect at 24-48 h p.a. and a regeneration as well as a reduction of immunostaining thereafter. Of importance, however, is the observation that a high number of pathological changes, whether observed in the liver, kidney or gastrointestinal tract, were also immunopositive for microcystin, suggesting a causal relationship of tissue damage with presence of microcystins as also indicated for other fish species, i.e. trout and carp (Fischer et al., 2000; Fischer & Dietrich, 2000). The present low-dose exposure experiment in coregonids also suggests that subchronic and chronic exposure to low doses of P. rubescens can cause enhanced physiological stress as well as continued pathological changes in the gastrointestinal tract, liver and kidneys possibly leading to decreased functionality of these organs. The combination of stress and organ damage might thus explain the reduced weight and hence fitness observed in coregonids subjected to stratified and dispersed P. rubescens blooms such as those occurring in Lake Ammersee.

ACKNOWLEDGMENTS We would like to thank Jussi Meriluoto, Abo Akademi University Turku, Finland and Judith Blom, University Zurich, Switzerland for providing microcystin standards. Helmut Segner, University of Bern, Switzerland, Christoph Kueng, Fish and Wildlife Service Canton Bern and H. Zieri, Fish and Wildlife Service Canton Uri, Switzerland for the assistance in obtaining sufficient numbers of one-year-old coregonids. RCC Ltd. for histological tissue processing and substantial support. Iris Töpfer for experimental assistance. We would also like to acknowledge the Arthur and Aenne Feindt Foundation (Germany) for kindly funding this study.

95 3. EXPOSURE EXPERIMENTS ______

3.2. PHYSIOLOGICAL STRESS AND PATHOLOGY IN EUROPEAN WHITEFISH (COREGONUS LAVARETUS) INDUCED BY SUBCHRONIC EXPOSURE TO ENVIRONMENTALLY RELEVANT DENSITIES OF PLANKTOTHRIX RUBESCENS

Bernhard Ernst, Stefan J. Hoeger, Evelyn O’Brien, Daniel R. Dietrich

Environmental Toxicology, University of Konstanz, P.O. Box X918, 78457 Konstanz, Germany

Published in Aquatic Toxicology 82 (2007) 15–26

ABSTRACT Planktothrix rubescens belongs to the most ubiquitous cyanobacterial species in mesotrophic and oligotrophic lakes in the pre-alpine regions. In most of these lakes, coregonids are among the dominant species of the ichthyofauna with great importance for the professional fishery. A possible link between the occurrence of toxic Planktothrix blooms and the recurrent slumps in coregonid yields has been suggested. Indeed, acute toxic effects of microcystins and other cyanobacterial toxins have been shown for various fish species. However, chronic exposure scenarios appear to be more common and thus more environmentally realistic than acute intoxications. The aim of this study was therefore to investigate the physiological stress response and organ pathology in coregonids sub-chronically exposed to ambient water containing low, medium and high P. rubescens densities, known to be typical of pre-alpine lakes. Coregonid hatchlings were exposed in four tanks containing 0 (sham-control) and approximately 1500 (low), 15,000 (medium) and 55,000 (high) P. rubescens cells/ml for up to 28 days. Temperature, oxygen concentration, pH-value, P. rubescens cell density and microcystin concentration were recorded and the fish were observed for behavioural changes and examined for parasite infestations. Gill ventilation rates, general condition factors and mortalities were determined and liver, kidney, gut and gill were assessed histopathologically and immunhistologically. Depending on the cell density, exposed fish showed behavioural changes, including increased ventilation rates possibly representing a physiological stress response. Susceptibility to ectoparasitic infestation and increased mortality in exposed fish suggested P. rubescens associated effects on fish fitness. Histopathological alterations in liver, gastrointestinal tract and kidney, which were also immunopositive for microcystin suggested causality of tissue damage and the presence of microcystins. In contrast, observed gill pathology appeared to result primarily from mechanical abrasion and irritation due to ectoparasitic infestation. The current exposure experiment confirmed the hypothesis that subchronic and chronic exposure to low cyanobacterial cell densities and hence microcystins can exacerbate physiological stress and sustained pathological alterations in exposed coregonids. The study therefore supports the theory that P. rubescens blooms may be causal to the observed weight reduction and hence fitness of coregonids in pre-alpine lakes such as Lake Ammersee (Germany).

KEYWORDS: Fish; Microcystin; Planktothrix; Coregonid; Cyanobacteria; Whitefish; Stress; Pathology

96 3. EXPOSURE EXPERIMENTS ______

INTRODUCTION Toxic cyanobacteria occur ubiquitously in fresh and coastal waters. Due to their ability to produce highly toxic metabolites, i.e. various neurotoxins and potent protein phosphatase inhibitors, e.g. microcystins, mass occurrences of cyanobacteria have been associated with mortality of wild and domestic animals including fish (Briand et al., 2003; Jewel et al., 2003; Rodger et al., 1994; Toranzo et al., 1990). To date, at least 46 cyanobacterial species are known to produce toxins (Sivonen & Jones, 1999), of which the microcystin-producing Planktothrix is one of the most important genera in temperate climates. Planktothrix occur in eutrophic waters, producing surface or whole water-body blooms during winter circulation and metalimnic bloom layers during lake stratification in summer (Walsby et al., 1998). However, Planktothrix rubescens can also occur in mesotrophic and even oligotrophic lakes that have recently undergone an anthropogenically induced phase of re-oligotrophication (Ernst et al., 2006b; Jacquet et al., 2005). As a consequence, P. rubescens has been a predominant species of the phytoplankton community in several monomictic and dimictic European lakes, especially in the alpine and pre-alpine regions for several decades (Gammeter et al., 1997; Krupa & Czernas, 2003; Kucklentz et al., 2001; Wehrli & Wüest, 1996). P. rubescens blooms can reach and maintain cell densities of more than 50,000 cells/ml over several months, calculated based on filament length and a unit cell size and spectrofluorimetric measurements (Ernst et al., 2001; Hoeger et al., 2005; Jacquet et al., 2005). In many pre-alpine lakes, coregonids (Coregonus sp.) are among the dominant species of the ichthyofauna with great importance for the professional fishery. Over the past decades, recurrent slumps in yearly yields of coregonids caused by reduced fish weight and fitness have been observed to coincide with the appearance of P. rubescens blooms in Swiss and German lakes (Braun, 1953; Ernst et al., 2001). Therefore, a possible link between the occurrence of toxic Planktothrix blooms and the recurrent changes in growth and population dynamics of coregonids has been suggested (Ernst et al., 2001; Ernst et al., 2006a). Toxic effects of microcystins and other cyanobacterial toxins occur in various fish species, including salmoniformes, siluriformes, cypriniformes as well as perciformes (summarised in Malbrouck & Kestemont, 2006). However, differences in fish species sensitivity to toxic cyanobacteria and cyanobacterial toxins (primarily microcystin-LR (MC-LR)) are known to exist (Fischer & Dietrich, 2000; Tencalla, 1995). Uptake of cyanobacterial toxins by fish has been shown to occur primarily via oral ingestion of toxins or toxic cyanobacterial cells and to a reduced if not negligible extent from toxin uptake by the gill epithelium (Bury et al., 1995; Tencalla et al., 1994). Previous studies have shown that fish exposed to acutely toxic concentrations of microcystins or bloom material show pathological changes in the liver, kidney, gut and gill pathology, and associated effects, such as increased serum liver enzyme values. In addition, behavioural changes and changes in development of fish have been reported (summarised in Malbrouck & Kestemont, 2006). However, most of these findings were restricted to acute studies,

97 3. EXPOSURE EXPERIMENTS ______often employing toxin or bloom exposure routes that do not realistically reflect environmentally relevant situations. The microcystin concentrations found in water bodies (Hoeger et al., 2005 and references therein) are, with some exceptions, generally lower than the known acute lethal concentrations for fish (Landsberg, 2002). Furthermore, cyanobacterial blooms especially those of P. rubescens can last up to several months (Ernst et al., 2001; Jacquet et al., 2005; Salmaso, 2000). Thus, a subchronic or chronic exposure scenario appears to be quite common and hence more environmentally realistic, while acute intoxications may have less ecological relevance. However, apart from a few studies suggesting oxidative stress (direct assessment of reactive oxygen species overproduction, assessment of oxidation products or determination of changes in antioxidant mechanisms and detoxication) as a causative factor in cyanobloom or cyanotoxin induced fish pathology (Bláha et al., 2004; Jos et al., 2005; Li et al., 2005), more in-depth information on subchronic or chronic effects of microcystin exposure in fish are rare. Coregonids exposed to a single oral dose of 80,000 P. rubescens cells show an enhanced physiological stress (e.g. increased ventilation rates) as well as pathological changes characteristic of microcystin intoxications (Ernst et al., 2006a). This suggests that prolonged (subchronic or chronic) exposure of coregonids to high P. rubescens filament densities could cause increased physiological stress and thus also reduced growth. The aim of this study was therefore to investigate the physiological stress response (ventilation rate, behavioural changes, blood glucose levels) and organ pathology in hatchling coregonids (Coregonus lavaretus L.) sub- chronically immersed in P. rubescens densities known to occur in pre-alpine lakes, i.e. • a low density of approximately 1500 cells/ml, regularly occurring in large zones of the waterbody for the majority of the year without the overt presence of P. rubescens blooms/layers; • a medium density of approximately 15,000 cells/ml which is often observed in metalimnic layers and sometimes in large zones of the water body during winter circulation, and • a high density of approximately 55,000 cells/ml, often occurring in metalimnic P. rubescens layers.

MATERIAL & METHODS

Chemicals All reagents and solvents were of analytical or chromatographic grade and were purchased from Fluka (Switzerland), Merck (Germany), Riedel de Haen (Germany), Roth (Germany) or Sigma- Aldrich (Germany).

98 3. EXPOSURE EXPERIMENTS ______

Cultivation and P. rubescens Toxin Characterisation P. rubescens was originally isolated from a Lake Ammersee seston sample in autumn 2002 (Hoeger et al., 2007) and cultivated in BG11 medium according to the method described by Rippka et al. (1979). The microcystin-LR equivalent content and microcystin congener composition of this P. rubescens culture was previously described by Ernst et al. (2006a) and was shown to contain 2-3 µg MC-LRequiv./mg dw and to predominantly consist of demethylated microcystin-RR variants.

Fish Exposure Hatchlings of European whitefish (Coregonus lavaretus L.; approx. weight: 6 g; approx. length: 10 cm) were obtained from the Lake Ammersee Fisheries Cooperative (Germany). Fish were distributed randomly into four 100 l tanks (24 coregonids/tank). Five coregonids sampled immediately prior to random distribution were taken as an initial control and used for comparison with the sham-control and exposure groups (see below). Tanks were filled with tap water and aerated using a commercial aquarium aeration pump. Ten percent of the tank water volume was replaced by fresh water daily. Fish were fed (15% of body weight) with frozen daphnia and chironomid larvae (Honka, Germany) once per day. After acclimatisation (24 h), three tanks (A, B & C) were exposed to P. rubescens filament densities corresponding to approximately 1500 (low), 15,000 (medium) and 55,000 P. rubescens cells/ml (high), respectively. The remaining tank (tank Co) served as corresponding sham-control, i.e. without addition of P. rubescens filaments.

Monitoring of Experimental Parameters throughout the Experiment Tank temperatures were measured twice daily, while oxygen concentrations were determined at least three times per week. pH-values were determined on two separate occasions during the exposure.

P. RUBESCENS CELL DENSITIES: P. rubescens cell densities were determined in the P. rubescens culture and daily in the exposure tanks via image processing as described by Ernst et al. (2006b). Subsequent to water renewal, densities were readjusted to the initial filament density via replenishment of P. rubescens from the stock culture.

MICROCYSTIN CONCENTRATION: The MC-LR equivalent concentration (MC-LRequiv.) in the tank water was determined on days 3, 6, 9, 13, 16, 17, 20, 22 and 24 in water from tank, A, B and C and on day 6, 13, 17 and 24 in water from the sham-control tank. For this purpose 1.5 ml tank water were taken through a 0.2 µm syringe filter and stored at –20 °C. Total MC-LR equivalent concentration (MC-LRequiv.) in a given water sample was analysed using the Adda-ELISA kit (Abraxis; USA; see also Fischer et al. 2001) according to the manufacturer’s instructions. The analytical results (MC-LRequiv. concentration) were finally combined to give a time-weighted average MC-LRequiv. concentration per tank over the duration of the experiment.

99 3. EXPOSURE EXPERIMENTS ______

Exposure day

0 7 14 21 28

Ventilation rate x 1 x x xxx x x xx x x x xx x x x x Condition factor x2 xxxx Serum glucose x2 xxxx Histopathology 2 x xxxx Immunhistology x2 xxxx

1 Basal rate 2 Initial control samples

Fig. 3.4: Time points of parameter determination during the 28 day experiment.

Mortality, Parasite Infestations, Behavioural Observations and Ventilation Rates Fish were observed daily for behavioural changes and overt parasite infestation. Dead fish were registered, removed from the tanks, and withdrawn from the experimental cohort used for subsequent analyses. The ventilation rate of control and exposed fish was determined before (group specific basal rate) and at 18 different time-points during the exposure (Fig. 3.4), by counting the opercular movement of four individuals per tank for 15 s each. Ventilation rates determined were multiplied by four, thus giving the ventilation rate per min.

Sampling and experimental procedures On days 7, 14, 21 and 28, six fish were sampled from each tank (sham-control and exposure groups). All fish (also including fish of the initial control) were assessed for macroscopical anomalies and then analysed using the experimental parameters shown in Figure 3.4 and described below:

WEIGHT: Wet weight was determined on a fine balance (PB3002; Mettler, Germany).

LENGTH: Length (nose-tip to tail) was determined to the nearest mm using a centimetre ruler.

CONDITION FACTOR: Condition factor (CF) was calculated according to Barton et al. (2002) as CF = weight [g] x (length [mm])-3 x 105.

SERUM GLUCOSE: Serum glucose was determined in blood taken from the caudal vein, using a 1 ml syringe and a 0.40 x 20 mm needle (Braun, Germany). Blood serum was isolated via coagulative precipitation at 4 °C for 12 h and subsequent centrifugation (4000 x g; 10 min, 4°C). The supernatant (serum) was frozen and stored at –20 °C until measurement. Serum glucose was determined using an automatic sensor (Accu–Chek; Roche, USA) according to the manufacturer’s instructions.

HISTOPATHOLOGY: A representative tissue sample of liver, kidney, gill and hindgut was dissected from each fish, placed into a labelled tissue cassette and briefly fixed in 4% PBS-buffered formalin. In addition, fresh (non-fixed) gill samples were inspected microscopically for

100 3. EXPOSURE EXPERIMENTS ______accumulation of P. rubescens filaments between the gill lamellae. PBS-buffered formalin fixed tissues were routinely embedded in paraffin, sectioned to 3-5 µm and mounted on microscope slides (RCC Ltd., Switzerland). Histopathological assessment was carried out using haematoxylin and eosin (H&E) stained sections by light microscopy at 40 to 400-fold magnification. Pathological changes were classified as none (0), mild (1), moderate (2), strong (3) and severe (4), including intermediate classes (e.g. 0.5, 1.5).

IMMUNOHISTOCHEMICAL DETERMINATION OF MICROCYSTIN: Sections of liver, kidney, gill and intestine mounted on polysin-coated glass slides were deparaffinised in 100% xylol, rehydrated in descending ethanol concentrations (100%, 95% and 70%) and then incubated with 1 mg/ml type XIV bacterial protease (Sigma-Aldrich, Germany) in PBS for antigen-demasking at 37 °C for

10 min. Endogenous peroxidase was blocked with 3% H2O2 at room temperature for 15 min. Endogenous biotin was blocked using a commercial blocking kit (Avidin/Biotin Blocking Kit; BioGenex, USA). Slides were further blocked with normal goat serum (1:500 in PBS; Vector Laboratories, U.K.) for 20 min and a casein solution (Power BlockTM; BioGenex, USA) for 10 min at room temperature. Adda-antiserum (see Fischer et al. 2001) was diluted 1:5000 in Power BlockTM and applied to the tissue section in a humidified atmosphere for 16 hours at 4 °C. Antigen-antiserum complexes were visualised using a HRP-labelled, biotin-streptavidin amplified detection system with the 3-amino-9-ethylcarbazole (AEC) chromogen (Super SensitiveTM; BioGenex, USA). Sections were counterstained at room temperature for 6 min with Mayer’s haematoxylin (Sigma-Aldrich, Germany), rinsed with tap water and mounted using Crystal/MountTM (Biomeda, USA) and Shandon HistomountTM (Thermo Electron Corporation, Germany). A section was classified to be microcystin-positive when microcystin-positive areas were above background chromogen staining and the positive staining areas were congruently observable in two independently stained serial sections from the same tissue sample block. Immunopositive staining was classified as none (0), sporadic (+), pronounced (++), and extensive (+++).

Statistical Analyses Normally, animals exposed in groups can potentially influence each other and as such, cannot be considered as independent replicates. However, increasing the number of replicate tanks with a low number of fish was impracticable and would have provided for additional variables (stress, dispersion of filament densities). Therefore each individual fish of a given treatment group (sham- control or exposure group) was treated as if it represented an independent experimental replicate of that treatment group. Statistical analyses were carried out using GraphPad Prism 4® (USA) Software. Values were given as the mean ± standard deviation (SD) for condition factor and ventilation rate. Pathological changes in the respective tissues are given as median ± mean absolute deviation (MAD) of the individual ranks of at least three individuals examined.

VENTILATION RATE: A one-way ANOVA followed by Tukey’s Multiple Comparison Test was employed to compare the ventilation rates observed in each tank before addition of P. rubescens.

101 3. EXPOSURE EXPERIMENTS ______

Tab. 3.5: Hydrological parameters and time-weighted microcystin concentrations in the tanks containing no (sham-control), low, medium and high P. rubescens densities

Tank Co, sham-control Tank A, low Tank B, medium Tank C, high

Temperature [°C] 18.7 ± 1.0 (57) 18.3 ± 1.0 (57) 18.4 ± 1.1 (57) 17.1 ± 1.2 (57) Oxygen [%] 97.4 ± 2.7 (14) 95.9 ± 2.4 (14) 98.4 ± 2.3 (14) 96.6 ± 2.5 (14) P. rubescens [cells/ml] 1709 ± 1372 (57) 14575 ± 3536 (57) 55984 ± 12091 (57) MC-LRequiv. [µg/l] n.d. (04) 0.3 ± 0.19 (26) 1.8 ± 1.46 (26) 11.0 ± 4.10 (26)

Values are given as mean ± SD; the number of analysed samples (n) is given in parentheses.

Statistical differences in ventilation rates were analysed using a one-way ANOVA followed by Dunnett’s Multiple Comparison Test to compare ventilation rates of the control fish and the rates of the fish immersed in low, medium and high P. rubescens densities at each time point.

CONDITION FACTOR: Condition factors were tested for statistical differences using an one-way ANOVA followed by Dunnett’s Multiple Comparison Test to compare condition factors of control fish and the condition factor of the fish exposed to low, medium and high P. rubescens densities at each time-point.

HISTOPATHOLOGY: Ranking of pathological changes was tested for statistical differences using the non-parametric Mann-Whitney U-Test. Significant differences were determined at the p <0.01 and p <0.05 level for all analyses and are indicated as ** for p <0.01 and * for p <0.05.

RESULTS

Hydrological Parameters, Microcystin Concentrations and P. rubescens Cell Densities Temperature in the tanks ranged between 15 °C and 20 °C. Oxygen consistently exceeded 8 mg/l (corresponding to more than 90% saturation) in all tanks during the experiment (Tab. 3.5). The pH-values determined were 7.9 and 8.0 in tanks Co & A and in tanks B & C, respectively. The 95% confidence interval of P. rubescens mean cell densities determined were 1353-2066 cells/ml in tank A, 13,657-15,493 cells/ml in tank B, and 52,845-59,122 cells/ml in tank C (Tab. 3.5). The time-weighted mean MC-LRequiv. concentrations determined were 0.3, 1.8 and 11.0 µg/l for exposure tanks A, B, and C, respectively. No microcystin was detectable in water samples of tank Co (sham-control).

Mortality, Parasite Infestation, Behavioural Observations and Ventilation Rates Some coregonids died during the course of the experiment: Two (8% mortality) in the low dosed (tank A), three (13% mortality) in the medium dosed (tank B) and six (25% mortality) in the high dosed (tank C) treatment. In contrast to this, only one fish (4% mortality) died in the corresponding sham-control tank. Fish mortality primarily occurred during the first two weeks of the exposure experiment.

102 3. EXPOSURE EXPERIMENTS ______

Tab. 3.6: Ventilation rates determined in coregonids exposed in no (sham-control), low, medium and high P. rubescens cell densities for up to 28 days

Ventilation rate [counts/min] Tank Co, sham-control Tank A, low Tank B, medium Tank C, high

Week 1 Basal rate 92 ± 3.3 95 ± 3.8 96 ± 3.3 95 ± 2.0 Day 1 113 ± 3.8 109 ± 6.0 114 ± 2.3 129 ± 3.8**

Day 3 116 ± 3.3 119 ± 6.8 128 ± 5.7 145 ± 8.9** Day 7 116 ± 5.7 112 ± 6.5 112 ± 3.3 154 ± 17.7**

Week 2 Day 8 129 ± 5.0 122 ± 5.2 137 ± 2.0 153 ± 14.7** Day 9 127 ± 6.0 122 ± 5.2 144 ± 7.3** 181 ± 6.0**

Day 10 111 ± 2.0 127 ± 11.9* 143 ± 8.9** 185 ± 6.0** Day 13 127 ± 5.0 154 ± 20.3* 171 ± 5.0** 182 ± 6.9** Day 14 129 ± 6.8 143 ± 10.5 151 ± 8.9** 165 ± 5.0**

Week 3 Day 15 133 ± 6.0 162 ± 10.1 176 ± 4.6** 174 ± 7.7** Day 16 123 ± 5.0 152 ± 6.5** 160 ± 15.7** 175 ± 6.0**

Day 17 128 ± 5.7 145 ± 6.0** 149 ± 7.6** 159 ± 6.0** Day 20 136 ± 8.6 152 ± 5.7* 171 ± 8.9** 177 ± 8.2** Day 21 123 ± 9.5 153 ± 6.0** 172 ± 9.8** 174 ± 5.2**

Week 4 Day 22 127 ± 17.4 149 ± 7.6 173 ± 11.9** 187 ± 6.8** Day 23 134 ± 4.0 181 ± 2.0** 184 ± 5.7** 193 ± 11.5**

Day 24 120 ± 8.6 156 ± 11.3** 164 ± 5.7** 189 ± 21.3** Day 27 117 ± 7.6 127 ± 7.6 153 ± 8.9** 155 ± 13.2** Day 28 118 ± 5.2 144 ± 7.3** 153 ± 10.0** 169 ± 10.5**

Values are given as mean ± SD of 4 observed individuals for each time point. Significant differences between control fish and exposed fish are indicated (*p <0.05 and **p <0.01) for each time point.

In the third week of exposure, some fish presented with an ectoparasitical infestation with Ichthyophthirius sp. The severity of the infestation was highest in tank C (highest P. rubescens density) with approximately 50% of individuals being affected. No infestation was observed in the sham-control tank. In order to avoid ectoparasitical infestation associated fish mortality compromising the completion of the exposure experiment, remaining coregonids of all treatments (including the sham-control) were therapeutically treated via immersion in a 4% NaCl-solution for 0.5 min. Following exposure to medium and high P. rubescens cell densities (tank B and C), the coregonids immediately responded with broad, irregular gaping and a few of the exposed fish appeared to regurgitate (initiated flow-reversal) through the gills. Broad, irregular gaping was also observed in coregonids exposed to low P. rubescens densities (tank A), however, not until day 13 and to a noticeably lower degree than in tanks B and C. Coregonids in tanks B and C displayed a more hectic swimming behaviour and appeared more susceptible to startling when compared to the low dose and the corresponding sham-control fish. Furthermore, fish in tanks B and C appeared increasingly disoriented and presented with an upwardly inclined posture over the duration of the experiment. The ventilation rates determined before the P. rubescens filament exposure (basal rates) were statistically indistinguishable in all four tanks (Tab. 3.6). The 95% confidence interval of the overall mean ventilation rate determined in tank Co (sham-control) was 120-127 beats/min. The 95% confidence interval of the overall mean ventilation rates determined in the P. rubescens containing tanks were 131-150, 143-163, and 161-178 beats/min in tank A, B and C, respectively. Statistical analyses demonstrated that ventilation rates of the coregonids in the P. rubescens 103 3. EXPOSURE EXPERIMENTS ______

Tab. 3.7: Condition factors of coregonids immersed in no (sham-control), low, medium and high P. rubescens densities for up to 28 days

Condition, [g x mm-3 x 105] Tank Co, sham-control Tank A, low Tank B, medium Tank C, high

Initial 0.59 ± 0.035 (5) Day 7 0.55 ± 0.045 (5) 0.56 ± 0.014 (6) 0.56 ± 0.026 (6) 0.55 ± 0.019 (3) Day 14 0.53 ± 0.029 (6) 0.54 ± 0.032 (5) 0.54 ± 0.048 (5) 0.50 ± 0.047 (4) Day 21 0.50 ± 0.037 (6) 0.52 ± 0.031 (6) 0.49 ± 0.030 (6) 0.51 ± 0.030 (6) Day 28 0.50 ± 0.053 (6) 0.50 ± 0.047 (5) 0.47 ± 0.038 (4) 0.51 ± 0.044 (5)

Values are given as mean ± SD; the number of analysed individuals (n) is given in parentheses.

containing tanks were significantly elevated compared to the corresponding sham-control fish. Ventilation rates in tank C were significantly elevated from the beginning of the experiment at all time points, while in tank B this was the case as of day nine of the exposure only. In tank A there was also a significant increase of ventilation rates as of day ten, however, the relative increase appeared not as pronounced as in tanks B and C (Tab. 3.6).

Condition Factor and Serum Glucose The condition factors decreased with increasing time in all treatments including the sham-control tank (Tab. 3.7). This decrease appeared to occur most rapidly in the highest dosed treatment (tank C), being already apparent at day 14. However, differences between condition factors of the sham-control and exposed coregonids were not statistically significant due to the high inherent variability of this parameter. The limited serum sample volumes resulted in no reliable determination of serum glucose levels (data not shown).

Histopathology

LIVER: Neither initial control fish nor coregonids from the sham-control presented with histopathological changes in the liver beyond the normal range observed in coregonids of this age group. In contrast, P. rubescens exposed coregonids presented with focal liver pathology consisting of hepatocytes with granulated cytosol, reduced glycogen stores, disintegration of the parenchymal liver architecture, cell dissociation, heterochromaty, chromatin margination, necrosis and apoptosis, ruptured vessels, and dilated sinusoids, predominantly peripheral to central veins (Fig. 3.5). The range and severity of pathological changes were exposure time and P. rubescens cell density dependent. The degree of pathological change observed in the P. rubescens exposed fish was significantly different from the corresponding sham-control on day 14 and 21 in tank A, day 7, 14, 21 and 28 in tank B and day 14, 21 and 28 in tank C (Tab. 3.8).

KIDNEY: Coregonids from the sham-control showed histopathological changes in the kidney within the normal range observed in coregonids of this age group. The observed changes included sporadic occurrence of apoptotic and regenerating cells and epithelial cell exfoliation. In fish of the initial control group, these changes appeared marginally more pronounced than in fish of the sham-control tank (Tab. 3.8). In comparison to sham-controls, the kidneys of P. rubescens exposed fish showed more severe tubular degeneration including cell vacuolisation and cell shedding in

104 3. EXPOSURE EXPERIMENTS ______the proxima as well as proteinaceous casts in the tubular lumina. Progressive proximal tubuli degeneration with exposure duration resulted in hyalinisation of tubuli epithelia and interstitial cell lyses (Fig. 3.6). The degree of renal pathology observed in P. rubescens exposed fish was significantly different from the corresponding sham-controls on day 28 in tanks A and B and on day 14, 21 and 28 in tank C (Tab. 3.8). Significant differences were also suggested for exposure as of day 7 to 21 in tanks A and B, however, these effects, although statistically correct, are assumed to have little biological relevance as the values lie within the range of pathological changes observable in the initial- and sham-controls.

HINDGUT: Pathological alterations in the hindgut tissue of controls and exposed fish were characterised by the occurrence of frayed gut villi, exfoliation of epithelial cells, widespread cell lyses, infiltration of leucocytes and intraluminal protein cast deposition. Coregonids from all treatments (including initial- and sham-control) showed a sporadic Proteocephalus sp. infestation

A B

LA V

V LA

CD CD 25 µm 25 µm

C CI LA MC+ LA MC+

V V

MC+ 25 µm 25 µm

Fig. 3.5: Liver tissue of a sham-control coregonid (A) and coregonids exposed to low (B) and high (C) P. rubescens cell densities for 28 days. Sections were stained with H&E (A–C) or microcystin antibodies (CI). Exposed fish presented with disintegration of the parenchymal liver architecture (LA), cell dissociation (CD), dilated sinusoids (white arrow heads), heterochromaty and necrosis (black arrows) predominantly peripheral to blood vessels (V). Comparison of H&E and immunostained sections (C and CI) demonstrates the presence of microcystin (MC+) in histologically altered tissue sections.

105 3. EXPOSURE EXPERIMENTS ______

Tab. 3.8: Histopathological changes observed in various organs of coregonids, immersed in no (sham-control), low, medium and high P. rubescens cell densities up to 28 days

Day Liver Kidney Hindgut Gill

Initial 0 1.0 ± 0.36 (5) 1.5 ±0.24 (5) 0.5 ±0.24 (5) 1.0 ±0.32 (5) Sham-control 7 0.8 ± 0.28 (4) 0.5 ±0.36 (5) 0.5 ±0.36 (5) 1.0 ±0.40 (5) 14 0.5 ± 0.24 (6) 1.0 ±0.27 (6) 1.5 ±0.80 (5) 1.3 ±0.61 (6) 21 1.0 ± 0.21 (6) 1.0 ±0.51 (6) 2.0 ±0.68 (6) 1.3 ±0.32 (6) 28 0.8 ± 0.44 (6) 1.5 ±0.46 (6) 2.0 ±0.85 (6) 0.8 ±0.41 (6) Low 7 1.0 ± 0.33 (6) 1.0 ±0.17 (6) 1.5 ±0.52 (6) 1.3 ±0.35 (6) 14 1.0 ± 0.36 (5)* 1.5 ±0.44 (5)(°) 1.5 ±0.56 (5) 2.0 ±0.32 (5)

21 1.5 ± 0.33 (6)* 2.0 ±0.52 (6) 2.3 ±0.52 (5) 2.3 ±0.65 (6)* 28 1.0 ± 0.44 (5) 2.5 ±0.40 (5)** 1.5 ±0.32 (5) 1.5 ±0.68 (5) Medium 7 1.5 ± 0.41 (6)* 2.0 ±0.44 (6)(°°) 0.5 ±0.44 (5) 1.5 ±0.68 (6) 14 1.5 ± 0.20 (5)** 2.0 ±0.44 (5)(°) 1.5 ±0.96 (5) 1.5 ±0.60 (5) ( ) 21 2.5 ± 0.56 (6)** 2.0 ±0.40 (5) ° 2.3 ±0.67 (6) 2.5 ±0.52 (6)* 28 1.8 ± 0.28 (4)* 2.8 ±0.61 (4)* 2.0 ±0.33 (3) 1.0 ±0.44 (4) High 7 1.5 ± 0.24 (3) 1.5 ±0.21 (3) 1.0 ±0.21 (3) 2.0 ±0.33 (3) 14 2.3 ± 0.61 (4)* 2.5 ±0.25 (4)** 2.8 ±0.50 (4) 2.5 ±0.33 (3)*

21 2.3 ± 0.28 (6)** 2.0 ±0.33 (6)* 2.8 ±0.71 (6) 2.8 ±0.52 (6)** 28 2.0 ± 0.40 (5)** 3.0 ±0.32 (5)** 1.5 ±0.76 (5) 1.5 ±0.20 (5)

Changes were ranked from none (0) to severe (4) including intermediate ranks (e.g. 0.5, 1.5). Values are presented as median ± mean absolute deviation for each time point; the number of fish assessed (n) is given in parentheses. Significant differences between control and immersed fish are indicated (*p <0.05, **p <0.01) for each time point. (°) Statistical significance, although mathematically correct was assumed to have little biological relevance due to values that lie within the range of pathological changes observed in the initial- and sham-controls.

after the first week. The severity of infestation appeared to be comparable across all treatments and was no longer detectable at the end of the experiment (data not shown). In contrast to corresponding sham-controls, gastro-intestinal tissue of P. rubescens exposed fish showed an enhanced loss of the mucosal structure and extensive epithelial degeneration (Fig. 3.7). These pathological changes appeared to be most predominant after the third week of exposure, especially in fish of tank C. However, the range and severity of pathological change observed in P. rubescens exposed fish was not significantly different from the corresponding sham-controls (Tab. 3.8).

GILLS: P. rubescens filaments were observable in gills of exposed fish ranging from single filaments in tank A to accumulation filament bundles in fish of tanks B and C (Fig. 3.8). Neither initial- or sham-control fish showed histopathologically relevant alterations in the gill beyond the normal range. In contrast, macroscopically the gills of P. rubescens exposed coregonids often appeared congested and frequently showed pathological changes including fusion, partial rupture and hyperplasia (clubbing) of the tips of the secondary lamellae, enhanced vacuolisation, oedema, apoptosis and focal necrosis in the lamellar epithelium and a partial dissociation of the epithelium sometimes leading to an exfoliation of epithelial cells into the gill mucus (Fig. 3.8). Parasitic Ichthyophthirius sp. were observable in the gills of all P. rubescens exposed fish after the third exposure week, and were particularly evident in fish of tank C. The range and severity of pathological changes observed in P. rubescens exposed fish was significantly different from the corresponding sham-control tank as of the second week in tank C, and after the third week in

106 3. EXPOSURE EXPERIMENTS ______

A B D

D L L

25 µm 25 µm H C D BI

H H

L MC+ MC+

MC+ L

25 µm 25 µm

Fig. 3.6: Kidney tissue of a sham-control coregonid (A) and coregonids exposed to low (B) and high (C) P. rubescens cell densities for 28 days. Sections were stained with H&E (A–C) or microcystin antibodies (BI). Exposed fish presented with enhanced tubuli degeneration (D) including cell vacuolisation (black arrows) and cell shedding in the proxima (white arrow heads), hyalinisation (H) of tubular epithelia and interstitial cell lysis (L). Comparison of H&E and immunostained sections (B and BI) demonstrates the presence of microcystin (MC+) in histologically altered tissue sections.

tanks A and B (Tab. 3.8). However, overall pathological scores were not significantly different between P. rubescens exposed and sham-control fish at the end of the experiment (Tab. 3.8). Despite a reduced survival rate at the highest P. rubescens exposure treatment, none of the observed individual pathological changes were deemed to be of a severity that would imply being critical for coregonid survival.

Microcystin-Immunhistology Microcystin-immunhistology is summarised in Table 3.9. Microcystin-positive staining was most distinct in the liver of the exposed coregonids, with regard to both the number of positive sections (individual fish) and the area of immunopositive staining. Livers of P. rubescens treated coregonids showed immunopositive staining predominantly consisting of dispersed positive foci after just one week of exposure in tanks B and C and as of week two of exposure in tank A (Fig. 3.8). Only irregular immunopositive staining was detectable in kidney and gut tissue of fish

107 3. EXPOSURE EXPERIMENTS ______

A MC+ AI

MC+ MC+

50 µm 50 µm

Fig. 3.7: Gastro-intestinal section of a coregonid exposed to high P. rubescens cell densities for 28 days. Sections were stained with H&E (A) and microcystin antibodies (AI). Exposed fish presented with extensive epithelial degeneration and exfoliation (white arrow heads) of epithelial cells. Comparison of H&E and microcystin-immunostained sections demonstrates the presence of microcystin (MC+) in histologically altered tissue sections.

A B CD CD CD

F F

25 µm 25 µm

C D

E 25 µm 25 µm

Fig. 3.8: Gill tissue of a sham-control coregonid (A) and coregonids exposed to low (B) and high (C & D) P. rubescens cell densities for 28 days. Sections A–C were stained with H&E, while section D is a fresh (non- fixed) gill sample. Exposed fish presented with fusion (F), partial rupture and hyperplasia (clubbing) of the tips (white arrow heads) of the secondary lamellae, enhanced vacuolisation, focal necrosis (black arrows) in the lamellar epithelium and partial dissociation (CD) of the epithelium sometimes leading to exfoliation of epithelial cells (E). P. rubescens filaments were observable in gills of P. rubescens exposed fish ranging from single filaments to accumulation of filament bundles (FB). 108 3. EXPOSURE EXPERIMENTS ______

Tab. 3.9: Microcystin-immunohistochemical positivity in organ sections of coregonids, immersed in no (sham- control), low, medium and high P. rubescens cell densities up to 28 days

Day Liver Kidney Hindgut Gill

Initial 0 n.d. (4) n.d. (4) n.d. (4) n.d. (4)

Sham-control 7 n.d. (5) n.d. (5) n.d. (5) n.d. (5) 14 n.d. (4) n.d. (4) n.d. (4) n.d. (4) 21 n.d. (5) n.d. (5) n.d. (5) n.d. (5) 28 n.d. (4) n.d. (4) n.d. (4) n.d. (4)

Low 7 i.ip. (6) i.ip. (6) i.ip. (6) n.d. (6) 14 + (5) + (5) i.ip. (5) n.d. (5)

21 + (6) i.ip. (6) i.ip. (6) n.d. (6) 28 + (5) i.ip. (5) + (5) n.d. (5)

Medium 7 + (6) i.ip. (6) + (6) n.d. (6) 14 + (5) + (5) + (5) n.d. (5)

21 + (6) + (5) i.ip. (6) n.d. (6) 28 ++ (4) + (4) + (4) n.d. (4)

High 7 + (3) + (3) + (3) n.d. (3) 14 ++ (4) + (4) + (4) n.d. (3) 21 + (6) + (5) + (6) n.d. (6) 28 + (5) + (5) + (4) n.d. (5)

Immunopositive staining was ranked as none (n.d.), sporadic (+), cumulative (++) and extensive (+++). Classification is presented as median for each time point; the number of fish assessed (n) is given in parentheses. When only a minority of assessed fish showed sporadic immunopositivity, these samples were denoted as irregular immunopositive (i.ip.). None of the assessed fish showed immunopositivity when denoted as not detectable (n.d.).

in tank A, while immunopositive foci in kidney and gut tissue of fish of tanks B and C were detected consistently (Fig. 3.6 and 3.7). Microcystin-immunopositive staining in liver, kidney and gut tissue co-localised to histopathological changes in the serially sectioned tissues (Fig. 3.5-3.7), whereas no microcystin staining was detectable in gills of any of the exposed fish.

DISCUSSION

Hydrological Parameters, Microcystin Concentrations and P. rubescens Cell Densities Temperature and oxygen were stable and comparable between control and exposure tanks. Observed effects were therefore considered, to be the consequence of P. rubescens exposure (and microcystin contaminations therein). P. rubescens cell densities and corresponding microcystin-

LR equivalent concentrations (MC-LRequiv.) determined in the tanks throughout the experiments were stable, i.e. 1500 cells/ml (0.3 µg MC-LRequiv./l) for tank A, 15,000 cells/ml (2 µg MC-LRequiv./l) for tank B, and 55,000 cells/ml (11 µg MC-LRequiv./l) for tank C. These cell densities represent the environmental situations in pre-alpine lakes quite closely. Indeed, P. rubescens cell densities of more than 1500 cells/ml were permanently present throughout the entire period from January 2001 to November 2001 in Lake Ammersee, Germany (Ernst et al., 2001 & Ernst, unpublished data). For Lake Bourget, France, Jacquet et al. (2005) reported P. rubescens cell densities up to 20,000 cells/ml uniformly distributed during winter circulation and densities of more than 50,000

109 3. EXPOSURE EXPERIMENTS ______cells/ml in the metalimnic layer during stratification in summer. Microcystin-concentrations, predominantly dmRR variants, reached 6.7 µg MC/l (Briand et al. 2004).

Behavioural Observations and Ventilation Rates P. rubescens exposed coregonids appeared to actively avoid P. rubescens uptake as suggested by irregular gaping and regurgitation. This was especially prominent for coregonids exposed to medium (15,000 cells) and high (55,000 cells/ml) cell densities. This observation is corroborated by similar findings by Tencalla & Dietrich (1997) in trout and Ernst et al. (2006a) in coregonids. In addition, fish exposed to the two highest densities presented with increased startle response, hectic swimming and disorientation which appeared to become more pronounced with increasing exposure duration. As this type of stress response (Little, 2002) was reported earlier for fish exposed to P. rubescens (Ernst et al., 2006a) and other, non-filamentous cyanobacterial species, known to produce microcystins (Baganz et al., 2004; Baganz et al., 1998; Carbis et al., 1996a; Råbergh et al., 1991), it is suggested that this type of stress is unrelated to the filamentous form of P. rubescens but could be a direct consequence of exposure to microcystins and probably other P. rubescens components as summarised by Welker & von Döhren (2006). Indeed, this interpretation is supported by observations of Bury et al. (1996a), who reported increased physiological stress, as measured by elevated plasma cortisol and plasma glucose levels, in brown trout (Salmo trutta L.) exposed to lysed toxic M. aeruginosa cells. P. rubescens induced stress can also be inferred from the elevated ventilation observed in the exposed coregonids, as indicated by the significantly increased ventilation rates. Increased ventilation frequency is an indicator for enhanced stress, irrespective of the type of stressor present (Barreto & Volpato, 2004). Especially in the third exposure week elevated ventilation rates may have also resulted from the parasitical Ichthyophthirius sp. infestation, however, as significant increased ventilation rates were observed already prior to the Ichthyophthirius sp. infestation, the observed effect on the coregonids ventilation rate is assumed to be causally related to P. rubescens exposure. As increased ventilation rates were also observed in coregonid yearlings treated with a single oral dose of an environmentally relevant P. rubescens concentration (Ernst et al., 2006a), this may suggest that the increased ventilation is more likely the result of the systemic microcystin exposure rather than the mechanical irritation of P. rubescens filaments on the coregonid gills.

Condition Factor, Fitness & Mortality Growth and condition decreased in all fish in the study presented here, irrespective of the exposure type (sham-control or exposed). This may be ascribed to the fact that the planktivorous coregonids are non-domesticated fish with limited cultivation capabilities. However, stress related effects on fish fitness appeared more prominent with regard to the susceptibility of the individual treatment and control groups to ectoparasitic infestation and mortality. Indeed, the observed Ichthyophthirius infestation occurred only in the P. rubescens exposed fish and most severely in

110 3. EXPOSURE EXPERIMENTS ______the highest treatment group, suggesting that the P. rubescens exposed coregonids were less resistant to the parasitic infestation. Despite that the pathological alterations were of limited severity, a general reduction of fitness was suggested by the fact that a dose-response relationship between filament density and fish mortality was noted.

Histopathology The observed pathological changes in liver and kidney are characteristic of pathological lesions as described earlier for cyanobacteria and microcystin intoxications in coregonids (Ernst et al., 2006a) and other fish species (Carbis et al., 1996a; Fischer & Dietrich, 2000; Råbergh et al., 1991; Tencalla & Dietrich, 1997). A causal relationship of tissue damage with the presence of microcystin-containing P. rubescens is additionally suggested by the observation that a high number of pathological changes, whether observed in liver or kidney, were also immunopositive for microcystin as already shown for coregonids orally exposed to P. rubescens (Ernst et al., 2006a) and for trout and carp gavaged with M. aeruginosa (Fischer et al., 2000; Fischer & Dietrich, 2000). In the medium dose treatment (tank B), overt liver pathology was already observable on day seven of exposure. At this time point, a comparable pathology was also observable in liver of fish exposed to the highest P. rubescens density (tank C), although this was not statistically significant (primarily due to reduced sample numbers as a result of mortality). The overall severity of pathological alterations observed in P. rubescens exposed coregonid kidneys was comparable to that observed in the livers, and was also dose- and time-dependent. This observation corroborates earlier findings by Fischer & Dietrich (2000) for carp orally exposed to bloom material containing microcystin. In contrast to the pathology observed in the gut of coregonids orally exposed to P. rubescens (Ernst et al., 2006a) and carp intraperitoneally treated with 550 µg MC-LR/kg bw (Råbergh et al., 1991), the degree of pathological changes observed in the hindgut of exposed coregonids in this study was not significantly different to that observed in the sham control. However, a number of pathological changes co-localised with immunopositive staining for microcystin, suggesting that the observed pathology may be the consequence of the actual microcystin exposure. An explanation for the comparatively high degree of tissue alteration in hindgut across all treatments is offered by the fact that coregonids from all treatments presented with sporadic Proteocephalus sp. infestation at the beginning of the experiment. As the progression and severity of the Proteocephalus infestation was comparable across all treatment groups (including initial- and sham-control fish) the effect on the overall pathology examined in the study appears to be negligible. Proteocephalus sp. infestation is relatively common in salmonids of the pre-alpine lakes (Hanzelova et al., 1995). As the coregonids used in this experiment were wild-hatched, the observed Proteocephalus infestation appears to be inherent to this coregonid population and already present prior to the start of the experiment.

111 3. EXPOSURE EXPERIMENTS ______

Gill pathology in P. rubescens exposed coregonids was slightly more severe than that observed in the concurrent control fish and was comparable to the changes reported for carp treated intraperitoneally with 25 and 50 µg MC/kg bw or immersed in 1.7 µg MC/ml (Carbis et al., 1996a). However, the observed alterations in gill tissue were not microcystin-immunopositive. In view of the latter observation and the pathology described for an Anabaena sp. associated fish kill (Toranzo et al., 1990), it appears most likely that the gill pathology observed in the P. rubescens exposed coregonids resulted from mechanical abrasion and irritation from P. rubescens filaments and Ichthyophthirius infestation rather than microcystin associated effects. This interpretation is further corroborated by the fact that the severity of the observed gill pathology declined subsequent to therapeutic treatment with 4% saline and abatement of Ichthyophthirius infestation. Furthermore, in a previous study oral dosing of coregonids with P. rubescens resulted in no significant gill pathology (Ernst et al., 2006a). It is interesting to note, however, that the degree of Ichthyophthirius infestation appeared P. rubescens density dependent, thus suggesting that the severity of Ichthyophthirius infestation may have been dependent on the general fitness or immunological capability of the individual coregonids as already discussed above.

Conclusion The present subchronic exposure experiment confirmed the initial hypothesis that subchronic and chronic exposure to low cyanobacterial cell densities and microcystin contaminations can enhance physiological stress and sustained pathological alterations in exposed coregonids. Moreover, the progression and severity of the observed adverse effects in P. rubescens exposed coregonids occurred in a dose-dependent manner, indicating that the higher the P. rubescens cell densities and hence the microcystin-concentrations, the more pronounced and earlier the occurrence of the adverse effects. However, even very low cell densities (1500 P. rubescens cells/ml and

0.3 µg MC-LRequiv./l) resulted in significant physiological stress, affecting coregonid fitness and resulting in substantial pathology, i.e. tissue damage and therefore sustained alteration in normal organ function. As all of the P. rubescens densities employed in this subchronic toxicity study represent realistic environmental situations in pre-alpine lakes which may be sustained for weeks or months, the subchronic effects described by the four week experiment presented here probably underestimate the potential adverse effects encountered by feral coregonid populations in lakes afflicted with P. rubescens blooms. This is particularly problematic, when fish cannot actively avoid P. rubescens exposure, e.g. during whole water-body blooms in winter circulation. Altogether the study presented here supports the theory that stratified and dispersed P. rubescens blooms may be implicated in the observed reduced weight and hence fitness of coregonids in pre-alpine lakes such as Lake Ammersee.

ACKNOWLEDGEMENTS We would like to thank RCC Ltd. for excellent technical support and Daniela Schmid for experimental assistance. The financial support and unrelenting interest of the Arthur and Aenne Feindt Foundation (Germany) in this topic is highly appreciated.

112

4. FIELD STUDIES

4.1 ABUNDANCE AND TOXICITY OF PLANKTOTHRIX RUBESCENS IN THE PRE-ALPINE LAKE AMMERSEE, GERMANY

Bernhard Ernst, Stefan J. Hoeger, Evelyn O’Brien, Daniel R. Dietrich

Environmental Toxicology, University of Konstanz, P.O. Box X918, 78457 Konstanz, Germany

Submitted for publication in Harmful Algae

ABSTRACT Regular occurrences of the cyanobacterium Planktothrix rubescens have been observed in several lakes that have undergone recent re-oligotrophication, e.g. Lake Ammersee. Planktothrix species are known to produce microcystins, potent phosphatase inhibitors that have been associated with morbidities and mortalities in humans and animals. The aim of this study was to characterise the temporal and spatial abundance and toxicity of P. rubescens in Lake Ammersee. Exemplified by the indigenous coregonid population, the findings were finally discussed in the context of possible adverse effects on aquatic organisms. P. rubescens cell densities and biovolumes were calculated via fluorescence image analyses. P. rubescens was present during the whole observation period from 1999-2004, albeit at different cell densities. Maximum biovolumes of 45 cm3/m2 were observed in May 2001. Filaments were regularly distributed over the entire water column during winter and stratified in distinct metalimnic layers during summer, reaching maximum cell densities of ≤15,000 (winter) and ≤77,000 cells/ml (summer). The results demonstrate that P. rubescens mass occurrence is strongly influenced by water transparency, i.e. illumination in the metalimnion. Moreover, the P. rubescens abundance appears to result from regular phosphate depletion in the epilimnion, possibly additionally benefiting from high nitrogen loads. Microcystin (MC) was detectable in 27 and 38 of 54 seston samples via HPLC and Adda-ELISA measurements, respectively. The main microcystin congeners in the seston samples were [Asp3]-MC-RR and [Asp3, Dhb7]-MC-RR. Microcystin concentrations correlated significantly with the respective phycoerythrin (PE)- concentrations. The variation in the MC/PE-ratios was low suggesting that the microcystin production of P. rubescens in Lake Ammersee is consistent and indicating that the appearance of P. rubescens coincides with measurable microcystin levels. Moreover, the observation of pronounced metalimnic oxygen depletions appears to be causally related to recurring high P. rubescens abundance. In conclusion the results suggest that aquatic organisms such as coregonids are regularly confronted with potentially adverse P. rubescens densities, which might provide a possible explanation for the often observed challenge of coregonid populations in P. rubescens containing pre-alpine lakes.

KEYWORDS: Planktothrix; Cyanobacteria; Re-oligotrophication; Phosphorous; Nitrogen; Transparency; Secchi depth; Microcystin; Coregonids; Fish

113 4. FIELD STUDIES ______

INTRODUCTION Cyanobacteria are important constituents of phytoplankton communities and ubiquitous in lakes of different nutritional status. Approximately 50 of 2000 known cyanobacterial species are recognised to produce toxic molecules, e.g. alkaloids and peptides (Sivonen & Jones, 1999). Many of these toxins have been associated with mortalities of wild and domestic animals as well as severe human intoxications. Among these toxins, microcystins14 are most frequently found and have gained attention due to their potent inhibition of protein phosphatases and associated morbidities and mortalities in humans and animals (Briand et al., 2003; Falconer, 2001). Microcystin-producing cyanobacteria are present in coastal and inland waters, primarily in naturally eutrophic waterbodies and waters that have experienced nutritional enrichment due to anthropogenic influences (eutrophication) (Bartram et al., 1999). However, in contrast to presently eutrophicated water bodies with Anabaena sp., Aphanizomenon sp. and Microcystis aeruginosa blooms, regular mass occurrences of the cyanobacterium Planktothrix rubescens have been observed in lakes that had undergone recent re-oligotrophication. This especially includes lakes in the pre-mountainous areas of the Alps characterised by an ice-age modulated landscape of hills and valleys (pre-alpine regions) (Tab. 4.1).

Tab. 4.1: Size and nutritional status in (pre-) alpine lakes with documented P. rubescens abundance during the last decade

Country Lake Size [km2] Trophic status Abundance References

AUSTRIA Wörthersee 19.4 mesotrophic 2000-2006 Schulz et al. (2000-2007) Ossiacher See 10.8 mesotrophic 2000, 2003 -2006 Schulz et al. (2000-2007) Millstätter See 13.3 oligo-mesotrophic 2000-2003, 2005-2006 Schulz et al. (2000-2007) Mondsee 16.6 mesotrophic 1994-1997, 2001 Kurmayer 2004; Teubner 2004

FRANCE Lac du Bourget 42 mesotrophic 1996, 1999, 2001 Jacquet 2005; Leboulanger 2002

GERMANY (BAVARIA) Ammersee 46.6 mesotrophic 1996, 1998-2001 Teubner 2004; Ernst 2001 Chiemsee 79.9 mesotrophic 2004, 2006 Ernst unpublished data Starnberger See 56.4 mesotrophic 1997, 2005 Ernst unpublished data; Nixdorf 2004

ITALY Lago Maggiore 212.5 oligo-mesotrophic 1995-1999 Morabito 2002 Lago di Garda 368.0 oligo-mesotrophic 1995-2000 Salmaso 2000; Salmaso 2002 Lago d’Iseo 62 mesotrophic regularly Salmaso 2000 Lago di Como 146.0 mesotrophic 1997-1999 Buzzi 2002; Bettinetti 2000 Lago di Pusiano 5.3 eutrophic 2001, 2002 Legnani 2005

SWITZERLAND Lac de Neuchatel 217.9 oligo-mesotroph 1999-2004 www.die3seen.ch Zürichsee 65.1 mesotrophic 1993-2000 Hoeger 2005; Walsby 2002 Lago di Lugano 48.8 meso-eutroph regularly Salmaso 2000 Thunersee 47.8 oligo-mesotroph regularly Ochsenbein 2003 Bielersee 37.8 meso-eutroph 2000-2006 www.die3seen.ch Murtensee 22.8 mesotroph 2004-2006 www.die3seen.ch Sempachersee 14.4 eutroph 1989-1997 Bürgi 2002; Mez 1998

14 cyclic heptapeptides, sharing the common structure cyclo(-Adda-D-Glu-Mdha- D -Ala-L-X- D -MeAsp- L -Z) where X and Z are variable L -amino acids, Adda is an uncommon amino acid 3-amino-9-methoxy-2,6,8,-trimethyl-10-phenyl-4,6,-decadienoic acid, D -MeAsp is 3-methylaspartic acid, and Mdha is N-methyl-dehydroalanine 114 4. FIELD STUDIES ______

The mass occurrence of P. rubescens is predominantly ascribed to two forms of specialisation providing for an ecological niche as well as a competitive advantage over green algae: i.e. an efficient regulation of buoyancy via semi-permeable gas vesicles enabling P. rubescens filaments to stratify effectively in the water column and the production of allophycocyanin, phycocyanin and phycoerythrin, photopigments enabling maximum utilisation of light energy and existence under low light conditions (Feuillade, 1994; Walsby & Schanz, 2002). Consequently, P. rubescens stratify in compact metalimnic layers overshadowed by the epilimnic biocoenosis during summer stratification. Moreover, P. rubescens filaments can grow at low light conditions during circulation in the late autumn to early spring months or even below an ice cover during winter (Blikstad-Halstvedt et al., 2007). Metalimnic blooms of Planktothrix species are often observed to co-occur with marked oxygen deficiencies in the metalimnion (Buzzi, 2002; Ernst et al., 2001; Krupa & Czernas, 2003; Lindholm & Meriluoto, 1991; Salmaso, 2000). Indeed, the senescence of cyanobacterial blooms may generate an increased oxygen demand and consequently result in massive oxygen depletion. In addition, cyanobacterial senescence provides for the release of cyanobacterial toxins (Malbrouck & Kestemont, 2006). In comparison to other cyanobacterial species, Planktothrix sp. have been shown to contain the highest concentrations of microcystin per gram dry weight (Fastner et al., 1999b). Depending on the P. rubescens abundance, toxicity and distribution, both the release of cyanobacterial toxins as well as metalimnic oxygen deficiencies may result in adverse effects on aquatic organisms (Sivonen & Jones, 1999; Wiegand & Pflugmacher, 2005), especially on coregonids (Ernst et al., 2007; Ernst et al., 2006a). The aims of this study were therefore • to characterise the spatial and seasonal abundance of P. rubescens in Lake Ammersee, • to characterise the variability of microcystin content in P. rubescens and • to investigate the temporal co-occurrence of alterations in P. rubescens abundance and metalimnic oxygen depletion. Using the indigenous coregonid population as an example, the findings were finally discussed in the context of possible adverse effects on aquatic organism.

MATERIALS & METHODS

Chemicals All reagents and solvents employed were of analytical or chromatographic grade and quality and purchased from Fluka (Switzerland), Merck (Germany), Riedel de Haen (Germany), Roth (Germany) or Sigma (Germany). Water was purified to 18.2 MΩ/cm.

115 4. FIELD STUDIES ______

J F M A M J J A S O N D 1999 2000 2001

2002

2003

2004

Fig. 4.1: Time-points and intervals of sampling during the six-year survey. Sampling was shortened in 2000 (no Secchi measurements; sampling in August 2000 consisted of only seston-sampling).

Study Location Lake Ammersee is a typical pre-alpine lake, located in the south of Germany at 553 m altitude which arose following the retreat of the glaciers at the end of the last ice age. The lake is elongated in south-north orientation (16 km length and 2.9 km average width) and formed like a tube with steep shores on both west and east sides. Lake Ammersee is dimictic with a surface area of 46.6 km2, a total volume of 1750 x106 m3 and a maximum and average depth of 81.1m and 37.5m, respectively (Grimminger, 1982). Complete winter circulation, including the lake bottom, is regularly achieved. Water residence time is 2.7 years, whereby the lake’s principal water source is the river Ammer – with a mean flow rate of 16.6 m3/s. Due to the large catchment area of the river Ammer the lake collects water from an area of 993 km2, including widely natural and agricultural, but also urban and industrial influenced regions. Until the late 1970’s, the lake underwent a distinct phase of eutrophication, primarily as a result of increased urbanisation, detergent use and intensification of agriculture in the catchment area, reaching yearly mean total phosphorous concentrations of 60 µg/l. Due to a reduction of anthropogenic influences, the continued eutrophication was halted and reversed, i.e. a re- oligotrophication process was initiated. In consequence, the yearly mean total phosphorous concentrations decreased to approximately 10 µg/l (Kucklentz et al., 2001).

Sample Sites and Sampling Intervals Samples and field measurements were generally taken in the middle of the lake (47°59’(32)N; 11°07’(95)E). Sampling time-points and intervals are depicted in Figure 4.1. Sampling consisted of water sampling for the determination of P. rubescens cell densities, measurement of water transparency, determination of temperature and oxygen profiles and in addition, monthly seston sampling for the determination of the P. rubescens microcystin content. In addition, two identically sampled previous seston samples, taken in August and November 1998, were included in the sample cohort.

P. rubescens Abundance For the determination of the P. rubescens cell densities, water samples (50 ml) were taken using a Ruttner flask sampler (Richter & Wiese, Germany). Samples were taken at 0, 3, 5, 25 and 40 m

116 4. FIELD STUDIES ______

depth. Eight additional metalimnion-specific samples were taken, ranging between 6 to 15 m depth, always adjusted to the respective metalimnic temperature and thus depth stratification. Water samples were immediately fixed with Lugol’s iodine solution and stored in darkness for at least 24 h until sample filtration. Defined sample volumes were filtered on nitrocellulose membranes (pore size 8 µm - diameter 25 mm, Schleicher & Schuell, Germany). Filters were dried in darkness at room temperature and subsequently analysed via fluorescence microscopy and image processing according to the protocol published in Ernst et al. (2006b). P. rubescens biovolumes were estimated for each time-point and sample depth via multiplication of the cell density with the average cross section surface of 25.6 µm2 as given for P. rubescens (Walsby et al., 1998) and a mean cell length of 3 µm (Ernst et al., 2006b). Using GraphPad Prism 4® (USA), the biovolumes obtained were further integrated from the water surface to 40 m depth to provide for the P. rubescens biovolume per m2. Finally, the development of the P. rubescens biovolume was compared on a ten-day scale throughout the investigated period (missing time-points were interpolated from data pre- and post- missing data-points).

Limnological Parameters Water transparency was determined via a Secchi disk (diameter 25 cm). Secchi measurements were always carried out by the same person. Yearly mean values of the Secchi depth (ØZs) were determined via calculation of a time-weighted average of the Secchi depths, determined during the vegetation period i.e. from the beginning of May until the end of October each year. No Secchi measurements were carried out in 2000.

The individual euphotic depths (Zeu) were estimated from the respective Secchi depths (Zs) as

Zeu = 2.5 x Zs (Lemmin, 1978). Seasonal variations of metalimnic light conditions were estimated using the trends of the euphotic depth measurements throughout the vegetation period. Temperature, dissolved oxygen and oxygen saturation were determined using a portable oxy- meter (Oxi-197, WTW, Germany). Measurements were performed at intervals of 1 m starting from the surface to 20 m depth and in 5 m intervals starting from 20 to 40 m depth. The upper limit of the metalimnic layer (Zmeta) was defined as the depth with a decrease in temperature of

≥1°C/m and is therefore approximately 1 m below the mixing zone (Zmix); Zmeta = Zmix + 1. The seasonal change of the upper metalimnic limit was estimated using the Zmeta determined from the beginning of May until the end of October each year.

In order to characterise the environmental conditions at the depth of P. rubescens stratification, the depths of maximum P. rubescens cell densities (peakmax) were compared to the Secchi depths

(ZS) and the upper limits of the metalimnic layer (Zmeta). For this, all measurements from the beginning of May until end of October obtained as of 1999 until 2004 were employed. In addition variations in light conditions and metalimnic stratification were compared within each year and amongst the years in order to illustrate seasonal and annual differences of metalimnic conditions.

117 4. FIELD STUDIES ______

Seston Sample Preparation P. rubescens microcystin and phycoerythrin contents were analysed in monthly seston samples (see also Fig. 4.1). Samples were taken with a ballasted 40 µm gauze net hauled between 5 and 15 m depth horizontally through the water column. The taxonomical composition of the seston samples was determined via light microscopy using fresh sample aliquots. Classification of cyanobacterial genus was performed in accordance with Anagnostides & Komárek (1988), Suda et al. (2002), and Geitler (1932). The remaining sample volumes were immediately frozen (–20 °C) and stored until lyophilisation. For analysis, frozen samples were thawed and lyophilised via speed-vac evaporation (Alpha RVC, Christ, Germany), weighed and portioned for methanolic microcystin and aqueous phycoerythrin extraction.

Seston Sample Microcystin Contents Seston sample microcystin (MC) contents were determined in order to characterise the variability of microcystin content in P. rubescens of Lake Ammersee. As the majority of Planktothrix cells lyse due to freezing and lyophilisation, Planktothrix cell counts were inappropriate as reference for toxin quantification. Thus, P. rubescens microcystin content was determined using the phycoerythrin (PE) concentration of the respective seston sample as reference and expressed as MC/PE. Use of phycoerythrin as reference - a widely Planktothrix specific photopigment (Ernst et al., 2006b) – has the advantage of minimising the interference by zooplankton and algae often reported for chlorophyll and biomass measurements. The use of phycoerythrin as reference additionally minimises the interference due to variations within the seston sample composition.

QUANTIFICATION OF PHYCOERYTHRIN: Biliprotein-concentrations in the lyophilised seston samples were determined via extraction of defined sample quantities (≤50 mg dw) in 1 ml phosphate buffered saline (pH 7.0) by three freeze-thaw cycles using liquid nitrogen. Each extract was centrifuged (45 min at 40,000 x g) and the absorption (A) of the resulting supernatants was determined at wavelengths of 562 nm, 615 nm and 652 nm. Absorption was additionally determined at 750 nm for nullification (N). The optical density for the respective wavelength

(ODxxx) was calculated as ODxxx = A – N. Phycocyanin (PC), allophycocyanin (APC) and phycoerythrin (PE) concentrations were calculated according to the description of (Tandeau de Marsac, 1977) using the following equations:

PC [mg/l] = (OD615 – 0.747 x OD652) / 5.34

APC [mg/l] = (OD652 – 0.208 x OD615) / 5.09

PE [mg/l] = (OD562 – 2.41 x PC – 0.849 x APC) / 12.7 The extraction procedure was repeated at least three times and the mean phycoerythrin concentration for each seston sample calculated. The analytical protocol provided for a quantification limit of ≤0.1 µg PE mg/l dw.

QUANTIFICATION OF MICROCYSTIN: Microcystin was extracted from a defined sample quantity (≤70 mg dw) via alternate shaking of the sample suspended in 10 ml of 75% methanol and

118 4. FIELD STUDIES ______ultrasonication at 35 kHz for 1 h. Subsequently the suspension was centrifuged (10 min at 10,000 x g) and the resulting supernatant stored at room temperature while the remaining pellet was re-extracted with 75% methanol. The extraction procedure was repeated three times for each seston sample and the supernatants were pooled to give one extract for each seston sample. For further purification and microcystin concentration, sample extracts were reduced to their aqueous component (approximately 7.5 ml) under a gentle nitrogen stream, replenished to 30 ml with deionised water and applied to preconditioned C-18 end-capped solid phase extraction (SPE) cartridges (mass: 500mg; Isolute® C18(EC), Germany). Microcystins in the extracts were eluted from the cartridges with 12 ml methanol. Eluents were dried under a gentle nitrogen stream and re-dissolved in 3 ml 20% MeOH for HPLC- and anti-Adda ELISA-analyses (see below).

HPLC: HPLC was performed using a Shimadzu (Germany) HPLC equipment (including controller (SCL-10AVP), autosampler (SIL 10ADVD), two pumps (LC-10ATVD), degasser (DGU-14A) and diode array detector (SPD-M10AVP)), with an analytical C18 column (Grom Sil 120 ODS-4 HE, 5 µm, 250 x 4 mm, Stagroma, Germany). A gradient with water (0.05% TFA) and acetonitrile (0.05% TFA) was used as the mobile phase according to the method described by Lawton et al. (1994). Microcystin congeners were detected and identified via retention time and typical spectra in comparison with MC-LR, MC-RR, MC-YR, MC-LF, MC-LW standards (all purchased from Alexis, Switzerland), [D-Asp3]-MC-RR and [D-Asp3]-MC-LR (both kindly provided by Prof. Meriluoto, Åbo Akademi University, Turku, Finland) and [Asp3,Dhb7]-MC-RR (kindly provided by Dr. J. Blom, University of Zurich, Switzerland). Microcystin concentrations were calculated based on the peak area of the internal MC-LR standard employing a factor of 0.79 for quantification of [Asp3,Dhb7]-MC-RR (Hoeger et al., 2007). Based on a detection limit of 10 ng microcystin per injection and an injection volume of 20 µl, the limit of quantification was estimated at 0.05 µg MC-LRequiv./mg dw.

ELISA: The anti-Adda ELISA Kit (Abraxis LLC, USA) is based on an antiserum raised against the unique C20 amino acid 3-amino-9-methoxy-2,6,8-trimethyl-10-phenyl-4,6-decadieonic acid (Adda), which is common to the majority of known microcystin variants (Fischer et al., 2001). Therefore, Adda-ELISA analyses represent a good approach for the determination of the overall microcystin concentration in the seston samples. The ELISA was performed according to the manufacturer’s instructions. Each sample extract was analysed twice using duplicate measurements yielding a mean microcystin content per sample. Due to unspecific matrix- interfering compounds in the highly concentrated seston sample extracts, the limit of quantification was estimated at 0.05 µg/mg dw.

Statistical Analyses Statistical analyses were carried out using GraphPad Prism 4® (USA) Software. Values are given as mean ± SEM, unless indicated otherwise. The yearly mean values of the Secchi depth (ØZs) were analysed using a two-tailed one-way ANOVA followed correction of multiple analyses using a Bonferroni’s post-test.

119 4. FIELD STUDIES ______

The depth of maximum P. rubescens cell densities (peakmax) was analysed for statistical correlation versus the Secchi depth (ZS) and the upper limit of the metalimnic layer (Zmeta) via Pearson’s correlation test. Statistical significance of the correlation was determined at the p <0.05 level and indicated as *** for p <0.001 and * for p <0.05. Seasonal tendencies in light conditions were estimated via polynomial trends of the euphotic depth (Zeu) from May until October (n = 26 in 1999, n = 17 in 2001, n = 18 in 2002 and 2003). No trends were given for 2000 and 2004 due to missing samples or limited sampling intervals (n = 4), respectively. The seasonal change in lake stratification was estimated via determination of polynomial trends of the upper limit of the metalimnic layer (Zmeta) determined from May to October (n = 25 in 1999, n = 18 in 2001, n = 17 in 2002 and n = 15 in 2003). No trends were given for 2000 and 2004 due to limited sampling intervals (n ≤4). The microcystin congener composition of microcystin-positive Lake Ammersee seston samples as determined via HPLC analyses is given as mean percentage of total microcystin ± SEM. Correlation of the determined MC- and PE-contents were performed using the Pearson’s correlation test. Correlation analyses included all microcystin-positive samples acquired from August 1998 to September 2004 and were performed for HPLC- and ELISA-determinations independently. Statistical significance of the correlation was determined at the p <0.05 level and indicated as *** for p <0.001. MC/PE ratios were analysed for outliers using the Tukey box plot rule.

RESULTS

P. rubescens Abundance During the 261-week observation period, starting in April 1999 and ending September 2004, P. rubescens was always present, albeit at varying cell densities (Fig. 4.2). The distribution patterns observed included phases with a distribution over the entire 40 meter water column investigated, as well as phases with distinct metalimnic layers. Maximum cell densities during winter circulation, whereby P. rubescens was mostly distributed throughout the entire water column, reached 15,000 cells/ml. Maximum P. rubescens cell densities in the metalimnic layer reached 77,000 cells/ml and 45,000 cells/ml as observed in 8 –10 m depth in the end of August 2001 and 2000, respectively. P. rubescens cell densities of ≥55,000 cells/ml occurred during 6 weeks, densities of ≥15,000 cells/ml during 53 weeks, and densities of ≥1500 cells/ml during 123 weeks, corresponding 2%, 22% and 47% of the 261-week observation period, respectively (Tab. 4.2). Integrated P. rubescens biovolumes are summarised in Figure 4.3. Maximum biovolumes were observed at the onset of lake stratification in May 2001, reaching 45 cm3/m2. Yearly onset of lake thermal stratification was recurrent between 20th of April and the 10th of May. Layers of P. rubescens filaments were regularly observed as of the beginning of thermal lake stratification (Fig. 4.2). During stratification, the depth of maximum P. rubescens cell

120 4. FIELD STUDIES ______

[cells/ml]

1000-3000 3000-5000 5000-7000 7000-9000 9000-11000 11000-13000 13000-15000 15000-17000 17000-19000 19000-21000 21000-23000 23000-25000 25000-27000 27000-29000 29000-31000>29000

0 0

5 5

10 10

15 15 depth [m] depth [m] depth 20 20

25 25 no measurements no measurements

30 30

35 35 1999 2000 40 40 JFMAMJJASOND JFMAMJJASOND

0 0

5 5

10 10

15 15 depth [m] depth [m] depth 20 20

25 25

30 31

35 35 2001 2002 40 40 JFMAMJJASOND JFMAMJJASOND

0 0

5 5

10 10

15 15 depth [m] depth [m] depth 20 20

25 25 no measurements 30 30

35 35 2003 2004 40 40 JFM AMJJASOND JFM AMJJASOND

Fig. 4.2: The spatial and seasonal abundance of P. rubescens in the upper 40m of Lake Ammersee from 1999 to 2004.

121 4. FIELD STUDIES ______

Tab. 4.2: Summary of periods displaying P. rubescens cell densities which have experimentally been shown to affect exposed coregonids (Ernst et al. 2007), and the time-weighted, yearly mean values of the Secchi depths (ØZS) determined during the vegetation period i.e. from the beginning of May until end of October for each year. ØZS values are given as mean ± SEM (n ≥15). No Secchi measurements were carried out in 2000

Ø P. rubescens ZS ≥55,000 [cells/ml] ≥15,000 [cells/ml] ≥1500 [cells/ml] [m]

1999° - Jul Apr – Oct 3.3 ±0.19 2000° - Aug – Sep & Oct – Dec* Aug –Dec* 2001 Jul - Aug Jan* & Mar – Oct Jan – Oct* 3.3 ±0.34 2002 - - May – Sep 2.7 ±0.21 2003 - - Aug - Sep 2.9 ±0.26 2004° - Jul Jun – Sep 3.6 ±0.15

* distributed over the entire water column ° measurements were not performed throughout the whole year

densities (peakmax) ranged between 7-13 m depth and significantly correlated with the Secchi depth and the upper limit of the metalimnic layer (Fig. 4.4).

The time-weighted, yearly mean values of the Secchi depths (ØZS) are depicted in Table 4.2.

Although differences in ØZS were not statistically significant, metalimnic conditions differed from May to October as well as from year to year. There were vegetation periods with a high water transparency and sustained euphotic depths reaching the upper limit of the metalimnic layer (i.e. 1999, 2001 & 2004), as shown in Figure 4.5. However, there were also vegetation periods, where the euphotic depths did not or only temporarily reach the upper limit of the metalimnic layer (i.e. 2002 & 2003). With the exception of 1999, the seasonal development of Lake Ammersee thermal stratification was roughly identical throughout the observation period. However, there were marked differences regarding water transparency and respective extent of the euphotic depth (Fig. 4.6).

Seston Sample Phycoerythrin- & Microcystin Contents Phycoerythrin (PE) was detectable in all Lake Ammersee seston samples. The overall mean seston PE-content was 3.4 ±0.7 µg PE/mg dw (n=54). Microcystins (MC) were detectable in 27 and 38 of 54 seston samples via HPLC and Adda-ELISA measurements, respectively. In all of the 27 Lake Ammersee seston samples that were microcystin-positive, as determined via HPLC analyses, the main microcystin congener (79 ±4% of the total microcystin) had a retention time and spectrum consistent with [Asp3]-MC-RR. Seventeen samples contained a microcystin variant with a retention time and spectrum consistent with [Asp3Dhb7]-MC-RR (14 ±1% of the total microcystin). Twenty two samples contained a microcystin variant with a retention time and spectrum consistent with [Asp3]-MC-LR (9 ±2% of the total microcystin), and 18 samples contained a further MC-RR-variant with a retention time which was not comparable to the

122 4. FIELD STUDIES ______employed internal standards, however presenting with a spectrum characteristic for microcystin congeners (8 ±2% of the total microcystin) (Fig. 4.7).

50 Irrespective of the analytical method, the ] 2

- 40 m

3 detected microcystin amounts significantly 30 correlated with the PE-contents 20 1999 10 determined (Fig. 4.8). MC/PE ratios are

0 summarised in Table 4.3. The variation in 0J 3F 6M 9A 12M 15J 18J 21A 24 S 27O 30N 33D 36 50 MC/PE ratios throughout the six-year ] biovol. [cm ] biovol. 2 - 40 observation period was low, with statistical m 3 30 outliers only in May and December 1999 20 2000 (for both HPLC and ELISA analyses), and 10 in June and August 1999 and August 2000 0 0J 3F 6M 9A 12M 15J 18J 21A 24 S 27O 30N 33D 36 (for ELISA analyses only). The mean of the 50 ] biovol. [cm ] biovol.

2 determined seston microcystin contents - 40 m 3 30 was 0.21 ±0.03 µg MC-LRequiv./µg PE (n=27)

20 2001 via HPLC- and 0.43 ±0.06 µg MC- 10 LRequiv./µg PE (n=38) via ELISA 0 determination, corresponding 0.85 ±0.11 µg 0J 3F 6M 9A12 M 15J 18J 21A24 S 27O 30N 33D 36 50

] biovol. [cm ] biovol. MC-LRequiv./mg dw (n=27) and 1.50 ±0.22 2 - 40 m 3 µg MC-LRequiv./mg dw (n=38), respectively. 30

20 2002 10 Oxygen

0 The water column was saturated with 0J 3F 6M 9A 12M 15J 18J 21A24 S 27O 30N 33D 36 50 oxygen as of the beginning of circulation ] biovol. [cm ] biovol. 2 - 40 usually starting at the end of November m 3 30 (± 2-3 weeks) and attaining full circulation 20 2003 by the end of December (Fig. 4.9). With the 10 beginning of lake stratification in May, the 0 0J 3F 6M 9A 12 M 15J 18 J 21 A 24 S 27 O 30 33N 36D epilimnion (corresp. approx. 0 – 8 m depth) 50 ] biovol. [cm ] biovol.

2 was often supersaturated (e.g. 120% - 40 m 3 30 saturation; data not shown), occasionally

20 2004 reaching maximum levels of 150% 10 biovol. [cm saturation (August 2001). Subsequently 0 and simultaneous with prolonged lake 0J 3F 6M 9A 12M 15J 18J 21A24 S 27O 30N 33D 36 month stratification and a sustained shift of the Periods without regular sampling are shaded thermocline into deeper layers, oxygen Fig. 4.3: Integrals of the P. rubescens biovolume measurements suggested decreasing in Lake Ammersee, as determined from the surface to 40m depth. oxygen saturation in the metalimnion (as

123 4. FIELD STUDIES ______

14 A 14 B 12 12

10 10 [m] [m] 8 8 max max 6 6 peak 4 peak 4

2 2 0 0 0 2 4 6 8 10 12 14 16 0 1 2 3 4 5 6 7

Zmeta [m] ZS [m]

Fig. 4.4: Correlation (± 95% confidence interval of the mean) of the depth of maximum P. rubescens densities (peakmax) to (A) the upper metalimnic limit (Zmeta) and (B) the Secchi depth (ZS). Correlation was statistically significant (Pearson’s correlation test) for Zmeta (n = 72, r = 0.40, p <0.001) as well as for ZS (n = 71, r = 0.25, p <0.05).

of June in 2002 and July in 1999, 2001, 2003 and 2004). Metalimnic oxygen values regularly were minimal at the end of September and in October, i.e. 3.7 mg/l in 1999, 4.0 mg/l in 2001, 5.1 mg/l in 2002, and 4.6 mg/l in 2003.

Tab. 4.3: Microcystin contents and potentially toxic cyanobacterial species present in Lake Ammersee seston samples collected from August 1998 until September 2004. Microcystin-amounts were analysed via HPLC and ELISA and expressed as microcystin-LR equivalents (MC-LRequiv.) per µg phycoerythrin (PE). Abundant cyanobacterial species were Planktothrix rubescens (Ptx), Microcystis sp. (Mic) and Anabaena sp. (Ana). Cyanobacteria marked in bold represented a main component of the phytoplankton community

[µg MCequiv./ µg PE] abundant [µg MCequiv./ µg PE] abundant

species HPLC ELISA HPLC ELISA species

Aug 0.2 0.2 Ptx Jan * * Ptx

1998 Feb n.d. n.d. Ptx Nov 0.2 0.7 Ptx Mar n.d. n.d. Ptx Apr 0.1 0.2 Ptx Apr n.d. n.d. Ptx May 0.8 1.2 Ptx May n.d. 0.3 Ptx Jun 0.3 1.2 Ptx Jun 0.1 0.3 Ptx Jul 0.3 0.5 Ptx Jul 0.1 0.2 Ptx 2002 Aug 0.3 1.1 Ptx Aug n.d. 0.3 Ptx , Mic, Ana

1999 Sep 0.1 0.4 Ptx Sep n.d. n.d. Ptx , Mic, Ana Oct 0.3 0.7 Ptx Oct n.d. n.d. Ptx , Mic, Ana Nov 0.2 0.6 Ptx Nov n.d. n.d. Ptx Dec 0.7 1.9 Ptx Dec n.d. n.d. Ptx Aug 0.4 1.1 Ptx Jan n.d. n.d. Ptx

Oct 0.2 0.5 Ptx Feb n.d. n.d. Ptx Mar n.d. n.d. Ptx 2000 Nov 0.3 0.5 Ptx Dec 0.1 0.2 Ptx Apr n.d. n.d. Ptx May n.d. 0.2 Ptx Jan 0.1 0.3 Ptx Jun n.d. n.d. Ptx Feb 0.2 0.4 Ptx Jul n.d. n.d. Ptx Mar 0.1 0.1 Ptx 2003 Aug n.d. 0.1 Ptx , Mic, Ana Apr 0.1 0.3 Ptx Sep n.d. 0.1 Ptx , Mic, Ana May 0.1 0.2 Ptx Oct n.d. 0.1 Ptx , Mic, Ana Jun 0.1 0.3 Ptx Nov n.d. n.d. Ptx Jul 0.1 0.4 Ptx

2001 2001 Dec n.d. 0.2 Ptx Aug 0.1 0.2 Ptx Jan n.d. 0.1 Ptx Sep 0.1 0.3 Ptx Oct n.d. 0.4 Ptx Mar n.d. n.d. Ptx Nov * * Ptx Apr n.d. n.d. Ptx Dec * * Ptx May n.d.* * Ptx

2004 2004 Jul n.d.0.1 0.3 Ptx n.d. = not detectable (< quant. limit) Aug n.d. 0.1 Ptx , Mic, Ana

* deficient sample volumes Sep n.d. 0.1 Ptx , Mic, Ana 124 4. FIELD STUDIES ______

May Jun Jul Aug Sep Oct 0 DISCUSSION 2 4

6 r2=0.82 P. rubescens Distribution & Succession: 8 R2 = 0,8241 10 Comparable to P. rubescens occurrences 12 observed in deep pre-alpine lakes in 14 1999 2 16 rR=0.592 = 0,5916 Italy (Buzzi, 2002; Legnani et al., 2005; Morabito et al., 2002; Salmaso, 2000), 0 2 Switzerland (Walsby et al., 1998) and 4 France (Jacquet et al., 2005) the 6 8 P. rubescens assemblage in Lake 10 r2=0.73 Ammersee appears to proceed in a 12 R2 = 0,7331 14 2001 uniform annual pattern. As 16 2 demonstrated by Walsby et al. (1998), Rr2 =0.89= 0,8852 P. rubescens cells can remain viable 0 2 during winter mixing down to 80 m 4 6 depth. Considering the maximum depth r2=0.37 8 of 82 m, the P. rubescens population in 10 R2 = 0,3714 Lake Ammersee thus obviously can 12 14 2002 endure winter circulation. This is 16 2 Rr2=0.82 = 0,8154 confirmed by the constantly high 0 integrated P. rubescens biovolume 2 during mixing in winter 2000/01. The 4 6 winter endurance of the P. rubescens 8 population may provide P. rubescens 10 R2r =2=0.14 0 , 14 17 12 with an early competitive advantage in 14 2003 exploiting the resources available at the 16 rR2=0.662 = 0,659 beginning of the vegetation period 0 (Walsby et al. 1998, Legnani et al. 2005). 2 4 Despite this, a regular short-term 6 decline in P. rubescens abundance is 8 10

observed during early spring in Lake [m] depth [m] depth [m] depth [m] depth [m] depth 12 Ammersee, as is also observed in other 14 2004* 16 lakes (Bettinetti et al., 2000; Salmaso, * Trends were not estimated due to deficient sample intervals 2000; Walsby et al., 1998; Walsby & Schanz, 2002). The latter is assumed to Fig. 4.5: Spatial and seasonal development of the euphotic depth Zeu (black) and the upper stem from the change in hydrostatic metalimnic limit Zmeta (grey) in Lake Ammersee, conditions. As shown by Walsby et al. (depicted as polynomial trends), from the beginning of May until end of October (periods of (1998), the beginning of lake vegetation), in 1999, 2001, 2002, 2003 and 2004. 125 4. FIELD STUDIES ______

May Jun Jul Aug Sep Oct May Jun Jul Aug Sep Oct

0 A 0 B 2 2 4 4 6 6 8 8 2004 * 2003 2004 * 10 2002 10

depth [m] [m] depth 12 [m] depth 12 1999 1999 14 14 2001 2002 16 16 2003 2001 2000 *

* Trends were not estimated due to deficient sample intervals

Fig. 4.6: Comparison of the spatial and seasonal development of the euphotic depth Zeu (A) and the upper metalimnic limit Zmeta (B) in Lake Ammersee, (depicted as polynomial trends), during the periods of vegetation (beginning of May until end of October) in 1999, 2001, 2002, 2003 and 2004.

stratification provides a natural selection for P. rubescens cells with strong gas vesicles, enabling filaments to remain buoyant after winter mixing and consequently remain able to stratify in the water column. Filaments that float to the surface mixed layer or remain within the hypolimnion disappear at the beginning of lake stratification, most likely due to high UV irradiance and nutritional competition in the surface layer and cold temperature as well as light limitation in the hypolimnion (Mur et al., 1999; Walsby & Schanz, 2002).

9.47 min / Bgnd (DAD-248 nm) 9.97 min / Bgnd (DAD-248 nm) A B

220 240 260 220 240 260

15.89 min / Bgnd (DAD-248nm nm) 16.54 min / Bgnd (DAD-248nm nm) C D

220 240 260 220 240 260 A nm nm B CD

098 1010 1211 14 12 1613 1814 2015 Minutes

Fig. 4.7: HPLC-chromatogram of a Lake Ammersee seston sample (November 2000) with four microcystin variants and respective retention times and spectra consistent with [Asp3]-MC-RR (A), [Asp3Dhb7]-MC-RR (B) and [Asp3]-MC-LR (D). The fourth spectrum (C) being comparable to those characterised for microcystin congeners, is proposed to represent another MC-RR variant.

126 4. FIELD STUDIES ______

A B

Fig. 4.8: Correlation (± 95% confidence interval of the mean) of the microcystin (MC)-content with the phycoerythrin (PE)-content in Lake Ammersee seston samples. Microcystin analyses were performed using (A) HPLC and (B) ELISA analyses. Correlation was statistically significant (Person’s correlation test) for both HPLC (n = 27, r = 0.70, p <0.001) and ELISA analyses (n = 38, r = 0.74, p <0.001).

Maximum P. rubescens cell densities are observed in Lake Ammersee, as in other lakes, within compact stratified layers as of late spring (Walsby & Schanz 2002, Jacquet et al. 2005, Blikstad-

0 0 0 1999 2000 2001

5 5 5

10 10 10

15 [m] depth 15 15

20 20 20

25 25 25 no measurements 30 30 30

35 35 35

40 40 40 JASONDJ A S O N D JASOND J A SONDJASOND JA SOND 0 0 0 0 2002 2003 2004

5 5 5 5 [% saturation]

10 10 10 10 100-110>100 90-100 15 15 15 15 depth [m] [m] depth 80-90 20 20 20 20 70-80 60-70 25 25 25 25 no measurements no measurements 50-60 30 30 30 30 40-50 30-40 35 35 35 35

40 40 40 40 JASONDJ A S O N D JASOND J A SONDJASONDJASOND JA SOND month month month

Fig. 4.9: Oxygen conditions in the upper 40m of Lake Ammersee from July until December (1999 to 2004). 127 4. FIELD STUDIES ______

Halstvedt et al. 2007). As indicated by the correlation of peakmax and Zmeta

(Zmeta = 0.45 x peakmax + 3.3), the P. rubescens peaks - regularly located at 7-13 m depth - are situated approximately 0.5 to 4 m below the upper limit of the metalimnion, thus closely related to the strong thermal gradient stabilising the stratified water column. Moreover, the depth of P. rubescens stratification broadly corresponds (for peaks in 7-13 m) with the triple of the Secchi depths (peakmax ≈ 3 x ZS), as demonstrated by the correlation of peakmax and

ZS (ZS = 0.11 x peakmax + 2.1). Thus the depth of P. rubescens stratification is slightly below the standard estimation for the euphotic depth (Zeu = 2.5 ZS, Lemmin 1978), albeit in accordance with the approximation given by Dokulil & Teubner (2000) and euphotic depths recently estimated in other meso- and oligotrophic lakes (Salmaso 2000, Buzzi 2002, Morabito et al. 2002). The light intensity at the observed P. rubescens stratification depth probably corresponds to ≥1% of the surface irradiance, which is suggested to be sufficient for the low light requirements of P. rubescens (Mur et al. 1999). The latter data thus confirm that the border between epilimnion and metalimnion represents an ecological niche that prevents P. rubescens filaments from mixing within the surface layer while simultaneously providing sufficient light for P. rubescens growth (Micheletti et al., 1998; Morabito et al., 2002; Walsby et al., 1998). The decrease of the thermal gradient - due to cooling of the surface following lower daytime temperatures and convective cooling during night in autumn – provides for an increased mixing of the near-surface layers, which may include near metalimnic layers (Blikstad-Halstvedt et al., 2007; Buzzi, 2002; Salmaso, 2000; Teubner et al., 2004; Walsby et al., 1998). Consequently, P. rubescens filaments may be mixed within the surface layer, where they can accumulate and provide for the surface blooms occasionally observed in stratified lakes (Walsby et al. 1998, Walsby & Schanz 2002, Blikstad- Halstvedt et al. 2007).

P. rubescens Cell Densities and Biovolumes P. rubescens occurred in Lake Ammersee continuously, throughout the observation period of six years, as shown via P. rubescens cell counts and presence of detectable phycoerythrin. Cell densities and integrated biovolumes were in the range of those previously observed in other pre- alpine lakes, irrespective of whether they were determined during winter circulation or summer stratification (Hoeger et al., 2005; Jacquet et al., 2005; Walsby & Schanz, 2002). Spatial differences in P. rubescens abundance within Lake Ammersee appear to be minimal and to occur only temporarily as suggested by sporadic sampling at a second sampling point in the north of the lake (Ernst et al., unpublished data). In contrast, regarding both maximum cell densities and the integrated P. rubescens biovolume, considerable differences in P. rubescens abundances were observed between different years. Based on the cell densities and integrated biovolumes, the highest P. rubescens abundance was observed in 2001 followed by 2000 > 1999 > 2004 > 2002, 2003. This raises the question which factors influence these apparent differences in annual P. rubescens abundance.

128 4. FIELD STUDIES ______

Obviously, P. rubescens growth in deep pre-alpine lakes is likely to be affected by multiple factors and/or processes (Dokulil & Teubner 2000, Walsby & Schanz 2002, Jacquet et al. 2005, Blikstad- Halstvedt et al. 2007, etc.) amongst which are nutritional conditions. As nutrient concentrations were not part of the current investigation, the discussion on the influence of the nutritional conditions remains largely hypothetical. However, P. rubescens dominates the phytoplankton biomass in stratified, re-oligotrophicated lakes frequently at total phosphorous-concentrations of approximately 10 µg/l (Dokulil & Teubner 2000, Salmaso 2000, Morabito et al. 2002). This applies to Lake Ammersee, where the mean annual total phosphorous concentration never exceeded 15 µg/l during the past decade (Teubner et al. 2004). In re-oligotrophicated lakes with <15 µg total phosphorous-concentration per litre, phosphate depletion strongly influences seasonal phytoplankton succession in the epilimnion in the course of a year (Anneville et al., 2002). Indeed, Jacquet et al. (2005) observed P. rubescens mass occurrence to arise when phosphate had severely been depleted in Lake Bourget, France. This appears possible due to the ability of P. rubescens to grow at the border of the euphotic zone allowing it to benefit from the nutrient rich metalimnion (Anneville et al. 2002, Walsby & Schanz 2002). Moreover, P. rubescens, being capable of excreting alkaline phosphatases into the ambient water, has the physiological advantage of utilising organic phosphorous when inorganic phosphates are growth limiting (Feuillade 1994). Indeed growth limitation for phytoplankton species other than P. rubescens will also reduce the degree of shading in the surface mixed layer, therefore providing for sufficient illumination in deeper zones, as discussed below (Bürgi & Stadelmann, 2002; Legnani et al., 2005; Walsby & Schanz, 2002). Blikstad-Halstvedt et al. (2007) observed that the nutritional conditions determined in Lake Steinsfjorden, Norway, were often below the minimum N and P levels which allowed maximum P. rubescens growth rate in culture studies. This applied to both, the epi- and metalimnion, particularly with regard to nitrogen levels. Thus, P. rubescens populations may thrive under high nitrogen concentrations. This interpretation is supported by the observation that P. rubescens mass occurrence primarily arises in lakes with low phosphate and high nitrogen loads as observed in many oligo- and mesotrophic pre-alpine lakes (Jacquet et al., 2005; Salmaso, 2000; Teubner et al., 2004; Zotina et al., 2003), including Lake Ammersee, where mean nitrogen concentrations regularly exceed 1 mg/l (Nixdorf et al., 2004). Another important factor influencing P. rubescens growth appears to concern the temperature regime (Blikstad-Halstvedt et al. 2007). Jacquet et al. (2005) suggested that a possible reason for the increasing success of P. rubescens may be an earlier onset of lake stratification and thus longer water column stratification in lakes due to global warming. However, it is unlikely that the annual variation in P. rubescens abundance observed in Lake Ammersee was caused by differences in the thermal stratification, especially as the onset of lake thermal stratification was comparable throughout the investigation period. In addition, the latitude of the mixed surface layer and the temporal development of the thermal stratification were roughly identical throughout the study (Zmeta exceeded the long-standing average only in 1999, which can without doubt be ascribed to an extreme flood, which raised the water level by approximately 2 m;

129 4. FIELD STUDIES ______www.hnd.bayern.de). However, the suggestion by Jacquet et al. (2005) cannot be totally dismissed as decisively longer observation periods would be required to definitively assign global warming induced small changes in onset of lake temperature stratification to P. rubescens abundance.

The observation that time-weighted yearly mean values of the Secchi depths (ØZS) were >3 m in 1999, 2001 and 2004 and contrastingly <3 m in 2002 and 2003 however, suggests that P. rubescens growth might primarily be related to differences in water transparency and therefore to light conditions. Indeed, the comparison of the annual P. rubescens abundances with the respective seasonal trend of the upper metalimnic limit (Zmeta) and the latitude of the euphotic depth demonstrated that P. rubescens abundance was high, when the vegetation periods were characterised by high water transparency and sustained euphotic depths reaching the upper limit of the metalimnic layer (i.e. 1999, 2001 & 2004). In contrast, P. rubescens abundance was low, when the euphotic depths did not or only temporarily reached the upper limit of the metalimnic layer (i.e. 2002 & 2003). Moreover, the development of the thermal stratification was comparable throughout the observation period, further corroborating the hypothesis that P. rubescens abundance in Lake Ammersee is primarily affected by water transparency and Zeu/Zmix-ratios ≥1

(whereby Zmix is the latitude of the mixed surface layer (Zmix = Zmeta - 1). Indeed, Mur & Schreurs

(1995) noted that stratifying Planktothrix species usually grow in water columns with Zeu/Zmix ratios close to 0.7 - 1.2. An increased average Zeu/Zmix ratio to 0.9 – 1 is also discussed as a key to the success of P. rubescens in Lake Bourget, France (Jacquet et al. 2005). The importance of light conditions for P. rubescens growth is also supported by findings of Dokulil & Teubner (2000), showing that dominance of stratifying P. rubescens in Lake Mondsee, Austria, is associated with light climate correlatives (Zmix/ZS) of approximately 4 (Dokulil & Teubner 2000), which corresponds Zeu/Zmix ratios of ≥0.7, largely depending on the Zeu-approximation used.

P. rubescens Microcystin Contents The average microcystin content in microcystin-positive seston samples in this study was

0.85 ±0.11 µg MC-LRequiv./mg dw via HPLC and 1.50 ±0.22 µg MC-LRequiv./mg dw via ELISA determination. This is within the range of microcystin contents previously determined in P. rubescens extracts and seston sample extracts of P. rubescens dominated lakes (Fastner et al., 1999a; Jann-Para et al., 2004; Kurmayer et al., 2005). The results obtained in Lake Ammersee thus confirm that Planktothrix species contain comparably high amounts of microcystin (Akcaalan et al., 2006; Fastner et al., 1999a). The HPLC-analyses demonstrated that P. rubescens in Lake Ammersee primarily produce one main microcystin congener i.e. [Asp3]-MC-RR, far lower concentrations of two additional microcystin congeners, i.e. [Asp3, Dhb7]-MC-RR and [Asp3]-MC-LR, and one putative uncharacterised microcystin variant. This observation concurs with previous studies, describing demethylated variants of MC-RR to be the predominant microcystin congeners of P. rubescens (Blom et al., 2001; Fastner et al., 1999a; Kurmayer et al., 2005; Luukkainen et al., 1993), accompanied by a varying number of characterised microcystin variants such as [Asp3]-MC-LR,

130 4. FIELD STUDIES ______

[Asp3]-MC-HtyR and [Asp3]-MC-YR, and as yet uncharacterised congeners (Fastner et al., 1999a; Grach-Pogrebinsky et al., 2003; Kurmayer et al., 2005). It is thus not surprising that differences were observed between microcystin determination via ELISA and HPLC. Indeed, HPLC analysis of microcystins is entirely dependent on the availability of standards, whereas the Adda-ELISA recognises nearly all microcystin congeners with comparable sensitivity. Nevertheless, microcystins were detectable in 50% and 70% of the investigated seston samples via HPLC and Adda-ELISA measurements, respectively. The results also demonstrated frequent presence and high microcystin concentrations in samples taken during periods of high P. rubescens abundance, whereas samples taken during periods of low P. rubescens abundance were less frequently microcystin-positive and, when microcystin was present, the quantities detected were low. The latter data therefore suggest that the presence and quantity of microcystin detected in Lake Ammersee seston is strongly if not exclusively correlated with the abundance of P. rubescens and moreover that a nearly constant MC : P. rubescens biomass ratio predominates in Lake Ammersee. Consequently, variations in microcystin content per unit biomass of seston samples appear strongly dependent on the species composition of the seston sample itself (P. rubescens representation within the sample) rather than a result of variations in P. rubescens microcystin content. P. rubescens phycoerythrin content has been shown to be relatively constant; especially as P. rubescens shows no photopigment adaptation subsequent to environmental alterations (Skulberg & Skulberg, 1985). Moreover, high intracellular phycoerythrin concentrations appear to be specifically a Planktothrix inherent trait (Ernst et al. 2006b). Therefore, the ratio of MC : phycoerythrin (PE) would appear a better indicator for detection of P. rubescens microcystin content than seston biomass in seston samples. Mez (1998) determined PE-concentrations of up to 4 µg/l during a survey on the abundance of toxic cyanobacteria in Swiss lakes, which would correspond to approximately 4 µg/mg dw seston. This compares well with the overall mean PE- content of 3.4 ±0.7 µg PE/mg dw detected in the Lake Ammersee seston samples. The average MC/PE ratio as analysed in the Lake Ammersee seston samples was 0.21 ±0.03 µg

MC-LRequiv./µg PE and 0.43 ±0.06 µg MC-LRequiv./µg PE when analysed via HPLC and Adda- ELISA, respectively. MC/PE ratios varied only little, suggesting that the variation in the P. rubescens microcystin content was limited. This concurs with observations in other German freshwaters where temporal variability of microcystin variants as well as the variability of relative microcystin amounts within persisting P. rubescens blooms was rather low (Fastner et al. 1999a).

Oxygen Concurrent to prolonged lake stratification and a sustained shift of the thermocline into deeper layers, the oxygen measurements suggested decreasing oxygen-saturation in the metalimnion regularly leading to minimal metalimnic oxygen values in September and October. The observed decrease in available oxygen is assumed to result from a shift of the metalimnion below the

131 4. FIELD STUDIES ______euphotic depth, limiting metalimnic photosynthetic activity and thus metalimnic oxygen release, while oxygen consumption due to decomposition of organic matter remained constant (Kucklentz et al. 2001). In this study, a ranking of the metalimnic oxygen minima according to the lowest values of individual years yielded the following result: 1999 < 2001 < 2003 < 2002. This ranking compared very well with the ranking obtained with ordering yearly P. rubescens abundance from highest to lowest values. The results thus demonstrated that metalimnic oxygen depletion is more pronounced in years of high P. rubescens abundance while lower metalimnic depletion could be observable in years with lower P. rubescens abundance. Repeated observable metalimnic oxygen depletions therefore appear to be causally related to recurring high P. rubescens abundance. This apparent coherence of observations may be assumed to result from oxygen consuming decomposition of senescent P. rubescens cells in the metalimnion e.g. a decline of oxygen saturation was observed subsequent to a decrease in P. rubescens abundance coincident with increasing microcystin concentrations in the metalimnic water of Lake Ammersee in August 1999 (Ernst et al., 2001). It may also be caused by a shift of the stratified P. rubescens population into deeper layers below the photosynthetic compensations point thus enforcing a change from oxygen producing photosynthesis to respiration of oxygen (Walsby et al., 2001). Similarly, marked metalimnic oxygen depletions have been observed in several other P. rubescens-containing lakes (Lindholm & Meriluoto 1991, Salmaso 2000, Buzzi 2002, Krupa & Czernas 2003), lending more weight to above interpretation.

Possible Consequences Microcystin-containing cyanobacteria have been demonstrated to detrimentally affect various aquatic organisms (Sivonen & Jones 1999, Wiegand & Pflugmacher 2005) including fish (Carbis et al., 1996a; Fischer & Dietrich, 2000; Tencalla & Dietrich, 1997). This is very interesting as in many pre-alpine P. rubescens-containing lakes, coregonids15 have recently suffered massively reduced growth and changed population demographics resulting in recurrent slumps in yearly yields in coregonid fishery (Gammeter & Forster, 2002; Kirchhofer, 2004; Müller, 2003; Wißmath, 2000). Laboratory experiments clearly demonstrated adverse effects, including dose- and time- dependent physiological stress and progressive organ damage with subsequent elevated susceptibility to ectoparasitic infestations and increased mortality, following subchronic exposure of coregonids to P. rubescens (Ernst et al. 2006a, Ernst et al. 2007). This has been shown to apply to P. rubescens densities greater than 1500 cells/ml, which occurred in Lake Ammersee during 47% of the 261-week observation period. The results of this study thus suggest that Lake Ammersee coregonids are indeed regularly confronted with potentially adverse P. rubescens densities. As the microcystin content of P. rubescens appeared to be constant, detrimental effects

15 Coregonids (Coregonus sp.) are in most pre-alpine lakes among the dominant species of the ichthyofauna, and, due to their high commercial value, of great importance for the professional fishery 132 4. FIELD STUDIES ______on the coregonid population resulting from exposure to prolonged P. rubescens abundance thus cannot be excluded. This pertains specifically to periods with P. rubescens filaments distributed over the entire water column - as for example continuously observable from December 2000 to May 2001 – and thus making avoidance of P. rubescens impossible. As in addition, pronounced metalimnic oxygen depletion has regularly been observed to result in oxygen concentrations at the limit of the coregonid tolerance (Kischnik, 1992; Müller & Stadelmann, 2004), pronounced oxygen depletion may provide a further stress factor in the metalimnic layer, in addition to stratified microcystin-containing P. rubescens. Consequently, it is plausible that the observed challenge of coregonids in P. rubescens containing lakes is a result of adverse environmental conditions that are causally linked to the appearance of microcystin-containing P. rubescens.

Conclusion In summary P. rubescens mass occurrence appears to be strongly influenced by water transparency, i.e. illumination in the metalimnion. In addition, the current data allow the assumption that P. rubescens abundance in the re-oligotrophicated Lake Ammersee result from regular phosphate depletion in the epilimnion and possibly benefits additionally from high nitrogen loads. The results as presented thus confirm that the increasing P. rubescens mass occurrence in pre-alpine lakes is a paradoxical outcome of lake restoration efforts (Morabito et al. 2002, Jacquet et al. 2005). In addition, the variation in the MC/PE-ratios was low suggesting that the microcystin production of P. rubescens in Lake Ammersee is consistent thus indicating that the appearance of P. rubescens coincides with measurable microcystin levels. Aquatic organisms such as coregonids hence appear to be regularly confronted with potentially adverse P. rubescens densities which might provide a possible explanation for the often observed challenge to coregonid populations in P. rubescens containing pre-alpine lakes.

ACKNOWLEDGEMENTS We would like to thank Prof. Dr. K.O. Rothhaupt (University of Konstanz, Germany) for his help in the discussion of limnological issues. The financial support and unflagging interest of the Arthur and Aenne Feindt Foundation (Germany) is highly appreciated.

133 4. FIELD STUDIES ______

4.2. THE ADVERSE EFFECTS OF PLANKTOTHRIX RUBESCENS ON COREGONIDS (COREGONUS LAVARETUS) IN LAKE AMMERSEE – ADDITIONAL FIELD OBSERVATIONS –

Parts of this chapter have previously been published in Ernst et al. (2001): Presence of Planktothrix sp. and cyanobacterial toxins in lake Ammersee, Germany and their impact on whitefish (Coregonus lavaretus L.). Environmental Toxicology 16, 483-488

ABSTRACT

Experimental investigations into the ichthyotoxicity of microcystin-containing Planktothrix rubescens suggest that coregonid exposure to P. rubescens can cause physiological stress and pathological damage, substantially affecting coregonid fitness and elevating mortality rates. Adverse effects have been shown for P. rubescens densities known to occur in pre-alpine lakes, thus providing a possible explanation for recurrent slumps in coregonid yields observable in several of those lakes, e.g. in Lake Ammersee. Hence, this study aimed to elucidate whether there is evidence for P. rubescens exposure of coregonids in Lake Ammersee. Feral coregonids were obtained between 2001 and 2004 from monthly catches carried out with gill nets of various mesh sizes. The sample cohort was completed by samples from catches of local fishermen, including samples taken during bloom episodes in 1998 and 1999. Fish were investigated microscopically for accumulation of ingested P. rubescens filaments within the intestine. Gut contents were analysed for cyanobacterial biliprotein and gut contents as well as liver homogenates were analysed for microcystin accumulation. P. rubescens filaments were observable in the coregonids gut contents giving evidence for P. rubescens exposure of coregonids in Lake Ammersee. The results demonstrate this exposure to cause an accumulation of P. rubescens components within the coregonids intestine, as 8% of the investigated coregonids showed prominent blue colouration of gut contents probably resulting from a significant accumulation of cyanobacterial biliproteins. From the coregonids sampled during bloom episodes in August 1998 and April 1999, four out of ten contained significant microcystin accumulations, which unambiguously demonstrate microcystin exposure of feral coregonids in Lake Ammersee. The detection of covalently-bound microcystin in liver tissue furthermore demonstrates microcystins to traverse the ileal membrane and to accumulate in the liver. This makes substantial detrimental effects on the coregonids appear inevitable and thus substantiates the initially proposed suggestion of a causal relationship between P. rubescens mass occurrence and challenged coregonid populations in pre-alpine lakes such as Lake Ammersee.

KEYWORDS: Cyanobacteria; Planktothrix; Microcystin; Coregonids; Fish

134 4. FIELD STUDIES ______

INTRODUCTION Occurrences of Planktothrix rubescens blooms in pre-alpine lakes have been observed to coincide with recurrent slumps in coregonid yields causally associated with reduced fish weight and fitness (Braun, 1953; Ernst et al., 2001). The current knowledge on the ichthyotoxicity of cyanobacteria in general (summarised in Malbrouck & Kestemont, 2006) and recently published information specifically on the toxicity of P. rubescens on coregonids (Ernst et al., 2007; Ernst et al., 2006a), suggest that coregonid exposure to microcystin-containing P. rubescens can cause enhanced physiological stress as well as continuous organ damage. Elevated susceptibility to ectoparasitic infestations and increased mortality in coregonids experimentally exposed to P. rubescens filaments additionally corroborate effects on the coregonids fitness (Ernst et al., 2007). Hence, a causal relationship between the occurrence of P. rubescens containing microcystin and changes in growth and population dynamics of coregonids appears likely and provides a possible explanation for recurrent slumps in coregonid yields in pre-alpine lakes, such as in Lake Ammersee. Compared to other cyanobacteria, Planktothrix sp. have been shown to contain very high amounts of microcystin per gram dry weight, primarily consisting of various demethylated variants of MC-RR (Blom et al., 2001; Ernst et al., submitted; Fastner et al., 1999a; Kurmayer et al., 2005). This has been shown to apply also to Lake Ammersee (Ernst et al., submitted), where more than 50% of monthly surveyed seston samples contained microcystin, the concentration of which correlated with the abundance of P. rubescens, thus indicating that in Lake Ammersee the appearance of P. rubescens coincides with measurable microcystin concentrations. This was confirmed by the observation of a concurrence between microcystin and biliproteins specific to cyanobacteria as demonstrated by a correlation of microcystin amounts and the seston sample phycoerythrin content (Ernst et al., submitted). In Lake Ammersee, P. rubescens has been shown to persist over several years (Ernst et al., submitted), whereby P. rubescens filaments were evenly distributed over the entire water column during winter and stratified in distinct metalimnic layers during summer. Adverse effects on coregonids have been shown experimentally for P. rubescens cell densities greater than 1500 cells/ml (Ernst et al., 2007). Such P. rubescens cell densities were observed in Lake Ammersee during 47% of the total observation period of 261 weeks (Ernst et al., submitted), including periods with filaments distributed over the entire water column. This consequently implies that exposure levels employed in the laboratory study occur naturally in Lake Ammersee. The aim of this study thus was to determine whether there is evidence for P. rubescens exposure of feral coregonids in Lake Ammersee, via examination of the uptake of toxic P. rubescens and microcystin accumulations in wild catch coregonids.

135 4. FIELD STUDIES ______

MATERIAL & METHODS

Investigation of the Gut Content of Lake Ammersee Coregonids Gut contents of Lake Ammersee coregonids obtained from random samples from catches of the local fishery cooperative were examined microscopically for P. rubescens filament accumulation. From June 2001 to December 2004, Lake Ammersee coregonids were further regularly examined for elevated biliprotein (i.e. phycocyanin and allophycocyanin) concentrations in gut content. Required coregonids were obtained from monthly catches carried out (in cooperation with the fisheries advisory board of Upper-Bavaria, Germany) with gill nets in the pelagic zone of the lake. Fish were caught in nets with mesh sizes of 20, 25, 30, 35, 40 and 45 mm (length: 100 m per mesh size) and thus representative of the coregonid population structure in the lake (with the exception of yearlings which are normally caught in mesh sizes <20 mm). Previous coregonid gut content samples, taken during bloom episode in August 1998 and April 1999, were included in the sample cohort. At least six individuals per month (maximum 14 fish) were assessed, except for December 2002 (4 fish), April 1999 (2 fish), March 2003 (3 fish), March (5 fish) and October 2004 (4 fish), as at these time points the gut contents of the majority of coregonids were insufficient to allow accurate assessment. In December 2001, January 2003, February 2003, January 2004 and February 2004 guts of the coregonids were totally empty and thus no assessment could be carried out. Coregonids were dissected and gut contents were removed from the intestine, dried via speed vac evaporation and stored at –20 °C and darkness until extraction and biliprotein analysis. Biliprotein concentrations in the lyophilised gut content samples were determined via extraction of defined sample quantities (≤250 mg dw) in phosphate buffered saline (≤30 ml/mg dw) by two freeze-thaw cycles using liquid nitrogen. Each extract was centrifuged (15 min, 16,000 x g) and the absorption (A) of the resulting supernatants was determined at wavelengths of 615 nm and 652 nm. Absorption was additionally determined at 750 nm for nullification (N). The optical density for the respective wavelength (ODxxx) was calculated as ODxxx = A – N. Phycocyanin (PC) and allophycocyanin (APC) concentrations were calculated according to the description of Tandeau de Marsac (1977) using the following equations:

PC [mg/l] = (OD615 – 0.747 x OD652) / 5.34

APC [mg/l] = (OD652 – 0.208 x OD615) / 5.09 The analytical protocol provided for a quantification limit of ≥0.9 µg/mg dw and ≥0.75 µg/mg dw for phycocyanin and allophycocyanin, respectively. The procedure was carried out once for each gut content sample.

The gut content samples, taken in August 1998 and April 1999, were additionally analysed for microcystin contamination. Gut content extracts were obtained by sonication in 100% methanol (60 min) and subsequent centrifugation (68,000 x g, 60 min). Methanol was removed via

136 4. FIELD STUDIES ______speed-vac evaporation and the resulting extract re-dissolved in water for further purification by solid phase extraction (see Ernst et al. 2005). The resulting eluents were finally dissolved in a defined volume of water. Microcystins were quantified via anti-Adda ELISA in comparison with internal MC-LR standards (Alexis, Switzerland) and given as the microcystin-LR equivalent (MC-

LRequiv.) concentration. The MC-LRequiv. concentration in Lake Ammersee coregonid gut contents were compared using a one-way ANOVA and Tukey’s multiple comparison test. Gut content samples were classified to be microcystin-positive when significant elevations were determined at the p <0.05 level.

Determination of Microcystin-Adducts in Liver Homogenates Lake Ammersee coregonids were additionally investigated for microcystin accumulation in liver via random examination of fish liver homogenates. Fish, caught in August and November 1998, as well as April and August 1999, were dissected and liver tissue was homogenised in extraction buffer containing 1 mM PMSF, 5 mM EDTA, 1 mM DTT, 140 mM NaCl, 1% Triton X-100 and 10 mM Tris (pH 7.5). For qualitative detection of covalently-bound microcystin adducts, liver homogenates were separated via 10% SDS PAGE in accordance with Laemmli (1970). Separated proteins were transferred onto a nitrocellulose membrane via Western blot technique. The membranes were blocked using TTBS + 1% BSA for 30 min and MC-LR adducts were detected via incubation with polyclonal sheep anti-Adda serum (diluted 1:1000 in blocking buffer; see also Fischer et al., 2001) at room temperature for one hour according to Fischer & Dietrich (2000). Membranes were washed using TTBS (3x5 min) and incubated with secondary antibody (anti sheep IgG-AP, Sigma-Aldrich, Germany, diluted 1:5000 in TTBS) at room temperature for one hour. After washing with TTBS (3x5 min) and TBS (1x15 min), specific bands were visualised using AEC chromogen (BioGenex, USA) according to the manufacturer’s instructions. The molecular weights of detected adducts were estimated by comparison with full range rainbow marker proteins RPN 800 (Amersham, UK).

RESULTS P. rubescens filaments were observable in blue coloured gut content of Lake Ammersee coregonids (Fig 4.10). Coregonids showing prominent blue colouration of gut contents contained biliprotein (i.e. phycocyanin and allophycocyanin) concentrations above the respective limits of quantification. Altogether, 8% of the analysed gut content samples (n=289) contained significant amounts of biliproteins. The 95% confidence interval of the mean phycocyanin (PC) and allophycocyanin (APC) concentrations determined in biliprotein-positive gut contents were 1.5-2.6 µg/mg and 1.2-2.2 µg/mg for PC and APC, respectively. The highest biliprotein concentrations detected were 5.4 µg PC/mg and 5.7 µg APC/mg in a single coregonid caught in May 2002. Biliproteins were observed in gut content samples in all years investigated, whereby gut content samples never contained biliproteins during November, January and February, but

137 4. FIELD STUDIES ______

a b

2 cm 200 µm

c d

P P

50 µm 25 µm

Fig. 4.10: Lake Ammersee coregonid (a) and copepod from a Lake Ammersee seston sample (b) containing conspicuous blue coloured gut content (inserts). Microscopical examination of gut contents (c & d) further demonstrates ingestion of P. rubescens filaments (P) by Lake Ammersee coregonids and indicates biliprotein release during filament decomposition within the coregonids intestine (d).

regularly and with highest frequency in April and May (Tab. 4.4). A sporadic microscopical investigation of seston samples demonstrated that in Lake Ammersee also copepods may exhibit conspicuous blue gut colouration (Fig. 4.10). From the coregonids sampled during the bloom episodes in August 1998 and April 1999, four out of ten samples contained significantly elevated microcystin levels corresponding to a mean concentration of 30 ±3 µg MC-LRequiv./g dw (Fig. 4.11).

Tab. 4.4: Monthly catches showing coregonids in Lake Ammersee with (+) and without (–) biliprotein- positive gut contents. Sampling was carried out from June 2001 to December 2004. Previous coregonid gut content samples, taken during bloom episodes in August 1998 and April 1999, were included in the sample cohort

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec

-1999 +1998 2001 + - + + - - - 2002 - - + + + ------+ 2003 - - + + + - - - - + - - 2004 - - - - + + + + + + - + 138 4. FIELD STUDIES ______

0,0550 * * * * 0,0440 /g dw /g 0,03

equiv. 30

0,0220

µg MC / mg

0,01µg MC-LR 10

0,000 //

Apr 99 - 1 Apr 99 - 2 Aug 98 - 1 - 98 Aug 2 - 98 Aug 3 - 98 Aug 4 - 98 Aug 5 - 98 Aug 6 - 98 Aug 7 - 98 Aug Aug 98 - mix

Fig. 4.11: Determination of microcystin in blue coloured gut contents of Lake Ammersee coregonids caught during Planktothrix bloom episodes in 1998 and 1999 (error bars = 95% CI, n = 10). Gut content samples were classified as microcystin-positive when significant elevations were determined at the *p <0.05 significance level using ANOVA and Tukey’s Multiple Comparison test. Previously published in Environmental Toxicology 16: 483-488, 2001

Qualitative detection of covalently bound microcystin in liver tissue of Lake Ammersee coregonids sampled during bloom episodes in August 1998 and 1999 furthermore revealed microcystin adducts in liver homogenates. Those adducts had molecular weights between 28 and 39 kD, which is characteristically for microcystin covalently bound to protein phosphatases (Fig. 4.12).

DISCUSSION The documentation of P. rubescens filaments in gut content of Lake Ammersee coregonids demonstrates that feral coregonids indeed ingest P. rubescens from seston, thus giving evidence for natural exposure of coregonids to P. rubescens in Lake Ammersee. Once ingested it appears likely that P. rubescens cells were ruptured by the coregonid digestive process. However, rupture of cyanobacterial cells in the intestine of fish has been demonstrated to vary between different cyanobacteria species with cells of various Aphanizomenon sp. being almost totally broken while others, predominantly cyanobacteria species with cell walls including

Aug 98 Nov 99 Apr 99 Aug 99 Marker protein 38 kD

33 kD

29 kD

Fig. 4.12: Detection of microcystin-binding-protein adducts in liver homogenates of Lake Ammersee coregonids by immunoblotting. Samples of August 1998 and 1999 show bands with a molecular weight of 38 kD which is characteristically for microcystin covalently bound to protein phosphatases. Previously published in Environmental Toxicology 16: 483-488, 2001

139 4. FIELD STUDIES ______an exopolysaccharide sheath (e.g. Microcystis sp.), remaining largely intact (Carbis et al., 1997; Cazenave et al., 2005; Gavel et al., 2004; Kamjunke et al., 2002a; Kamjunke et al., 2002b; Lewin et al., 2003). P. rubescens lacks a mucilaginous sheath (Anagnostides & Komárek, 1988; Feuillade, 1994) and should thus in principal be susceptible to rupture in the fish intestine. The results of this study confirm the breakdown of P. rubescens filaments within the gastrointestinal tract of feral coregonids, as demonstrated by the regular detection of intracellular biliproteins specific for cyanobacteria (i.e. phycocyanin and allophycocyanin) in the gut content. This indicates that P. rubescens exposure of Lake Ammersee coregonids indeed results in a release of cyanobacterial components within the coregonid intestine thus allowing exposure of coregonids to toxic P. rubescens components. Analyses of coregonid gut contents further demonstrated that gut contents showing prominent blue colouration contained biliprotein concentrations above the respective limits of quantification. Corroborated by microscopical examinations, thus, the blue colouration of gut content appears to be predominantly caused by the presence of cyanobacterial biliproteins, released following the rupture of filaments within the gastrointestinal tract of coregonids. As such, this may represent a marker for the exposure of feral coregonids to P. rubescens. The occurrence of significant biliprotein concentrations in Lake Ammersee coregonid gut contents and thus exposure of coregonids to P. rubescens components varied with season. Indeed, no biliprotein-positive gut content samples were observed in samples taken in November, January and February. This appears to be due to the naturally reduced activity and lower food intake due to spawning behaviour and slowed winter metabolism. This is in agreement with previous observations (Enz et al., 2001; Mookerji et al., 1998; Skurdal et al., 1985) and is additionally corroborated by the predominantly empty coregonid guts during the winter months. In contrast to this, biliproteins were observable in the coregonids gut contents with highest frequency and concentration during April and May. This applied to all years investigated. Interestingly, this was also true for 2002 and 2003, when P. rubescens cell densities in Lake Ammersee were very low and never exceed 3000 cells/ml (Ernst et al., submitted). The field observations carried out in this study thus demonstrate that biliprotein accumulations in coregonids gut content occurred independently from annual variations in P. rubescens abundances and thus to a certain extent also independent of prevailing P. rubescens cell densities. This indicates that P. rubescens exposure of Lake Ammersee coregonids is not solely dependent on P. rubescens abundance and that biliprotein accumulation in gut contents and accordingly P. rubescens exposure of Lake Ammersee coregonids might also originate from sources other than direct ingestion of P. rubescens. An additional source for the incorporation of P. rubescens components could be copepods that may temporarily play an important role in the coregonid diet (Enz et al., 2001; Mayr, 1998). As observations of this study demonstrate that not only Lake Ammersee coregonids, but also copepods in Lake Ammersee seston samples exhibit conspicuous blue gut colouration, it is likely

140 4. FIELD STUDIES ______that copepods also accumulate P. rubescens components within their digestive tract. It thus appears possible, that Lake Ammersee coregonids may be also exposed to P. rubescens components via feeding on copepods.

The determination of biliproteins in gut content of Lake Ammersee coregonids suggests that other cyanobacterial components, including toxic metabolites (e.g. microcystins), may also accumulate in the digestive tract of exposed coregonids. This was confirmed by the detection of significantly elevated microcystin levels in coregonid gut contents sampled during P. rubescens bloom episodes in 1998 and 1999 which are in agreement with previous investigations demonstrating microcystin contaminations in faeces of cyanobacterial exposed fish (Chen et al., 2007; Xie et al., 2004). These results thus prove that feral coregonids in Lake Ammersee are exposed to P. rubescens and in consequence also to microcystins. Microcystin exposure of Lake Ammersee coregonids was additionally confirmed by the presence of Adda-positive bands in liver homogenates indicating the presence of covalently-bound microcystin adducts in livers (Fischer & Dietrich, 2000; Hitzfeld et al., 1999; Mikhailov et al., 2003). These hepatic microcystin adducts furthermore revealed that microcystins are not only released within the coregonid intestine, but also traverse the ileal membrane and enter the coregonids metabolism. The exposure of feral Lake Ammersee coregonids to microcystin containing P. rubescens can thus account for the experimentally observed toxicological effects (i.e. physiological stress and organ pathology, effects fish growth and fitness and enhanced fish mortality) previously reported (Ernst et al., 2007; Ernst et al., 2006a).

In summary, this study gives evidence for naturally occurring exposure of coregonids in Lake Ammersee to P. rubescens. The results demonstrate this exposure to cause an accumulation of P. rubescens components within the coregonid gastrointestinal tract, by direct ingestion and subsequent rupture of P. rubescens filaments and/or feeding of coregonids on copepods which have accumulated P. rubescens components. The P. rubescens components accumulating within the coregonids intestine have been shown not only to include biliproteins, which apparently cause a prominent blue colouration of coregonid faeces, but also microcystins. This unambiguously demonstrates the exposure of feral coregonids in Lake Ammersee to ichthyotoxic microcystins. Since microcystins have also been demonstrated to cross the ileal membrane and to accumulate in the liver of Lake Ammersee coregonids, substantial detrimental effects on coregonids seem inevitable. The results presented here thus substantiate the initially-proposed suggestion of a causal relationship between P. rubescens mass occurrence and challenged coregonid populations in pre- alpine lakes such as Lake Ammersee.

141

5. GENERAL DISCUSSION

5.1. ASSESSMENT OF THE IMPACT OF PLANKTOTHRIX RUBESCENS ON FERAL COREGONID IN LAKE AMMERSEE

INITIAL SITUATION Pelagic coregonids in Lake Ammersee have regularly suffered prominent growth reduction apparently resulting in reduced fish fitness which appears to be associated with the regular disappearance of coregonid age groups (Wißmath, 2000). Twofold annuli, as regularly observable in scales of Lake Ammersee coregonids, suggest that this growth retardation results from an additional starvation period, beyond the one normally occurring during the winter months (Wißmath et al., 1992). Both reduced growth and loss of coregonid age groups cause pronounced slumps in fishery yields. The latter may prove to existentially challenge to the Lake Ammersee fishery (Wißmath, 2000). The observed phenomena have been heatedly discussed in connection with deficient zooplankton food-species as a consequence of intentional nutritional re-depletion of waters (re- oligotrophication) and high fish densities (Mayr, 2001; Morscheid & Morscheid, 2001), and also in relation to recurrent metalimnic oxygen deficiencies evidently forcing the pelagic coregonids to habitats of inadequate temperature, darkness and insufficient food supply (Wißmath et al., 1992; Wißmath et al., 1993).

Additionally, slumps in coregonid yields in Lake Ammersee have been observed to coincide with the increased occurrence of the cyanobacterium P. rubescens: This applied to the 1970s, when P. rubescens proliferated following nutritional enrichment (eutrophication) of the lake, but also to the period since the 1990s, as P. rubescens has experienced a renaissance as a consequence of re- oligotrophication of the lake (see chapter 1.4 & 4.1). This apparent correlation appeared even more interesting, since previous studies revealed cyanobacteria of the Planktothrix genera to contain toxic metabolites that cause water quality problems and have detrimental effects on fish (Berg et al., 1986; Lindholm et al., 1989; Sivonen et al., 1990; Skulberg, 1984). As Lake Ammersee coregonids repeatedly displayed blue coloured gut contents, presumably resulting from cyanobacterial biliproteins and thus indeed indicating coregonid contact with cyanobacteria, it appeared plausible that the challenge to the coregonid population in Lake Ammersee might be causally related to the occurrence of toxic P. rubescens. Hence, one main goal of this study was to elucidate whether, and if yes, how the cyanobacterium P. rubescens affects feral coregonids in the pelagic zone of Lake Ammersee.

142 5. GENERAL DISCUSSION ______

PLANKTOTHRIX RUBESCENS

direct impacts on coregonids indirect impacts on coregonids i.e. toxicity and irritation i.e. effects on coregonid environment

coregonids do spatial impacts on impacts on not avoid avoidance food exogenous factors of P. rubescens P. rubescens filaments organisms e.g. pH value, oxygen filaments

e.g. detrimental coregonids coregonids effects on consequence stop grazing ingest bioaccumulation selection of consequence embryonal- and migration to avoid P. rubescens of toxins inferior food migration larval filament intake filaments species development

inadequate toxicological deficient food inadequate habitation, effects supply habitation, deficient food physiological stress deficient food supply organ damage supply

deficient decreased organ deficient nutrition functionality nutrition

retarded growth irregularities in exhaustion of increased reduced fitness population energy mortality rates dynamics elevated susceptibility reserves to other negative influences

Fig. 5.1: Plausible routes by which microcystin containing P. rubescens may impact feral coregonids in the pelagic zone of lakes, based on the current knowledge of the ichthyotoxicity of microcystin-containing cyanobacteria and specific information on the toxicity of P. rubescens in coregonids as ascertained in this study.

143 5. GENERAL DISCUSSION ______

THE TOXICITY OF P. RUBESCENS IN LAKE AMMERSEE – BASIC CONSIDERATIONS Initial investigations confirmed P. rubescens in Lake Ammersee to contain considerable amounts of microcystin (see chapter 4.1). The assessment of P. rubescens toxicity carried out in this study is in agreement with Kurmayr et al. (2005), demonstrating P. rubescens isolates from Lake Ammersee to contain the highest microcystin content when compared to Planktothrix isolates from 16 other European lakes. These initial investigations, the assessment of current knowledge on the ichthyotoxicity of microcystin-containing cyanobacteria (summarised in Malbrouck & Kestemont, 2006) and specific information on the toxicity of P. rubescens in coregonids as determined in this study (see chapter 3.1 & 3.2), gives rise to several conceivable means by which P. rubescens might impact feral coregonids in the pelagic zone of Lake Ammersee (Fig 5.1).

DIRECT IMPACT OF TOXIC P. RUBESCENS ON LAKE AMMERSEE COREGONIDS

Effects on the Embryonic and Larval Development of Coregonids Studies investigating cyanobacterial toxicity on the embryo and larval development of fish have demonstrated microcystins to reduce embryo survival and hence hatching rates, and furthermore, to cause significant malformations restricting the chance of survival for larvae hatched subsequent to cyanobacterial exposure (summarised in chapter 1.3). Although this issue was not explicitly examined in this study, some indications of detrimental effects on the embryonic development of coregonids in Lake Ammersee do exist. In winter 2000/01, when P. rubescens in Lake Ammersee consistently maintained cell densities between 5000 and 15,000 cells/ml for almost five months (see chapter 4.1), sporadic sampling revealed the presence of P. rubescens filaments as well as microcystin contamination in the hatching water pumped from the lake, thus indicating exposure of coregonid spawn to microcystin and microcystin-containing P. rubescens through hatching (Ernst et al., 2001). Interestingly, in that winter, coregonid hatching in the hatchery attained a rate of just 25 %, which is low compared to normal average hatching success of ≥70% (Eckmann, 2003 and references therein). This suggests that the exposure of coregonid spawn to microcystin might have been the reason for the significantly reduced embryo survival. Furthermore, of the remaining, hatched coregonid larvae, 4% showed malformations comparable to those typical for exposure to cyanobacterial toxins (Ernst et al., 2001), thus additionally corroborating the apparent exposure of coregonid embryos to microcystin and/or microcystin-containing P. rubescens in the hatchery. Since in winter 2000/01, P. rubescens and most probably also microcystins were homogenously distributed over the entire water column, it appears likely that also coregonid embryo developing in natural habitats may have suffered sustained microcystin exposure. Thus, it can be expected

144 5. GENERAL DISCUSSION ______that reduced survival and increased larval malformations also occurred in natural developing coregonid embryos. This is corroborated by the observation that the strength of the coregonid age group born in 2001 did not reach even half of that born in 2000 subsequent to a winter without elevated P. rubescens densities (see chapter 4.1). It thus appears thinkable, that the disappearance of coregonid age groups in Lake Ammersee might also be a result of a disturbance of embryonal development, caused by microcystin exposure of the spawn.

Effects on Metabolism, Grazing and Coregonid Habitation The exposure experiments carried out in this study clearly demonstrate that subchronic exposure of coregonids to P. rubescens causes several detrimental effects (see chapter 3.1 & 3.2). This has been shown to apply to P. rubescens densities greater than 1500 cells/ml, which, in Lake Ammersee occurred during 50% of the 261 weeks observed (see chapter 4.1). The results of this study thus suggest that Lake Ammersee coregonids are indeed regularly confronted with detrimental P. rubescens exposure situations. In such situations, coregonids have principally two possibilities of dealing with adverse P. rubescens abundances: (i) they may remain in water layers with elevated P. rubescens density, or (ii) they may avoid P. rubescens exposure via migration to strata lacking P. rubescens filaments.

COREGONIDS REMAIN IN STRATA WITH HIGH P. RUBESCENS DENSITY: When remaining in layers with higher P. rubescens densities, coregonids may actively avoid P. rubescens filament ingestion via a general reduction in feeding activity as has been observed for other planktivorous fish species in the presence of toxic cyanobacteria (Beveridge et al., 1993; Keshavanath et al., 1994). Empty gastrointestinal tracts as repeatedly presented by Lake Ammersee coregonids unambiguously reveal those coregonids actually to fast during the growing season (Mayr, 2001; Negele et al., 2000; Ernst et al., unpublished data). Based on the current data it cannot be assessed with certainty whether this fasting indeed results from an active avoidance of P. rubescens filaments ingestion. However, such fasting would have the inevitable consequence of deficiencies in the nutritional status of coregonids in Lake Ammersee. The results of this study further indicate that P. rubescens filaments can become ensnarled between coregonid gill lamellae and gillraker, as a result of water flow through the gills (see chapter 3.2). Hence, coregonids remaining in water strata with elevated P. rubescens densities can be expected to incorporate toxic P. rubescens filaments and consequently to suffer microcystin exposure despite reduced grazing (Fig. 5.2). Field observations carried out in this study certainly confirm that Lake Ammersee coregonids occasionally incorporate P. rubescens filaments and thus indicate that they indeed stay within water layers containing critical P. rubescens densities (see chapter 4.2). This may be ascribed to either deficient P. rubescens perception and/or due to a lack of avoidance opportunities (e.g. when

145 5. GENERAL DISCUSSION ______

Fig. 5.2: One year old coregonid in a Lake Ammersee water sample (sampled from 11 m depth in June 2001) containing P. rubescens density of approximately 11,000 cells/ml. Illumination of P. rubescens filaments by visualising the autofluorescence of biliproteins via

epifluorescence microscopy (green light excitation) allows a comparison of the environmentally relevant proportions. The constellation demonstrates an inevitable uptake of cyanobacterial filaments by coregonids foraging in water containing comparable P.

rubescens densities.

filaments are distributed over the entire water column during winter circulation). Blue coloured gut contents were observed in Lake Ammersee coregonids with highest frequency during April and May (see chapter 4.2), when coregonids are forced to feed due a generalised depletion of energy stores subsequent to overwintering (Negele et al., 2000; Reshetnikov et al., 1970; Valtonen, 1974). It is plausible that during April and May the coregonids regularly must remain in water layers which provide sufficient food despite the presence of high P. rubescens cell densities. The detection of microcystin in gut contents and liver homogenates of Lake Ammersee coregonids further indicates that the ingestion of P. rubescens results in microcystin exposure (see chapter 4.2), which may lead to toxic symptoms such as physiological stress and organ damage. According to the experimentally demonstrated injuries (see chapter 1.3 & 3), this damage can be expected to affect organ functionality and also coregonid glycogen levels. In conclusion, also toxicological consequences of continuous coregonid exposure to toxic P. rubescens may be supposed to substantially impair the nutritional status of feral coregonids in Lake Ammersee.

COREGONIDS AVOID P. RUBESCENS EXPOSURE VIA MIGRATION: As experimental observations reveal P. rubescens to cause noticeable stress in coregonids (see chapter 3), it appears probable that feral coregonids try to avoid P. rubescens exposure via migration between water layers. Such migration is particularly relevant during summer stratification, when P. rubescens filaments are usually stratified into distinct metalimnic layers where they concur with the coregonids typical habitats (Mookerji et al., 1998; Wißmath et al., 1993). Previous observations on the Lake Ammersee coregonid population indeed documented those coregonids to regularly withdraw from the metalimnion into atypical depths during the growing season in summer (Wißmath et al., 1993). This has to date been attributed to insufficient metalimnic oxygen conditions (Wißmath, 2004; Wißmath et al., 1993). Since coregonid migration was also evident during periods of sufficient metalimnic oxygen, it could however also originate from avoidance of P. rubescens (Fig 5.3). This migration from metalimnic layers during summer also provides additional stressors to the coregonids as the prevailing water temperatures above the metalimnion are at the limit of coregonid tolerance and significantly lower temperatures,

146 5. GENERAL DISCUSSION ______

[°C] [% saturation] 10 20 30 10 20 30 10 20 30 0 50 100 150 200 0 50 100 150 200 0 50 100 150 200

Depth [m] 20

Jun Jul Aug 2001 2001 2001

0 25 50 0 25 50 0 70 103 cells/ml] P. rubescens [103 cells/ml]

Fig. 5.3: Snapshots from Lake Ammersee during summer 2001: Acoustic recordings (background) demonstrated a continuously migration of coregonids (grey sickles) beneath the metalimnion which is characterised by the temperature (green), oxygen conditions (red) and P. rubescens layers (purple).

Acoustic campaigns were conducted with a Lowrance X-16/192 kHz-echo sounder (Lowrance, USA) immediately after sun down. The surface bubble layer in the range of 0-6 m depth has to be excluded. Temperature, oxygen profiles and P. rubescens cell densities were determined as carried out in chapter 4.1 and 2.1, respectively.

darkness and reduced food resources below the metalimnic layer result in an unavoidable reduction of metabolism (see chapter 4.1). Hence it appears inevitable, that Lake Ammersee coregonids suffer sustained nutritional constraints when migrating from the metalimnic layers in order to escape elevated P. rubescens densities, which, depending on the duration of the P. rubescens stratification, could result in a deficient nutrition of the fish.

In conclusion, both avoidance of P. rubescens (either as a consequence of reduced food intake or migration from typical habitats) and exposure to toxic P. rubescens can be expected to cause detrimental effects on coregonid metabolism and thus to result in sustained impairment of their nutrition. This suggests that the occurrence of toxic P. rubescens per se causes a deficient nutrition of feral coregonids in Lake Ammersee.

P. RUBESCENS-INDUCED ENVIRONMENTAL CHANGES: INDIRECT EFFECTS Cyanobacteria have been shown to affect various biotic and abiotic factors in lakes, hence suggesting indirect effects on coregonids living in the pelagic zone of lakes. This particularly concerns possible impacts on coregonid food organisms (such as cladoceran plankton) and effects on exogenous factors (e.g. effects on oxygen conditions, pH values, ammonium content etc.), which may result in an insufficient sometimes even hostile environment for fish (Wiegand & Pflugmacher, 2005 and references therein).

147 5. GENERAL DISCUSSION ______

Impacts on Coregonids following P. rubescens-induced Changes in Oxygen Regime One of the most critical exogenous parameters, which can be affected by cyanobacteria are oxygen conditions, as the decomposition of senescent cyanobacterial bloom material often causes substantial oxygen depletion and consequentially extensive mortalities of feral fish (Jewel et al., 2003; Nascimento & Azevedo, 1999; Pollux & Pollux, 2004; Toranzo et al., 1990). The oxygen regime in Lake Ammersee is characterised by the regular occurrence of a pronounced metalimnic oxygen depletion (Kucklentz et al., 2001; Lenhart, 2000), which, as demonstrated in this study (see chapter 4.1), appears to be causally related to recurring high P. rubescens abundance16. As these metalimnic oxygen depletions regularly cause levels to fall below coregonid tolerance (see chapter 4.1), they may provide, in addition to stratified microcystin-containing P. rubescens, for an additional stress factor in the metalimnic layer. This suggests that metalimnic stratified P. rubescens could force the coregonids in Lake Ammersee to withdraw from their typical habitat not only directly, in order to avoid contact with filaments but also indirectly, due to enhanced oxygen depletion. As discussed above, such migration out of the metalimnic layer would result in a reduction of coregonid nutrition (see above) and thus coregonid starvation.

Impacts on Coregonids via P. rubescens-induced Effects on Zooplankton Toxic cyanobacteria have been demonstrated to influence the zooplankton composition of waterbodies, as there is remarkable variation in the response of diverse zooplankton species to toxic and even to non-toxic cyanobacteria (summarised in Sivonen & Jones, 1999). Small and selective foraging species (e.g. Bosmina sp., various copepods) seemed to be less affected, while larger, filter feeding cladoceran species (mostly Daphnia sp.) are negatively selected as they cannot avoid cyanobacterial uptake except by stopping grazing. Since pelagic coregonids are planktivorous, feeding preferentially on large non-evasive cladoceran species (e.g. Daphnia sp., Bythotrephes sp.; see Mayr, 2001; Mookerji et al., 1998 and Skurdal et al., 1985), cyanobacteria thus in principle appear able to change the zooplankton composition in a manner disadvantageous for coregonids. In comparison with other pre-alpine lakes, the feeding habits of Lake Ammersee coregonids are unusually often based on Bosmina and copepod species (Mayr, 2001). It therefore appears possible, that the occurrence of toxic P. rubescens may shift the balance of zooplankton from Daphnia sp. toward Bosmina and copepods. As a consequence a qualitative and, when the regular appearance of empty coregonid guts during the growth season are considered, even a general loss of coregonid food organisms is possible. Such insufficient food supply can be assumed to substantially impair coregonid nutritional status. In addition, several organisms serving as fish food have themselves been shown to accumulate considerable amounts of cyanobacterial toxins and for this reason represent potential vectors for the ingestion of cyanobacterial toxins by fish (Engström-Öst et al., 2002; Kankaanpää et al.,

16 Resulting from oxygen-dependent decomposition of senescent metalimnic P. rubescens cells or from a shift of the stratified P. rubescens population into deeper layers that enforces P. rubescens to respiration (see chapter 4.1) 148 5. GENERAL DISCUSSION ______

2005b; Karjalainen et al., 2005; Smith & Haney, 2006). Observations of this study demonstrate copepods in Lake Ammersee to accumulate P. rubescens components within their digestive tract and furthermore show copepods to play an important role in the diet of Lake Ammersee coregonids (chapter 4.2). It therefore appears plausible, that coregonids in Lake Ammersee may be exposed to P. rubescens toxins via feeding on copepods which have accumulated toxic P. rubescens components. This in principle might cause toxicological damage and effects on the coregonid nutrition comparable to that arising subsequent to ingestion of toxin-containing P. rubescens filaments.

CONSEQUENCES In reviewing the various routes by which microcystin-containing P. rubescens may impact feral coregonids in the pelagic zone of Lake Ammersee, it is obvious that in addition to influences causing direct adverse effects (e.g. effects on embryonal development and organ damage) to the fish, all conceivable factors of P. rubescens exposure finally culminate in a deficient nutrition for the coregonids. Prolonged P. rubescens abundances, as observable in Lake Ammersee, thus appear to result in substantial deficiencies in the coregonids nutrition. This can be expected to result in significant growth reduction hence providing a possible explanation for anomalous annuli arising on the scales of Lake Ammersee coregonids during the growing season and the associated growth retardation.

Furthermore, when considering that in pre-alpine lakes the coregonids growing season normally has a duration of just five months (Enz et al., 2001, Mookerji et al., 1998), it appears clear that prolonged nutritional deficiencies severely restrict the ability of coregonids to stockpile the energy reserves that are crucial for spawning and overwintering. Accordingly, prolonged P. rubescens abundances as observable in Lake Ammersee can be expected to result in incomplete and even exhaustion of energy reserves, most likely additionally reducing coregonid fitness and increasing their susceptibility to other detrimental effects. In 1993 and 1994, subsequent to the elevated P. rubescens abundances which occurred in 1992 and 1993 (Lenhart, 2000), Lake Ammersee coregonids were indeed demonstrated to lack sufficient fat and glycogen reserves (Negele et al., 2000). These observations corroborate the expected causality of prolonged P. rubescens abundance and deficient energy reserves. As coregonids have been shown to counterbalance their energy deficiencies with progressive degradation of muscle protein, it moreover appears likely, that inadequate nutrition ensuing from prolonged occurrence of toxic P. rubescens might not only result in reduced growth but also in significant weight reduction. It can be assumed, that the regular lack of energy most likely causes problems for coregonids during late winter and early spring, when after overwintering, depletion of energy stores is maximal, sufficient food supply is still lacking and furthermore, increasing water temperatures

149 5. GENERAL DISCUSSION ______stimulate coregonid metabolism thus increasing energy demands (Negele et al., 2000; Reshetnikov et al., 1970; Valtonen, 1974). Hence it is not surprising, that coregonid age groups in Lake Ammersee repeatedly disappear during the months of April and May. This is supported by the fact that the disappearance of coregonids predominantly involves mature age groups (Wißmath, 2004), which need an energy surplus for spawning. It is thus plausible that the loss of coregonid age groups in Lake Ammersee results from energy deficiencies caused by insufficient coregonid nutrition in the summer prior to spawning linking the disappearance of coregonid age groups to detrimental impacts of prolonged occurrence of toxic P. rubescens.

CONCLUDING ASSESSMENT The intentional re-depletion of phosphorous loads (re-oligotrophication) of Lake Ammersee provoked a renaissance of the microcystin-producing cyanobacterium P. rubescens. The prolonged occurrence of this toxic P. rubescens can be expected to affect feral coregonids in the pelagic zone in a variety of ways including immediate detrimental effects but also indirectly, primarily resulting in an insufficient nutrition of the coregonids causing reduced growth and deficient energy reserves. These energy deficiencies appear to cause the increased mortalities of Lake Ammersee coregonids, primarily observed after spawning and overwintering, when energy reserves are required. As a result of this, whole age groups of Lake Ammersee coregonids disappear subsequent to first spawning, i.e. in the first month of their third year. Regular disappearance of three-year-old coregonids represents a sustained detrimental effect on the population dynamics of coregonids, since this restricts the continuity of the pelagic Lake Ammersee population to spawning of a single age group. Furthermore, as sustainable fishery aims to catch only fish which have spawned at least once in their lifetime, coregonids dying after first spawning thus disappear before being suitable for sustainable fishery. This represents a huge loss of the fishing base for the coregonid fishery and dramatic slumps in fishery yields, which correspond, due to the importance of coregonids for the local fishery, to an existential threat to professional Lake Ammersee fishermen.

In conclusion, the current investigations reveal a significant impact of toxic P. rubescens on coregonids unambiguously confirming the supposed link between the occurrence of toxic P. rubescens and the observed irregulatories in growth and population dynamics of coregonids in Lake Ammersee.

150 5. GENERAL DISCUSSION ______

5.2. ASSESSMENT OF HUMAN HEALTH HAZARD ARISING FROM THE ABUNDANCE OF TOXIC P. RUBESCENS IN LAKE AMMERSEE

Following the argumentation introduced by Dietrich & Hoeger (2005), human risk rising from the abundance of toxic P. rubescens in pre-alpine lakes, can principally be restricted to three main routes of intoxication: • dermal, nasal or oral contact to cyanobacteria and/or cyanobacterial toxins during recreational use of water (accidental ingestion), • the ingestion of cyanobacteria and/or cyanobacterial toxins via contaminated drinking water • and the uptake of cyanobacterial toxins via contaminated food, e.g. fruit and vegetable accumulating cyanobacterial toxins due to irrigation with contaminated water, as well as cyanobacterial toxins accumulated in fish, crayfish and shellfish.

IRRITATION & ACCIDENTAL INTOXICATION DURING RECREATIONAL WATER ACTIVITIES A number of toxic cyanobacteria may develop mass occurrences during periods where people use water bodies for recreational purposes. Cyanobacteria are thus increasingly included in bathing water monitoring programs which sometimes results in the closing of public bathing sites and the banning of water sport activities (Chorus et al., 2000). This also applies to the pre-alpine region, where mass development of predominantly Microcystis, Aphanizomenon and Anabaena species regularly forces authorities to safeguard the public from cyanobacterial blooms with appropriate precautions. In order to induce adverse human health effects during recreational water activity, the level of P. rubescens needs to fulfil two requirements: (i) toxic P. rubescens filaments must appear in the surface layer (i.e. the only layer accessible to humans) and (ii) the P. rubescens occurrence must attain cell densities that are critical for human health. In order to reduce the likelihood of adverse effects on human health during recreational use of cyanobacteria-containing waters, the World Health Organisation (WHO) suggests a first guidance level advising recommendations for visitors at bathing sites of a maximum of 20,000 cells/ml (Falconer et al., 1999). Throughout the investigation carried out in Lake Ammersee from 1999 until 2004, continuous appearance of P. rubescens filaments in the near surface layer was widely restricted to the periods from September until end of May (chapter 4.1). With the exception of a short period during December 2000, P. rubescens abundances however did not reach cell densities ≥20,000 cells/ml and were thus below the WHO guideline recommendation. Therefore the risk arising for those few people pursuing water sport activities during the autumn and winter season 151 5. GENERAL DISCUSSION ______is assumed to be low despite the continuous presence of P. rubescens in the surface layer. In contrast to the situation in autumn and winter, P. rubescens cell densities regularly exceeded 20,000 cells/ml in Lake Ammersee from June until end of August (e.g. in 1999, 2000, 2001 and 2004). During that time, P. rubescens however was generally restricted to the compact metalimnic layers below 7 m depth and was thus inaccessible to the public (see chapter 4.1). Disturbances of P. rubescens stratification may nevertheless result in temporary shifts of toxic P. rubescens to the near surface layers as observed during high season in June and July 2001. However, even here densities exceeding 5000 cells/ml were not observed and as such, a minimal to negligible risk to human health was assumed. Nevertheless, such P. rubescens shifts during the summer season as well as the temporary occurrence of high surface P. rubescens densities during winter constitute a residual risk, as they are almost impossible to predict with regard to either cell density or duration. In order to recognise those situations in time and thus to enable authorities to safeguard people via the implementation appropriate precautions, a routine assessment of cyanobacterial surface cell densities in Lake Ammersee is to be recommended.

P. rubescens toxicity is associated with the production of microcystins. Serious human health risk arising from the occurrence of P. rubescens is thus generally restricted to the ingestion of microcystins. Analysis of random Lake Ammersee water samples demonstrate that microcystins may accumulate in surface layers although appreciable P. rubescens cell densities are not observable (Ernst et al., 2001). This indicates that health risks may arise even when P. rubescens filaments are not discernible. As microcystin concentrations exceeding 200 ng/l have not yet been observed, health risks appear low when considering appropriate WHO-guideline recommendations (Falconer et al., 1999). However, as this assessment is based on an inadequate number of sporadic samples, leaving long periods without appropriate investigation, definitive conclusions are impossible. Thus, to exclude human health hazard rising from accidental ingestion of microcystin during recreational activity further investigations on microcystin concentrations in the surface water of Lake Ammersee would be advantageous.

INTOXICATION VIA INGESTION OF CONTAMINATED DRINKING WATER AND FOOD As Lake Ammersee is neither used as a drinking water reservoir nor for irrigation of agricultural areas, no health risks arise from the ingestion of P. rubescens and/or cyanobacterial toxins via contaminated drinking water and or uptake of contaminated fruit and vegetables. Thus, human health hazard rising from microcystin ingestion via food uptake is restricted to a possible accumulation of microcystin in fish, crayfish and shellfish. This can further be restricted to fish, as crayfish and shellfish are very rare in Lake Ammersee and thus have no relevance for human consumption.

152 5. GENERAL DISCUSSION ______

Lake Ammersee fish, especially coregonids are fished commercially by professional fishermen and subsequent gutting and further processing sold for human consumption. Fish can accumulate cyanobacterial toxins (especially microcystin) in amounts presenting a possible human health risk (Deblois et al., Toxicon in press; Magalhaes et al., 2003; Magalhaes et al., 2001; Soares et al., 2004; Xie et al., 2005). The possibility that Lake Ammersee fish prepared for human consumption might contain critical microcystin contaminations hence needs to be assessed: Microcystins have been shown to accumulate primarily in the liver of exposed fish, whereas small amounts have also been shown to accumulate in kidney, blood, gill, bile, intestine and brain (see chapter 1.3). Correctly gutted fish thus represent minimal to no health hazard, as microcystin- accumulating organs do not reach human consumption. Hence, any health risk stemming from the consumption of fish exposed to cyanobacterial toxins and/or toxic cyanobacteria is restricted to the remaining risk due to the consumption of muscle tissue which usually contains no or very low amounts of microcystin.

Regarding the risk to humans presented by microcystin toxicity, the WHO proposes a daily tolerable intake for microcystin of 0.04 µg MC-LRequiv./kg a day. In accordance with the publication by Falconer et al. (2001), the interim maximum acceptable concentration (IMAC) in fish used for human consumption can be determined as IMAC = TDI x BW x POT/AFC where BW is average human body weight (i.e. 60 kg), POT is the proportion of toxin consumed in the form of contaminated fish, and AFC is the average fish consumption. Based on an average German freshwater fish consumption of approximately 10 g/day (AFC = 10g) and the assumption that contaminated Lake Ammersee fish is the main source of microcystin intoxication (POT ≥0.5) the IMAC of Lake Ammersee fish used for consumption can be estimated to ≥0.12 µg/g. Microcystin contaminations in muscle tissue of greater than 0.12 µg/g have only very rarely been determined and then without exception in fish caught during bloom episodes in tropical or subtropical waters (Cazenave et al., 2005; Chen et al., 2007; Gkelis et al., Aquatic Toxicology in press; Magalhaes et al., 2001). Such cyanobacterial bloom episodes generate much higher cell densities and a multiple of the microcystin concentrations observed in Lake Ammersee (e.g. a bloom of M. aeruginosa in the Jacarepaguá Lagoon, Brazil, generating 107 cells/ml and

0.3 mg MC-LRequiv./l resulted in ≤0.34 µg MC-LRequiv./g in muscle tissue of tilapia; see Magalhaes et al., 2001). As such high cell densities and subsequently high microcystin concentrations are not expected for Lake Ammersee, it appears unlikely that muscle tissue of Lake Ammersee fish contain the level of microcystin contamination necessary to present a hazard to human health. However, for a conclusive risk assessment, regular analyses of Lake Ammersee fish are necessary.

153

6. ABBREVIATIONS

Adda 3-amino-9-methoxy-2,6,8,- MC-LR microcystin-LR trimethyl-10-phenyl-4,6,- (including leucine and arginine) decadienoic acid MC-LRequiv. microcystin-LR equivalents AFC average fish consumption MC-LW microcystin-LW ALT alaninaminotransferase (including leucine and ANOVA analysis of variance tryptophan) APC allophycocyanin MC-RR microcystin-RR AST aspartateaminotransferase (including two arginine) Asp3-MC-RR [D-Asp3]-microcystin-RR MC-YR microcystin-YR (including two arginine) (including tyrosine and arginine) Asp3-Dhb7-MC-RR Mdha N-methyldehydroalanine [D-Asp3-(E)-Dhb7]-microcystin- Mdhb N-methyldehydrobutyrine RR (including two arginine) Measp D-methylaspartic acid ATP adenosine triphosphate MeOH methanol b.a. before application MQ water purified to 18.2 MΩ/cm BMAA Beta-N-methylamino-L-alanine MW molecular weight BSA bovine serum albumin NOD nodularin BuOH butanol OD optical density bw body weight OATP organic anion transporter BW average human body weight protein CF condition factor p.a. post application CI confidence interval PBS phosphate buffered saline D Dalton PC phycocyanin DNA desoxyribonucleinacid PE phycoerythrin DTT dithiothreitol PMSF phenylmethanesulphonylfluoride dw dry weight POT proportion of toxin (consumed in EDTA ethylenediaminetetraacetic acid form of contaminated fish) ELISA enzyme linked immunosorbent PP protein phosphatase assay (c)PPAssay (colourimetric) protein GST glutathione S-transferase phosphatase inhibition assay H&E haematoxylin & eosin PSP paralytic shellfish poison(ing) HPLC high performance liquid ROS reactive oxygen species chromatography SD standard deviation HSI hepatosomatic index SDS-PAGE sodium dodecyl sulfate IC inhibiting concentration polyacrylamide gel IMAC interim maximum acceptable electrophoresis concentration SEM standard error of the mean i.p. intraperitoneal SPE solid phase extraction i.v. intravenous TBS Tris buffered saline LD lethal dose TDI tolerable daily intake LDH lactatedehydrogenase TFA trifluoroacetic acid LPS lipopolysaccharide TPP total plasma protein MAD mean absolute deviation TTBS Tris buffered saline + MC microcystin Tween20TM MC-LF microcystin-LF UV ultraviolet (including leucine and WHO World Health Organisation phenylalanine) ww wet weight

154

7. REFERENCES

Adamovsky, O., Kopp, R., Hilscherova, K., Babica, P., Palikova, M., Paskova, V., Navratil, S., Marsalek, B. & Blaha, L. (2007): Microcystin kinetics (bioaccumulation and elimination) and biochemical responses in common carp (Cyprinus carpio) and silver carp (Hypophthalmichthys molitrix) exposed to toxic cyanobacterial blooms. Environmental Toxicology & Chemistry 26, 2687-2693. Adams, D. G. & Duggan, P. S. (1999): Heterocyst and akinete differentiation in cyanobacteria. New Phytologist 144, 3-33. Akcaalan, R., Young, F. M., Metcalf, J. S., Morrison, L. F., Albay, M. & Codd, G. A. (2006): Microcystin analysis in single filaments of Planktothrix spp. in laboratory cultures and environmental blooms. Water Research 40, 1583-1590. Albay, M., Akcaalan, R., Tufekci, H., Metcalf, J. S., Beattie, K. A. & Codd, G. A. (2003): Depth profiles of cyanobacterial hepatotoxins (microcystins) in three Turkish freshwater lakes. Hydrobiologia 505, 89-95. Albay, M., Matthiensen, A., Gurevin, C., Akcaalan, R., Aykulu, G. & Codd, G. A. (2004): Occurrence of toxic blue-green algae in brackish water lake (Kucuk Cekmece Golu, Istanbul). 6th International Conference on Toxic Cyanobacteria, pp. 25. Amorim, A. & Vasconcelos, V. (1999): Dynamics of microcystins in the mussel Mytilus galloprovincialis. Toxicon 37, 1041-1051. Anagnostides, K. & Komárek, J. (1988): Modern approach to the classification of cyanophytes: 3. Oscillatoriales. Archiv für Hydrobiologie Suppl. 80, 327-472. Andersen, R. J., Luu, H. A., Chen, D. Z. X., Holmes, C. F. B., Kent, M. L., Le Blanc, M., Taylor, F. J. R. & Williams, D. E. (1993): Chemical and biological evidence links microcystins to salmon 'netpen liver disease'. Toxicon 31, 1315-1323. Anderson, D. M. (1994): Red tides. Scientific American 271, 62-68. Annadotter, H., Cronberg, G., Lawton, L., Hansson, H. B., Göthe, U. & Skulberg, O. (2001): An extensive outbreak of gastroenteritis associated with the toxic cyanobacterium Planktothrix agardhii (Oscillatoriales, Cyanophyceae) in Scania South Sweden, pp. 200-208. In I. Chorus (Ed.): Cyanotoxins: Occurrence, Causes, Consequences, Springer, Berlin. Anneville, O., Souissi, S., Ibanez, F., Ginot, V., Druart, J. C. & Angeli, N. (2002): Temporal mapping of phytoplankton assemblages in Lake Geneva: Annual and interannual changes in their patterns of succession. Limnology & Oceanography 47, 1355-1366. Arcelli, C., Cordella, I. & Levialdi, S. (1975): Parallel thinning of binary pictures. Electronics Letters 11, 148-149. Azevedo, S. (2001): New Brazilian regulation for cyanobacteria and cyanotoxins in drinking water. 5th International Conference on Toxic Cyanobacteria. Baganz, D., Staaks, G., Pflugmacher, S. & Steinberg, C. E. (2004): Comparative study of microcystin-LR- induced behavioral changes of two fish species, Danio rerio and Leucaspius delineatus. Environmental Toxicology 19, 564-70. Baganz, D., Staaks, G. & Steinberg, C. (1998): Impact of the cyanobacterial toxin, microcystin-LR on behaviour of zebra fish, Danio rerio. Water Research 32, 948-952. Bagu, J. R., Sönnichsen, F. D., Williams, D., Andersen, R. J., Sykes, B. D. & Holmes, C. F. B. (1995): Comparison of the solution structures of microcystin-LR and motuporin. Nature Structural Biology 2, 114-116. Bailey-Watts, A. E. & Kirka, A. (1981): The assessment of size variation in Loch Leven phytoplankton: methodology and some of its uses in the study of factors influencing size. Journal of Plankton Research 3, 261-282. Barford, D., Das, A. K. & Egloff, M. P. (1998): The structure and mechanism of protein phosphatases: insights into catalysis and regulation. Annual Reviews in Biophysics and Biomolecular Structure 27, 133-164. Barreto, R. E. & Volpato, G. L. (2004): Caution for using ventilatory frequency as an indicator of stress in fish. Behavioural Processes 66, 43-51. Barton, B. A., Morgan, J. D. & Vijayan, M. M. (2002): Physiological and condition-related indicators of environmental stress in fish, pp. 111-148. In S. M. Adams (Ed.): Biological Indicators of Aquatic Ecosystem Stress, American Society of Fisheries, Bethesda, Maryland. Bartram, J., Carmichael, W. W., Chorus, I., Jones, G. & Skulberg, O. (1999): Eutrophication, cyanobacterial blooms and surface scums, pp. 5-7. In I. Chorus & J. Bartram (Eds): Toxic Cyanobacteria in Water, E & FN Spon, London. Belov, A. P., Giles, J. D. & Wiltshire, R. J. (1999): Toxicity in water column following the stratification of a cyanobacterial population development in a calm lake. IMA Journal of Mathematics Applied in Medicine and Biology 16, 93-110.

155 7. REFERENCES ______

Bentley, R. (1999): Secondary metabolite biosynthesis: the first century. Critical Reviews in Biotechnology 19, 1-40. Berg, K., Skulberg, O. M., Skulberg, R., Underdal, B. & Willen, T. (1986): Observations on toxic blue-green algae (cyanobacteria) in some Scandinavian lakes. Acta Veterinaria Scandinavia 27, 440-452. Best, J. H., Eddy, F. b. & Codd, G. A. (2001): Effects of purified microcystin-LR and cell extracts of Microcystis strains PCC 7813 and CYA 43 on cardiac function in brown trout (Salmo trutta) alevins. Fish Physiology and Biochemistry 24, 171-178. Best, J. H., Pflugmacher, S., Wiegand, C., Eddy, F. B., Metcalf, J. S. & Codd, G. A. (2002): Effects of enteric bacterial and cyanobacterial lipopolysaccharids, and of microcystin-LR, on glutathione S- transferase activities in zebra fish (Danio rerio). Aquatic Toxicology 60, 223-231. Bettinetti, R., Morabito, G. & Provini, A. (2000): Phytoplankton assemblage structure and dynamics as indicator of the recent trophic and biological evolution of the western basin of Lake Como (N. Italy). Hydrobiologia 435, 177-190. Beveridge, M. C. M., Baird, D. J., Rahmatullah, S. M., Lawton, L. A., Beattie, K. A. & Codd, G. A. (1993): Grazing rates on toxic and non-toxic stains of cyanobacteria by Hypophthalmichthys molitrix and Oreochromis niloticus. Journal of Fish Biology 43, 901-907. Bláha, L., Kopp, R., Simkova, K. & Mares, J. (2004): Oxidative stress biomarkers are modulated in silver carp (Hypophthalmichthys molitrix Val.) exposed to microcystin-producing cyanobacterial water bloom. Acta Veterinaria Brunensis 73, 477-482. Blikstad-Halstvedt, C., Rohrlack, T., Andersen, T., Skulberg, O. & Edvardsen, B. (2007): Seasonal dynamics and depth distribution of Planktothrix spp. in Lake Steinsfjorden (Norway) related to environmental factors. Journal of Plankton Research 29, 471-482. Blom, J. F., Bister, B., Bischoff, D., Nicholson, G., Jung, G., Süssmuth, R. D. & Jüttner, F. (2003): Oscillapeptin J, a grazer toxin of the freshwater cyanobacterium Planktothrix rubescens. Journal of Natural Products 66, 431-434. Blom, J. F., Robinson, J. A. & Jüttner, F. (2001): High grazer toxicity of [D-Asp3, (E)-Dhb7]microcystin-RR of Planktothrix rubescens as compared to different microcystins. Toxicon 39, 1923-1932. Boaru, D. A., Dragos, N. & Schirmer, K. (2006): Microcystin-LR induced cellular effects in mammalian and fish primary hepatocyte cultures and cell lines: a comparative study. Toxicology 218, 134-148. Botes, D. P., Kruger, H. & Viljoen, C. C. (1982): Isolation and characterization of four toxins from the blue- green alga, Microcystis aeruginosa. Toxicon 20, 945-954. Botes, D. P., Wessels, P. L., Kruger, H., Runnegar, M. T. C., Santikarn, S., Smith, R. J., Barna, J. C. J. & Williams, D. H. (1985): Structural studies on cyanoginosin-LR, -YR, -YA and -YM: peptide toxins from Microcystis aeruginosa. Journal of the Chemical Society Perkin Transactions 1, 2747-2752. Bradford, M. M. (1976): A rapid and sensitive method for the quantification of microgram quantities of protein utilizing the principle of protein-dye binding. Analytical Biochemistry 72, 248-254. Braun, R. (1953): Vom "Burgunderblut", pp. 12: Naturkundliche Skizze, Schweiz. Briand, J. F., Jacquet, S., Bernard, C. & Humbert, J. F. (2003): Health hazards for terrestrial vertebrates from toxic cyanobacteria in surface water ecosystems. Veterinary Research 34, 361-377. Briand, J.-F., Jacquet, S., Filinois, C., Avois-Jacquet, C., Maisonnette, C., Freissinet, C., Leberre, B., Bosse, J.-P. & Humbert, J. F. (2004): Variations in the microcystin production of Planktothrix rubescens (Cyanobacteria) assessed by a four years in situ survey of Lac du Bourget (France). 6th International Conference on Toxic Cyanobacteria, pp. 35-36. Briand, J. F., Robillot, C., Quiblier-Llobéras, C. & Bernard, C. (2002): A perennial bloom of Planktothrix agardhii (Cyanobacteria) in a shallow eutrophic French lake: limnological and microcystin production studies. Archiv für Hydrobiologie 153, 605-622. Bürgi, H. R. & Stadelmann, P. (2002): Change of phytoplankton composition and biodiversity in Lake Sempach before and during restoration. Hydrobiologia 469, 33-48. Bury, N. R., Eddy. F.B. & Codd, G. A. (1996a): Stress responses of brown trout, Salmo Trutta L., to the cyanobacterium, Microcystis aeruginosa. Environmental Toxicology and Water Quality 11, 187-193. Bury, N. R., Codd, G. A., Wendelaar Bonga, S. E. & Flik, G. (1998a): Fatty acids from the cyanobacterium Microcystis aeruginosa with potent inhibitory effects on fish gill Na+/K+-ATPase activity. The Journal of Experimental Biology 201, 81-89. Bury, N. R., Eddy, F. B. & Codd, G. A. (1995): The effects of the cyanobacterium Microcystis aeruginosa, the cyanobacterial hapatotoxin microcystin-LR, and ammonia on growth rate and ionic regulation of brown trout. Journal of Fish Biology 46, 1042-1054. Bury, N. R., Flik, G., Eddy, F. B. & Codd, G. A. (1996b): The effect of cyanobacteria and the cyanobacterial toxin microcystin-LR on Ca2+ transport and Na+/K+-ATPase in tilapia gills. The Journal of Experimental Biology 199, 1319-1326. Bury, N. R., McGeer, J. C., Eddy, F. B. & Codd, G. A. (1997): Liver damage in brown trout, Salmo trutta L., and rainbow trout, Oncorhynchus mykiss (Walbaum), following administration to the cyanobacterial hepatotoxin microcystin-LR via the dorsal aorta. Journal of Fish Disease 20, 209-215. Bury, N. R., Newland, A. D., Eddy, F. B. & Codd, G. A. (1998b): In vivo and in vitro intestinal transport of 3H-microcystin-LR, a cyanobacterial toxin in rainbow trout (Oncorhynchus mykiss). Aquatic Toxicology 41, 139-148. 156 7. REFERENCES ______

Buzzi, F. (2002): Phytoplankton assemblages in two sub-basins of Lake Como. Journal of Limnology 61, 117-128. Byth, S. (1980): Palm Island mystery disease. Medical Journal of Australia 2, 40, 42. Carbis, C., Rawlin, G., Mitchell, G., Anderson, J. & McCauley, I. (1996a): The histopathology of carp, Cyprinus carpio L., exposed to microcystin by gavage, immersion and intraperitoneal administration. Journal of Fish Diseases 19, 199-207. Carbis, C. R., Mitchell, G. F., Anderson, J. W. & McCauley, I. (1996b): The effects of microcystins on the serum biochemistry of carp, Cyprinus carpio L., when the toxins are administered by gavage, immersion and intraperitoneal routes. Journal of Fish Disease 19, 151-159. Carbis, C. R., Rawlin, G. T., Grant, P., Mitchell, G. F., Anderson, J. W. & McCauley, I. (1997): A study of feral carp, Cyprinus carpio L., exposed to Microcystis aeruginosa at Lake Mokoan, Australia, and possible implications for fish health. Journal of Fish Disease 20, 81-91. Carmichael, W. W., Biggs, D. F. & Gorham, P. R. (1975): Toxicology and pharmacological action of Anabaena flos-aquae toxin. Science 187, 542-544. Carmichael, W. W., Evans, W. R., Yin, Q. Q., Bell, P. & Moczydlowski, E. (1997): Evidence for paralytic shellfish poisons in the freshwater cyanobacterium Lyngbya wollei (Farlow ex Gomont). Applied and Environmental Microbiology 63, 3104-10. Cazenave, J., Bistoni, M. D., Pesce, S. F. & Wunderlin, D. A. (2006a): Differential detoxification and antioxidant response in diverse organs of Corydoras paleatus experimentally exposed to microcystin-RR. Aquatic Toxicology 76, 1-12. Cazenave, J., Bistoni M. D., L., Zwirnmann, E., Wunderlin, D. A. & Wiegand, C. (2006b): Attenuating effects of natural organic matter on microcystin toxicity in zebra fish (Danio rerio) embryos - benefits and costs of microcystin detoxication. Environmental Toxicology 21, 22-32. Cazenave, J., Nores, M. L., Miceli, M., Diaz, M. P., Wunderlin, D. A. & Bistoni, M. L. (in press): Changes in the swimming activity and glutathione S-transferase activity of Jenynsia multidentata fed with microcystin- RR. Water Research. Cazenave, J., Wunderlin, D. A., Bistoni, M. D., Ame, M. V., Krause, E., Pflugmacher, S. & Wiegand, C. (2005): Uptake, tissue distribution and accumulation of microcystin-RR in Corydoras paleatus, Jenynsia multidentata and Odontesthes bonariensis. A field and laboratory study. Aquatic Toxicology 75, 178-190. Chen, J., Xie, P., Zhang, D. & Lei, H. (2007): In situ studies on the distribution patterns and dynamics of microcystins in a biomanipulation fish-bighead carp (Aristichthys nobilis). Environmental Pollution 147, 150-157. Chorus, e. a. (2006): PEPCY - Toxic and bioactive peptides in cyanobacteria (Summary), pp. 6, EU contract Nr: QLK4-CT-2002-02634, Berlin. Chorus, I. & Bartram, J. (1999): Toxic cyanobacteria in water. A guide to their public health consequences, monitoring and management. E & FN Spon. London. Chorus, I., Falconer, I. R., Salas, H. J. & Bartram, J. (2000): Health risk caused by freshwater cyanobacteria in recreational waters. Journal of Toxicology and Environmental Health, Part B 3, 323-347. Chriswell, R. K., Shaw, G. R., Eaglesham, G. K., Smith, K. F., Norris, R. L., Seawright, A. A. & Moure, R. A. (1999): Stability of cylindrospermopsin, the toxin from the cyanobacterium Cylindrospermopsis raciborskii, effects of pH, temperature and sunlight on decomposition. Environmental Toxicology 14, 155-161. Codd, G. A., Morrison, L. F. & Metcalf, J. S. (2005): Cyanobacterial toxins: risk management for health protection. Toxicology & Applied Pharmacology 203, 264-272. Cohen, P. (1989): The structure and regulation of protein phosphatases. Annual Review of Biochemistry 58, 453-508. Cook, W. O., Beasley, V. R., Dahlem, A. M., Dellinger, J. A., Harlin, K. S. & Carmichael, W. W. (1988): Comparison of effects of anatoxin-a(s) and paraoxon, physostigmine and pyridostigmine on mouse brain cholinesterase activity. Toxicon 26, 750-753. Cook, W. O., Beasley, V. R., Lovell, R. A., Dahlem, A. M., Hooser, S. B., Mahmood, N. A. & Carmichael, W. W. (1989): Consistent inhibition of peripheral cholinesterases by neurotoxins from the freshwater cyanobacterium Anabaena flos-aquae: studies on duck, swine, mice and a steer. Environmental Toxicology & Chemistry 8, 915-922. Cox, P. A., Banack, S. A. & Murch, S. J. (2003): Biomagnification of cyanobacterial neurotoxins and neurodegenerative disease among the Chamorro people of Guam. Proceedings of the National Academy of Science USA 100, 13380-13383. Cox, P. A., Banack, S. A., Murch, S. J., Rasmussen, U., Tien, G., Bidigare, R. R., Metcalf, J. S., Morrison, L. F., Codd, G. A. & Bergman, B. (2005): Diverse taxa of cyanobacteria produce beta-N-methylamino- L-alanine, a neurotoxic amino acid. Proceedings of the National Academy of Science USA 102, 5074-5078. Craig, M. & Holmes, C. F. B. (2000): Freshwater hepatotoxins: Microcystin and nodularin, mechanisms of toxicity and effects on health, pp. 643-671. In L. M. Botana (Ed.): Seafood and Freshwater Toxins: Pharmacology, Physiology and Detection, Marcel Dekker Inc., New York.

157 7. REFERENCES ______

Craig, M., Luu, H. A., McCready, T. L., Williams, D., Andersen, R. J. & Holmes, C. (1996): Molecular mechanisms underlying the interaction of motuporin and microcystins with type-1 and type-2A protein phosphatases. Biochemistry and Cell Biology - Biochimie et Biologie Cellulaire 74, 569-578. Dalmo, R. A., Kjerstad, A. A., Arnesen, S. M., Tobias, P. S. & Bogwald, J. (2000): Bath exposure of Atlantic halibut (Hippoglossus hippoglossus L.) yolk sac larvae to bacterial lipopolysaccharide (LPS): absorption and distribution of the LPS and effect on fish survival. Fish and Shellfish Immunology 10, 107-128. Davidson, F. F. (1959): Poisoning of wild and domestic animals by a toxic waterbloom of Nostoc rivulare Kütz. Journal of the American Water Work Association 51, 1277-1287. Davis, P. A., Dent, M., Parker, J., Reynolds, J. & Walsby, A. E. (2003): The annual cycle of growth rate and biomass change in Planktothrix spp. in Belham Tarn, English Lake District. Freshwater Biology 48, 852-867. Deblois, C. P., Aranda-Rodriguez, R., Giani, A. & Bird, D. F. (in press): Microcystin accumulation in liver and muscle of tilapia in two large Brazilian hydroelectric reservoirs. Toxicon. DeMott, W. R., Zhang, Q.-X. & Carmichael, W. W. (1991): Effects of toxic cyanobacteria and purified toxins on the survival and feeding of a copepod and three species of Daphnia. Limnology and Oceanography 36, 1346-1357. Devidze, M. (1998): Harmful algae events in Georgian waters, pp. 91. In B. Reguera, J. Blanco, M. L. Fernández & T. Wyatt (Eds): Harmful Algae, IOC, Paris. Devlin, J. P., Edwards, O. E., Gorham, P. R., Hunter, M. R., Pike, R. K. & Stavric, B. (1977): Anatoxin-a, a toxic alkaloid from Anabaena flos-aquae NCR-44h. Canadian Journal of Chemistry 55, 1367-1371. Dietrich, D. R., Fischer, A., Michel, C. & Hoeger, S. J. (2008): Toxin mixture in cyanobacterial blooms - a critical comparison of reality with current procedures employed in human health risk assessment, pp. 885-912. In H. K. Hudnell (Ed.): Cyanobacterial Harmful Algal Blooms: State of the Science and Research Needs, Springer-Verlag, New York. Dietrich, D. R. & Hoeger, S. J. (2005): Guidance values for microcystin in water and cyanobacterial supplement products (blue-green algae supplements): a reasonable or misguided approach? Toxicology and Applied Pharmacology 203, 273-289. Dokulil, M. & Teubner, K. (2000): Cyanobacterial dominance in lakes. Hydrobiologia 438, 1-12. Dow, C. S. & Swoboda, U. K. (2000): Cyanotoxins, pp. 613-632: The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Druvietis, I. (1998): Observation on cyanobacteria blooms in Latvia's inland waters, pp. 35-36. In B. Reguera, J. Blanco, M. L. Fernández, and T. Wyatt (Eds): Harmful Algae, IOC, Paris. Eckmann, R. (2003): Alizarin marking of whitefish, Coregonus lavaretus otoliths during egg incubation. Fisheries Management and Ecology 10, 1-7. Edwards, C., Beattie, K. A., Scrimgeour, C. M. & Codd, G. A. (1992): Identification of anatoxin-a in benthic cyanobacteria (blue-green algae) and in associated dog poisonings at Loch Insh, Scotland. Toxicon 30, 1165-1175. Edwards, D. J., Marquez, B. L., Nogle, L. M., McPhail, K., Goeger, D. E., Roberts, M. A. & Gerwick, W. H. (2004): Structure and biosynthesis of the jamaicamides, new mixed polyketide-peptide neurotoxins from the marine cyanobacterium Lyngbya majuscula. Chemistry & Biology 11, 817-833. Egaas, E., Sandvik, M., Fjeld, E., Kallqvist, T., Goksoyr, A. & Svensen, A. (1999): Some effects of the fungicide propiconazole on cytochrome P450 and glutathione S-transferase in brown trout (Salmo trutta). Comparative Biochemistry and Physiology Part C: Pharmacology, Toxicology and Endocrinology 122, 337-44. Embleton, K. V., Gibson, C. E. & Heaney, S. I. (2003): Automated counting of phytoplankton by pattern recognition: a comparison with a manual counting method. Journal of Plankton Research 25, 669-681. Engström-Öst, J., Lethiniemi, M., Green, S., Kozlowsky-Suzuki, B. & Viitasalo, M. (2002): Does cyanobacterial toxin accumulate in mysid shrimps and fish via copepods? Journal of Experimental Marine Biology and Ecology 276, 95-107. Enz, C. A., Bürgie, H. R., Stössel, F. & Müller, R. (2001): Food preference of adult whitefish in eutrophic Lake Hallwil (Switzerland), and the question of cannibalism. Archiv für Hydrobiologie 152, 81-98. Eriksson, J. E., Grönberg, L., Nygård, S., Slotte, J. P. & Meriluoto, J. A. O. (1990): Hepatocellular uptake of 3H-dihydromicrocystin-LR, a cyclic peptide toxin. Biochimica et Biophysica Acta 1025, 60-66. Eriksson, J. E., Meriluoto, J., Kujari, H. P., Osterlund, K., Fagerlund, K. & Hallbom, L. (1988): Preliminary characterization of a toxin isolated from the cyanobacterium Nodularia spumigena. Toxicon 26, 161-166. Eriksson, J. E., Meriluoto, J. A. O. & Lindholm, T. (1989): Accumulation of a peptide toxin from the cyanobacterium Oscillatoria agardhii in the freshwater mussel Anodonta cygnea. Hydrobiologia 183, 211-216. Ernst, B. (2000): Das Microcystin-haltige Cyanobakterium Planktothrix agardhii und seine Auswirkungen auf die Coregonen im Ammersee, diploma thesis, Department of Natural Science, University of Konstanz, Konstanz. Ernst, B., Dietz, L., Hoeger, S. J. & Dietrich, D. R. (2005): Recovery of MC-LR in fish liver tissue. Environmental Toxicology 20, 449-458. 158 7. REFERENCES ______

Ernst, B., Hitzfeld, B. C. & Dietrich, D. R. (2001): Presence of Planktothrix sp. and cyanobacterial toxins in Lake Ammersee, Germany and their impact on whitefish (Coregonus lavaretus L.). Environmental Toxicology 16, 483-488. Ernst, B., Hoeger, S., O'Brien, E. & Dietrich, D. (submitted): Abundance and toxicity of Planktothrix rubescens in the pre-alpine Lake Ammersee, Germany. Harmful Algae. Ernst, B., Hoeger, S., O'Brien, E. & Dietrich, D. R. (2007): Physiological stress and pathology in European Whitefish (Coregonus lavaretus) induced by subchronic exposure to environmental relevant densities of Planktothrix rubescens. Aquatic Toxicology 82, 15-26. Ernst, B., Hoeger, S. J., O'Brien, E. & Dietrich, D. R. (2006a): Oral toxicity of the microcystin-containing cyanobacterium Planktothrix rubescens in European whitefish (Coregonus lavaretus). Aquatic Toxicology 79, 31-40. Ernst, B., Neser, S., O'Brien, E., Hoeger, S. J. & Dietrich, D. R. (2006b): Determination of the filamentous cyanobacteria Planktothrix rubescens in environmental water samples using an image processing system. Harmful Algae 5, 281-289. Falconer, I., Dornbusch, M., Moran, G. & Yeung, S. (1992): Effects of the cyanobacterial (blue-green algal) toxins from Microcystis aeruginosa on isolated enterocytes from the chicken small intestine. Toxicon 30, 790-793. Falconer, I. R. (2001): Toxic cyanobacterial bloom problems in Australian waters: risk and impacts on human health. Phycologia 40, 228-233. Falconer, I. R. (2005): Cyanobacterial Toxins of Drinking Water Supplies, CRC Press, Boca Raton. Falconer, I. R., Bartram, J., Chorus, I., Kuiper-Goodman, T., Utkilen, H., Burch, M. & Codd, G. A. (1999): Safe levels and safe practice, pp. 155-178. In I. Chorus & J. Bartram (Eds): Toxic Cyanobacteria in Water, F & FN Spon, London. Falconer, I. R., Buckley, T. & Runnegar, M. T. (1986): Biological half-life, organ distribution and excretion of 125-I-labelled toxic peptide from the blue-green alga Microcystis aeruginosa. Australian Journal of Biological Science 39, 17-21. Falconer, I. R., Jackson, A. R. B., Langley, J. & Runnegar, M. T. C. (1981): Liver pathology in mice in poisoning by the blue-green alga Microcystis aeruginosa. Australian Journal of Biological Science 34, 179-187. Falconer, I. R. & Yeung, D. S. K. (1992): Cytoskeletal changes in hepatocytes induced by Microcystis toxins and their relation to hyperphosphorylation of cell proteins. Chemico-Biological Interactions 81, 181-196. Fastner, J., Erhard, M., Carmichael, W. W., Sun, F., Rinehart, K. L., Rönicke, H. & Chorus, I. (1999a): Characterization and diversity of microcystins in natural blooms and strains of the genera Microcystis and Planktothrix from German freshwater. Archiv für Hydrobiologie 145, 147-163. Fastner, J., Neumann, U., Wirsing, B., Weckesser, J., Wiedner, C., Nixdorf, B. & Chorus, I. (1999b): Microcystins (hepatotoxic heptapeptides) in German fresh water bodies. Environmental Toxicology 14, 13 - 22. Fawell, J. K., Mitchell, R. E., Everett, D. J. & Hill, R. E. (1999): The toxicity of cyanobacterial toxins in the mouse: I microcystin-LR. Human & Experimental Toxicology 18, 162-167. Feuillade, J. (1994): The cyanobacterium (blue-green alga) Oscillatoria rubescens D.C. Archiv für Hydrobiologie Beih. Ergebn. Limnol. 41, 77-93. Fischer, W., Hitzfeld, B. C., Tencalla, F., Eriksson, J. E., Mikhailov, A. & Dietrich, D. R. (2000): Microcystin- LR toxicodynamics, induced pathology, and immunhistochemical localisation in livers of blue-green algae exposed rainbow trout (Oncorhynchus mykiss). Toxicological Sciences 54, 365-373. Fischer, W. J., Altheimer, S., Cattori, V., Meier, P. J., Dietrich, D. R. & Hagenbuch, B. (2005): Organic anion transporting polypeptides expressed in liver and brain mediate uptake of microcystin. Toxicology and Applied Pharmacology 203, 257-263. Fischer, W. J. & Dietrich, D. R. (2000): Pathological and biochemical characterisation of microcystin-induced hepatopancreas and kidney damage in carp. Toxicology and Applied Pharmacology. 164, 73-81. Fischer, W. J., Garthwaite, I., Miles, C. O., Ross, K. M., Aggen, J. B., Chamberlin, A. R., Towers, N. R. & Dietrich, D. R. (2001): Congener-independent immunoassay for microcystins and nodularins. Environmental Science & Technology 35, 4753-4757. Fladmark, K. E., Serres, M. H., Larsen, N. L., Yasumoto, T., Aune, T. & Doskeland, S. O. (1998): Sensitive detection of apoptogenic toxins in suspension cultures of rat and salmon hepatocytes. Toxicon 36, 1001-1014. Forti, A. (1907): Myxophyceae. In D. Toni (Ed.): Sylloge Algarum Omnium, Padua. Fournie, J. W. & Courtney, L. A. (2002): Histopathological evidence of regeneration following hepatotoxic effects of the cyanotoxin microcystin-LR in the Hardhead Catfish and Gulf Killfish. Journal of Aquatic Animal Health 14, 273-280. Francis, G. (1878): Poisonous Australian lake. Nature 18, 11-12. Friedland, K. D., Ahrenholz, D. W. & Haas, L. W. (2005): Viable gut passage of cyanobacteria through the filter-feeding fish Atlantic menhaden, Brevoortia tyrannus. Journal of Plankton Research 27, 715-718. Fu, J. & Xie, P. (2006): The acute effects of microcystin LR on the transcription of nine glutathione S- transferase genes in common carp Cyprinus carpio L. Aquatic Toxicology 80, 261-266. 159 7. REFERENCES ______

Fuentes, J. & Eddy, F. B. (1997): Drinking in marine, eurhyaline and freshwater teleost fish, pp. 135-149. In N. Hazon, F. B. Eddy & G. Flik (Eds): Ionic Regulation in Animals, Springer, Berlin. Fujiki, H., Sueoka, E. & Suganuma, M. (1996): Carcinogenesis of microcystins, pp. 203-232. In M. F. Watanabe, K. Harada, W. W. Carmichael & H. Fujiki (Eds): Toxic Microcystis, CRC Press, Tokyo. Furey, A., Crowley, J., Shuilleabhain, A. N., Skulberg, O. M. & James, K. J. (2003): The first identification of the rare cyanobacterial toxin, homoanatoxin-a, in Ireland. Toxicon 41, 297-303. Gaete, V., Canelo, E., Lagos, N. & Zambrano, F. (1994): Inhibitory effects of Microcystis aeruginosa toxin on ion pumps of the gill of freshwater fish. Toxicon 32, 121-127. Gammeter, S. & Forster, R. (2002): Langzeituntersuchungen im Zürichobersee, pp. 46, Wasserversorgung Zürich, Zürich. Gammeter, S., Forster, U. & Zimmermann, U. (1997): Limnologische Untersuchungen im Zürichsee 1972- 1996, Wasserversorgung Zürich (WVZ), Zürich. Gavel, A., Marsalek, B. & Adamek, Z. (2004): Viability of Microcystis colonies is not damaged by silver carp (Hypophthalmichthys molitrix) digestion. Algological Studies 153, 189-194. Gehringer, M. M. (2004): Microcystin-LR and ocadaic acid-induced cellular effects: a dualistic response. FEBS Letters 557, 1-8. Geitler, L. (1932): Cyanophyceae. Johnson Reprint Cooperation. Berlin. Gessner, F. (1950): Das Phytoplankton der Seen Oberbayerns in seiner quantitativen Entfaltung. Berichte der Bayerischen Botanischen Gesellschaft 28, 180-194. Gill, M. D. (1982): Bacterial toxins: a table of lethal amounts. Microbiological Reviews 46, 86-94. Giovannardi, S., Pollegioni, L., Pomati, F., Rosseti, C., Sacchi, S., Sessa, L. & Calamari, D. (1999): Toxic cyanobacterial blooms in Lake Varese (Italy): A multidisciplinary approach. Environmental Toxicology 14, 127-135. Gjolme, N., Utkilen, H. & Rohlack, T. (2004): A suggestion for a common way to express cyanobacteria biomass in culture work. 6th International Conference on Toxic Cyanobacteria, pp. 31. Gkelis, S., Lanaras, T. & Sivonen, K. (in press): The presence of microcystins and other cyanobacterial bioactive peptides in aquatic fauna collected from Greek freshwaters. Aquatic Toxicology. Glazer, A. N. (1985): Light harvesting by phycobilisomes. Annual Review of Biophysics and Biophysical Chemistry. 14, 47-77. Goldberg, J., Huang, H. B., Kwon, Y. G., Greengard, P., Nairn, A. C. & Kuriyan, J. (1995): Three- dimensional structure of the catalytic subunit of protein serine/threonine phosphatase-1. Nature 376, 745-753. Gorham, P. R., MCLachlan, J., Hammer, U. T. & Kim, W. K. (1964): Isolation and culture of toxic strains of Anabaena flos-aquae (Lyngb.) de Breb. Verhandlungen der Internationalen Vereinigung für Theoretische & Angewandte Limnologie 15, 796-804. Graber, M. A. & Gerwick, W. H. (1998): Kalkipyrone, a toxic gamma-pyrone from an assemblage of the marine cyanobacteria Lyngbya majuscula and Tolypothrix sp. Journal of Natural Products 677-680. Grach-Pogrebinsky, O., Sedmak, B. & Carmeli, S. (2003): Protease inhibitors from Slovenian Lake Bled toxic waterbloom of the cyanobacterium Planktothrix rubescens. Tetrahedron 59, 8329-8336. Grimminger, H. (1982): Verzeichnis der Seen in Bayern - Teil 1. Bayerisches Landesamt für Wasserwirtschaft München. München. Gubbins, M. J., Eddy, F. B., Gallacher, S. & Stagg, R. M. (2000): Paralytic shellfish poisoning toxins induce xenobiotic metabolising enzymes in Atlantic salmon (Salmo salar). Marine Environmental Research 50, 479-483. Gugger, M., Lenoir, S., Berger, C., Ledreux, A., Druart, J. C., Humbert, J. F., Guette, C. & Bernard, C. (2004): First record of anatoxin-a in a French river associated with dog neurotoxicosis. 6th International Conference on Toxic Cyanobacteria, pp. 43. Gunn, G. J., Rafferty, A. G., Rafferty, G. C., Cockburn, N., Edwards, C., Beattie, K. A. & Codd, G. A. (1992): Fatal canine neurotoxicosis attributed to blue-green algae (cyanobacteria). The Veterinary Record 130, 301-302. Hanzelova, V., Snabel, V., Kralova, M. & Fagerholm, H. P. (1995): A comparative study on fish parasites Proteocephalus exiguus and P. percae (Cestoda: Proteocephalidae): morphology, isoenzyms and karyotype. Canadian Journal of Zoology 73, 1191-1198. Harada, K. (1996): Chemistry and detection of microcystins, pp. 103-148. In M. Watanabe, K. Harada, W. Carmichael & H. Fujiki (Eds): Toxic Microcystis, CRC Press Inc., Boca Raton. Hardegree, M. C. & Tu, A. T. (1988): Bacterial toxins. Dekker. New York. Hastie, C. J., Borthwick, E. B., Morrison, L. F., Codd, G. A. & Cohen, P. T. (2005): Inhibition of several protein phosphatases by a non-covalently interacting microcystin and a novel cyanobacterial peptide, nostocyclin. Biochimica et Biophysica Acta 1726, 187-193. Hawkins, P. R., Novic, S., Cox, P., Neilan, B. A., Burns, B. P., Shaw, G., Wickramasinghe, W., Peerapornpisal, Y., Ruangyuttikarn, W., Itayama, T., Saitou, T., Mizuochi, M. & Inamori, Y. (2005): A review of analytical methods for assessing the public health risk from microcystin in the aquatic environment. Journal of Water Supply: Research and Technology 54, 509-518.

160 7. REFERENCES ______

Hawkins, P. R., Runnegar, M. T., Jackson, A. R. & Falconer, I. R. (1985): Severe hepatotoxicity caused by the tropical cyanobacterium (blue-green alga) Cylindrospermopsis raciborskii (Woloszynska) Seenaya and Subba Raju isolated from a domestic water supply reservoir. Applied and Environmental Microbiology 50, 1292-1295. Henriksen, P. (2001): Toxic freshwater cyanobacteria in Denmark, pp. 49-56. In I. Chorus (Ed.): Cyanotoxins: Occurrence, Causes, Consequences, Springer Verlag, Berlin. Henriksen, P., Carmichael, W. W., An, J. & Moestrup, O. (1997): Detection of an anatoxin-a(s)-like anticholinesterase in natural blooms and cultures of cyanobacteria/blue-green algae from Danish lakes and in the stomach contents of poisoned birds. Toxicon 35, 901-913. Heresztyn, T. & Nicholson, B. C. (2001): Determination of cyanobacterial hepatotoxins directly in water using a protein phosphatase inhibition assay. Water Research 35, 3049-3056. Hitzfeld, B. C., Fischer, W. J., Eriksson, J. E., Mikhailov, A. & Dietrich, D. R. (1999): Immunochemical detection of microcystin-LR in tissues and cells of rainbow trout. Toxicological Science 48, 33. Hitzfeld, B. C., Höger, S. J. & Dietrich, D. R. (2000): Cyanobacterial toxins: Removal during drinking water treatment, and human risk assessment. Environmental Health Perspectives 108 Supplement 1, 113-122. Hoeger, S. J., Hitzfeld, B. C. & Dietrich, D. R. (2005): Occurrence and elimination of cyanobacterial toxins in drinking water treatment plants. Toxicology and Applied Pharmacology 203, 231-42. Hoeger, S. J., Schmid, D., Blom, J. F., Ernst, B. & Dietrich, D. R. (2007): Analytical and functional characterization of microcystins [Asp3]MC-RR and [Asp3,Dhb7]MC-RR: consequences for risk assessment? Environmental Science & Technology 41, 2609-2616. Hoeger, S. J., Shaw, G., Hitzfeld, B. C. & Dietrich, D. R. (2004): Occurrence and elimination of cyanobacterial toxins in two Australian drinking water treatment plants. Toxicon 43, 639-49. Hofer, R. & Lackner, R. (1995): Fischtoxikologie. Gustav Fischer Verlag. Jena. Holschuh, A. (2001): Danger on the beach: Is toxic algae killing dogs at Big Lagoon? pp. 7: North Coast Journal. Honkanen, R. E., Zwiller, J., Moore, R. E., Daily, S. L., Khatra, B. S., Dukelow, M. & Boynton, A. L. (1990): Characterization of microcystin-LR, a potent inhibitor of type 1 and type 2A protein phosphatases. Journal of Biological Chemistry 265, 19401-19404. Hoogveld, H. L. & Moed, J. R. (1993): A digitising tablet for determining the length distribution of filamentous cyanobacteria. European Journal of Phycology 28, 59-61. Hormazabal, V., Ostensvik, O., Underdal, B. & Skulberg, O. M. (2000): Simultaneous determination of the cyanotoxins anatoxin a, microcystin desmethyl-3, LR, RR, and YR in fish muscle using liquid chromatography-mass spectrometry. Journal of Liquid Chromatography and Related Technologies 23, 185-196. Humpage, A. R., Rositano, J., Baker, P. D., Nicholson, B. C., Steffensen, D. A., Bretag, A. H. & Brown, R. K. (1993): Paralytic shellfish poisons from freshwater blue-green algae. The Medical Journal of Australia 159, 423. Huynh-Delerme, C., Edery, M., Huet, H., Puiseux-Dao, S., Bernard, C., Fontaine, J. J., Crespeau, F. & Luze, A. d. (2005): Microcystin-LR and embryo-larval development of medaka fish, Oryzias latipes. I. Effects on the digestive tract and associated systems. Toxicon 46, 16-23. Ibelings, B. W., Bruning, K., de Jonge, J., Wolfstein, K., Dionisio Pires, L. M., Postma, J. & Burger, T. (2005): Distribution of microcystins in alake Foodweb: No evidence for biomagnification. Microbial Ecology 49, 487-500. Ishida, K., Matsuda, H., Murakami, M. & Yamaguchi, K. (1997): Micropeptins 478-A and -B, plasmin inhibitors from the cyanobacterium Microcystis aeruginosa. Journal of Natural Products 60, 184-187. Ito, E., Kondo, F. & Harada, K. (2000): First report on the distribution of orally administered microcystin-LR in mouse tissue using an immunostaining method. Toxicon 38, 37-48. Itou, Y., Ishida, K., Shin, H. J. & Murakami, M. (1999): Oscillapeptins A to F, serine protease inhibitors from the three strains of Oscillatoria agardhii. Tetrahedron 55, 6871-6882. Jacquet, C., Thermes, V., Luze de, A., Puiseux-Dao, S., Bernard, C., Joly, J. S., Bourrat, F. & Edery, M. (2004a): Effects of microcystin-LR on the development of medaka fish embryos (Oryzias latipes). Toxicon 43, 141-147. Jacquet, S., Briand, J. F., Leboulanger, C., Avois-Jacquet, C., Oberhaus, L., Tassin, B., Vincon-Leite, B., Paolini, G., Druart, J. C., Anneville, O. & Humbert, J. F. (2005): The proliferation of the toxic cyanobacterium Planktothrix rubescens following restoration of the largest natural French lake (Lac du Bourget). Harmful Algae 4, 651-672. Jann-Para, G. C., Schwob, I. & Feuillade, M. (2004): Occurrence of toxic Planktothrix rubescens blooms in Lake Nantua, France. Toxicon 43, 279-285. Janssens, V. & Goris, J. (2001): Protein phosphatase 2A: A highly regulated family of serine/threonine phosphatases implicated in cell growth and signalling. Biochemical Journal 353, 417-439. Jewel, M. A. S., Affan, A. & Khan, S. (2003): Fish mortality due to cyanobacterial bloom in an aquaculture pond in Bangladesh. Pakistan Journal of Biological Sciences 6, 1046-1050. Jones, G. & Negri, A. P. (1997): Persistence and degradation of cyanobacterial paralytic shellfish poisons (PSPs) in freshwaters. Water Research 31, 525-533. 161 7. REFERENCES ______

Jones, G. & Orr, P. T. (1994): Release and degradation of microcystin following algicide treatment of a Microcystis aeruginosa bloom in a recreational lake, as determined by HPLC and protein phosphatase inhibition assay. Water Research 28, 871-876. Jones, G. J., Bourne, D. G., Blakeley, R. L. & Doelle, H. (1994): Degradation of the cyanobacterial hepatotoxin microcystin by aquatic bacteria. Natural Toxins 2, 228-235. Jos, A., Pichardo, S., Prieto, A. I., Repetto, G., Vazquez, C. M., Moreno, I. & Camean, A. M. (2005): Toxic cyanobacterial cells containing microcystins induce oxidative stress in exposed tilapia fish (Oreochromis sp.) under laboratory conditions. Aquatic Toxicology 72, 261-271. Jungblut, A. D., Hawes, I., Mountfort, D., Hitzfeld, B., Dietrich, D. R., Burns, B. P. & Neilan, B. A. (2005): Diversity within cyanobacterial mat communities in variable salinity meltwater ponds of McMurdo Ice Shelf, Antarctica. Environmental Microbiology 7, 519-529. Kamjunke, N., Mendonca, R., Hardewig, I. & Mehner, T. (2002a): Assimilation of different cyanobacteria as food and the consequences for internal energy stores of juvenile roach. Journal of Fish Biology 60, 731-738. Kamjunke, N., Schmidt, K., Pflugmacher, S. & Mehner, T. (2002b): Consumption of cyanobacteria by roach (Rutilus rutilus): useful or harmful to the fish? Freshwater Biology 47, 243-250. Kankaanpää, H., Holiday, J., Schröder, H., Godard, T. J., Fister von, R. & Carmichael, W. W. (2005a): Cyanobacteria in prawn farming in northern New South Wales, Australia - a case study on cyanobacteria diversity and hepatotoxin bioaccumulation. Toxicology and Applied Pharmacology 203, 243-256. Kankaanpää, H., Turunen, A. K., Karlsson, K., Bylund, G., Meriluoto, J. & Sipiä, V. (2005b): Heterogeneity of nodularin bioaccumulation in northern Baltic Sea flounders in 2002. Chemosphere 59, 1091-1097. Kankaanpää, H., Vuorinen, P. J., Sipiä, V. & Keinänen, M. (2002a): Acute effects and bioaccumulation of nodularin in sea trout (Salmo trutta m. trutta L.) exposed orally to Nodularia spumigena under laboratory conditions. Aquatic Toxicology 61, 155-168. Kankaanpää, H. T., Vuorensola, K. M., Sipiä, V. O. & Meriluoto, J. A. O. (2002b): Chromatographic and spectral behaviour and detection of hepatotoxic nodularin in fish, clam, mussel and mouse tissues using HPLC analyses. Chromatographia 55, 157-162. Karjalainen, M., Engstrom-Ost, J., Korpinen, S., Peltonen, H., Paakkonen, J. P., Ronkkonen, S., Suikkanen, S. & Viitasalo, M. (2007): Ecosystem consequences of cyanobacteria in the northern Baltic Sea. Ambio 36, 195-202. Karjalainen, M., Reinikainen, M., Spoof, L., Meriluoto, J. A., Sivonen, K. & Viitasalo, M. (2005): Trophic transfer of cyanobacterial toxins from zooplankton to planktivores: consequences for pike larvae and mysid shrimps. Environmental Toxicology 20, 354-62. Karlsson, K., Sipiä, V., Kankaanpää, H. & Meriluoto, J. (2003a): Mass spectrometric detection of nodularin and desmethylnodularin in mussels and flounders. Journal of Chromatography B 784, 243-253. Karlsson, K., Sipiä, V., Krause, E., Meriluoto, J. & Pflugmacher, S. (2003b): Mass spectrometry detection and quantification of nodularin-R in flounder livers. Environmental Toxicology 18, 284-288. Kaya, K. (1996): Toxicology of microcystins, pp. 175-202. In M. Watanabe, K. Harada, W. Carmichael & H. Fujiki (Eds): Toxic Microcystis, CRC Press Inc., Boca Raton. Keil, C., Forchert, A., Fastner, J., Szewzyk, U., Rotard, W., Chorus, I. & Krätke, R. (2002): Toxicity and microcystin content of extracts from a Planktothrix bloom and two laboratory strains. Water Research 36, 2133-2139. Kent, M. L. (1990): Nepten liver disease (NLD) of salmonid fishes reared in sea water: species susceptibility, recovery, and probable cause. Diseases of Aquatic Organisms 8, 21-28. Kent, M. L., Myers, M. S., Hinton, D. E. & R. A. Elston (1988): Suspected toxicopathic hepatic necrosis and megalocystis in pen-reared Atlantic salmon Salmo salar in Puget Sound, Washington, USA. Diseases of Aquatic Organisms 4, 91-100. Keshavanath, P., Beveridge, M. C. M., Baird, D. J., Lawton, L. A., Nimmo, A. & Codd, G. A. (1994): The functional grazing response of a phytoplanktivorous fish Oreochromis niloticus to mixtures of toxic and non-toxic strains of the cyanobacterium Microcystis aeruginosa. Journal of Fish Biology 45, 123-129. Khan, S. A., Ghosh, S., Wickstrom, M., Miller, L. A., Hess, R., Haschek, W. M. & Beasley, V. R. (1995): Comparative pathology of microcystin-LR in cultured hepatocytes, fibroblasts, and renal epithelial cells. Journal of Natural Toxins 3, 119-128. Kirchhofer, A. (2004): Monitoring der Felchenfänge der Berufsfischer von Brienzersee, Thunersee und Bielersee 1984 - 2003, pp. 41, WFN - Wasser Fisch Natur, Gümmenen. Kischnik, P. (1992): The coregonids of Lake Lacher See. Archiv für Hydrobiologie Beih. Ergebn. Limnol. 38, 273-294. Kolmakov, V. I. & Gladyshev, M. I. (2003): Growth and potential photosynthesis of cyanobacteria are stimulated by viable gut passage in crucian carp. Aquatic Ecology 37, 237-242. Kondo, F., Matsumoto, H., Yamada, S., Ishikawa, N., Ito, E., Nagata, S., Ueno, Y., Suzuki, M. & Harada, K. (1996): Detection and identification of metabolites of microcystin formed in vivo in mouse and rat livers. Chemical Research in Toxicology 9, 1355-1359.

162 7. REFERENCES ______

Kopp, R. & Hetesa, J. (2000): Changes of haematological indices of juvenile carp (Cyprinus carpio L.) under the influence of natural populations of cyanobacterial water blooms. Acta Veterinaria Brunensis 69, 131-137. Kotak, B. G., Semalulu, S., Fritz, D. L., Prepas, E. E., Hrudey, S. E. & Coppock, R. W. (1996): Hepatic and renal pathology of intraperitoneally administered microcystin-LR in rainbow trout (Oncorhynchus mykiss). Toxicon 34, 517-525. Kreitlow, S., Mundt, S. & Lindequist, U. (1999): Cyanobacteria-a potential source of new biologically active substances. Journal of Biotechnology 70, 61-63. Krienitz, L., Ballot, A., Kotut, K., Wiegand, C., Pütz, S., Metcalf, J. S., Codd, G. A. & Pflugmacher, S. (2003): Contribution of hot spring cyanobacteria to the mysterious deaths of Lesser Flamingos at Lake Bogoria, Kenya. FEMS Microbiology Ecology 43, 141-148. Krishnamurthy, T., Carmichael, W. W. & Sarver, E. W. (1986): Toxic peptides from freshwater cyanobacteria (blue-green algae). I. Isolation, purification and characterization of peptides from Microcystis aeruginosa and Anabaena flos-aquae. Toxicon 24, 865-873. Krupa, D. & Czernas, K. (2003): Mass appearance of cyanobacterium Planktothrix rubescens in Lake Piaseczno, Poland. Water Quality Research Journal of Canada 38, 141-152. Kucklentz, V., Hamm, A., Jöhnk, K., Tsang-Pi, C., Morscheid, H., Roth, D., Schmidt-Halewicz, S., Morscheid, H. & Mayr, C. (2001): Antwort bayerischer Voralpenseen auf verringerte Nährstoffzufuhr, pp. 272, Bayerisches Landesamt für Wasserwirtschaft, München. Kuiper-Goodman, T., Falconer, I. R. & Fitzgerald, J. (1999): Human health aspects, pp. 113-153. In I. Chorus & J. Bartram (Eds): Toxic Cyanobacteria in Water, F&FN Spon, London. Kurmayer, R., Christiansen, G., Fastner, J. & Börner, T. (2004): Abundance of active and inactive microcystin genotypes in populations of the toxic cyanobacterium Planktothrix spp. Environmental Microbiology 6, 831-841. Kurmayer, R., Christiansen, G., Gumpenberger, M. & Fastner, J. (2005): Genetic identification of microcystin ecotypes in toxic cyanobacteria of the genus Planktothrix. Microbiology 151, 1525-1533. Laemmli, U. (1970): Cleavage of structural proteins during assembly of the head of bacteriphage T4. Nature 227, 680-685. Lahti, K., Niemi, R. M., Rapala, J. & Sivonen, K. (1997): Biodegradation of cyanobacterial hepatotoxins - characterisation of toxin degrading bacteria. 8th International Conference on Harmful Algae, pp. 363-365. Lampert, W. (1981): Toxicity of blue-green Microcystis aeruginosa: Effective defence mechanism against grazing pressure by Daphnia. Verhandlungen der Internationalen Vereinigung für Theoretische & Angewandte Limnologie 21, 1436-1440. Landsberg, J. H. (2002): Effects of harmful algal blooms on aquatic organism. Reviews in Fishery Science 10, 113-390. Laurén-Määttä, C., Hietala, J. & Walls, M. (1997): Response of Daphnia pulex populations to toxic cyanobacteria. Freshwater Biology 37, 635-647. Lawrence, J. F. & Menard, C. (2001): Determination of blue green-algae, fish and water using liquid chromatography with ultraviolet detection after sample clean-up employing immunoaffinity chromatography. Journal of Chromatography 922, 111-117. Lawton, L. A., Edwards, C. & Codd, G. A. (1994): Extraction and high-performance liquid-chromatographic method for the determination of microcystins in raw and treated waters. Analyst 119, 1525-1530. Leboulanger, C., Dorigo, U., Jacquet, S., Le Berre, B., Paolini, G. & Humbert, J. F. (2002): Application of a submersible spectrofluorometer for rapid monitoring of freshwater cyanobacterial blooms: a case study. Aquatic Microbial Ecology 30, 83-89. Lecoz, N., Malecot, M., Quiblier, C., Puiseux-Dao, S., Bernard, C., Crespeau, F. & Edery, M. (2008): Effects of cyanobacterial crude extracts from Planktothrix agardhii on embryo-larval development of medaka fish, Oryzias latipes. Toxicon 51, 262-269. Lefebvre, K. A., Trainer, V. L. & Scholz, N. L. (2004): Morphological abnormalities and sensorimotor deficits in larval fish exposed to dissolved saxitoxin. Aquatic Toxicology 66, 159-70. Legnani, E., Copetti, D., Oggioni, A., Tartari, G., Palumbo, M. T. & Morabito, G. (2005): Planktothrix rubescens seasonal dynamics and vertical distribution in Lake Pusiano (North Italy). Journal of Limnology 64, 61-73. Lemmin, U. (1978): Lakes, Chemistry, Geology, Physics. Springer Verlag. Lenhart, B. (2000): Langfristige Entwicklung (Eutrophierung und Reoligotrophierung) am Ammersee. Münchner Beiträge zur Abwasser-, Fischerei- und Flussbiologie 53, 97-114. Lewin, W. C., Kamjunke, N. & Mehner, T. (2003): Phosphorus uptake by Microcystis during passage through fish guts. Limnology & Oceanography 48, 2392-2396. Li, X., Liu, Y. & Song, L. (2001): Cytological alterations in isolated hepatocytes from common carp (Cyprinus carpio L.) exposed to microcystin-LR. Environmental Toxicology, 517-522. Li, X., Liu, Y., Song, L. & Liu, J. (2003): Response of antioxidant systems in the hepatocytes of common carp (Cyprinus carpio L.) to the toxicity of microcystin LR. Toxicon 42, 85-89. Li, X. Y., Chung, I. K., Kim, J. I. & Lee, J. A. (2004): Subchronic oral toxicity of microcystin in common carp (Cyprinus carpio L.) exposed to Microcystis under laboratory conditions. Toxicon 44, 821-827.

163 7. REFERENCES ______

Li, X. Y., Chung, I. K., Kim, J. I. & Lee, J. A. (2005): Oral exposure to Microcystis increases activity- augmented antioxidant enzymes in the liver of loach (Misgurnus mizolepis) and has no effect on lipid peroxidation. Comparative Biochemistry and Physiology C-Pharmacology Toxicology and Endocrinology 141, 292-296. Li, X. Y., Wang, J., Liang, J. B. & Liu, Y. D. (2007): Toxicity of microcystins in the isolated hepatocytes of common carp (Cyprinus carpio L.). Ecotoxicology and Environmental Safety 67, 447-451. Liang, X. F., Li, G. G., He, S. & Huang, Y. (2007): Transcriptional responses of alpha- and rho-class glutathione S-transferase genes in the liver of three freshwater fishes intraperitoneally injected with microcystin-LR: relationship of inducible expression and tolerance. Journal of Biochemistry and Molecular Toxicology 21, 289-298. Lilleheil, G., Andersen, R. A., Skulberg, O. M. & Alexander, J. (1997): Effects of a homoanatoxin-a- containing extract from Oscillatoria formosa (Cyanophyceae/cyanobacteria) on neuromuscular transmission. Toxicon 35, 1275-1289. Lindholm, T., Degerlund, M., Spoof, L. & Meriluoto, J. (2002): A century of water quality changes in coastal lake with toxic Planktothrix. Verhandlungen der Internationalen Vereinigung für Theoretische & Angewandte Limnologie 28, 984-987. Lindholm, T., Eriksson, J. E. & Meriluoto, J. A. O. (1989): Toxic cyanobacteria and water quality problems - Examples from a eutrophic lake on Åland, South West Finland. Water Research 23, 481-486. Lindholm, T. & Meriluoto, J. A. O. (1991): Recurrent depth maxima of the hepatotoxic cyanobacterium Oscillatoria agardhii. Canadian Journal of Fisheries and Aquatic Sciences 48, 1629-1634. Lindholm, T., Öhman, P., Kurki-Helasmo, K., Kineaid, B. & Meriluoto, J. (1999): Toxic algae and fish mortality in a brackish-water lake in Aland, SW Finland. Hydrobiologia 397, 109-120. Linné, C. (1753): Species Plantarum. Stockholm. Liras, V., Lindberg, M., Nystrom, P., Annadotter, H., Lawton, L. A. & Graf, B. (1998): Can ingested cyanobacteria be harmful to the signal crayfish (Pacifastacus leniusculus). Freshwater Biology 39, 233-242. Little, E. E. (2002): Behavioral measures of environmental stressors in fish, pp. 431-472. In M. S. Adams (Ed.): Biological Indicators of Aquatic Ecosystem Stress, American Fisheries Society, Bethesda, Maryland. Liu, Y., Song, L. & Li, X. (2002): The toxic effects of microcystin LR on embryo-larval and juvenile development of Loach, Misgurnus mizolopis Gunthe. Toxicon 40, 395-399. Luukkainen, R., Sivonen, K., Namikoshi, M., Fardig, M., Rinehart, K. L. & Niemela, S. I. (1993): Isolation and identification of eight microcystins from thirteen Oscillatoria agardhii strains and structure of a new microcystin. Applied and Environmental Microbiology 59, 2204-2209. Mackenthum, K. M. & Herman, E. F. (1948): A heavy mortality of fishes resulting from the decomposition of algae in the Yahara river, Wisconsin. American Fisheries Society 75, 175-180. MacKintosh, C., Beattie, K. A., Klumpp, S., Cohen, P. & Codd, G. A. (1990): Cyanobacterial microcystin-LR is a potent and specific inhibitor of protein phosphatases 1 and 2A from both mammals and higher plants. FEBS Letters 264, 187-192. Magalhaes, V. F., Marinho, M. M., Domingos, P., Oliviera, A. C., Costa, S. M., Azevedo, L. O. & Azevedo, S. M. F. O. (2003): Microcystins (cyanobacteria hepatotoxins) bioaccumulation in fish and crustaceans from Septiba Bay (Brazil, RJ). Toxicon 42, 289-295. Magalhaes, V. F. d., Soares, R. M. & Azevedo, S. M. F. O. (2001): Microcystin contamination in fish from the Jcarepaguá Lagoon (Rio de Janeiro, Brazil): Ecological implication and human health risk. Toxicon 39, 1077-1085. Mahmood, N. A. & Carmichael, W. W. (1987): Anatoxin-a(s), an anticholinesterase from the cyanobacterium Anabaena flos-aquae NRC-525-17. Toxicon 25, 1221-1227. Malbrouck, C. & Kestemont, P. (2006): Effects of microcystins on fish. Environmental Toxicology and Chemistry 25, 72-86. Malbrouck, C., Trausch, G., Devos, P. & Kestemont, P. (2003): Hepatic accumulation and effects of microcystin-LR on juvenile goldfish Carassius auratus L. Comparative Biochemistry and Physiology 135, 39-48. Malbrouck, C., Trausch, G., Devos, P. & Kestemont, P. (2004a): Effect of microcystin-LR on protein phosphatase activity and glycogen content in isolated hepatocytes of fed and fasted juvenile goldfish Carassius auratus L. Toxicon 44, 927-932. Malbrouck, C., Trausch, G., Devos, P. & Kestemont, P. (2004b): Effect of microcystin-LR on protein phosphatase activity in fed and fasted juvenile goldfish Carassius auratus L. Toxicon 43, 295-301. Marionnet, D., Chambras, C., Taysse, L., Bosgireaud, C. & Deschaux, P. (1998): Modulation of drug- metabolizing systems by bacterial endotoxin in carp liver and immune organs. Ecotoxicology and Environmental Safety 41, 189-194. Marquardt, H. & Schäfer S. G. (2004): Lehrbuch der Toxikologie. Wissenschaftliche Verlagsgesellschaft, Stuttgart. Matsunaga, S., Moore, R. E., Niemczura, W. P. & Carmichael, W. W. (1989): Anatoxin-a(s), a potent anticholinesterase from Anabaena flos-aquae. Journal of the American Chemical Society 111, 8021-8023.

164 7. REFERENCES ______

Maynes, J. T., Luu, H. A., Cherney, M. M., Andersen, R. J., Williams, D., Holmes, C. F. & James, M. N. (2006): Crystal structures of protein phosphatase-1 bound to motuporin and dihydromicrocystin-LA: elucidation of the mechanism of enzyme inhibition by cyanobacterial toxins. Journal of Molecular Biology 356, 111-120. Mayr, C. (1998): Zum Einfluß von Trophie, Fischdichte und Habitatwahl auf die Nahrungs- und Wachstumsbedingungen von Renken (Coregonus lavaretus) in vier oberbayerischen Seen, pp. 221, PhD Thesis, Ludwig Maximilian Universität, München. Mayr, C. (2001): The influence of population density on growth of whitefish (Coregonus lavaretus L.) in four prealpine lakes. Limnologica 31, 53-60. Mazur-Marzec, H., Tyminska, A., Szafranek, J. & Plinski, M. (2007): Accumulation of nodularin in sediments, mussels, and fish from the Gulf of Gdansk, southern Baltic Sea. Environmental Toxicology 22, 101-111. Meier-Abt, F., Hammann-Hanni, A., Stieger, B., Ballatori, N. & Boyer, J. L. (2007): The organic anion transport polypeptide 1d1 (Oatp1d1) mediates hepatocellular uptake of phalloidin and microcystin into skate liver. Toxicology and Applied Pharmacology 218, 274-279. Meriluoto, J. (1997): Chromatography of microcystins. Analytica Chimica Acta 352, 277-298. Meriluoto, J. (2004): Detection methods for cyanobacterial toxins. 6th International Conference on Toxic Cyanobacteria, pp. 39. Metcalf, J. S., Beattie, K. A., Pflugmacher, S. & Codd, G. A. (2000): Immuno-crossreactivity and toxicity assessment of conjugation products of the cyanobacterial toxin, microcystin-LR. FEMS Microbiology Letters 189, 155-158. Metcalf, J. S. & Codd, G. A. (2005): Anatoxin-a. Acta Academiae Aboensis, Ser. B 65, 41-42. Mez, K. (1998): Erste Untersuchungen über toxische Cyanobakterien in Schweizer Mittelland- und Voralpenseen, pp. 39, Institut für Pflanzenbiologie / Mikrobiologie, Zürich. Mez, K., Beattie, K. A., Codd, G. A., Hanselmann, K., Hauser, B., Naegeli, H. & Preisig, H. R. (1997): Identification of a microcystin in benthic cyanobacteria linked to cattle deaths on alpine pastures in Switzerland. European Journal of Phycology 32, 111-117. Mezhoud, K., Praseuth, D., Puiseux-Dao, S., Francois, J. C., Bernard, C. & Edery, M. (in press): Global quantitative analysis of protein expression and phosphorylation status in the liver of the medaka fish (Oryzias latipes) exposed to microcystin-LR I. Balneation study. Aquatic Toxicology. Micheletti, S., Schanz, F. & Walsby, A. E. (1998): The daily integral of photosynthesis by Planktothrix rubescens during summer stratification and autumnal mixing in Lake Zürich. New Phytologist 139, 233-246. Mikhailov, A., Härmälä-Braskén, A. S., Hellman, J., Meriluoto, J. & Eriksson, J. E. (2003): Identification of ATP-synthase as a novel intracellular target for microcystin-LR. Chemico-Biological Interactions 142, 223-237. Mohamed, Z. A., Carmichael, W. W. & Hussein, A. A. (2003): Estimation of microcystin in the freshwater fish Oreochromis niloticus in an Egyptian fish farm containing a Microcystis bloom. Environmental Toxicology 18, 137-141. Mohamed, Z. A. & Hussein, A. A. (2006): Depuration of microcystins in tilapia fish exposed to natural populations of toxic cyanobacteria: a laboratory study. Ecotoxicology and Environmental Safety 63, 424-429. Monserat, J. M., Yunes, J. S. & Bianchini, A. (2001): Effects of Anabaena spiroides (cyanobacteria) aqueous extracts on the acetylcholinesterase activity of aquatic species. Environmental Toxicology and Chemistry 20, 1228-1235. Mookerji, N., Heller, C., Meng, H. J., Bürgi, H. R. & Müller, R. (1998): Diel and seasonal patterns of food intake and prey selection by Coregonus sp. in re-oligotrophicated Lake Lucerne, Switzerland. Journal of Fish Biology 52, 443-457. Moore, d. F. C. & Scott, W. E. (1985): Digestion of Microcystis aeruginosa by Oreochromis mossambicus. Journal of the Limnological Society of South Africa 11, 14-19. Morabito, G., Ruggiu, D. & Panzani, P. (2002): Recent dynamics (1995-1999) of the phytoplankton assemblages in Lago Maggiore as a basic tool for defining association patterns in the Italian deep lakes. Journal of Limnology 61, 129-145. Morscheid, H. & Morscheid, H. (2001): Ökosystemare Zusammenhänge am Beispiel des Ammersees, pp. 499. In M. Dokulil, A. Hamm & J. G. Kohl (Eds): Ökologie und Schutz von Seen, Facultas -Univ.-Verlag, Wien. Müller, R. (2003): Populationsdynamische Untersuchungen an den Felchen des Brienzersees, EAWAG, Forschungszentrum für Limnologie, Kastanienbaum. Müller, R. & Stadelmann, P. (2004): Fish habitat requirements as the basis for rehabilitation of eutrophic lakes by oxygenation. Fisheries Management and Ecology 11, 251-260. Mur, L. R. & Schreurs, H. (1995): Light as a selective factor in the distribution of the phytoplankton species. Water Science and Technology 32, 24-34. Mur, L. R., Skulberg, O. & Utkilen, H. (1999): Cyanobacteria in the environment, pp.15-40. In I. Chorus (Ed.): Toxic Cyanobacteria in Water, Springer, Berlin.

165 7. REFERENCES ______

Murch, S. J., Cox, P. A. & Banack, S. A. (2004): A mechanism for slow release of biomagnified cyanobacterial neurotoxins and neurodegenerative disease in Guam. Proceedings of the National Academy of Science USA 101, 12228-12231. Namikoshi, M., Murakami, T., Watanabe, M. F., Oda, T., Yamada, J., Tsujimura, S., Nagai, H. & Oishi, S. (2003): Simultaneous production of homoanatoxin-a, anatoxin-a, and a new non-toxic 4-hydroxyhomoanatoxin-a by the cyanobacterium Raphidiopsis mediterranea Skuja. Toxicon 42, 533-538. Nascimento, S. M. & Azevedo, S. M. O. (1999): Changes in cellular components in a cyanobacterium (Synechocystis aquatilis f. salina) subjected to different N/P ratios - An ecophysiological study. Environmental Toxicology 14, 37 - 44. Navratil, S., Palikova, M. & Vajkova, V. (1998): The effects of pure microcystin LR and biomass of blue-green algae on blood indices of carp (Cyprinus carpio L.). Acta Veterinaria Brunensis 67, 273-279. Negele, R. D., Braunbeck, T., Berbner, T., Schwaiger, J., Mallow, U., Ferling, H. & Ott, B. (2000): Licht und elektronenmikroskopische Untersuchungen an Fischlebern zur Beurteilung des Gesundheitszustandes von Renken aus dem Ammersee. Münchner Beiträge zur Abwasser-, Fischerei- und Flußbiologie 53, 245-281. Negri, A. P., Jones, G. & Hindmarsh, M. (1995): Sheep mortality associated with paralytic shellfish poisons from the cyanobacterium Anabaena circinales. Toxicon 33, 1321-1329. Nixdorf, B., Hemm, M., Hoffmann, A. & Richter, P. (2004): Dokumentation von Zustand und Entwicklung der wichtigsten Seen Deutschlands; Teil 11: Bayern, pp. 112. In Umweltbundesamt (Ed.): Dokumentation von Zustand und Entwicklung der wichtigsten Seen Deutschlands, Brandenburgische Technische Universität Cottbus, Lehrstuhl Gewässerschutz, Berlin. Nobre, A. C., Jorge, M. C., Menezes, D. B., Fonteles, M. C. & Monteiro, H. S. (1999): Effects of microcystin- LR in isolated perfused rat kidney. Brazilian Journal of Medical and Biological Research 32, 985-988. Nobre, A. C., Nunes-Monteiro, S. M., Monteiro, M. C., Martins, A. M., Havt, A., Barbosa, P. S., Lima, A. A. & Monteiro, H. S. (2004): Microcystin-LR promote intestinal secretion of water and electrolytes in rats. Toxicon 44, 555-559. Nogle, L. M., Okino, T. & Gerwick, W. H. (2001): Antillatoxin B, a neurotoxic lipopeptide from the marine cyanobacterium Lyngbya majuscula. Journal of Natural Products 64, 983-5. Oberemm, A. & Becker, J. (1997): Wirkungen cyanobakterieller Inhaltstoffe auf die Embryo- Larvalentwicklung von Fischen und Amphibien. Wasser Boden Luft 4/97, 116-119. Oberemm, A., Becker J., Codd G.A. & C., S. (1999): Effects of cyanobacterial toxins and aqueous crude extracts of cyanobacteria on the development of fish and amphibians. Environmental Toxicology 14, 77-88. Oberemm, A., Fastner, J. & Steinberg, C. (1997): Effects of microcystin-LR and cyanobacterial crude extracts on embryo-larval development of zebrafish (Danio rerio). Water Research 31, 2918-2921. Ochsenbein, U. & Mattmann, B. (2003): Gewässerbericht 1997-2000, pp. 112, Amt für Gewässerschutz und Abfallwirtschaft, Bern. Ohta, T., Nishiwaki, R., Yatsunami, J., Komori, A., Suganuma, M. & Fujiki, H. (1992): Hyperphosphorylation of cytokeratins 8 and 18 by microcystin-LR, a new liver tumor promoter, in primary cultured rat hepatocytes. Carcinogenesis 13, 2443-2447. Ohtani, I., Moore, R. E. & Runnegar, M. T. (1992): Cylindrospermopsin: A potent hepatotoxin from the blue- green alga Cylindrospermopsis raciborskii. Journal of the American Chemical Society 114, 7941-7942. Ojaveer, E., Simm, M., Balode, M., Purina, I. & Suursaar, U. (2003): Effect of Microcystis aeruginosa and Nodularia spumigena on survival of Eurytemora affinis and the embryonic and larval development of the Baltic herring Clupea harrengus membras. Environmental Toxicology 18, 236-242. Oliver, C. J. & Shenolikar, S. (1998): Physiological importance of protein phosphatase inhibitors. Frontiers in Bioscience 3, D 961-972. Oliver, R. L. & Ganf, G. G. (2000): Freshwater Blooms, pp. 149-194. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Olson, F. C. W. (1950): Quantitative estimates of filamentous algae. Transactions of the American Microscopical Society 59, 272-279. Onodera, H., Oshima, Y., Henriksen, P. & Yasumoto, T. (1997a): Confirmation of anatoxin-a(s), in the cyanobacterium Anabaena lemmermannii, as the cause of bird kills in Danish lakes. Toxicon 35, 1645-1648. Onodera, H., Satake, M., Oshima, Y., Yasumoto, T. & Carmichael, W. W. (1997b): New saxitoxin analogues from the freshwater filamentous cyanobacterium Lyngbya wollei. Journal of Natural Toxins 5, 146-151. Oren, A. (2000): Salts and brines, pp. 281-301. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Orjala, J. & Gerwick, W. H. (1996): Barbamide, a chlorinated metabolite with molluscicidal activity from the Caribbean cyanobacterium Lyngbya majuscula. Journal of Natural Products 59, 427-430.

166 7. REFERENCES ______

Orjala, J., Nagle, D. & Gerwick, W. H. (1995a): Malyngamide H, an ichthyotoxic amide possessing a new carbon skeleton from the Caribbean cyanobacterium Lyngbya majuscula. Journal of Natural Products 58, 764-768. Orjala, J., Nagle, D. G., Hsu, V. L. & Gerwick, W. H. (1995b): Antillatoxin: an exceptionally ichthyotoxic cyclic lipopeptide from the tropical cyanobacterium Lyngbya majuscula. Journal of the American Chemical Society 117, 8281-8282. Oshima, Y. (1995): Chemical and enzymatic transformation of shellfish toxins in marine organisms, pp. 475-480. In P. Lassus, G. Arzul, E. Erard-Le-Denn, P. Gentien & C. Marcaillou-Le-Baut (Eds): Harmful Marine Algal Blooms, Lavoisier, Paris. Osswald, J., Rellan, S., Carvalho, A. P., Gago, A. & Vasconcelos, V. (2007): Acute effects of an anatoxin-a producing cyanobacterium on juvenile fish-Cyprinus carpio L. Toxicon 49, 693-698. Palikova, M., Kavaru, F., Navratil, S., Kubala, L., Pesak, S. & Vajcova, V. (1998): The effects of pure microcystin LR and biomass of blue-green algae on selected immunological indices of carp (Cyprinus carpio L.) and silver carp (Hypophthalmichthys molitrix Val.). Acta Veterinaria Brunensis 67, 265-272. Palikova, M., Krejci, R., Hilscherova, K., Babica, P., Navratil, S., Kopp, R. & Blaha, L. (2007): Effect of different cyanobacterial biomasses and their fractions with variable microcystin content on embryonal development of carp (Cyprinus carpio L.). Aquatic Toxicology 81, 312-318. Palíkova, M., Navratil, S., Marsalek, B. & Blaha, L. (2003): Toxicity of crude extract of cyanobacteria for embryos and larvae of carp (Cyprinus carpio L.). Acta Veterinaria Brunensis 72, 437-443. Palikova, M., Navratil, S., Tichy, F., Marsalek, B. & Blaha, L. (2004): Histopathology of carp (Cyprinus carpio L.) larvae exposed to cyanobacteria extract. Acta Veterinaria Brunensis 73, 253-257. Papendorf, O., Konig, G. M., Wright, A. D., Chorus, I. & Oberemm, A. (1997): Mueggelone, a novel inhibitor of fish development from the fresh water cyanobacterium Aphanizomenon flos-aquae. Journal of Natural Products 60, 1298-1300. Penaloza, R., Rojas, M., Vila, I. & Zambrano, F. (1990): Toxicity of a soluble peptide from Microcystis sp. to zooplankton and fish. Freshwater Biology 24, 233-240. Pentecost, A. & Whitton, B. A. (2000): Limestones, pp. 257-279. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Penzlin, H. (2002): Lehrbuch der Tierphysiologie. Elsevier. München. Persson, P.-E., Sivonen, K., Keto, J., Kononen, K., Niemi, M. & Vilajamaa, H. (1984): Potentially toxic blue- green algae (cyanobacteria) in Finish natural waters. Acta fennica 14, 147-154. Pflugmacher, S., Wiegand, C., Oberemm, A., Beattie, K. A., Krause, E., Codd, G. A. & Steinberg, C. (1998): Identification of an enzymatically formed glutathione conjugate of the cyanobacterial hepatotoxin microcystin-LR: the first step of detoxication. Biochemica et Biophysica Acta, 527-533. Phillips, M. J., Roberts, R. J. & Stewart, J. A. (1985): The toxicity of the cyanobacterium Microcystis aeruginosa to rainbow trout, Salmo gairdneri Richardson. Journal of Fish Diseases 8, 339-344. Pichardo, S., Jos, A., Zurita, J. L., Salguero, M., Camean, A. M. & Repetto, G. (2005): The use of the fish cell lines RTG-2 and PLHC-1 to compare the toxic effects produced by microcystins LR & RR. Toxicology In Vitro 19, 865-873. Pichardo, S., Jos, A., Zurita, J. L., Salguero, M., Camean, A. M. & Repetto, G. (2007): Acute and subacute toxic effects produced by microcystin-YR on the fish cell lines RTG-2 and PLHC-1. Toxicology In Vitro 21, 1460-1467. Pietsch, C., Wiegand, C., Amé, M. V., Nicklisch, A., Wunderlin, D. & Pflugmacher, S. (2001): The effect of a cyanobacterial crude extract on different aquatic organisms: Evidence for cyanobacterial toxin modulating factors. Environmental Toxicology 16, 535-542. Pittman, S. J. & Pittman, K. M. (2005): Short term consequences of benthic cyanobacterial bloom (Lyngbya majuscula Gomont) for fish and penaeid prawns in Moreton Bay (Queensland, Australia). Estuarine Coastal & Shellfish Science 63, 619-632. Pollux, B. J. A. & Pollux, P. M. J. (2004): Fish and waterfowl mortality at Romeinenweerd due to cyanobacterial bloom. Natuurhistorisch Maandblad 93, 207-209. Pomati, F., Sacchi, S., Rossetti, C., Giovannardi, S., Onodera, H., Oshima, Y. & Neiland, B. A. (2000): The freshwater cyanobacterium Planktothrix sp. FP1: molecular identification and detection of paralytic shellfish poisoning toxins. Journal of Phycology 36, 553-562. Pouria, S., de Andrade, A., Barbosa, J., Cavalcanti, R., Barreto, V., Ward, C., Preiser, W., Poon, G., Neild, G. & Codd, G. (1998): Fatal microcystin intoxication in haemodialysis unit in Caruaru, Brazil. Lancet 352, 21-26. Prakash, A., Medcof, J. C. & Tennant, A. D. (1971): Paralytic shellfish poisoning in Eastern Canada. Fisheries Research Board of Canada. Bulletin 177, 87. Prepas, E. E., Kotak, B. G., Campbell, L. M., Evans, J. C., Hrudey, S. E. & Holmes, C. F. B. (1997): Accumulation and elimination of cyanobacterial hepatotoxins by the freshwater clam Anodonta grandis simpsoniana. Canadian Journal of Fishery and Aquatic Science 54, 41-46. Prieto, A. I., Jos, A., Pichardo, S., Moreno, I. & Camean, A. M. (2006): Differential oxidative stress responses to microcystins LR and RR in intraperitoneally exposed tilapia fish (Oreochromis sp.). Aquatic Toxicology 77, 314-321. 167 7. REFERENCES ______

Qiu, T., Xie, P., Ke, Z., Li, L. & Guo, L. (2007): In situ studies on physiological and biochemical responses of four fishes with different trophic levels to toxic cyanobacterial blooms in a large Chinese lake. Toxicon 50, 365-376. Råbergh, C. M. I., Bylund, G. & Eriksson, J. E. (1991): Histopathological effects of microcystin-LR, a cyclic peptide toxin from the cyanobacterium (blue-green alga) Microcystis aeruginosa, on common carp (Cyprinus carpio L.). Aquatic Toxicology 20, 131-146. Rapala, J., Berg, K. A., Lyra, C., Niemi, R. M., Manz, W., Suomalainen, S., Paulin, L. & Lahti, K. (2005): Paucibacter toxinivorans gen. nov., sp. nov., a bacterium that degrades cyclic cyanobacterial hepatotoxins microcystins and nodularin. International Journal of Systematic and Evolutionary Microbiology 55, 1563-1568. Rapala, J., Erkomaa, K., Kukkonen, J., Sivonen, K. & Lathi, K. (2002): Detection of microcystins with protein phosphatase inhibition assay, high performance liquid chromatography-UV detection and enzyme-linked immunosorbant assay: Comparison of methods. Analytica Chimica Acta 466, 213-231. Rauch, J., Schlamp, A., Gastl, P. & Ernst, A. (1961-1999): Protokollbuch der Fischereigenossenschaft Ammersee. Germany. Reimann, K. (1955): Bericht über die Untersuchung des Ammersees im Jahr 1955 (unpublished), Bayerische Biologische Versuchsanstalt. Cited in Steinberg, 1980. Reshetnikov, Y. S., Paranyshkina, L. P. & Kiyashko, V. I. (1970): Seasonal changes of blood serum protein composition and fat content in whitefishes. Voprosy Ikhtiologii. 10, 804-815. Rinehart, K. L., Harada, K., Namikoshi, M., Chen, C., Harvis, C. A., Munro, M. H. G., Blunt, J. W., Mulligan, P. E., Beasly, V. R., Dahlem, A. M. & Carmichael, W. W. (1988): Nodularin, microcystin and the configuration of Adda. Journal of the American Chemical Society 110, 8557-8558. Rippka, R., Deruelles, J., Waterbury, J. B., Herdman, M. & Stanier, R. Y. (1979): Generic assignments, strain histories and properties of pure cultures of cyanobacteria. Journal of General Microbiology 111, 1-61. Robinson, N. A., Miura, G. A., Matson, C. F., Dinterman, R. E. & Pace, J. G. (1989): Characterisation of chemical tritiated microcystin-LR and its distribution in mice. Toxicon 27, 1035-1042. Robinson, N. A., Pace, J. G., Matson, C. F., Miura, G. A. & Lawrence, W. B. (1991): Tissue distribution, excretion and hepatic biotransformation of microcystin-LR in mice. Journal of Pharmacology and Experimental Therapeutics 256, 176-182. Rodger, H. D., Turnbull, T., Edwards, C. & Codd, G. A. (1994): Cyanobacterial (blue-green-algal) bloom associated pathology in brown trout, Salmo trutta L., in Loch Leven, Scotland. Journal of Fish Diseases 17, 177-181. Rohrlack, T., Christoffersen, K., Hansen, P. E., Zhang, W., Czarnecki, O., Henning, M., Fastner, J., Erhard, M., Neilan, B. A. & Kaebernick, M. (2003): Isolation, characterization, and quantitative analysis of Microviridin J, a new Microcystis metabolite toxic to Daphnia. Journal of Chemical Ecology 29, 1757-1770. Rose, E. T. (1953): Toxic algae in Iowa lakes. Proceedings of the Iowa Academy of Science 60, 738-745. Rucker, J., Stuken, A., Nixdorf, B., Fastner, J., Chorus, I. & Wiedner, C. (2007): Concentrations of particulate and dissolved cylindrospermopsin in 21 Aphanizomenon-dominated temperate lakes. Toxicon 50, 800-809. Runnegar, M. T., Gerdes, R. G. & Falconer, I. R. (1991): The uptake of the cyanobacterial hepatotoxin microcystin by isolated rat hepatocytes. Toxicon 29, 43-51. Rymuszka, A., Sieroslawska, A., Bownik, A. & Skowronski, T. (2007): In vitro effects of pure microcystin-LR on the lymphocyte proliferation in rainbow trout (Oncorhynchus mykiss). Fish and Shellfish Immunology 22, 289-292. Sahin, A., Tencalla, F. G., Dietrich, D. R., Mez, K. & Naegeli, H. (1995): Enzymatic analysis of liver samples from rainbow trout for diagnosis of blue-green algae-induced toxicosis. American Journal of Veterinary Research 56, 1110-1116. Sahin, A., Tencalla, F. G., Dietrich, D. R. & Naegeli, H. (1996): Biliary excretion of biochemically active cyanobacteria (blue-green algae) hepatotoxins in fish. Toxicology 106, 123-130. Saker, M. L. & Eaglesham, G. K. (1999): The accumulation of cylindrospermopsin from the cyanobacterium Cylindrospermopsis raciborskii in tissues of the Redclaw crayfish Cherax quadricarinatus. Toxicon 37, 1065-1077. Saker, M. L., Metcalf, J. S., Codd, G. A. & Eaglesham, G. K. (2004): Accumulation and depuration of the cyanobacterial toxin cylindrospermopsin in the freshwater mussel Anodonta cygnea. Toxicon 43, 185-194. Salierno, J. D., Snyder, N. S., Murphy, A. Z., Poli, M., Hall, S., Baden, D. & Kane, A. S. (2006): Harmful algal bloom toxins alter c-fos protein expression in the brain of killifish, Fundulus heteroclitus. Aquatic Toxicology 78, 350-357. Salmaso, N. (2000): Factors affecting the seasonality and distribution of cyanobacteria and chlorophytes: A case study from the large lakes south of the Alps with special reference to Lake Garda. Hydrobiologia 438, 43-63. Salmaso, N. (2002): Ecological patterns of phytoplankton assemblages in lake Garda: seasonal, spatial and historical features. Journal of Limnology 61, 95-115. 168 7. REFERENCES ______

Sano, T. & Kaya, K. (1995): Oscillamide Y, A chymotrypsin inhibitor from toxic Oscillatoria agardhii. Tetrahedron Letters 36, 5933-5936. Sano, T. & Kaya, K. (1996): Oscillatorin, a chymotrypsin inhibitor from the cyanobacterium Oscillatoria agardhii. Tetrahedron Letters 37, 6873-6876. Sawyer, P. J., Gentile, J. H. & Sasner, J. J. (1968): Demonstration of a toxin from Aphanizomenon flos-aquae (L.) Ralfs. Canadian Journal of Microbiology 14, 1199-1204. Schlenk, D. & Di Giulio, R. T. (2002): Biochemical response as indicator of aquatic ecosystem health, pp. 13-42. In M. Adams (Ed.): Biological Indicators of Aquatic Ecosystem Stress, American Fisheries Society, Bethesda, Maryland. Schmid, D., Ernst, B., Hoeger, S. J. & Dietrich, D. R. (2004): Characterisation and differentiation of two microcystin from Planktothrix spec. isolated from a pre-alpine lake in Europe. 6th International Conference on Toxic Cyanobacteria, pp. 79. Schopf, J. W. (2000): The fossil record: tracing the roots of the cyanobacterial lineage, pp. 13-35. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer, Dordrecht. Schopf, W. & Packer, B. (1987): Early archean (3.3 billion to 3.5 billion - year old) microfossils from Warrawoona Group, Australia. Science 237, 70-72. Schulz, L., Reichmann, M., Ambros, M., Rauter, A., Fresner, R., Troyer-Mildner, J., Wirkner, W., Grund, E., Mairitsch, M., R., L., B., S., Loderer, A., Wieser, G. & Sampl, H. (2000 -2007): Käntner Seenberichte: Veröffentlichungen des Kärntner Instituts für Seenforschung 15-21, Käntner Institut für Seenforschung, Klagenfurt. Schwimmer, D. & Schwimmer, M. (1968): Medical aspects of phycology. In D. Jackson (Ed.): Algae, Man and the Environment, Syracuse University Press, Syracuse. Sephton, D. H., Haya, K., Martin, J. L., LeGresley, M. M. & Page, F. H. (2007): Paralytic shellfish toxins in zooplankton, mussels, lobsters and caged Atlantic salmon, Salmo salar, during a bloom of Alexandrium fundyense of Grand Manan Island, in the Bay of Fundy. Harmful Algae 6, 745-758. Sevrin-Reyssac, J. & Pleticosic, M. (1990): Cyanobacteria in fish ponds. Aquaculture 88, 1-20. Seydel, E. (1913): Fischsterben durch Wasserblüte. Mitteilungen des Fischereivereins für die Provinz Brandenburg 5, 87-91. Sieracki, M. E. & Wah Wong, M. (1999): Image Cytometry: Fluorescence images of Georges Bank nanoplankton (http://www.bigelow.org/cytometry/Image_gallery/ImageCyto.html) Sieroslawska, A., Rymuszka, A., Bownik, A. & Skowronski, T. (2007): The influence of microcystin-LR on fish phagocytic cells. Human & Experimental Toxicology 26, 603-607. Singh, S., Kate, B. N. & Banerjee, U. C. (2005): Bioactive compounds from cyanobacteria and microalgae: an overview. Critical Reviews in Biotechnology 25, 73-95. Sipiä, V., Kankaanpää, H., Lathi, K., Carmichael, W. W. & Meriluoto, J. (2001a): Detection of nodularin in flounders and cod from the Baltic Sea. Environmental Toxicology 16, 121-126. Sipiä, V. O., Kankaanpää, H. T., Flinkmann, J., Lahti, K. & Meriluoto, J. A. O. (2001b): Time-dependent accumulation of cyanobacterial hepatotoxins in flounders (Platichthys flesus) and mussels (Mytilus edulis) from the Northern Baltic Sea. Environmental Toxicology 16, 330-336. Sipiä, V. O., Kankaanpää, H. T., Pflugmacher, S., Flinkman, J., Furey, A. & James, K. J. (2002): Bioaccumulation and detoxication of nodularin in tissues of flounder (Platichthys flesus), mussels (Mytilus edulis, Dreissena polymorpha), and clams (Macoma balthica) from the northern Baltic Sea. Ecotoxicology and Environmental Safety 53, 305-311. Sipiä, V. O., Sjovall, O., Valtonen, T., Barnaby, D. L., Codd, G. A., Metcalf, J. S., Kilpi, M., Mustonen, O. & Meriluoto, J. A. (2006): Analysis of nodularin-R in eider (Somateria mollissima), roach (Rutilus rutilus L.), and flounder (Platichthys flesus L.) liver and muscle samples from the western Gulf of Finland, northern Baltic Sea. Environmental Toxicology & Chemistry 25, 2834-2839. Sivonen, K. & Jones, G. (1999): Cyanobacterial Toxins, pp. 41-111. In I. Chorus & J. Bartram (Eds): Toxic Cyanobacteria in Water, E & FN Spon, London. Sivonen, K., Niemelä, S. I., Niemi, R. M., Lepistö, L., Luoma, T. H. & Räsänen, L. A. (1990): Toxic cyanobacteria (blue-green) algae in Finnish fresh and coastal waters. Hydrobiologia 190, 267-275. Skulberg, O. (1964): Algal problems related to the eutrophication of European water supplies, and a bioassay method to assess fertilizing influence of polluted waters., pp. 262-299. In D. F. Jackson (Ed.): Algae and Man, Plenum Press, New York. Skulberg, O. & Skulberg, R. (1985): Planktic species of Oscillatoria (Cyanophyceae) from Norway. Archiv für Hydrobiologie Suppl. 71, 157-174. Skulberg, O. M. (1984): Toxic blue green algal blooms in Europe: a growing problem. Ambio 13, 397-406. Skulberg, O. M. (1994): Osciallatoialean cyanoprokaryotes and their application for algal culture technology. Archiv für Hydrobiologie Suppl. 105 Algological Studies 75, 265-278. Skulberg, O. M., Carmichael, W. W., Andersen, R. A., Matsunuga, S., Moore, R. E. & Skulberg, R. (1992): Investigations of a neurotoxic oscillatorialean strain (Cyanophyceae) and its toxin. Isolation and characterisation of homoanatoxin-a. Environmental Toxicology and Chemistry 11, 321-329. Skurdal, J., Hessen, D. O. & Berge, D. (1985): Food selection and vertical distribution of pelagic whitefish Coregonus lavaretus (L.) in Lake Tyrifjorden, Norway. Fauna Norvegica Series 6, 18-23.

169 7. REFERENCES ______

Smith, J. L. & Haney, J. F. (2006): Foodweb transfer, accumulation, and depuration of microcystins, a cyanobacterial toxin, in pumpkinseed sunfish (Lepomis gibbosus). Toxicon 48, 580-589. Snyder, G. S., Goodwin, A. E. & Freeman, D. W. (2002): Evidence that channel catfish, Ictalurus punctatus (Rafinesque), mortality is not linked to ingestion of the hepatotoxin microcystin-LR. Journal of Fish Disease 25, 275-285. Soares, R. M., Magalhaes, V. F. & Azevedo, S. M. F. O. (2004): Accumulation and depuration of microcystins (cyanobacteria hepatotoxins) in Tilapia rendalli (Cichlidae) under laboratory conditions. Aquatic Toxicology 70, 1-10. Spencer, P. S., Nunn, P. B., Hugon, J., Ludolph, A. C., Ross, S. M., Roy, D. N. & Robertson, R. C. (1987): Guam amyotrophic lateral sclerosis-parkinsonism-dementia linked to a plant excitant neurotoxin. Science 237, 517-522. Spoof, L. (2004): High performance liquid chromatography of microcystins and nodularins, cyanobacterial peptide toxins, pp. 73: Department of Biochemistry and Pharmacy, PhD Thesis, Abo Akademi Univeristy, Turku-Abo. Spoof, L., Berg, K. A., Rapala, J., Lahti, K., Lepisto, L., Metcalf, J. S., Codd, G. A. & Meriluoto, J. (2006): First observation of cylindrospermopsin in Anabaena lapponica isolated from the boreal environment (Finland). Environmental Toxicology 21, 552-560. Spoof, L., Klimova, S., Mikahailov, A., Eriksson, J. E. & Meriluoto, J. (2003a): Synthesis and organotropism of 3H-dehydro derivates of the cyanobacterial peptide hepatotoxin nodularin. Toxicon 41, 153-162. Spoof, L., Vesterkvist, P., Lindholm, T. & Meriluoto, J. (2003b): Screening for cyanobacterial hepatotoxins, microcystins and nodularin in environmental water samples by reversed-phase liquid chromatography-electrospray ionisation mass spectrometry. Journal of Chromatography A 1020, 105-119. Stal, L. J. (2000): Cyanobacterial mats and stromatolites, pp. 61-120. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Steinberg, C. & Lenhart, B. (1991): Zur Trophieentwicklung des Ammersees mit besonderer Berücksichtigung der Trophieanzeige durch Cyanobakterien, pp. 89-106. In Rundengespräche der Kommission für Ökologie: Ökologie der oberbayerischen Seen, München. Steinberg, C. E. (1980): Nutrient enrichment in a subalpine lake: its degree an effects on the phytoplankton of Lake Ammersee. Gewässer und Abwässer 66/67, 175-187. Stephens, E. L. (1948): Microcystis toxica sp. nov., a poisonous alga from the Transvaal and Orange Free State. Hydrobiologia 1, 14. Stryer, L. (1996): Biosynthese der Bausteine, pp. 721-823. In L. Stryer (Ed): Biochemie, Spektrum Akademischer Verlag, Heidelberg. Suda, S., Watanabe, M. M., Otsuka, S., Mahakahant, A., Yongmanitchai, W., Nopartnaraporn, N., Liu, Y. & Day, J. D. (2002): Taxonomic revision of water-bloom forming species of oscillatorioid cyanobacteria. International Journal of Systematic and Evolutionary Microbiology 52, 1577-1595. Sugaya, Y., Yasuno, M. & Yanai, T. (1990): Effects of toxic Microcystis viridis and isolated toxins on goldfish. Japanese Journal of Limnology 51, 149-153. Svobodová, Z., Kaláb, P., Dusek, L., Vykosová, B., Kolárová, J. & Janousková, D. (1999): The effect of handling and transport on the concentration of glucose and cortisol in blood plasma of common carp. Acta Veterinaria Brunensis 68, 265-274. Takenaka, S. (2001): Covalent glutathione conjugation to cyanobacterial hepatotoxin microcystin LR by F344 rat cytosolic and microsomal glutathione S-transferases. Environmental Toxicology and Pharmacology 9, 135-139. Tan, L. T., Okino, T. & Gerwick, W. H. (2000): Hermitamides A and B, toxic malyngamide-type natural products from the marine cyanobacterium Lyngbya majuscula. Journal of Natural Products 63, 952-955. Tandeau de Marsac, N. (1977): Occurrence and nature of chromatic adaptation in cyanobacteria. Journal of Bacteriology 130, 82-91. Teixeira, M., Nascimento Costa, M., Lucia Pires de Carvalho, V., Santos Pereira, M. & Hage, E. (1993): Gastroenteritis epidemic in the area of the Itaparica Dam, Bahia, Brazil. Bulletin of the Pan American Health Organization 27, 244-253. Tencalla, F. (1995): Toxicity of cyanobacterial peptide toxins to fish, PhD Thesis, Swiss Federal Institute of Technology, Zürich. Tencalla, F. & Dietrich, D. (1997): Biochemical characterization of microcystin toxicity in trout (Oncorhynchus mykiss). Toxicon 35, 583-595. Tencalla, F. G., Dietrich, D. R. & Schlatter, C. (1994): Toxicity of Microcystis aeruginosa peptide toxin to yearling rainbow trout (Oncorhynchus mykiss). Aquatic Toxicology 30, 215-224. Teneva, I., Asparuhova, D., Dzhambazov, B., Mladenov, R. & Schirmer, K. (2003): The freshwater cyanobacterium Lyngbya aerugineo-coerulea produces compounds toxic to mice and to mammalian and fish cells. Environmental Toxicology 18, 9-20. Teubner, K., Morscheid, H., Tolloti, M., Greisberger, S., Morscheid, H. & Kucklentz, V. (2004): Bedingungen für Auftreten toxinbildender Balualgen (Cyanobakterien) in bayerischen Seen und anderen stehenden Gewässern, pp. 105, Bayerisches Landesamt für Wasserwirtschaft, München. 170 7. REFERENCES ______

Toranzo, A. E., Nieto, F. & Barja, J. L. (1990): Mortality associated with cyanobacterial bloom in farmed rainbow trout in Galicia (Northwestern, Spain). Bulletin of the European Association of Fish Pathology 10, 106-107. Tu, A. T. (1991): Reptile Venoms and Toxins. Dekker. New York. Turner, P. C., Gammie, A. J., Hollinrake, K. & Codd, G. A. (1990): Pneumonia associated with contact with cyanobacteria. British Medicine Journal 300, 1440-1441. Utkilen, H., Skulberg, O., Skulberg, R., Gjolme, N. & Underdal, B. (2001): Toxic cyanobacterial blooms of inland waters in Southern Norway 1978-1998, pp. 46-49. In I. Chorus (Ed.): Cyanotoxins: Occurrence, Causes, Consequences, Springer Verlag, Berlin. Vajcová, V., Navrátil, S. & Palíková, M. (1998): The effect of intraperitoneally applied pure microcystin LR on haematological, biochemical and morphological indices of silver carp (Hypophthalmichthys molitrix Val.). Acta Veterinaria Brunensis 67, 281-287. Valtonen, T. (1974): Seasonal and sex-bound variation in the carbohydrate metabolism of the liver of whitefish. Comparative Biochemistry and Physiology 47, 713-722. van Apeldoorn, M. E., van Egmond, H. P., Speijers, G. J. & Bakker, G. J. (2007): Toxins of cyanobacteria. Molecular Nutrition & Food Research 51, 7-60. van Buynder, P. G., Oughtred, T., Kirkby, B., Phillips, S., Eaglesham, G., Thomas, K. & Burch, M. (2001): Nodularin uptake by seafood during a cyanobacterial bloom. Environmental Toxicology 16, 468-471. van den Hoek, C. & Jahns, H. M. (2002): Algae - an introduction to phycology. University Press. Cambridge. Vaucher, J. P. (1803): Historie des Conferves déau douce, Geneva. Vega, A. & Bell, E. A. (1967): α-Amino-β-methylaminopropionic acid, a new amino acid from seeds of Cycas circinalis. Phytochemistry 6, 759-762. Vesterkvist, P. S. M. & Meriluoto, J. A. O. (2003): Interaction between microcystins of different hydrophobicities and lipid monolayers. Toxicon 41, 349-355. Viaggiu, E., Melchiorre, S., Volpi, F., Corcia Di, A., Manchini, R., Garibaldi, L., Circhigno, G. & Bruno, M. (2004): Anatoxin-a toxin in the cyanobacterium Planktothrix rubescens from a fishing pond in northern Italy. Environmental Toxicology 19, 191-197. Vincent, W. F. (2000): Cyanobacterial dominance in the polar regions, pp. 321-340. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Vollenweider, R. A. (1976): Advances in defining critical loading levels for phosphorus in lakes eutrophication. Annali Dell Istituto Superiore di Sanita 35, 53. Walsby, A. E. & Avery, A. (1996): Measurement of filamentous cyanobacteria by image analysis. Journal of Microbiological Methods 26, 11-20. Walsby, A. E., Avery, A. & Schanz, F. (1998): The critical pressures of gas vesicles in Planktothrix rubescens in relation to the depth of winter mixing in Lake Zürich, Switzerland. Journal of Plankton Research 20, 1357-1375. Walsby, A. E., Dubinsky, Z., Kromkamp, J. C., Lehman, C. & Schanz, F. (2001): The effect of diel changes in photosynthetic coefficients and depth of Planktothrix rubescens on the daily integral of photosynthesis in Lake Zürich. Aquatic Science 63, 326-349. Walsby, A. E. & Schanz, F. (2002): Light-dependent growth rate determines changes in the population of Planktothrix rubescens over the annual cycle in Lake Zürich, Switzerland. New Phytologist 154, 671-687. Wang, L., Liang, X. F., Liao, W. Q., Lei, L. M. & Han, B. P. (2006): Structural and functional characterization of microcystin detoxification-related liver genes in a phytoplanktivorous fish, Nile tilapia (Oreochromis niloticus). Comparative Biochemistry and Physiology Part C: Toxicology and Pharmacology 144, 216-227. Wang, P. J., Chien, M. S., Wu, F. J., Chou, H. N. & Lee, S. J. (2005): Inhibition of embryonic development by microcystin-LR in zebrafish, Danio rerio. Toxicon 45, 303-308. Ward, D. M. & Castenholz, R. W. (2000): Cyanobacteria in geothermal habitats, pp. 37-59. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Wehrli, B. & Wüest, A. (1996): Zehn Jahre Seenbelüftung: Erfahrungen und Optionen, pp. 128, Eidgenössische Anstalt für Wasserversorgung, Abwasserreinigung und Gewässerschutz, Zürich. Weise, G., Drews, G., Jann, B. & Jann, K. (1970): Identification and analysis of a lipopolysaccharide in cell walls of the blue-green alga Anacystis nidulans. Archiv für Mikrobiologie 71, 89-98. Weiss, J. H., Koh, J. Y. & Choi, D. W. (1989): Neurotoxicity of beta-N-methylamino-L-alanine (BMAA) and beta-N-oxalylamino-L-alanine (BOAA) on cultured cortical neurons. Brain Research 497, 64-71. Welker, M. & Döhren, H. v. (2006): Cyanobacterial peptides - Nature's own combinatorial biosynthesis. FEMS Microbiology Reviews 30, 530-563. Welker, M., Steinberg, C. E. & Jones, G. (2001): Release and persistence of microcystin in natural waters, pp. 83-101. In I. Chorus (Ed.): Cyanotoxins: Occurrence, Causes, Consequences, Springer, Berlin. White, A. W. (1984): Paralytic shellfish toxins and finfish, pp. 257-269. In E. P. Ragelis (Ed.): Seafood Toxins, American Chemical Society Symposium Series, Washington.

171 7. REFERENCES ______

White, S. H., Duivenvoorden, L. J., Fabbro, L. D. & Eaglesham, G. K. (2006): Influence of intracellular toxin concentrations on cylindrospermopsin bioaccumulation in a freshwater gastropod (Melanoides tuberculata). Toxicon 47, 497-509. White, S. H., Duivenvoorden, L. J., Fabbro, L. D. & Eaglesham, G. K. (2007): Mortality and toxin bioaccumulation in Bufo marinus following exposure to Cylindrospermopsis raciborskii cell extracts and live cultures. Environmental Pollution 147, 158-167. Whitton, B. A. (2000): Soils and rice-fields, pp. 233-255. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Whitton, B. A. & Potts, M. (2000): Introduction to the cyanobacteria, pp. 1-11: The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Wickstrom, M. L., Khan, S. A., Haschek, W. M., Wyman, J. F., Eriksson, J. E., Schaeffer, D. J. & Beasley, V. R. (1995): Alterations in microtubules, intermediate filaments, and microfilaments induced by microcystin-LR in cultured cells. Toxicological Pathology 23, 326-337. Wiegand, C. & Pflugmacher, S. (2005): Ecotoxicological effects of selected cyanobacterial secondary metabolites a short review. Toxicology and Applied Pharmacology 203, 201-218. Wiegand, C., Pflugmacher, S., Oberemm, A., Meems, N., Beattie, K. A., Steinberg, C. E. W. & Codd, G. A. (1999): Uptake and effects of microcystin-LR on detoxication enzymes of early life stages of the zebra fish (Danio rerio). Environmental Toxicology 14, 89-95. Williams, D. E., Craig, M., Dawe, S. C., Kent, M. L., Andersen, R. J. & Holmes, C. F. B. (1997a): 14C-labelded microcystin-LR administered to Atlantic salmon via intraperitoneal injection provides in vivo evidence for covalent binding of microcystin-LR in salmon livers. Toxicon 35, 985-989. Williams, D. E., Craig, M., Dawe, S. C., Kent, M. L., Holmes, C. F. B. & Andersen, R. J. (1997b): Evidence for a covalently bound form of microcystin-LR in salmon liver and Dungeness crab larvae. Chemical Research in Toxicology 10, 463-469. Williams, D. E., Dawe, S. C., Kent, M. L., Andersen, R. J., Craig, M. & Holmes, C. F. B. (1997c): Bioaccumulation and clearance of Microcystis from salt water mussels, Mytilus edulis, and in vivo evidence for covalently bound microcystin in mussel tissues. Toxicon 35, 1617-1625. Williams, D. E., Kent, M. L., Andersen, R. J., Klix, H. & Holmes, C. F. B. (1995): Tissue distribution and clearance of tritium-labelled dihydromicrocystin-LR epimers administered to Atlantic salmon via intraperitoneal injection. Toxicon 33, 125-131. Wißmath, P. (2000): Die Entwicklung der Renkenbestände im Ammersee und die Auswirkungen auf die fischereiliche Bewirtschaftung. Münchner Beiträge zur Abwasser-, Fischerei- und Flußbiologie 53, 399-401. Wißmath, P. (2004): Etwas über die Renkenfischerei im Ammersee. Fischer & Teichwirt 8, 769-771. Wißmath, P., Limburg, U., Wunner, U. & Huber, B. (1992): Drastische Einbrüche in der Entwicklung des Ammersee-Renkenbestandes: Bewirtschaftungsfehler, Nahrungsmangel oder andere Ursachen? Fischer & Teichwirt 5, 158-163. Wißmath, P., Limburg, U., Wunner, U. & Huber, B. (1993): Das Ammersee-Syndrom. Fischer & Teichwirt 3, 78-87. Wood, S. A., Selwood, A. I., Rueckert, A., Holland, P. T., Milne, J. R., Smith, K. F., Smits, B., Watts, L. F. & Cary, C. S. (2007): First report of homoanatoxin-a and associated dog neurotoxicosis in New Zealand. Toxicon 50, 292-301. Wright, A. D., Papendorf, O., Konig, G. M. & Oberemm, A. (2006): Effects of cyanobacterium Fischerella ambigua isolates and cell free culture media on zebrafish (Danio rerio) embryo development. Chemosphere 65, 604-608. Wynn-Williams, D. D. (2000): Cyanobacteria in deserts - life at the limit? pp. 341-366. In B. A. Whitton & M. Potts (Eds): The Ecology of Cyanobacteria: Their Diversity in Time and Space, Kluwer Academic Publishers, Dordrecht. Xie, L., Xie, P., Guo, L., Li, L., Miyabara, Y. & Park, H. D. (2005): Organ distribution and bioaccumulation of microcystins in freshwater fish at different trophic levels from the eutrophic Lake Chaohu, China. Environmental Toxicology 20, 293-300. Xie, L., Xie, P., Ozawa, K., Honma, T., Yokoyama, A. & Park, H. D. (2004): Dynamics of microcystin-LR and - RR in the phytoplanktivorous silver carp in a subchronic toxicity experiment. Environmental Pollution 127, 431-439. Xu, L., Lam, P. K. S., Chen, J., Zhang, Y. & Harada, K. (2000): Comparative study on in vitro inhibition of grass carp (Ctenopharyngodon idellus) and mouse protein phosphatases by microcystins. Environmental Toxicology 15, 71-75. Yoshida, T., Makita, Y., Satsuma, T., Nagata, S., T-shirt, F., Yoshida, F., Sequim, M., S.I., T., Harada, T., Mata, K. & Ueno, Y. (1998): Immunohistochemical localisation of microcystin-LR in the liver of mice: A study on the pathogenesis of microcystin-LR induced hepatotoxicity. Environmental Toxicologic Pathology 26, 411-418. Yoshida, T., Makita, Y., Tsutsumi, T., Yoshida, F., Sekijima, M., Tamura, S. & Ueno, Y. (1997): Acute oral toxicity of microcystin-LR, a cyanobacterial hepatotoxin, in mice. Natural Toxins 5, 91-95. Yoshizawa, S., Matsushima, R., Watanabe, M. F., Harada, K., Ichihara, A., Carmichael, W. W. & Fujiki, H. (1990): Inhibition of protein phosphatases by microcystins and nodularin associated with hepatotoxicity. Journal of Cancer Research and Clinical Oncology 116, 609-614. 172 7. REFERENCES ______

Yu, S.-Z. (1995): Primary prevention of hepatocellular carcinoma. Journal of Gastroenterology and Hepatology 10, 674-682. Yunes, J. S., Matthiensen, A., Pariese, M., Salomon, P. S., Caggett, S. L., Beattie, K. A. & Codd, G. A. (1998): Microcystis aeruginosa growth stages and the occurrence of microcystins in Patos Lagoon, Southern Brazil, pp. 18-21. In B. Reguera, J. Blanco, M. L. Fernández & T. Wyatt (Eds): Harmful Algae, IOC, Paris. Zambrano, F. & Canelo, E. (1996): Effects of Microcystin-LR on the partial reactions of the Na+ -K+ pump of the gill of carp (Cyprinus carpio Linneo). Toxicon 34, 451-458. Zhang, H., Zhang, J., Chen, Y. & Zhu, Y. (2007): Influence of intracellular Ca(2+), mitochondria membrane potential, reactive oxygen species, and intracellular ATP on the mechanism of microcystin-LR induced apoptosis in Carassius auratus lymphocytes in vitro. Environmental Toxicology 22, 559-564. Zhang, J., Zhang, H. & Chen, Y. (2006): Sensitive apoptosis induced by microcystins in the crucian carp (Carassius auratus) lymphocytes in vitro. Toxicology In Vitro 20, 560-566. Zhang, Y., Xie, P., Wang, W., Li, D., Li, L., Tang, R., Lei, H. & Shi, Z. (in press): Dose-dependent effects of extracted microcystins on embryonic development, larval growth and histopathological changes of southern catfish (Silurus meridionalis). Toxicon. Zhao, M., Xie, S., Zhu, X., yang, Y., Gan, L. & Song, L. (2006): Effect of inclusion of blue-green algae meal on growth and accumulation of microcystins in gibel carp (Carassius auratus gibelio). Journal of Applied Ichthyology 22, 72-78. Zhou, L., Yu, H. & Chen, K. (2002): Relationship between microcystin in drinking water and colorectal cancer. Biomedical & Environmental Sciences 15, 166-171. Zimba, P. V., Khoo, L., Gaunt, P. S., Brittain, S. & Carmichael, W. W. (2001): Confirmation of catfish, Ictalurus punctatus (Rafinesque), mortality from Microcystis toxins. Journal of Fish Disease 24, 41-47. Zohary, T. & Madeira, A. M. P. (1990): Structural, physical and chemical characteristics of Microcystis aeruginosa hyperscums from a hypertrophic lake. Freshwater Biology 23, 339-352. Zotina, T., Köster, O. & Jüttner, F. (2003): Photoheterotrophy and light-dependent uptake of organic and organic nitrogenous compounds by Planktothrix rubescens under low irradiance. Freshwater Biology 48, 1859-1872.

173

8. APPENDIX

174

ERKLÄRUNG

Die vorliegende Arbeit wurde ohne unzulässige Hilfe Dritter und ohne Benutzung anderer als der angegebenen Hilfsmittel angefertigt. Die aus anderen Quellen direkt oder indirekt übernommenen Daten und Konzepte sind unter Angabe der Quelle gekennzeichnet. Weitere Personen, insbesondere Promotionsberater, waren an der inhaltlich materiellen Erstellung dieser Arbeit nicht beteiligt. Die Arbeit wurde bisher weder im In- noch im Ausland in gleicher oder ähnlicher Form einer anderen Prüfungsbehörde vorgelegt.

Eigenabgrenzung / Kooperationen: Die technische Etablierung des computergestützten Bildverarbeitungssystems zur Bestimmung der P. rubescens Dichten in Wasserproben (siehe Kapitel 2.1) erfolgte in Zusammenarbeit mit der Visiometrics IPS GbR, Konstanz. Die in Kapitel 2.2 diskutierte Microcystinanalytik wurde im Rahmen einer Diplomarbeit von Lisa Dietz, in der AG Human- und Umwelttoxikologie an der Universität Konstanz unter meiner Betreuung durchgeführt. Alle weiteren Leistungen wurden, sofern nicht explizit angemerkt, von mir selbst erbracht.

Konstanz im Juni 2008,

Bernhard Ernst

175

DANKSAGUNG

Zu guter Letzt möchte ich mich bei all den lieben und hilfsbereiten Menschen bedanken, die durch ihre Unterstützung, durch Tipps und Tricks, durch die Bereitstellung von Material, Daten und Geräten und nicht zuletzt durch ihre Begeisterungsfähigkeit und Aufnahme in ein nettes Arbeitsklima zum Gelingen dieser Arbeit beigetragen haben.

Mein besonderer Dank gilt: Prof. Dr. Daniel Dietrich, für die Möglichkeit die Promotion in seiner Arbeitsgruppe anfertigen zu können, sein unermüdliches Engagement, seine wissenschaftliche und menschliche Unterstützung und sein Vertrauen in mein tägliches „Wirken“. Prof. Dr. Karl-Otto Rotthaupt, für die Übernahme des Koreferates und das Interesse an dieser Arbeit. Den Familien Jark, Franzen und Jenckel, Arthur und Aenne Feindt Stiftung Hamburg, für die finanzielle Trägerschaft des Projektes, die unkomplizierte Unterstützung und das große Interesse an der untersuchten Problematik. Dr. Stefan Höger, für die unzähligen „Saunadiskussionen“, die kritische Durchsicht meiner Manuskripte, die nachhaltige Motivation vor Kongresspräsentationen und nicht zuletzt für sein fürsorgliches Bemühen um mein nächtliches Wohlbefinden in Konstanz. Dr. Evelyn O’Brien, für die zahlreichen Englischkorrekturen und die hilfreichen Diskussionen über „Trojan copepods and stuff like this“. Karin Rieder, für die Unterstützung in den umfangreichen Angelegenheiten der Bürokratie und die unermüdliche Belieferung mit „wissenschaftlich erfülltem“ Papier! Dr. Werner Fischer, für seine Aufgewecktheit und Begeisterungsfähigkeit und die damit verbundene Initiierung dieser Arbeit. Dr. Bettina Hitzfeld, für die Hilfestellung bei meinen ersten „wissenschaftlichen Gehversuchen“. Prof. Dr. Billy Day, University of Pittsburgh, for his help in toxin-analyses and helpful discussions on mass spectrometry and chemical details - „Billy, it’s a good day, isn’t it?” Dr. Rolf Dieter Negele, Dr. Julia Schwaiger, Ulrike Mallow & Herrmann Ferling, Bayerisches Landesamt für Umwelt, Wielenbach, für die Unterstützung bei der Durchführung eines Expositionsversuches. Prof. Dr. Claudia Wiegand & Dr. Stefan Pflugmacher, Institut für Gewässerökologie und Binnenfischerei, Berlin, für die interessante Zeit am IGB, die Unterstützung und Zusammenarbeit und nicht zuletzt für die netten Abende in St. Petersburg und Berlin. Prof. Dr. Stephan Neser, für die Unterstützung in Sachen Bildverarbeitungssystem und seine unermüdliche Geduld im Umgang mit einem nur begrenzt „Physik-kompatiblen“ Biologen. Anke, Alex, Biggi, Jörg, Anne, Lisa, Dani, Kerstin, Susanne, Heiko, Niki, Andi, Daniel und die vielen anderen Kolleginnen und Kollegen, die mich über die vergangenen Jahre mit Rat und Tat durch den Laboralltag begleitet haben und vor allem Miriam, für ihr Verständnis, ihre ausdauernden Anfeuerungsrufe und für die Linderung der Nebenwirkungen dieser Arbeit.

176