Toxic Secondary Metabolites from Plants: a New Aspect of Water Quality

www.natoxaq.eu

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 722493

University College Copenhagen, University of Copenhagen

Toxic secondary metabolites from plants: A new aspect of water quality

PhD thesis of Vaidotas Kisielius September 2020

Supervisors: Senior associate lecturer Lars Holm Rasmussen Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark Professor Hans Christian Bruun Hansen Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark Dr Michael Rodamer Agilent Technologies, Hewlett-Packard St. 8, 76337 Waldbronn, Germany

Toxic secondary metabolites from plants: A new aspect of water quality

PhD thesis of Vaidotas Kisielius

University College Copenhagen Faculty of Health Department of Technology University of Copenhagen Faculty of Science PhD School of Science

Period of work: April 2017 – September 2020

Main academic supervisor: Senior associate lecturer Lars Holm Rasmussen, University College Copenhagen (Denmark) Academic supervisor: Professor Hans Christian Bruun Hansen, University of Copenhagen (Denmark) Co-supervisor: Dr Michael Rodamer, Agilent Technologies (Germany)

Chair of assessment committee: Professor Nikoline Juul Nielsen, University of Copenhagen (Denmark) Assessment committee member: Dr Jørn Smedsgaard, Research and Development Consultant at FOSS Analytical A/S (Denmark) Assessment committee member: Dr Carmel Ramwell, Senior Environmental Scientist at Fera Science Ltd (United Kingdom)

Copenhagen, 20201

Water background image of the title page © istockphoto.com

Safe drinking-water is water that does not represent any significant risk to health over a lifetime of consumption, including different sensitivities that may occur between life stages.

WHO Guidelines for Drinking-water Quality (ISBN 978-92-4-154995-0)

Only the dose determines whether a substance is poisonous or not.

Paracelsus (1493 – 1541)

Dedicated to my mum

Foreword

This thesis summarises the main research activities performed to fulfil the requirements to obtain a PhD degree from the University of Copenhagen. It has been submitted to the PhD School of Science of the University in May 2020 and submitted in revised form in September 2020. The studies should be concluded with an oral defence approx. three months after the submission.

The experimental work has been conducted mainly at two research groups: 1: Department of Technology, University College Copenhagen (Supervisor – Senior associate lecturer Lars Holm Rasmussen) and 2: Department of Plant and Environmental Sciences, University of Copenhagen (Supervisor Professor Hans Christian Bruun Hansen). The project was fully financed by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska- Curie grant agreement No. 722493.

The thesis consists of three parts. Part 1: the aims, objectives, general hypotheses and the context of the PhD project in the scheme of an associated research network. Part 2: Synopsis providing an introductory risk identification of plant toxins as water contaminants (Chapter 1) and illustration about how the research findings incorporate into the current knowledge of the field (Chapters 2, 3 and General conclusions). Part 3: Four research manuscripts created for this the PhD project.

The thesis does not contain information about the courses passed as part of the PhD training, performed non-research academic activities, and manuscripts that were prepared during the PhD project period in the same research institutions but not primarily for this PhD project. For this and related information please refer to the Supervisor Report.

Thesis statement

Bracken ferns (Pteridium sp.) and butterburs (Petasites sp.) can contaminate nearby water resources with carcinogenic and/or hepatotoxic compounds, to levels exceeding the tolerable drinking-water contaminants.

Acknowledgements

I would like to begin the acknowledgements section with a big gratitude to the initiators of the NaToxAq network: Senior associate lecturer Lars Holm Rasmussen, Prof. Hans Christian Bruun Hansen and Prof. Bjarne Westergaard Strobel. Your efforts with little early guarantees of success have eventually transformed the professional lives of 16 ambitious young scholars. Your mission will be continued long after the date of the end of the project.

The quality of this thesis greatly benefited from constructive comments of members of the first assessment committee – to whom I am sincerely grateful – Prof. Nikoline Juul Nielsen, Dr Jørn Smedsgaard, and Prof. Miguel E. Alonso Amelot. Besides, I am thankful to people who directly contributed to my research with supervision, a helping hand, critical advice, or guidance through the publishing processes: Dr Michael Rodamer, Dan Nybro Lindqvist, Anette Kjeldal Lausten, Dr Lise Rølmer Nissen, Prof. Mikkel Boas Thygesen, Prof. Bo Markussen, Prof. Mary T. Fletcher, Prof. Elena María Fernández González, Prof. Franklin Riet-Correa, Jonas Halgren Mikkelsen, Michael Josefsen, Mikkel Drejer, Jimmy Kjellerup Dornhoff, Nicoline Wildenrath Brandtoft Hansen, Jimena Bertorello Martinez, Mette Rosenfjeld and Anita Schjødt Sandager. Likewise, I would like to thank Milda Kisielienė for linguistic advice.

I am also grateful to fellow researchers of the NaToxAq network for gathering from 5 continents and staying abroad for the period of our common study: Inés Rodríguez Leal, Xiaomeng Liang, Massimo Picardo, Bettina Gro Sørensen, Mulatu Yohannes Nanusha, Daria Filatova, Jawameer Hama, Carina Schönsee, Regiane Sanches Natumi, Daniel Bernardo Garcia Jorgensen, Marcel Schneider, Natasa Skrbic, Barbara Kubíčková, Bilal Tariq and Ellie Stone.

The network would not have prospered without the devoted professionals to whom I also wish to present my sincere acknowledgements: Dr Kristine Groth Kirkensgaard, Angelika Lene Rasmussen, Karin Norris, Dr Werner Brack, Dr Thomas Bucheli, Prof. Jan H. Christensen, Dr Carmel Ramwell, Dr Elisabeth Janssen, Prof. Luděk Bláha, Prof. Klára Hilscherová, Prof. Pavel Babica, Prof. Oscar Núñez Burcio, Dr Marinella Farré, Prof. Efstathios Diamantopoulos, Dr Ann-Katrin Pedersen, Dr Sarah Christensen, Prof. Matthew MacLeod, Barbara Franziska Günthardt and others whose input was essential but remained less spotlighted.

I am grateful to the University College Copenhagen for recruiting me during and beyond the period of the PhD project, and to over 30 companions of Laborantuddannelsen for the daily uplifting colleagueship. Also to the Waldbronn campus of Agilent Technologies for a unique, welcoming and well-organized internship. Moreover, I would like to thank my previous supervisors for scientific knowledge and research management skills provided me earlier that also helped me through the present project: Prof. Mindaugas Rimeika and Dr Loreen Ople Villacorte (supervisors of BSc and MSc theses, respectively).

Finally, I would like to thank the family Lina, Monika, Kotryna, Asta, Rimantas, Justinas, Denis, Aurimas, Milda, Elzė, Lukas, Marius, Olga and Kęstutis for the quality time so needed between the activities.

Table of Contents

Thesis statement ...... 6 Acknowledgements ...... 7 Summary ...... 10 Dansk resumé ...... 12 Abbreviations...... 13 Part I: PhD project ...... 14 Preface ...... 14 Project aims ...... 16 The context of the project ...... 17 Major achievements ...... 19 Part II: Synopsis ...... 21 Chapter 1. Natural toxins as water contaminants (Risk identification) ...... 21 Safe drinking-water ...... 21 Natural toxins ...... 22 Plant toxins ...... 23 Plant toxin categorisation ...... 24 Rationale for water monitoring ...... 25 Criteria for plant toxin monitoring in water ...... 27 Chapter 2. Terpene glycosides from Pteridium species (Case 1) ...... 32 Genus Pteridium ...... 33 Species and distribution ...... 33 Competitive advantages ...... 34 Toxicity to farm animals ...... 35 Toxicity to humans ...... 36 The terpene glycoside ptaquiloside...... 36 Ptaquiloside-like compounds ...... 37 Analytical standards ...... 39 Analytical methods ...... 40 Occurrence in plants ...... 42 Toxicity of CAU...... 48 Comparison of compound stability in water ...... 49 Occurrence in water ...... 55

Chapter 3. Pyrrolizidine alkaloids from Petasites species (Case 2) ...... 61 Genus Petasites ...... 61 Petasites hybridus ...... 61 Alkaloids...... 63 Pyrrolizidine alkaloids ...... 64 Occurrence in water ...... 66 General conclusions...... 69 References ...... 71 Part III: Manuscripts ...... 91 0. Introduction to manuscripts ...... 91 1. Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns ...... 92 Supplementary Material ...... 102 2. Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium ...... 121 Supplementary material ...... 141 3. Occurrence and stability of ptesculentoside, caudatoside and ptaquiloside from bracken ferns in surface waters ...... 143 Supplementary material ...... 165 4. Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids ...... 168 Supplementary material ...... 187 About the author ...... 189

Summary

The subject of water contamination by harmful plant toxins contains major knowledge gaps. It is hypothesized that toxic terrestrial plants producing compounds with certain physicochemical properties cause risk to the quality of adjacent drinking-water resources. The properties of the main groups of plant toxins are compared to emerging water contaminants of anthropogenic origin – Persistent and Mobile organic Compounds (PMOCs). Based on the properties of water solubility and aquatic persistency, intensively toxic compounds produced by common higher plants have been selected for further investigation: carcinogenic terpene glycosides from bracken fern (Pteridium sp.) and carcinogenic/hepatotoxic pyrrolizidine alkaloids from butterbur (Petasites sp.) Altogether, the presented work compiles four manuscripts (one published, one accepted for publication and two in preparation for submission) that are divided into two cases according to the subject matter.

Case 1 is centred on the terpene glycosides from Pteridium sp. The carcinogenic terpene glycoside ptaquiloside (PTA) can be released from Pteridium stands and contaminate nearby water resources. Moreover, the genus produce additional structurally nearly identical but more water soluble compounds – ptesculentoside (PTE) and caudatoside (CAU). Besides PTA, the compounds have been largely underinvestigated in regards to chemistry, toxicology, transfer to animals, humans and the environment. The research presented in this thesis created analytical standards of the compounds and of their hydrolysis products, and analytical methods needed to quantify them in plants and in water (Manuscripts 1 and 3). The methods applied to plant samples from 6 continents and to plants systematically sampled in Northern Europe demonstrated that the average combined contents of CAU and PTE in Pteridium nearly comprised the average contents of PTA (Manuscripts 1 and 2). Statistically significant geographical trends have been observed in geographic distribution between CAU and PTA contents in Pteridium in Northern Europe.

The study for the first time reported CAU, PTE and their hydrolysis products in surface waters (up to 5.3 µg/l of PTE, CAU and PTA in total) (Manuscript 3). Research performed simultaneously with the methods developed in this project for the first time demonstrated in vivo toxicity of CAU1. Assuming that tolerable levels of CAU and PTE in drinking-water should be the same as for PTA (2x10-3 μg/l estimated in previous research), the terpene glycosides reported in surface water in this thesis would violate the tolerable level of carcinogens in drinking water by up to 2650 times. Stability studies revealed that PTE, CAU and PTA in natural waters degrade fast under slightly basic pH (7.4) and high temperature (15 ˚C) but are significantly more stable under moderately acidic to neutral pH (5.2-6.5) and at low temperature (5 ˚C). Exposed risk of the three compounds to

1 Schneider de Oliveira L.G., Boabaid F.M., Kisielius, V., Rasmussen L.H., Buroni F., Lucas M., Schild C.O., Lopez F., Machado M., Riet-Correa F., Hemorrhagic diathesis in cattle due to consumption of Adiantopsis chlorophylla (Swartz) Fée (Pteridaceae). Toxicon X. 5 (2020) 100024.

10 drinking-water resources should be regarded of similar importance, as they were comparably stable under 16 different combinations of studied conditions.

Case 2 of this study focuses on pyrrolizidine alkaloids (PAs) from Petasites sp. By analysing water resources adjacent to the plants, up to 0.53 μg/l of these compounds were detected in a stream and up to 0.23 μg/l in seepage water in shallow groundwater wells (2.2–3.0 m depth, Denmark). No PAs were detected in studied deep groundwater (~60 m depth) in the same area nor in control seepage waters remote from the plants. This study reports the PAs in water wells for the first time (Manuscript 4). Taking into consideration tolerable levels of the PA ingestion proposed by the health authorities (0.007 μg of daily intake per kg of bodyweight), a daily intake of 1 L of the surface water or 2 L of the well water reported in this study would violate the safe limit for an average healthy human being. This research detected two groups plant toxins in water bodies out of two options carefully selected for investigation (terpene glycosides and pyrrolizidine alkaloids). Even though the specific waters in which the compounds were detected were not consumed for drinking, the results may have far-reaching implications due to the following reasons: 1) Pteridium is the fifth most abundant plant genus in the world. As indirect result of anthropogenic activities it is expanding, and large monocultures can be encountered in proximity to drinking-water resources; 2) Petasites is regarded invasive in the region where the PAs in water were detected. The species require moist soils and therefore proliferate very closely adjacent to freshwater resources; 3) It is estimated that 3% of all flowering plants contain at least 1 PA and many others produce toxins with chemical structures determining high water-solubility and environmental persistence (e.g. other heterocyclic alkaloids, steroids, triazoles, polyketides). The observations show that higher plants producing these toxins ought to be taken into consideration when assessing environmental risks to the quality of adjacent exploited water resources.

Dansk resumé

Denne ph.d.-afhandling argumenterer for at frigivelse af skadelige giftstoffer fra planter til vandmiljøet er et hyppigt forekommende men overset fænomen. Med udgangspunkt i de fysisk- kemiske egenskaber af naturlige giftstoffer og forekomsten af de plantearter som danner dem, er de kræftfremkaldende terpen glykosider fra planteslægten Pteridium (Ørnebregner) og de kræftfremkaldende/levertoksiske pyrrolizidin alkaloider fra slægten Petasites (Hestehov) blevet undersøgt. Som central del af projektet, blev der udviklet analytiske metoder og protokoller samt oprenset forskellige toksiske naturstoffer fra Ørnebregner. Undersøgelser viste for første gang at terpen glykosiderne caudatosid og ptesculentosid findes i overfladevand (samlet mængde op til 5.3 µg/l). Det vurderes at vand med de højeste observationer vil overskride den tolerable koncentration af kræftfremkaldende stoffer som terpen glykosider med 2650 gange. I en anden del af dette projekt blev der fundet op til 0.53 μg/l pyrrolizidin alkaloider i overfladevand og op til 0.23 μg/l i overfladenært brøndvand. Niveauet af pyrrolizidin alkaloider blev sammenlignet med de tolerable niveauer af tilsvarende stoffer i fødevarer, og det kunne konkluderes at 2L af det omtalte brøndvand ville overskride de eksisterende grænseværdier for mennesker. I alt blev to forskellige grupper af plantegiftstoffer fundet i vand ud af to omhyggeligt udvalgte og grundigt undersøgte stofgrupper. Selvom de undersøgte vandressourcer ikke umiddelbart anvendes til drikkevand, så kan resultaterne have vidttrækkende konsekvenser. Ørnebregner er i dag den femte mest hyppigt forekommende planteslægt i verden. Som et indirekte resultat af menneskelige aktiviter, øger bregnerne deres udbredelser og større områder kun dækket af Ørnebregner findes ofte tæt på vandressourcer. Den anden planteslægt - Hestehov - kræver fugtig jordbund og findes ofte direkte ved ferskvand. Derudover er det kendt at ca. 3% af alle blomsterplanter indeholder mindst 1 pyrrolizidin alkaloid og mange flere danner andre strukturelt lignende vandopløselige, persistente og giftige forbindelser.

Abbreviations

CAU – Caudatoside DOC – Dissolved Organic Carbon DW – Dry Weight EFSA – European Food Safety Authority ESR – Early Stage Researcher (refers to NaToxAq) EU – European Union HPLC – High Performance Liquid Chromatography IARC – International Agency for Research on Cancer under WHO LC-MS – Liquid Chromatography coupled with Mass Spectrometry LD50 – Median Lethal Dose LOD – Limit of Detection LOQ – Limit of Quantification MS – Mass Spectrometry NaToxAq – International Training Network “Natural Toxins and Drinking Water Quality - From Source to Tap”, funded by the EU Horizon 2020 research and innovation program under Marie Sklodowska-Curie grant agreement No. 722493 (www.natoxaq.eu) NMR – Nuclear Magnetic Resonance PAs – Pyrrolizidine Alkaloids PFAS – Persistent Polyfluoroalkyl Substances PMOCs – Persistent Mobile Organic Compounds PTA – Ptaquiloside PTA analogues – Compounds with chemical structures nearly identical to PTA (incl. PTE and CAU) PTE – Ptesculentoside PtrA, PtrB, PtrG – Pterosins A, B and G, respectively SPE – Solid Phase Extraction UN – United Nations WHO – World Health Organization of the United Nations WHO Guidelines – WHO Guidelines for drinking-water quality (ISBN: 978-92-4-154995-0) [1] WP – Work Package (refers to NaToxAq)

Part I: PhD project

Preface

Water is a vital component for every cell of every lifeform on Earth. It can dissolve and carry more substances than any other liquid. The same feature that makes water crucial for life makes it potentially dangerous: it is its exceptional ability to dissolve and carry contaminants. Besides the inherent properties of contaminants, their concentrations determine whether they can have adverse health effects. The father of toxicology Paracelsus (1493 – 1541) already in 1537 postulated that only the dose determines whether a substance is poisonous or not [2].

Today, water quality is monitored and regulated by authorities in most countries. However, occurrence of harmful levels of undesired drinking-water contaminants may still occur due to:

1. unavoidability of a contaminant, 2. unproportional costs of its removal, or 3. lack of risk identification.

A contemporary example of unavoidability of a contaminant in drinking-water could be dissolved persistent polyfluoroalkyl substances (PFAS) from a wide range of consumer and industrial products (e.g. surfactant component perfluorooctanoic acid). Detected in tap waters in 31 states of the US (total PFAS up to 0.19 µg/l) [3], these compounds are not removed because the technologies, at least in the required scale, are not yet available and/or implemented. The PFAS accumulate in the environment and do not degrade in the human body. Health authorities report PFAS as carcinogenic and harmful to the hormone and immune systems [4]. The legal limits are not in place, but administrative considerations in different countries recommend 0.1 µg/l as the threshold limit [5], while non-governmental organizations insist for 0.001 µg/l [6].

An example of unproportional costs of removal is fluoride in groundwaters in eastern Africa. The concentrations of the contaminant often exceed the recommended 1500 µg/l limit of the WHO Guidelines [1]. Fluoride is essential for teeth enamel. Paradoxically, overdosed it can cause severe dental and skeletal defects (fluorosis). Though it is technically possible to remove fluoride by anion exchange processes, the technologies are costly, in particular for low-income countries. The population is forced to use water exceeding the maximum acceptable level of the contaminant as governments prioritize allocation of financial resources to more acute healthcare and socio- economic issues.

Unfortunately, sometimes contaminated water can be regarded as safe, simply due to lack of knowledge of its composition. The third case of the presence of undesired water contaminant – lack of risk identification – could follow insufficient monitoring, limited sensitivity of analytical techniques and alike. Moreover, there are many unknown aspects of metabolites originating from

14 anthropogenic pollutants and industrial chemicals that have been used by humans since only recently. Previously unknown water contaminants are often identified by non-target screening. Many can be not monitored systematically because of the absence of knowledge about what exactly it is that needs to be monitored.

It is the purpose of this thesis to identify and characterize certain undesirable contaminants that can be present in drinking-water that fall into category of current lack of risk identification. It is proposed that plant toxins are not sufficiently identified as a risk. The absolute majority of the world’s toxic compounds are of natural origin, biosynthesized by living organisms [7,8]. Due to their omnipresence, the variety and the extent of generated biomass, inedible plants are a particularly large source of natural toxic compounds.

Many known compounds of plant origin, if detected in drinking-water, would be regarded as harmful. Many of them might not have been recognised because they simply have not been analysed for in water, or were monitored only as e.g. part of the DOC. The fraction of natural toxic components in DOC might not necessarily be high. However, individuals can consume water from the same reservoirs for very long time. Over long exposure periods the yet unidentified plant- origin water contaminants might lead or contribute to cumulative adverse health effects and chronic diseases, that are nowadays commonly attributed to diet, air pollution, lifestyle and ageing.

Although there is substantial knowledge in the field of risk identification and regulation of plant toxins in food and feed, there is little awareness about potential risks of plant toxin occurrence in drinking-water. The lack of risk identification may lead to absence of public demand for monitoring. It can in turn lead to shortage of development of proper analytical methods, and in turn reinforce the lack of risk identification (Figure 1).

Figure 1. Proposed potential lack of risk identification in the field of plant toxins as water contaminants. There is considerable development of knowledge in the fields of artificial toxins in water, food and animal feed; as well as plant toxins in food and animal feed (left graph, dark blue). The lack of risk identification in the field of plant toxins as water contaminants (left graph, red) might be reinforced by the lack of development in analytical methods, lack of monitoring initiatives and alike (right).

Project aims

The overall purpose of this PhD project is to identify risk of plant toxin occurrence as water contaminants and characterize the risk by empirical research. Where necessary, analytical methods have to be developed to perform the research.

The purpose is achieved by the following aims:

1. Referring to previous studies identifying problematic plant toxins and their groups, discuss the risk of plant toxin occurrence in water and select relevant plant toxin groups for investigation (Chapters 1–3).

2. Create analytical standards and develop analytical methods necessary to perform empirical studies of the selected compounds (Manuscript 1).

3. Apply the developed methods for measuring the selected compounds in plants and discuss their potential implications to ecosystems, health and water quality (Manuscripts 1 and 2. In addition to this thesis – a supporting publication1).

4. Adjust and apply these and other recently developed analytical methods for investigation of the selected plant toxins in water, characterize the risks and discuss implications of findings (Manuscripts 3 and 4, General conclusions).

It is not the purpose of this PhD project to provide substantial water monitoring data or determine the scale of occurrence of natural toxins in drinking-water, nor to suggest solutions. These and other tasks are addressed by a research network that this project is a part of (see the next section ”The context of the project”).

The context of the project

This project is an integral part of the European Training Network "NaToxAq" funded by the European Union’s Horizon 2020 research and innovation program under Marie Sklodowska-Curie grant agreement (No. 722493). The network aims to produce multidisciplinary knowledge about natural toxic substances in aquatic environments. The network comprises 14 nearly simultaneous PhD projects (ESRs 1–14) and 2 shorter projects of postgraduate research (ESRs 15 and 16) (Figure 2). The main activities of all projects have been performed at 10 institutions of 7 European countries, and in close collaboration with 12 public and private partner organizations in Europe and the United States (www.natoxaq.eu). To promote the inter-institutional communication, the main activities of the PhD projects belonging to the same work packages (WPs) were performed in different institutions.

Figure 2. Individual projects (ESRs) within Work Packages (WPs) and intended channels of information flow (blue lines) © 2016 NaToxAq. This project was performed as ESR7 under WP3 - Sources and distribution of natural toxins. This project falls under WP3 “Sources and distribution of natural toxins” (ESR7) and its working title has been Spatial and temporal variation in the production and release of glycosidic natural toxins in rangelands and forests. The project description focuses on the occurrence and release of plant toxins rather than extensive monitoring. The aims of the WP are reached by proper determination of specific plant toxins as potential water contaminants and identification of ecosystems that producing them.

Early intermediate results, developed analytical standards and methods of this project (ESR7) were transferred to projects ESR11 of WP4 (Fate and transport of natural toxins) and ESR13 of WP5 (Safe drinking-water supply). The compounds that were raised as concern in this project have

17 already been detected in drinking-water by project ESR131. Effects of the water treatment techniques for the toxin removal are also being studied within its scope. The overall knowledge and tools developed by the whole network should successively be combined and transferred to legislative and executive institutions and agencies via WP6 “Dissemination and outreach”.

This project created fruitful collaborations with the projects within WP3, particularly with the ESR8 based in the same country (Denmark). The performed sampling campaigns included collection of samples for the projects ESR4 and ESR8. In collaboration with the institution hosting the projects ESR4 and ESR5 (Helmholtz Centre for Environmental Research, Germany), mass spectral data of plant toxins and their metabolites discovered in this project have been included in European MassBank database allowing free data access, exchange and sharing [9]. This project also created collaborations with institutions outside the network: University of Queensland (Australia), University of Oviedo (Spain) and National Agricultural Research Institute of Uruguay, by sharing analytical samples and creating a joint research output2.

The main activities of this project have been performed in Denmark but with an international perspective. The quality of European drinking-water, particularly in Denmark, is generally good. Since 99% of the Danish centralized drinking-water supply is based on groundwater, the studied surface waters were not exploited as sources of drinking-water. Nevertheless, the identified concepts would work in other places of the world where surface waters are used for drinking water. The vegetation investigated in the project is also found across Europe and in other parts of the world. Surface water contamination in many parts of the world could be worse than in Europe, also regarding pollution with plant toxins. For example, in tropical and sub-tropical regions there are higher amounts of toxic plants, longer periods of vegetation cover and higher proportions of surface waters exploited as drinking-water resources. In low-income countries the issue of even suspected plant toxins could be down-prioritized due to the lack of technologies, methods, monitoring, resources and research.

It is an ambition of this project that high-risk ecosystems resulting in impaired water quality will be further studied using the analytical methods that have been developed. It is anticipated that the results of the NaToxAq network would serve legislative and regulatory bodies for making optimal decisions to initiate relevant further research, monitoring and mitigation of relevant water quality risks. It is also a long-term aspiration of this project that the created knowledge will find context beyond water research: in disciplines of plant science, food safety, human and veterinary medicine and environmental chemistry.

1 Skrbic et al. Monitoring of ptaquiloside, caudatoside and ptesculentoside from bracken ferns in groundwater. (Under preparation). 2 Schneider de Oliveira L.G., Boabaid F.M., Kisielius, V., Rasmussen L.H., Buroni F., Lucas M., Schild C.O., Lopez F., Machado M., Riet-Correa F., Hemorrhagic diathesis in cattle due to consumption of Adiantopsis chlorophylla (Swartz) Fée (Pteridaceae). Toxicon X. 5 (2020) 100024.

Major achievements

The work presented by this project comprised the following activities:

• Creation of analytical standards of novel compounds by applying SPE, preparative HPLC and other contemporary techniques of compound handling and purification, and purity determination applying NMR. • Creation, optimization and validation of an efficient LC-MS method for quantification of structurally nearly identical analytes in environmental samples (plants, water). • Application of LC-MS/MS for analysis of a group of compounds in environmental samples (plants, water). • Sampling and chemical analysis of plants in an extensive geographic area, including contemporary statistical analysis of the analysed chemical constituents.

The methods and their application resulted in:

• First report of plant toxins ptesculentoside, caudatoside and their hydrolysis products pterosins G and A as water contaminants (Manuscript 3). • First report of stability of ptesculentoside and caudatoside (Manuscript 3). • First report of pyrrolizidine alkaloids in water wells (Manuscript 4). • First indirect evidence of toxicity of caudatoside1 (not presented among Manuscripts).

The work in progress has been disseminated and discussed in the following outreach activities:

• Kisielius V., Hansen H.C.B., Rodamer M., Rasmussen L.H. Risk assessment of natural carcinogenic compounds from Bracken fern plants in surface waters with LC/MS. Platform presentation, Danish Water Forum Annual Water Conference. Lyngby, Denmark (2018). • Rasmussen L.H., Kisielius V., Hansen H.C.B., Rodamer M. A novel analytical method for simultaneous quantification of Bracken fern produced carcinogenic ptaquiloside-like compounds and their derivatives. Poster presentation, SETAC Europe 28th Annual Meeting. Rome, Italy (2018). • Kisielius V., Hansen H.C.B., Rodamer M., Lindqvist D.N., Rasmussen, L.H. A novel analytical method for simultaneous quantification of the carcinogens ptaquiloside, caudatoside and ptesculentoside. Poster presentation, 10th International Symposium of Poisonous Plants. St. George, Utah, USA (2018). • Platform presentations in the NaToxAq conferences held in Denmark, United Kingdom, Spain and Germany in 2017–2020 (www.natoxaq.eu).

As part of the outreach activities the project engaged in 230 hours of supervision of undergraduate students, including four final successfully defended professional bachelor projects of a laboratory technician program at University College Copenhagen: • Michael Josefsen “Optimering af metode til ekstraktion af carcinogenerne Ptaquilosid, Caudatosid, Ptesculentosid og deres derivater Pterosin A, Pterosin B og Pterosin C fra tørret bregnemateriale” (English: Optimization of method for extraction of the carcinogens Ptaquiloside, Caudatoside, Ptesculentoside and their derivatives Pterosin A, Pterosin B and Pterosin C from dried fern material) and Jonas Halgren Mikkelsen “Extraction of compounds from bracken ferns” (individual reports on a group project 2018–2019). Some results were published with the Manuscript 1.

• Mikkel Drejer and Jimmy Kjellerup Dornhoff “Stabilitetsanalyse af illudane glykosider i vand” (English: Stability analysis of illudane glycosides in water) (individual reports on a group project 2019–2020). The students became co-authors of the Manuscript 3.

Part II: Synopsis

Chapter 1. Natural toxins as water contaminants (Risk identification)

The following chapter is based on existing knowledge. Its purpose is to identify and discuss the potential risks that plant toxins can be overlooked potential harmful water contaminates, and select relevant plant toxin groups for further investigation.

Safe drinking-water

For the purpose of discussing harmful water contaminants, safe drinking-water has to be defined. Direction and coordination of health protection on a global scale, including management of environmental, food, toxicological and occupational safety, is the primary role of WHO. Its definition of safe drinking-water is: “water that does not represent any significant risk to health over a lifetime of consumption, including different sensitivities that may occur between life stages” [1]. In other words, safe drinking-water consumed for drinking and preparation of food throughout one's lifetime should not lead to intoxications or waterborne diseases. Safe drinking-water is an essential component of current policy for health protection and has been reflected in the UN Sustainable Development Goals [10]. Most of the drinking-water resources cannot be regarded as constantly safe. Water moves in hydrological cycle (Figure 3) potentially bypassing natural or anthropogenic pollution sources that can lead to perturbation of its quality.

Figure 3. The hydrological cycle. The movement of groundwater is generally slow, whereas precipitation- driven transport of surface water is generally rapid. The resulting water turnover rates partially determine vulnerability of water resources to pollution. © United States Geological Survey (labelled for reuse).

Natural toxins

Many forms of living organisms contain and/or release low molecular mass organic compounds that do not have a known role in the metabolism of the host organism. Such compounds are referred to as secondary metabolites. Various secondary metabolites can play various roles, but the most are currently recognised as crucial for defence, prey, deterrence of parasites and competition with other species. These biologically active compounds can be both gaseous and non-volatile, can be readily stored in the cells or activated in case of danger or environmental stress. By serving for both attack and defence they can damage or kill other life forms [11–14].

There are multiple ways of categorising natural toxic compounds - for example, by their origin, modes of toxicity, intensity of toxicity, molecular structures and functional groups, and other. For the purpose of chemical ecology, categorization of toxins by molecular structures will be further used in this thesis. However, the table below lists examples of toxic compounds by their source and highlights whether it is natural or anthropogenic (Table 1). The data should not be regarded as absolute, but as reflecting the best efforts of the author to find the lowest LD50 values reported for each of the source type. In fact, compounds produced by animals and plants include some of the most toxic substances known [15].

Table 1. Illustration of sources of toxic compounds in the order of decreasing acute toxicity of the most toxic known substance of a particular source.

Source type LD501 (mg/kg of bodyweight) of Origin the most toxic substance known (Natural/ anthropogenic) Bacterial toxins2 1–3 x10-6 [16,17] Natural (organic) Algal toxins3 0.13 x10-3 [18] Natural (organic) Nerve agents4 0.22 x10-3 [19] Anthropogenic Dioxins5 1 x10-3 [20] Natural/Anthropogenic Snake venoms6 1 x10-3 [11] Natural (organic) Frog venoms7 2 x10-3 [21] Natural (organic) Jellyfish venoms8 5 x10-3 [22] Natural (organic) Plant toxins9 0.02 [23] Natural (organic) Spider venoms10 0.02–0.04 [12,24] Natural (organic) Marine vertebrate venoms11 0.3 [25] Natural (organic) Fungal toxins12 0.3–0.7 [26] Natural (organic) Mould toxins13 0.5 [27] Natural (organic) Insect toxins14 1 [28] Natural (organic) Pharmaceuticals15 2.9 [29] Anthropogenic Non-metal elements16 3.0 [30] Natural Scorpion venoms17 4.3 [31] Natural (organic) Metallic elements18 5.2 [32] Natural Metalloid elements19 13–15 [33,34] Natural Illicit drugs20 22 [35] Anthropogenic Pesticides21 60 [36] Anthropogenic

1 Median lethal dose. A measure of the dose of a substance needed to kill half of a given population, usually of mice. Some authors assume that two doses of the LD50 are the minimum lethal dose, but there is no constant rule [37]. 2 Neurotoxins Botulinum from Clostridium botulinum, Tetanospasmin from Clostridium tetani, Shiga toxins from Shigella dysenteriae (Neurotoxic proteins). 3 Maitotoxin (high-molecular mass neurotoxin) from Gambierdiscus toxicus. 4 Novichok (organophosphate). 5 TCDD (2,3,7,8-Tetrachlorodibenzodioxin). By-product of burning. 6 Textilotoxin (neurotoxic protein) from Pseudonaja textilis. 7 Batrachotoxin (neurotoxic steroidal alkaloid) from Dendrobatidae. Besides, in smaller amounts of batrachotoxin are found in other species of frogs, birds and beetles. 8 CrTX-A (proteinaceous compound) from Carybdea rastoni. 9 Ricin (neurotoxic protein) from Ricinus communis. 10 Latrotoxin (high-molecular mass neurotoxin) from Latrodectus. 11 Tetrodotoxin (low-molecular mass potent neurotoxin) from Tetraodontiformes (fish) (also found in Hapalochlaena (invertebrate octopus)). 12 Amatoxins (bicyclic peptides) from Amanita phalloides. 13 Aflatoxin B1 from Aspergillus flavus. 14 Cantharidin (terpenoid ) from Meloidae family. 15 Anesthetic fentanyl (alkaloid). Controlled or illegal depending on country. 16 Tetraphosphorus (P4). 17 Chlorotoxin-like peptide from Buthus martensii. 18 Cadmium (Cd2+). 19 Arsenic (As3−). 20 Heroin (alkaloid). 21 Organothiophosphate insecticide Demeton-s-methyl.

Plant toxins

The most toxic chemicals ever discovered are produced by slow-moving lifeforms (snakes, spiders, scorpions, frogs, sessile marine animals) [13]. The reason might be that these organisms cannot rapidly flee or attack. In fast-moving animals the mechanisms of chemical defence and attack are not that broadly developed [13]. In contrast, plants are among the most sessile lifeforms. Many plants have passive modes of physical protection, but chemical defence mechanisms in the plant kingdom are particularly broadly developed. For example, ricin (Table 1), purified from the seeds of Ricinus communis, ingested or inhaled can kill an adult human by a dose the size of a few grains of table salt [38]. The compound is illegal to possess and trade and has been applied as a weapon of war and massive terror [39,40].

Another major difference between toxic animals and plants is that the former are more prone to secure venoms and release them only when the central nervous system signals danger. In contrast, plants have neither nervous nor immune systems. In order for the plant toxins to exhibit the defence function, most need to be maintained constantly and in relatively high concentrations. In some plant species the toxic compounds are concentrated in one or more parts, in others – occur throughout the whole plant [41].

Out of the known approximate number of 400,000 of plant species, only about 100 are used on a regular basis for nutrition as sources of protein, oil or carbohydrates. In addition, several thousand plants yield edible berries and fruits. The rest, that is over 99% of all plants, are either toxic (depending on the dose) or repellent [14,42,43]. The repellent plants can have an unpalatable or deterring taste or rigid microscopic structures (phytoliths) that make them indigestible. Even non- edible parts of the plants cultivated for food are often repellent or toxic. For example, leaves of potato are rich with acute toxic steroidal alkaloid solanine [44], while leaves and blossoms of lupin contain various carcinogenic quinolizidine alkaloids [45].

Biologically active plant compounds were undoubtedly the first medicines, poisons and protection agents of humans [46]. Plant toxin application for homicide was widespread at least from ancient Egyptian and Greco-Roman civilizations [47]. Arrows poisoned by plant extracts were used for hunting and warfare by ancient societies and are still used around the world [48,49]. Plant-based insecticidal and larvicidal bed sheets are archeologically proved to have been used even 77,000 years ago [50]. Potential of biologically active properties of plant secondary metabolites is being increasingly explored for the purposes of modern medicine, pest management and other fields.

Herbivores can feed on certain toxic plants or their parts, but only because throughout evolution they developed tolerance to particular toxins (specialist herbivores) [51]. Toxic plants that are consumed by certain species can exhibit full potential toxicity to other species and to generalist herbivorous, including mammalian organisms like humans [52]. Reports on mammalian animal intoxication are an important source of information for discovering the modes of plant toxicity and serve as supplementary evidence of plant toxicity to humans [53].

Plant toxin categorisation

No uniform categorization of all structural groups of plant toxins can be found. The following main structural groups are referred to in different informational sources (in alphabetical order): aliphatic acids, aliphatic nitro compounds, alkaloids, amines, anthracenes, coumarins, flavonoids, furocoumarins, glucosinolates, glycosides, isoflavones, methylxanthines, naphthalenes, nitriles, non-protein amino acids, oxalates, phenolics, phenylpropanoids, phytates, phytoestrogens, polyacetylenes, polyketides, polypeptides, proteinaceous compounds, quinones, tannins, terpenes, terpenoids, saponins, steroids and sulphur compounds [54–62].

Some structurally complex molecules can encompass several of the above-mentioned structural groups (alkaloid amines, terpenoid amines, terpenoid alkaloids, etc.), while the most structurally complex – proteinaceous compounds (molecules consisting of, containing or resembling protein) – are considered primary rather than secondary metabolites. Nearly among all of the structural groups, plant toxins could be found that cause acute poisoning or chronic damage of digestive, nervous, endocrine, reproductive systems, genotoxicity, mutagenicity and carcinogenicity, and

24 less frequently – inhibit protein synthesis, cause skin burn, respiratory damage, disorders of cardiac rhythm, allergies, mental confusion, convulsions [41,54,63–65].

Rationale for water monitoring

Plant toxicity is an inevitable and indispensable part of ecosystems and is not questioned as problematic in itself. A potential problem to humans and animals arises only if the toxic constituents are released to the environment and are ingested. Plant toxins are directly regulated in edible products, drugs, cosmetics and animal feed. However, their structural variety implies that many can be water-soluble and environmentally persistent – similar properties as for anthropogenic pollutants [54]. Released from plants, natural compounds ought to be regarded as environmental contaminants with associated health risks, regardless their natural origin.

The aspect of chemical ecology of plant toxins is generally more often neglected compared with assessment of anthropogenic compounds. For example, cultivation of the fern Pteris vittata is suggested for eco-friendly phytoremediation of arsenic from soil [66–68]. The species produce carcinogenic terpene glycoside ptaquiloside [69]. The compound is water-soluble, easily washable from plant biomass by precipitation and can be released from ferns to soils and water recipients [70–72]. No information about toxic plant secondary metabolites of the species can be found in the bioremediation literature. If the remediation was considered in relation to drinking-water resources, the use of the fern ought to be evaluated based on the ptaquiloside risk.

Due to well-established sources of food and protective measures, acute plant intoxication in humans is learned to be avoided to a considerable extent. In contrast, severe plant poisonings of farm animals occur often. For example, a direct economic loss of 245 million dollars has been calculated for plant-toxin poisoning deaths of cattle and sheep in the United States in 1989 [73]. (The estimate corresponds to 500 million dollars of 2020 costs, and did not cover loss in production and cost of veterinary services related to poisoned but surviving cattle). It is conceivable that water can carry lower concentrations of the same plant toxins that cause cattle death, and lead or contribute to chronic diseases to humans in the long term. Release of plant toxins to surrounding environment is apparent in case if the compounds are volatile [74–76]. However, many non-volatile biologically active plant compounds, especially allelopathic ones (serving host organisms for impeding or competing other species), are released into surrounding soil via roots [77–79]. Plant bioactive compounds be also released through trichomes, glands, ruptured or degrading tissues [80–82]. They can be also distributed through spores and dust (mostly known – allergens) [83–85]. Not many studies have been performed, but it was demonstrated that plant toxins can be released by contact of precipitation water [70,86,87].

Toxic compounds in plants can locally outcompete anthropogenic chemicals in quantity because of their continuous production and comparatively high concentrations [88]. For instance, the

25 average global intensity of annual pesticide spray per kg of cropland is 2.6 kg/ha (2017) [89]. Expanding and invasive plants like Pteridium aquilinum produce the above mentioned natural carcinogenic glycosidic terpene ptaquiloside in estimated constant load of up to 5.9 kg per hectare [90]. The compound is water-soluble and well-documented to cause cancers in ruminants and in humans that use the plant as food [91–93]. Estimated tolerable concentration of ptaquiloside in drinking-water is 0.002 μg/l [94]. If the plant toxins leach to water bodies and persist until they reach the drinking-water consumer, the natural contaminant should generally be regarded more harmful than contemporary pesticides (legally accepted concentration of total pesticides in EU is 0.5 µg/l [95]).

It is conceivable that plant toxins can naturally accumulate in their ecosystems of origin and be washed off by precipitation runoffs following rainstorms and snow melts. Plant toxins could enter water bodies from more possible sources than anthropogenic pollutants: from cropped fields and grasslands, from lands that are unaffected by human activities, from urban areas and from vegetation that readily grows directly in the water bodies. Because of a variety of the sources, plant toxins can exist as both point and diffuse source pollutants. For example, water-soluble cardiac glycosides in garden ornamental plants (wallflower, lily of the valley, yellow and white oleanders) can be equivalent to point source pollutants. Due to large dispersion, water-soluble alkaloids in agricultural crop plants and cyanogenic glycosides in red clover grasslands can be equivalent to diffuse source pollutants. It is conceivable that these water-soluble toxins can enter surface and groundwater bodies exploited for drinking-water supply (Figure 4).

Another motive to study plant toxins as potential water contaminants is generally underinvestigated composition of DOC in surface and groundwater resources. In drinking-water

26 production, dissolved organic matter is partially removed because it can affect the consumer satisfaction due to undesirable colour, tastes and odours. However, it is generally regarded as having no direct adverse impact on human health [96]. Knowing that the organic matter originates primarily from plants [96] and that approximately 99% of all plants are toxic [14,42,43] it is suggested that although the organic matter in water from plants does not cause apparent acute health effects, some of its constituents can cause adverse chronic health effects.

Regarding any types of organic natural toxins, the WHO Guidelines for drinking-water quality refer only to “breakdown of plant material or from algae and other microorganisms that grow in the water or on sediments” [1]. The document also elaborates on release of cyanotoxins from algae and indicates that control of algal blooms is the preferred control option. Neither toxic higher aquatic plants nor adjacently growing terrestrial plants or fungi are mentioned as a risk [1]. It is known that drinking-water can be contaminated with toxins of natural origin, like secondary metabolites of algae and fungi. Since algae readily grow in water, their high aquatic biomass and excreted compounds cause direct problems in drinking-water production. The issue of aquatic algal toxins has therefore been raised and investigated for decades. Mitigation techniques to prevent algal blooms have been applied worldwide [97]. The presence of fungal toxins has been reported in water resources only since recently [98]. Fungi biosynthesise toxic metabolites structurally similar to those biosynthesised by plants [98], but plants generate substantially higher biomass. For the proof of concept of natural toxic compound leaching to water it is therefore more rational to study plants.

Criteria for plant toxin monitoring in water

Physicochemical properties of plant toxins that can cause risk to water resources should match with the properties of their anthropogenic counterparts. An increasing concern in the field of drinking-water safety is associated with anthropogenic persistent mobile organic compounds (PMOCs). No precise definition of PMOCs exist but the compounds are frequently referred to persisting in the environment and, due to high polarity, not being removed from water by sorption processes [99]. The PMOCs typically originate from discharge of pharmaceuticals, cleaning agents and personal care products (Table 2). The combination of solubility and environmental persistency determine that toxic PMOCs may end up in drinking water and pose risk to human health. Since ever-increasing number of anthropogenic PMOCs is detected in drinking-water, it is recognized that many more of the yet unknown compounds can be present in water reservoirs [99–102].

Table 2. PMOCs from pharmaceuticals and personal care products commonly found in terrestrial and aquatic ecosystems (based on [99] and [103])

Type Subgroup Representative compounds Analgesics Aspirin, diclofenac, ibuprofen Antibiotics Ampicillin, chloramphenicol, ciprofloxacin, clarithromycin, clindamycin, erythromycin, norfloxacin, sulfamethoxazole Pharmaceuticals Antihypertensive agents Metoprolol, norverapamil, propranolol Blood-lipid regulators Clofibrate, gemfibrozil Cytostatic drugs Cyclophosphamide, ifosfamide Hormones Estrogens E1, E2 Stimulants Caffeine Antimicrobial, antifungal agents Triclosan, triclocarban Cleaning agents, Preservatives Parabens (alkyl-p-hydroxybenzoates) personal care products Sunscreen UV filters Enzacamene, octyl methoxycinnamate polychlorinated biphenyls methyl-tertbutylether Other ethyl-enediaminetetraacetic acid short-chain perfluoroalkyl acids

It is suggested that plant toxins that meet the general criteria of the PMOCs (high solubility and persistency) can also pose risks to drinking water quality. Plant toxins that are non-soluble shall be significantly less relevant as they should not diffuse from the plants to water with which they come into contact. Rapidly degrading or hydrolysing compounds should not persist in water long enough to reach the drinking-water consumer.

Some bioactive compounds of plant origin have been reported in water resources before. For instance, groundwater surveys reveal that the alkaloid caffeine is among the most frequently encountered compounds [104,105], (Table 2). The presence of bioactive compounds of plant origin in water resources is related to the outlets of treated wastewater. Caffeine, nicotine and other plant-origin pharmaceuticals and stimulants in water resources do not cause real concern because the primary sources of their exposure (coffee, smoking, pills) are in orders of magnitude higher. Nevertheless, their occurrence in water is important indicator that some of plant bioactive compounds can express the properties of PMOCs. Apparently these compounds are not fully degraded during human metabolism, mechanical and biological treatment of wastewater, in the waste water sludge, and by microbial activities in waters receiving the effluent [106,107,108]. More toxic compounds with similar physicochemical properties, if released to drinking-water resources directly from plants, can therefore cause direct adverse health risks.

A recent study performed in silico estimation of properties of 1586 plant toxins from 844 Central- European plant species. It concluded that over 34% of the compounds were relevant as potential

28 water contaminants based on intensity of their inherent toxicity, high water-solubility and aquatic persistence [54]. The types of the compounds that met the criteria (hereafter called prioritised toxins) are provided in Table 3. The study did not provide information about the abundance of the compounds, which should be investigated separately and ecosystem-specifically. Plant species with high contents of these toxins, high local abundance, high biomass and growing location close to water resources might be of higher concern. Bioactive plant compounds that are not abundant as well as not highly toxic would likely be diluted in drinking-water to the limits below the established thresholds of equivalent anthropogenic pollutants, and therefore cause no real concern.

Table 3. Highly toxic, water-soluble and environmentally persistent secondary metabolites of European plants [54].

Structural Prioritized toxins / Rank by sub-structures (number of prioritized toxins) groups total toxins in plants Alkaloids 270 / 459 terpenoid* (52), pyrrolizidine (48), steroidal* (45), isoquinoline (45), quinolizidine (34), indole (27), amaryllidaceae (19) Terpenes 83 / 261 Sesquiterpenes (31), triterpenes (26), diterpenes (26) Steroids 56 / 115 (not further categorized) Polyketides 30 / 110 (not further categorized) *terpenoid alkaloids and steroidal alkaloids were counted as separate classes from terpenes and steroids.

The results of the study summarized in Table 3 indicate that due to intrinsic physicochemical properties, plant toxins expressing the highest potential risk as water contaminants are alkaloids, terpenes, steroids and polyketides, respectively. General information about these structural groups is provided in Table 4, examples of molecular structures: in Figure 5.

Table 4. Description and examples of alkaloids, terpenes, steroids and polyketides. (Prepared by [54,59,60,61].

Structural Description Examples of structural Examples of Examples of groups of sub-groups commonly known potential water prioritised plant toxins contaminants toxins [54] according to [54] Alkaloids Organic compounds • Acridine atropine, aconitine, containing at least one • Diterpenoid caffeine, cyclobuxine D, nitrogen atom in • Imidazole cocaine, galanthamine, heterocyclic ring • Indole/indolizidine codeine, heliosupine, structure. The names of • Phenanthrene nicotine, lupanine, the structural sub- • Phenylamine morphine, lycopsamine groups are determined • Purine strychnine lycorine by the rings that the • Pyridine/piperidine protoveratrine A, alkaloids encompass. • Pyrimidine protopine, sparteine, • Pyrrolidine • Pyrrolizidine

• Quinolizidine taxine B, • Quinoline/isoquinoline vincamine • Steroidal • Tropane/atropine • Tropolone Terpenes Hydrocarbons based on • Monoterpenes (2 units) artemisinin, artabsin a cyclic molecular • Sesquiterpenes (3 units) eucalyptol, baccatin III, skeletons constructed • Diterpenes (4 units) retinol, cucurbitacin B, from isoprene (C5H8) • Triterpenes (6 units) alpha-, beta-, elaterinide, units. • Tetraterpenes (8 units) gamma- mezerein, carotenes. ptaquiloside Steroids Organic compounds • Androstanes Cholesterol, digitoxigenin, with three six-member • Cholanes estrogen, strophanthidin and one five-member • Cholestanes testosterone rings fused in a specific • Pregnanes molecular configuration Polyketides Organic compounds • Acetogenins – formononetin, containing alternating • Ansamycins lupulone carbonyl and methylene • Polyenes groups, or derived from • Polyethers precursors containing such alternating groups

Figure 5. Examples and explanations of the structural groups of highly toxic, water-soluble and environmentally persistent toxins of Central-European plants

Another prominent feature of PMOCs is that many are demonstrated to persist in food chains [109,110]. At least one group of plant toxins have also been studied with respect to their flux through different trophic levels – pyrrolizidine alkaloids (Table 3) (hereafter called PAs) [111–113]. It has been demonstrated that some PA specialist herbivores (moths, butterflies, leaf beetles) are able to accumulate and use the PAs for their own benefit against a range of predators (spiders, lizards, birds, mammals) [112]. Mobility of biosynthesis products through several tropic levels in natural environments is a notable feature. It suggests that, similarly to anthropogenic PMOCs, persistent plant toxins like PAs can pose yet unknown risks to the environment.

______

The following chapters will discuss specific compounds from the top 2 structural groups of plant toxins indicated as potential water contaminants (Table 3), produced by plants with high abundance and high proximity to freshwater resources.

Chapter 2. Terpene glycosides from Pteridium species (Case 1)

In the previous chapter it has been discussed that physicochemical properties of certain terpenes make them potential water contaminants. Terpenes are hydrocarbons based on cyclic molecular skeletons constructed from isoprene units (C5H8). The units can combine in many different ways, representing a very high variety of compounds with distinct properties, even for compounds with the same chemical formulas. Their primary function in plants is attraction of pollinators and defence against pathogens and herbivores. Many are acute toxic and some have been proven effective botanical insecticides against mosquitoes, flies, larvae and insect pests [74,114,115].

Based on the existing knowledge that: • Database on intensity of toxicity, water-solubility and aquatic persistence of toxins of European plants indicate terpenes as potential water contaminants [54] (discussed in Chapter 1), • The glucose moiety supposedly determines even higher water-solubility of terpene glycosides, • Carcinogenic terpene glycoside ptaquiloside (PTA) has been reported washable from plant biomass by precipitation and present in adjacent surface and shallow groundwater resources [70–72,87,94] • The main plant that produces PTA (genus Pteridium) grows densely in many regions (referred to as the fifth most abundant plant genus in the world [116], • More of presumably carcinogenic terpene glycosides structurally similar to PTA are produced by Pteridium and other plants of Pteridaceae family [117–122], but so far these compounds have not been looked for in aquatic environments, the following chapter provides a review of the main producer of these compounds (Pteridium) and investigates following research questions:

1. Are toxic terpene glycosides, other than PTA, present in Pteridium in significant contents, including in geographical region where PTA leaching to water resources has previously been identified (Northern Europe). If yes, which factors determine the spatial and temporal variation of the compounds in the plants?

2. Do these compounds occur in water resources.

3. How stable are these compounds in natural waters and how it compares to stability of PTA.

Genus Pteridium

Species and distribution

Pteridium (family Dennstaedtiaceae, generic name bracken) is an ancient and very widely encountered genus. According to fossil evidence, it achieved a worldwide distribution by the Oligocene geologic epoch (> 23-34 millions of years ago) [123] and is regarded the fifth most abundant plant genus in the world [116]. Historically it has been considered to comprise of a single species, but 17 species, variations and subspecies are taxonomically accepted today [124]. Among the most common are P. esculentum (G. Forst.) encountered in Oceania, P. arachnoideum (Kaulf.) in Central and South America, P. caudatum (L. Maxon) in Americas and P. aquilinum var. latiusculum (Desv.) in the Northern Hemisphere [125]. The most extensive by both the amount and the geographical distribution is P. aquilinum (L. Kuhn) [116,124] (Figure 6). P. aquilinum is widely distributed in all continents except Antarctica [125,126].

Figure 6. Pteridium aquilinum covering forest floor in Denmark

Originally bracken is a natural integral part of open forest plant communities [127]. Nowadays it spreads in various terrains and climates in many parts of the world [128,129] (Figure 7). The genus tends to especially rapidly colonise empty areas like clear-cut or burned-down forests, unused agricultural lands and pastures and similar. Unsustainable land management and anthropogenic activities therefore indirectly prepare the ground for an uncontrolled spread of bracken monocultures. In favourable conditions bracken can infest to extreme rates. E.g. in the United Kingdom the common bracken covers 1.7 million ha, which corresponds to 7.3% of the total land surface area of the country [130]. As an indirect result of human activities and climate change, bracken distribution on the global scale is predicted to increase [128–130].

Figure 7. Registered distribution of Pteridium aquilinum according to Global Biodiversity Information Facility [125]. The observation points are mainly registered by volunteers worldwide and accepted by the administration of the facility. Therefore, the locations with no registration do not indicate the absence of the species. (The records of other species of Pteridium are also widespread, but mainly only within the regions mentioned above in the text. E.g. Pteridium esculentum in Oceania).

The bracken-infested terrains concern farmers, foresters and land managers because the plants are especially difficult to control. By colonising areas of productive lands it can lead to a loss of agricultural production [131]. Mechanical removal of bracken from the surface of the ground does not eliminate the plant. Individual bracken fronds may visually appear like individual plants (Figure 6), however, the fronds of the same bracken stand originate from continuous underground rhizome systems. The whole bracken stand is therefore often the same clone, or mixed several clones [132]. The underground biomass (rhizomes and roots) usually greatly exceed the aboveground biomass (fronds), store nutrients and can stay dormant for many years [127,131]. Frond cutting can even promote the subsequent more intensive outbreak of new emerging fronds. To eradicate bracken, intense ploughing or a controlled multiple-stage interventions of cutting, herbicide application and replacement of vegetation are needed [131].

Competitive advantages Bracken effectively overtakes new terrains because of a combination of several competitive advantages over other plants. One of the advantage is competition for light by shading other vegetation and competition for water by powerful rhizome systems. Another advantage is a sophisticated reproductive system. The genus can spread both by dispersal of spores and by vegetative extension. The sporulation depends on the region and the season, and may not occur every year. However, the spores can be carried long distances and one successfully germinated spore can give rise to an entire bracken stand [123,133]. In contrast, the vegetative extension of powerful underground rhizome systems can progress and constantly overtake neighbouring areas, especially areas cleared from other plants. Another form of competitive advantage of the plant is chemical composition that makes it repellent and toxic to herbivores. In most ecosystems the

34 herbivores prefer other plants. Moreover, bracken can hinder germination, growth, survival and reproduction of neighbouring other plant species (allelopathy) [134].

Bracken defence is expressed by the following deterrent and toxic secondary metabolites: astragalin, coumarin, isoquercetrin, pteroquilin, tiliroside, catecholamines, carboxylic esters, kaempferol flavonoids, phenolic compounds, polyhydroxy sterols, steroids, sterols, tannins, thiaminases, carboxylic acids (benzoic, caffeic, carboxylic, cinnamic, dihydrociannimic, fumaric shikimic, succinic, 5-O-caffeoylshikimic), cyanogenic glycosides (prunasin, amygdalin, linamarin), pterosins, pterosides and bioactive illudane-type glycosides [131,135–139]. The list is not necessarily complete. The combined effect of the complex combination of these compounds is responsible for the effective deterrence of fungal pathogens, insect and ruminant herbivores and competing plants [127,134,140].

Toxicity to farm animals Farm animals do not prefer grazing bracken, but do so if they have little alternatives. Therefore, besides the discussed troubles for land managers and foresters, the plant can pose serious health risks and potential loss of livestock [141]. Acute poisoning of bracken-intoxicated farm animals (e.g. horses, cattle, sheep) can take the form of fever, epistaxis (nose bleeding), skin haemorrhage (escape of blood from a ruptured blood vessel), tachycardia and tachypnoea (abnormally high frequency of heart-beat and breath-rate, respectively) mild jaundice (yellowing of eye whites), dark and fetid diarrhea, depression and possibly death. A long term ingestion can lead to observed degenerative changes of rapidly growing cells (bone marrow, intestinal epithelium), mucosae (oral, ocular and vulvar), cancers of upper-gastrointestinal tracts and urinary bladder (bovine enzootic haematuria), gradual retinal degeneration and irreversible bright blindness [131] [142–144].

Due to bracken content of cyanogenic glycosides, animals risk acute poisoning by hydrogen cyanide. However, the symptoms of cyanide poisoning may be mixed with the symptoms of other bracken constituents and other factors, and have yet been reported as such [145]. Early symptoms of bracken intoxication by monogastric animals (horses, pigs) are largely related to toxins thiaminases – enzymes that readily cleave thiamine and inhibit essential thiamine-regulated pathways (e.g. metabolism of glucose) [146,147]. Resultant thiamine deficiency can express as lethargy, anorexia and neurological symptoms (ataxia, polioencephalomalacia) [142,148].

Ruminant animals (cattle, sheep) maintain thiamine-producing abdominal microflora and are less prone to develop the early symptoms of thiamine deficiency. The ruminants can therefore be more expected to develop symptoms associated with prolonged bracken ingestion (chronic diseases like cancers in cattle and sheep, and retinal degeneration in sheep). Reports of bracken-caused chronical symptoms date back more than 100 years [144, 149–152]. Due to the worldwide prevalence of the plant and the long-term study of its effects, bracken is also often referred to as the best documented plant to naturally cause cancers in animals [123,127,153,154].

Toxicity to humans The International Agency for Research on Cancer under WHO (hereafter called IARC) evaluated the evidence of bracken carcinogenicity in experimental animals as sufficient and since 1987 has classified the plant as possibly carcinogenic to humans [155,156]. Nevertheless, customary food, traditional medicine and culinary delicacies containing bracken can be found in several countries [157–159]. Increased odds-ratios between the direct human consumption of bracken and upper- gastrointestinal tract cancers have been observed at least in Japan, Brazil and Venezuela [92,93,160].

Even if the plant is not consumed, longer term residence in areas with dense bracken growth have been reported to correlate with the increased risk of human gastric cancer [135,161]. Humans can ingest bioactive secondary metabolites of bracken through farm animal products (dairy, meat [162–164]) and presumably by inhaling the air containing bracken spores [84]. The risk factors of meat and air-suspended spores have not been sufficiently investigated. However, the statistically significant increase of risk for developing gastric cancer (2.34 times) due to residence in the bracken-infested areas was theoretically related to dairy products of domestic farm animals occasionally grazing the plant [161].

The terpene glycoside ptaquiloside

The carcinogenicity of bracken could not be explained before a Dutch and a Japanese research groups in 1983 simultaneously identified terpene glycoside ptaquiloside (formerly also termed aquilide A) [165,166] (hereafter called PTA). The structure was elucidated by 1H NMR using the extracts from Pteridium aquilinum and Pteridium aquilinum var. latiusculum (Desv.), respectively. The discovery allowed further research to firmly establish the relationship of the plant carcinogenicity to this compound [177, 167–169]. An electrophilic cyclopropane group in an active form of PTA (conjugated dienone) readily reacts with nucleophiles (e.g. water, amines, DNA) [170,171]. By alkylating DNA it can cause genetic mutations [135,153,169,172,173] (Figure 8).

PTA can be found in all parts of the plant: fronds, rhizomes, roots, spores and recently emerged croziers [84,174]. The PTA content among the plants is very variable and has been reported from 0 up to 45,000 μg/g in the dry weight (DW) of the plant [175]. The most commonly reported median concentrations vary around 1,000 μg/g of PTA in DW. It has also been observed that PTA content in young bracken croziers, which are more vulnerable for being grazed, is higher than in the fully matured fronds [174]. The PTA concentration in the below-ground biomass has also been reported to increase with the seasonal loss of the above-ground biomass [174]. These observations demonstrate the mechanisms of storage and transport of PTA in the plant, indicating that for the host organism it is a valuable secondary metabolite. Water containing PTA causes in vivo gastric carcinogenesis of mice and in vitro DNA damage of human gastric epithelial cells [169,176]. The increased risk of human carcinogenesis in areas infested by Pteridium is also presumed due to

36 metabolites of PTA found in dairy products of domestic farm animals that occasionally graze the plant [161,162].

Ptaquiloside-like compounds

When the sesquiterpene glycoside PTA was discovered, it was universally associated with the carcinogenicity of the plant (Figure 21). However, at least one study showed that the extracts of Pteridium express higher genotoxicity to animals than their contents of purified PTA [169]. Since the genotoxicity of the plant has not been related to any other of its known constituents, the presence of additional DNA-reactive compounds in the plant was indicated. Moreover, PTA hydrolyses into more stable compound pterosin B (Figure 8), but other pterosins structurally nearly identical to pterosin B are known. It has been therefore suggested that additional compounds similar to PTA give rise to other pterosins, which may also contribute to the total carcinogenicity of the plant [139,177] (Figure 8).

Several compounds have been elucidated in Pteridium with chemical structures nearly identical to PTA: iso-ptaquiloside (PTA epimer) and caudatoside (CAU) in 1997 [178], ptaquiloside Z in 1998 [179], and in 2010 hydroxy-PTA ptesculentoside (PTE) [180] (hereafter called PTA analogues) (Figure 8). Nowadays PTA and/or structurally closely related compounds have been detected in over 50 species of ferns (Pteridaceae) [69,117–122,181]. PTA analogues comprise more of glycosides with cyclopropane group, e.g. hypolosides A, B and C in Hypolepis punctata and Dennstaedtia hirsta (Pteridaceae) [121]. However, these have never been reported in Pteridium. Most recent observations infer that the majority of all PTA-like compounds in Pteridium comprise PTA, PTE and CAU [182].

PTE and CAU contain the cyclopropane functional group that is ultimately responsible for DNA alkylation and genetic mutations in animals caused by the PTA. A cyclobutane is much less reactive functional group than cyclopropane [183]. Pteridanoside – a compound found in bracken is nearly identical to PTA, but with cyclobutane instead of cyclopropane (Figure 8). It was demonstrated to be sixteen times less toxic than PTA [184]. Similar compounds without the cyclopropane or cyclobutane groups are not reported carcinogenic. It is therefore reasonable to suggest that the carcinogenicity of bracken results from the combined effects of PTA and other compounds containing the cyclopropane functional groups [119]. However, due to the absence of analytical standards and lack of advanced analytical techniques, PTA has been studied much more extensively than its analogues in regards to chemistry, toxicology, transfer to animals, humans and the environment. Most studies therefore still commonly refer to PTA as the only carcinogenic compound in bracken (Figure 21).

Carcinogenicity of milk from bracken-fed cows to experimental mice was first published in 1972 [185]. It was explained only after PTA metabolites were detected in milk in 1992 [186] (corresponding to up to 110 μg/L of PTA in the raw milk of cow [187]; 1.63x10-3 and 3.14x10-3 μg/L in the raw milk of sheep and goat, respectively [162]). The similarity of the in-vivo biological

37 activities of PTA, CAU and PTE was demonstrated by detecting the corresponding pterosins B, A and G in the tissues of cattle fed by Pteridium esculentum [164]. Though the cattle developed cancer, it could not be confirmed to be caused by PTE and CAU, as the bracken also contained PTA. However, this study demonstrated similar metabolic reactivity of the three compounds in animals and raised health concerns for the meat consumers.

Figure 8. PTA-like compounds. The carcinogenicity is attributed to the highly reactive cyclopropane functional groups. Pteridanoside with no cyclopropane but with cyclobutane is reported 16 times less toxic than PTA [184]. It is therefore suggested that the carcinogenicity of bracken results from the combined reactivity of compounds with the cyclopropane functional groups.

Even though the structural framework of PTE, CAU and PTA is identical (Figure 8), the difference between the compounds lie in active hydrogen bridge promoters – hydroxy groups. The groups facilitate binding to e.g. amino acids, nitrogen-containing cell components and water molecules. The groups determine that PTE and CAU are more water soluble than PTA. Increased solubility of the three glycosides in that sequence (PTE the most soluble) has been also observed in the compound elution order in preparative and analytical HPLC (Manuscript 1), and in the processes of developing SPE methods (Manuscripts 1 and 3). Increasing hygroscopy of anhydrous glycosides in the same order (PTE>CAU>PTA) was also observed in the process of developing analytical standards (Manuscript 1). The estimated approximate Log P values (Table 5) and the abovementioned observations imply that PTE and CAU are more prone to be washed from plants to water resources by e.g. precipitation than PTA.

Table 5. Molecular formulas, CAS numbers and estimated Log P values of glycosides and corresponding pterosins.

Compound PTE CAU PTA Pterosin G Pterosin A Pterosin B

Molecular Formula C20H30O9 C21H32O9 C20H30O8 C14H18O3 C15H20O3 C14H18O2

CAS number - - 87625-62-5 35964-50-2 35910-16-8 34175-96-7 Estimated Log P Computed by -2 -1.4 -1.3 1.9 2.2 2.5 XLogP3 3.0 (Source: PubChem)

Analytical standards No analytical standards of PTE, CAU and PTA are commercially available. A report of Pteridium esculentum containing >1 mg/g DW of each of the three compounds was found (reported concentrations determined by deliberately converting the glycosides and measuring the formed corresponding pterosins [182]). Recently emerged bracken fronds were collected in the same location in 2017 (Bribie island, Australia) and presented to Denmark by post for creation of analytical standards (Manuscript 1).

An SPE-based method for isolation of PTE, CAU and PTA was created. The method was less labour- intensive and 2.9 times more efficient for PTA extraction from bracken than the most recent method published in literature based on PTA sorption to XAD resin [188]. A number of attempts performed in the presented study to isolate several milligrams of anhydrous PTE, CAU and PTA to determine their purity by 1H NMR were unsuccessful due to extreme hygroscopy of the compounds. The method of isolating and handling the glycosides through very dry conditions was gradually developed. The purity of PTA was reached 92% with 2% precision (Supplementary material of Manuscript 1). The analytical standards of pterosins G, A and B were created by adopting the most efficient previously published method of PTA conversion to pterosin B [69], and verifying it for conversion of PTE and CAU into pterosins G and A (Manuscript 1).

Figure 9. 1H NMR spectrum of PTA applying the developed anhydrous compound handling protocol reported in Manuscript 1 (left) and not applying it (right). The scale of the images is standardized according to the height of the red-coloured peak of internal standard (same concentration). The green- coloured peaks were used for the purity determination of PTA.

Analytical methods Previously published chromatographic methods for quantification of PTE and/or CAU in plants required conversion of these compounds in plant extracts into corresponding pterosins and measuring them in GC-MS, HPLC/UV-VIS or HPLC-DAD [164,182,189] (hereafter referred to as Conversion methods). The Conversion methods do not directly differentiate pterosins that form by the deliberate conversion process from pterosins that can be present in samples prior to the conversion. Nevertheless, the Conversion methods can be applied for estimation of the glycoside contents in living plants, based on assumption that all pterosins present in analytical samples originate from glycosides due to sample treatment and storage [190]. Cleaning of analytical samples with polyamide can be applied to remove the pterosins [90,189], but it demands additional sample treatment step and can introduce measurement uncertainties.

A Conversion method with no polyamide treatment would not be applicable for direct determination of glycoside content in processed plant material (e.g. in bracken-based food) for direct estimation of sample toxicity. Processed bracken-based food or other media (e.g. water) can be non-toxic because the glycosides can be naturally hydrolysed into pterosins prior to analysis. A method that can quantitively examine the glycosides in their indigenous forms in the samples was therefore developed. Performance of instrumental part of the method is compared with the instrumental part of the single mass spectrometric Conversion method, and other most

40 contemporary methods of quantification of PTA and pterosin B (Table 6: Manuscript 1). The reported direct quantification method also requires less sample pre-treatment and hence is likely more accurate.

Table 6. Comparison of the developed method (Manuscript 1) with the existing contemporary methods for analysis of PTA, pterosin B and other pterosins. (PtrB, PtrA, PtrG mean pterosins B, A and G, respectively).

Instrumental LOD (μg/L) Time of Instrument instrumental Reference PTA PtrB CAU PtrA PTE PtrG analysis (min) LC-MS 0.22 0.03 0.26 0.02 0.08 0.01 5 Manuscript 1 [191] not not not Fletcher et al. 2011 [182] GC-MS - - - 14 reported reported reported (Conversion method) LC-MS 0.08 1.10 - - - - 10 Rai et al. 2017 [192] LC-MS 0.32 0.26 - - - - 18 Aranha et al. 2014 [193] LC-MS/MS 0.13 0.08 - - - - 5 Clauson-Kaas et.al. 2016 [188] LC-MS/MS 0.19 0.15 - - - - 15 Jensen et al. 2008 [194] HPLC-DAD - 2.00 - - - - 10 Rasmussen et al. 2017 [69] HPLC/UV-VIS 60.00 - - - - - 14 Ovesen et al. 2008 [195] HPLC/UV-VIS 19.00 8.40 - - - - 15 Jensen et al. 2008 [194] HPLC/UV-VIS 2,000 130 - - - - 12 Ayala-Luis et. al. 2006 [196]

The reported LC-MS method performance for PTA and pterosin B is as good as reported performance of the methods applying the generally more advanced LC-MS/MS technique (Table 6). The low LODs of the reported method have been achieved by testing multiple adducts (H+, H-, + + + + Na , K , Li , NH4 ) forming due to different mobile phase additives, and full factorial experimental design of the MS electrospray ionisation parameters (Supplementary material of Manuscript 1). Short time of instrumental analysis was obtained by carefully developed step-gradient elution. The method of Manuscript 1 (Table 6) is validated according to Selected Ion Monitoring (SIM) mass spectrometry scanning mode. The instrument allowed a maximum number of 4 simultaneously scanned m/z values. In order to simultaneously monitor 6 compounds, only the m/z values representing the molar masses of protonated pterosins were monitored. These values match with the masses of certain protonated fragments of the corresponding glycosides (Table 7).

The approach to monitor only 3 values of m/z ratios is practically convenient because the 6 compounds can be simultaneously monitored, and report of each glycoside appears in the same report window together with its corresponding pterosin. However, under the given conditions the actual most abundant fragments for CAU and PTA were 231.1 and 201.1 m/z, respectively (Table 7, bold). It partially explains why LOD of PTA and CAU is higher than of the rest of the compounds (Table 6). In case of necessity of even higher sensitivity, the actual most abundant fragments of the glycosides (231.1 and 201.1) can be monitored.

Table 7. Mass fragments in chronological order of the compound retention times. The masses monitored in SIM are underlined. (Relative abundances are provided in Supplementary material of Manuscript 1).

Monoisotopic m/z 200 – 460 Monitored Compound mass (g/mol) (relative abundance ≥ 10%, excluding isotopes) m/z value + + 217.1 [M-glucose-H2O+H] , 235.1 [M-glucose+H] , PTE 414.2 437.2 [M+Na]+, 453.1 [M+K]+ 235.1 pterosin G 234.1 235.1 [M+H]+, 257.1 [M+Na]+

+ + 231.1 [M-glucose-H2O+H] , 249.1 [M-glucose+H] , CAU 428.2 451.2 [M+Na]+ 249.1 pterosin A 248.1 249.1 [M+H]+, 271.1 [M+Na]+

+ + 201.1 [M-glucose-H2O+H] , 219.1 [M-glucose+H] , PTA 398.2 421.2 [M+Na]+, 437.1 [M+K]+ 219.1 pterosin B 218.1 219.1 [M+H]+

Occurrence in plants To examine the contents of PTE, CAU and PTA in Pteridium plants, the analytical standards and method developed in Manuscript 1 were applied for two main sets of plant samples:

1. Samples of 18 ferns (17 of which Pteridium) from 6 continents (Study No. 1, Table 8), 2. Samples of Pteridium from 66 locations along a 1500 km geographical gradient in Northern Europe (three samples per each location, Study No. 2, Table 8). Table 8. Comparison of average proportions of PTA, CAU and PTE contents reported in Pteridium species.

Average proportion of the content Average measured Study of the total 3 glycosides (%) content (mg/g DW) ±SD Sampling range References No.* Σ PTA, CAU, PTA CAU PTE PTA PTE Six continents 0.5 1.2 1 41 37 22 Manuscript 1 (various Pteridium, n=17)** ±0.7 ±1.5 Northern Europe (P. 1.4 2.1 2 aquilinum and P. aquilinum 67 31 2 Manuscript 2 ±1.5 ±1.9 var. latiusculum, n=198) Australia 3.7 8.0 3 47 8 45 (P. esculentum, n=12) ±3.1 ±4.8 New Zealand 0.8 0.8 Fletcher et al. 4 95 3 2 (P. aquilinum, n=3) ±0.4 ±0.4 2011 [182] Europe 1.5 1.6 5 89 5 6 (P. aquilinum, n=3)** ±1.3 ±1.4 *In the studies No. 1–2 the glycosides were measured separately from corresponding pterosins. In studies No. 3–5 the glycosides were deliberately converted into the corresponding pterosins, and the original glycoside content was estimated by applying a 1:1 M ratio of the conversion. In order to make the data comparable, the figures in studies No 1–2 are obtained by summing the contents of the measured glycoside contents with the measured pterosins and applying 1:1 M conversion ratios. The original data is presented in the Manuscripts 1 and 2.

**Results of a study may be affected by different sampling techniques, sample treatment and storage conditions applied to the samples of the same study.

It can be seen from the Table 8 that PTA in samples of Pteridium from six continents comprised only 41% of the total content of the 3 glycosides. In the samples from Northern Europe the PTA comprised 67% of the 3 glycosides. The results imply that the content of the studied reactive illudane glycosides in Pteridium can roughly be double the PTA content.

The plants analyzed in the study No. 1 (Table 8) represent different Pteridium species at different growth stages. The samples used in the study were collected in non-systematically selected locations (Figure 10, left), by different sampling methods, applying different storage conditions and duration periods prior to analysis (Supplementary material of the Manuscript 1). Though the results provide estimation of the glycoside contents in living plants and their relative distribution (Figure 10, right), no statistical analysis of the environmental factors that determine this distribution can be performed.

Figure 10. Sampling locations of the study No. 1 presented in Table 8 (left) and ternary plot of estimated relative distribution of PTE, CAU and PTA in the plants prior to hydrolysis (right). (Sample No.3 represents a single non-Pteridium fern with non-detected glycosides, and does not appear in the ternary plot).

On the contrary to the study No. 1, the plants analyzed in the study No. 2 (Table 8) were consistently collected from a continuous area at the same plant growth stage (July 12–31, 2018). The samples were preserved by uniform conditions and stored for the same period of time prior to analysis (Manuscript 2 [197]). The geographical region of the sampling was selected based on growing locations of the world-common P. aquilinum and Northern-hemisphere specific P. aquilinum var. latiusculum, within practically accessible distance from the research laboratory (Figure 11). The plant locations were identified by registries in Global Biodiversity Information Facility [125], Danish Environmental Portal [198] and personal observations of the research group involved.

Figure 11. P. aquilinum (top left) and P. aquilinum var. latiusculum (bottom left) registered by Global Biodiversity Information Facility [125] (darker yellow dots indicate higher densities). The actual sampling locations are presented in the image on the right (background map ©deviantart.com). The red star indicates the location of the research laboratory (Copenhagen, Denmark).

Three samples were taken in each sampling location (Figure 11, right): 40 cm tip of a fully matured frond (frond No. 1), the whole fully matured lowest pinna of the same frond, and 40 cm tip of a fully matured frond remote from the frond No. 1 by 10–20 m (frond No. 2). The method of sampling was selected in order to reduce sample heterogenicity by sampling more than one plants in each location, within given practical constraints. Due to long average distance of the sampling locations from the research laboratory, the samples needed to be preserved in the field. An ordinary plant-press with largest commercially available blotting paper (28x43 cm) determined the limitation of the sample size. The samples were pressed between the blotting papers separated by ordinary cardboard sheets within several minutes after collection. The samples from the locations No 1-27 and 28-66 were collected and carried in a vehicle during the first 8 and the last 10 days of the sampling campaign, respectively. Afterwards the plant presses were stored for 4-5 months in room temperature [189]. The sampling period was characterized by a record-low precipitation in the region, indicating that intermediate precipitation events could not have affected the contents of the water-soluble compounds in the plants. Over the 20 day period of the sampling campaign only one marginal precipitation event was observed, barely wetting the surface of the sampled plants. According to the Danish Meteorological Institute 1.3 mm of precipitation was registered in Denmark that day (July 17). Over the whole period of sampling in the country, only 0.8 mm of more precipitation was registered, but not witnessed by the author. Chemical analysis revealed that out of the 198 plant samples only 18.7% contained detectable levels of PTE, whereas 98.0% and 98.5% contained detectable levels CAU and PTA, respectively. The average distribution and the total sum of contents of the glycosides is provided in Table 8. The access of the sampled fronds to direct sunshine and the species of adjacently growing trees were registered in order to verify whether these environmental factors have an effect on the contents of the glycosides. The tree species were categorized by deciduous and coniferous, and by the light

44 transparency of their canopies. Altogether, the contents of each of the three glycosides were studied as a function of the following factors: - Variations of Pteridium (P. aquilinum and P. aquilinum var. latiusculum), - Geographic location (registered WGS84 coordinates), - Variation between duplicate fronds collected 10–20 m apart in the same sampling location, - Variation between the upper 40 cm of the fronds and the lowest fully matured pinnae, - Adjacently growing tree species (deciduous/coniferous), - Light transparency of the canopies of the adjacent growing tree species (light/dark), - Access of the fronds to direct sunshine (access/no access). Computing software R was applied as statistical analysis tool. The computation revealed that statistically significant effect on the contents of the glycosides (p ˂ 0.05) depend only on the sampling location. The effects can be described by the two following models: • CAU content based on geographical gradient: Each sampling location moving from South-West to North-East contained 15.6% greater content of CAU per 1 distance unit provided in Figure 12 (p=0.045).

• PTA content in Denmark versus in other two countries: The PTA content was 3.12 times higher in Denmark than in Sweden (p=0.0031), and 3.60 times higher in Denmark than in Finland (p=0.0018). The model of constant linear change applies for CAU only, and the model of step-wise abrupt change applies for PTA only. The two different models cannot be applied for the opposite compound.

Figure 12. Distribution of the 66 sampling locations on equirectangular projection, linearly scattered around the straight South West – North East line. The colour-coded dots FI, SE, DK represent sampling locations in Finland, Sweden and Denmark, respectively.

The variation of the compounds between the triplicate samples within the same sampling locations was not statistically significant (p>0.05, therefore not selected by the model). This can be because the dataset comprises data of only two bracken fronds in each sampling location, separated only by 10-20 meters apart (the same bracken stand). It is therefore not sufficient to conclude from the given database that no significant small-scale spatial variation of the compounds occurs. Multiple previous studies demonstrated that PTA contents in bracken fronds exhibit large variation over small-scale geographical areas. E.g. In 2003 close to the sampling location No. 11 (Figure 11) local variation of PTA in fourteen P. aquilinum stands within 8 km2 was measured from 1– 50*10-3 mg/g DW in some stands to 1–4.14 mg/g DW in other [94]. Results of multiple other studies demonstrate significant local variations of PTA between Pteridium plants [71,175,199–201]. Local variation of PTA and CAU in sampling locations No. 4 and No. 12 (Figure 11) was investigated. Seven sites were selected in each location. Two whole fully matured bracken fronds in each site were collected over several hours of each sampling event. (A previous study identified that no daily variation of PTA content in Pteridium occurs [94]). The sampling sites and the measured contents of PTA and CAU in P. aquilinum fronds in July of 2018 are provided in Figure 13. The samples were taken once a month from June to November and the average contents measured in all sites are provided in Figure 14.

Figure 13. Contents of PTA/CAU (mg/g DW) measured in whole bracken fronds in Grip Skov and Humleore (locations No. 4 and 12 of Figure 11, respectively). Sampling dates: 05.07.2018 and 06.07.2018, respectively.

Each sampling event occurred in the first week of each month (Figure 14). The error bars represent standard deviations between the 7 sampling sites mapped in Figure 13. Statistical analysis of the average contents of compounds in the plants in all 14 sites demonstrated that regardless the high variation between the sites within the same month, a statistically significant trend between the

46 months has been determined. Contents of compounds during months not sharing a group letter are significantly different (P<0.0001).

Figure 14. Average contents of PTA and CAU (mg/g DW) measured in bracken fronds in 7 sites in locations Grip Skov and Humleore (top graphs). Regardless the high variations between the sites, statistically significant trend of compound decrease over time of the year was observed (table).

The overall statistical analysis reveals that regardless the high variations of the contents of compounds in the plants over small spatial and temporal scales, consistent variation is identified over extensive spatial scale (1500 km) and extensive time scale (summer versus autumn). The study did not explain the cause of the observed variation. The few limited biotic and abiotic factors monitored in each sampling location were not identified as having effect on the contents of the compounds. Nevertheless, it can be hypothesized that the extensive spatial scale represents gradient in climate or in genetic variation of the species, or occurs due to phenotypic plasticity. Different proveniences growing in the same conditions were demonstrated to contain different PTA contents, which implies effect of genetic heritage on the PTA contents [94,202,203]. Nevertheless, the presented Northern European study identified no statistically significant difference in compound content between the two distinct genetic variations (P. aquilinum versus P. aquilinum var. latiusculum). Within the limited constrains of the database, this observation makes the assumption that compound variation occurred due to genetic variations less probable.

Effect of climatic gradient on the PTA contents in Pteridium has already been hypothesized previously, observing decreased PTA contents in South America in higher altitudes, that correlate with lower temperatures [204]. Lower PTA contents at lower temperatures could be due to restricted metabolic rate of the plants. Abrupt decrease of PTA towards North East (cooler climate) observed in the presented study could be likened to abrupt decrease of PTA between summer and autumn months. Nevertheless, the temporal variation demonstrated decrease of CAU contents in

47 autumn (cooler climate) but increase towards North-East (cooler climate). Climatic effects therefore do not correspond for the measured contents of CAU.

Toxicity of CAU It has been discussed above that the cyclopropane functional groups of PTE and CAU infer carcinogenicity of the compounds (Figure 8). Clinical symptoms similar to bovine enzootic hematuria were observed in several hundreds of cattle in area with no Pteridium in Uruguay. The symptoms are normally associated with Pteridium due to its PTA content [192]. The cattle were grazing in paddocks infested with Adiantopsis chlorophylla ferns (Pteridaceae) (Figure 15). Feeding trials by the plant reproduced the same symptoms in spontaneous cases.

Figure 15. Adiantopsis chlorophylla [120].

Six specimens were collected in the area, preserved and sent to Denmark by Prof. Franklin Riet-Correa (National Institute of Agricultural Research of Uruguay). Analysis revealed pterosin A and minor contents of CAU (Table 9). No PTA, PTE or corresponding pterosins were detected. The absence of PTA in the ferns suggests that the symptoms emerged due to the intake of CAU [120].

Table 9. Data of measured glycosides and pterosins in six specimens of Adiantopsis chlorophylla. The samples were extracted and measured in duplicates (±SD %). No PTE, PTA or corresponding pterosins were detected. (The provided per-sample numerical data is not published).

Sample No.: 1 2 3 4 5 6

Measured CAU not trace trace trace trace trace (mg/g DW) detected

Measured pterosin A 0.23 0.09 0.23 0.12 0.24 0.03 Measured

(mg/g (mg/g DW) (mg/g DW) ± 0.4% ± 0.2% ± 0.3% ± 0.1% ± 0.3% ± 0.2%

Corresponding CAU in living plants by 0.40 0.16 0.39 0.20 0.41 0.04

(applied 1:1molar ± 0.7% ± 0.3% ± 0.5% ± 0.3% ± 0.5% ± 0.3% Estimated (mg/g (mg/g DW) conversion ratio)

The data of Figure 8 demonstrates practical application of the method. The sample that contained the least estimated amount of CAU in living plants (sample No. 6) presumably lost it over the handling of sample, whereas the samples with higher CAU content maintained its detectable limits. (The ferns were dried for 40 days under natural conditions and sent in ambient temperature.) It is also notable that detection of the compound in non-pteridium Pteridaceae illustrates higher importance of the compound as potential environmental contaminant.

Comparison of compound stability in water

It was discussed in Chapter 1 that properties determining whether plant toxins are potential water contaminants are their abundance, toxicity, water solubility and aquatic stability. Given that the PTA-like compounds meet all the other criteria, the remaining factor largely determining their risk as potential water contaminants is aquatic stability.

Aquatic stability of PTA has been studied before. The results of different studies are summarized in Table 10. Data of the studies No. 1–3 performed in aqueous buffered solutions is not directly comparable, but it consistently shows that PTA is the most stable in neutral to medium acidic pH and degrades rapidly under all other pH ranges. Some variation between the results may be due to buffer effect of different applied buffer solutions [196]. The PTA is also consistently reported more stable in water with ever lower temperatures. Besides, sensitivity of PTA to higher than room temperatures is also known from observed decrease of the compound content in oven-dried plants [69,189]. The kinetic properties therefore imply that PTA endurance in natural water resources shall be facilitated by slightly acidic waters and cold weather.

Table 10. Summary of the results of the studies examining Stability of PTA in water. The half-lives of PTA from the study No. 1 (Table 7) were determined from the reported reaction rate constants, whereas from the studies No. 2–4 were read from published graphs.

Study Reaction kinetics Conditions Half-lives Media References No. (best fitting) T 22 ˚C: pH 2.88 (lowest studied) 16 hours pH 4.46 (compared with other T) 19 days pH 5.47 (optimal for stability) 78 days pH 8.93 (highest studied) 3 hours Pure aqueous Ayala-Luis et 1 First-order buffered solutions al. 2006 [196] pH 4.46: T 5 ˚C 160 days T 13 ˚C 60 days T 35 ˚C 7 days T 25 ˚C: pH 2.0 3 hours Pure aqueous Ojika et al. 2 pH 9.0 40 minutes Not reported buffered solutions 1987 [170] pH 10.0 6 minutes pH 11.0 1 minute pH 4.0 (lowest studied) 7 days Pure aqueous Saito et al. 3 pH 11.5 (highest studied) 7 minutes Not reported buffered solutions 1989 [118] (Room temperature) T 8 ˚C: pH 4.8–5.0 23 hours Zero order Lake water1 pH 7.2–7.8 23 hours Zero order Lake water2 Sandersen et 4 pH 7.2–8.2 37 hours Zero/first order Lake water3 al. 2016 [205] pH 7.2–8.2 95 hours Zero order Filtered lake water3 pH 6–6.5 27 hours First order Rain water T 15 ˚C: pH 7.0–7.8 60 hours Lake water4 pH 7.0–7.8 55 hours Filtered lake water4 pH 6.3–6.7 11 days Filtered, buffered lake water4 5 Manuscript 3 T 5 ˚C: Zero order pH 7.0–7.8 Est. 29 days Filtered lake water4 pH 6.3–6.7 Est. 62 days Filtered, buffered lake water4 pH 5.2–5.8 Est. 175 days Filtered lake water5 1Bøllemose (Denmark), September 2015. 2Skodsbord Dam (Denmark), March 2016. 3Skodsbord Dam, September 2015. 4Skodsbord Dam, December 2019. 5Bøllemose, December 2019.

Figure 16. PTA hydrolysis in aqueous solution as a function of pH (symbols represent measured data) [196]. The graph demonstrates that PTA in pure aqueous solutions is the most stable in the pH range approx. 4.5 to 6, and degrades exponentially faster in both increasing and decreasing pH. Results of the studies No. 2–3 (Table 10) confirm this trend.

Besides the conclusions that aquatic stability of PTA highly depends on temperature and pH, the published data does not examine PTA degradation kinetics as a function of other physicochemical properties of natural waters. One study providing limited data demonstrates that PTA degrades slower in filtered waters and differently among unfiltered waters sampled during different times of the year (Study No. 4, Table 10). Compound degradation in natural waters is highly influenced by microbial activity. Although this subject is not studied, results imply that microorganisms metabolize PTA as a carbon source.

The study No. 5 (Table 10) investigated stability of PTA, CAU and PTE based on the presence of microbial activity and different physicochemical parameters represented by distinct natural waters (Table 11). Effect of temperature and pH was also studied, since it has never been tested on CAU and PTE before. Table 11. The parameters of natural waters selected for stability studies of PTE, CAU and PTA.

Name and location Bøllemosen lake Skodsborg Dam lake (WGS84) (55.8269, 12.5639) (55.8177, 12.5659) pH 5.2 7.4 Turbidity, FNU (appearance) 2.6 (peat-brown) 0.3 (clear) Non-purgeable organic carbon, mg/L 38.1 13.3 Electric conductivity, µS/cm 65 210 Colony forming units in 1 ml* 64 between 121 and 340** * The microorganisms in both waters were present, but largely dormant due to sampling time (December 10) ** Uncertainty due to different results obtained in 21˚C and 37˚C incubation.

Table 12. Full-factorial experimental design used in compound stability in water analysis.

Parameter Value 1 (symbol) Value 2 (symbol) water origin Skodsborg Dam (A) Bøllemosen (B) temperature °C 15 5 pH Normalized to ~6.5 (+) Original pH of the lake (-) sterilization sterilized (0) not sterilized (1)

Full-factorial experimental design: Test No. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 A A B B B A B A A A B B B A B A Values 5 5 5 5 15 15 15 15 5 5 5 5 15 15 15 15 (parameter symbols) + + + + + + + + ------1 0 1 0 0 0 1 1 1 0 1 0 0 0 1 1

The waters were spiked with aqueous bracken extract resulting in starting concentrations of the compounds in experiments to 2.4, 0.22 and 0.8 mg/l of PTE, CAU and PTA, respectively. The measurements were taken at 0, 0.75, 2, 4, 6, 23, 72, 96 and 504 hours after the spike of the glycosides (Manuscript 3 [206]).

Experiments show that under the studied conditions the degradation rates between the three glycosides were very similar regardless the combination of the physicochemical conditions (4 out of the total 16 combinations are provided in Figure 17). In 10 out of 16 studied combinations of the conditions the compounds degraded before the last measurement. The degradation rate constants between all experiments until the second-last measurement (96 hours) were therefore used for comparison. Comparison of degradation rate constants was made by operating with the best-fitting zero order reaction rates (Figure 17).

Figure 17. Example of compound degradation rates under different conditions Comparison of the rate constants of all compounds in all experiments revealed that the compound stability depend on temperature (p˂0.001), pH (p˂0.001) and sterilization (p=0.001). Regardless the difference of physicochemical parameters between the waters (Table 11Error! Reference source not found.), the waters of different origin with pH values unified to pH 6.5 did not have an effect on the compound degradation rates (p=0.895). Graphical representation of the data and the reaction rate constants show that the three glycosides degraded in very similar rates. However, statistical analysis of the data combined from all 16 experiments confirmed that there was no difference between the average degradation rates of PTE and CAU (p=0.783), but PTA degraded on average 20% faster than the other two glycosides (p<0.05).

Figure 18. Comparison of average degradation rate constants of all 3 compounds in all 16 experiments.

Figure 18 illustrates that the compounds on average degraded significantly faster in Skodsborg Dam than in Bøllemosen. The parallel lines demonstrate that sterilization of waters had the same

53 effect on the reaction rate constants (Figure 18, left). The difference in the compound degradation rates between the two waters was therefore not due to the difference in microbial activity of the lakes. This indicates that other parameter had the effect on the difference. The average compound degradation rate constants in waters with pH adjusted to uniform value (pH ~6.5) were equal (cross-point in the Figure 18). This demonstrates that besides pH no other distinct inherent properties of the waters had an effect on the glycoside degradation rates. (Comparison of all other pairs of parameters demonstrate neither parallel, nor crossing lines, indicating inter-dependency of the other parameters. Full data is presented in Manuscript 3.) Results demonstrate that half-lives of the compounds in filtered natural waters are longer than half-lives reported in pure aqueous solutions (Table 10). The study of Manuscript 3 did not include stability studies in pure natural waters. Nevertheless, comparison with other studies implies that besides a destabilising factor (microbial activity) a stabilising factor in natural waters also exists. The factor cannot be explained by organic carbon, electric conductivity or turbidity, as waters with distinct values of these parameters but adjusted pH to a uniform value proved to have no effect on the compound stability (Figure 18). Throughout the experiments, formation of pterosins G and B were monitored (Supplementary material of Manuscript 3). The natural glycoside dissipation did not directly result in pterosin formation. Moreover, pterosin concentrations sometimes also followed downward trend due to their degradation. One previous study discussed that decomposition of PTA follows a complex route, forming unstable intermediate PTA dienone (Figure 8) and possibly other intermediate compounds [207]. That intermediate compounds of unspecified identity form in aqueous solutions it has also been seen as a part of this research. 1H NMR spectra of glycosides kept in water reveal formation of pterosins and unidentified compounds with CH3 groups (Figure 19). Nevertheless, the main purpose of the compound stability study was to compare the reaction rates of the three glycosides. The study of the fate of degrading compounds should be addressed by further research.

Figure 19. 1H NMR spectra of PTA standard (A) and the same standard after 2–3 days in room temperature (B).

Occurrence in water PTA is leachable from bracken fronds and is present in soil and water below bracken [70–72,94]. One study reported constant base-flow concentration of PTA over the bracken vegetation season in a stream feeding an active drinking-water reservoir in the United Kingdom (up to 0.006 µg/l). The base concentration temporarily peaked shortly after precipitation event (up to 2.280 µg/l) [72] (Study No. 2, Table 14). Maximum tolerable concentration of PTA in drinking-water theoretically estimated by increased mortality of 1 individual per 1 million of the life-time exposed individuals is 0.002 µg/L [94]. In order to measure PTE and CAU in waters by analytical method presented in Manuscript 1, an SPE sample purification and pre-concentration method was developed. The method is compared with the single previously published method of PTA and pterosin B pre-concentration in water samples (Jensen et al. 2008 [194]) in Table 13.

Table 13. Comparison of developed sample pre-concentration method with the single published method developed for compound pre-concentration from water samples.

Reference Jensen et al. 2008 [194] Manuscript 3 (applied in [70–72]) Technique SPE SPE and Evaporation Sample loading volume (ml) 20 20 Final volume (ml) / pre-concentration factor (times) 1 / 20 0.2 / 100 PTA 59.0 (±7.0) 99.2 (±3.6) PtrB 86.5 (±2.5) 71.5 (±1.4) CAU - 91.2 (±2.7) Recovery (%) PtrA - 96.2 (±1.0) PTE - 74.2 (±2.3) PtrG - 91.8 (±1.3)

Two types of water samples were taken adjacent to P. aquilinum from June to October of 2019: natural surface waters in Denmark and Sweden (study No. 5, Table 14) and drainage waters in Spain (study No. 6, Table 14). The waters were sampled and preserved based on the method published by Clauson-Kaas et al. 2016 [188]. In short, 40 ml of a samples were drawn into a sterile syringe in approx. 1 cm depth and 0.5 m distance from the shore. The samples were filtered through a 0.45 µm pore size sterile cellulose acetate filter into sterile tube and buffered by adding 1 ml of 0.3 M ammonium acetate adjusted with glacial acetic acid to pH 5.0. The samples were taken in triplicate, placed on ice and frozen within 8 hours in -20˚C. The pH for the sample preservation was optimised for stabilizing PTA against hydrolysis (pH ~4.5 to 6) [188,196], (Figure 16). Stability studies of PTE CAU and PTA presented in the previous section demonstrate that compound stability depend identically to change in pH. The pH ~4.5 to 6 is therefore regarded optimal for stability of PTE and CAU. The study No. 5 (Table 14) included 9-time sampling of a stream in 3 sampling spots, 9-time sampling of a nearby pond, and 2-time grab sampling of additional 3 lakes and 1 stream. The study No. 6 (Table 14) included 1-2 time grab sampling of waters from drainage channels and pits in

Asturias region in Spain. The samples from Asturias were collected and preserved by the same protocol, and kindly sent frozen to Denmark by Prof Elena María Fernández González (University of Oviedo, Spain).

Figure 20. Locations of sampled water. A: East Denmark and South-West Sweden (from left to right: Humleore, Gribskov, Helsingor, Asljunga). B: North-West Spain (Asturias).

Table 14. Comparison of results of the studies reporting PTA or PTA-like compounds in water resources.

Study Countries Sampling sites Detected Remarks References No. (All associated with adjacently PTE/CAU/PTA in growing P. aquilinum) water (μg/l) 3 sites; ground- and surface PTA: 0–1.02 Detected in 21 Clauson-Kaas et 1 Denmark water out of 41 al. 2014 [70] samples 1 primary site, few results from PTA: 0–2.28 Base Clauson-Kaas et other sites; surface streams concentration al. 2016 [72] United 0.006 µg/l. 2 Kingdom peaking to 2.2 µg/l after precipitation 3 sites; ground- and surface PTA: 0–1.25 Detected in 4 O'Driscoll et. al. 3 Ireland water out of 18 2016 [71] samples 2 sites, shallow groundwater PTA: 4–45 Detected in 2 Rasmussen 4 Denmark out of 2 2003 [94] samples 3 sites; surface water PTA: 0–0.02 Detected in 2 5* Denmark out of 22 samples Manuscript 3* 7 sites; surface water CAU: 0–0.007 Detected in 7 6* Spain PTE: 0–5.3 out of 9 samples * Detection of pterosins in Studies No. 5 and 6 is listed in Manuscript 3)

The comprehensive data of the studies No. 5 and 6 is provided in Table 15. The Humleore stream was sampled in 3 points (upstream, middle and downstream of the adjacent P. aquilinum stands), but the compounds were detected only in the downstream. The results in the Table 15 are therefore provided of the downstream measuring point only. It is notable that the results are unexpected, inconsistent and difficult to interpret. Since in some occasions marginal concentrations of certain bracken toxins were detected, it is suggested that the other compounds could also have been present, though below the detectable limits. The study No. 5 was made with intention to replicate the study No. 2 (Table 14) by monitoring a stream water. Unlike the results of the Study No. 2, the study No. 5 did not identify base or pulse concentrations of the compounds in the stream. The reason might be that the stream sampled in the study No. 2 was located beneath steep shores of dense bracken stands. Each rain event must have washed the PTA directly from Pteridium leaves into the water. The stream monitored in the study No. 5 was located in a flat terrain, with bracken fronds remote by 20-30 meters. Different hydraulic conditions can have resulted in different retention times between the contact of precipitation water with the plants to its entry into the surface water body. Another reason why the results of the Study No. 2 were not replicated by the study No. 5 might be different weather conditions. The season of sampling of the Study No. 5 was characterized by very frequent mild precipitation. The washable part of the compounds on the surface of the plants might not have had time enough to recover between the very frequent precipitation events, accumulated in soil, and therefore did not cause peak concentrations in the nearby stream. PTA has been detected in soil nearby Pteridium aquilinum by several previous studies in the range of 0–5 µg/g [175], 0.01–7.70 µg/g [208] and 0.008–0.52 µg/g [70], all studies reporting decrease of concentration with depth. No soil samples were analysed in the presented study, but it is suggested that the particular weather conditions could have facilitated the compound retention in soil. On the other hand, the waters in the Study No. 6 (Table 14) were not permanent surface water bodies, but precipitation-driven drainage channels and pits. According to the sample provider, the waters are accessible and occasionally used for drinking by cattle and horses in otherwise dry terrain. The water is also running down into the valleys, where it can possibly infiltrate water that is being utilised by humans. Detection of the reactive illudane glycosides indicate a possible pathway of animal poisoning apart from direct contact with the plants. Results of the Study No. 6 suggest that the particular plants present near the sampled water bodies contained significant contents of PTE. However, no plants from the region were analysed. It is notable that the samples in which PTE was detected (0.4 – 5.3 μg/), also contained 1.0 – 2.3 μg/ of pterosin G. Besides, the samples with non-detectable PTE contained 0.003 – 0.027 μg/ of pterosin G, implying marginal concentrations of PTE in those samples prior to its hydrolysis.

Table 15. Illudane glycosides and corresponding pterosins detected in water resources. Empty values represent ˂LOD

Water body Date Comments µg/l ± SD (DD.MM (precipitation PTE PtrG CAU PtrA PTA PtrB of 2019) over past 1/3 days (mm))1 07.06 8.2/8.8 trace 0.003 0.006 (±0.002) (±0.004) 14.06 1.9/19.0 0.002 (±0.001) 16.06 8.8/14.6 Stream 16.08 8.4/19.3 (Humleore)1 12.09 5.2/30.8 01.10 10.9/19.6 0.003 (±0.002) 08.10 0.2/0.5 18.10 5.9/13.7 10.11 0.0/2.0 07.06 8.2/8.8 0.014 0.021 (±0.004) (±0.009) 14.06 1.9/19.0 16.06 8.8/14.6 Pond 16.08 8.4/19.3 (Humleore)2 12.09 5.2/30.8 01.10 10.9/19.6 trace 08.10 0.2/0.5 18.10 5.9/13.7 10.11 0.0/2.0 14.08 4.9/6.0 trace Lake 13.10 20.9/55.9 0.002 (Gribskov)3 (±0.002) Lake 13.10 14.8/49.4 (Helsingor)4 10.11 0.6/3.2 Stream 31.07 0.0/22.7 trace (Asljunga)5 13.10 0.5/41.1 Lake 31.07 0.0/22.7 (Asljunga)6 13.10 0.5/41.1

Table 15 (continuation)

Water Date Comments µg/l ± SD body (DD.MM (Distinct sampling PTE PtrG CAU PtrA PTA PtrB of 2019) location No. and WGS 84 coordinates) 01.10 1 (43.22904, 0.001 -5.65816) 13.10 1 (43.22904, 0.013 0.007 0.003 -5.65816) (±0.002) (±0.001) (±0.001) 13.10 2 (43.22994, 5.3 1.2 -5.67083) (±4,6) (±1,5) 13.10 3 (43.22915, 0.027 Drainage -5.6754) (±0.003) ditches 13.10 4 (43.2272 0.003 0.001 and pits -5.65708) (±0.002) (Asturias)7 13.10 5 (43.22934, 0.4 1.2 -5.67141) (±0.2) (±1.9) 13.10 6 (43.22749, 0.4 2.2 0.007 -5.67983) (±0.09) (±0,8) (±0.004) 13.10 7 (43.22739, -5.68088) 13.10 8 (43.23083, 0.4 1.0 -5.67636) (±0.2) (±0.4) 1 Compounds measured in nearby Pteridium in previous season by Manuscript 2 (prior to hydrolysis): 0.02, 0.29 and 0.45 of PTE, CAU and PTA respectively. The period with no measurements between 16.06 and 16.08 is due to dry bed. 2 Compounds measured in nearby Pteridium in previous season by Manuscript 2 (prior to hydrolysis): 0.04, 0.29 and 1.01 of PTE, CAU and PTA respectively. 3 Compounds measured in nearby Pteridium in previous season by Manuscript 2 (prior to hydrolysis): 0.02, 0.36 and 0.68 of PTE, CAU and PTA respectively. 4 Compounds measured in nearby Pteridium in previous season by Manuscript 2 (prior to hydrolysis): 0.22, 1.79 and 6.53 of PTE, CAU and PTA respectively. 5,6 Compounds measured in nearby Pteridium in previous season by Manuscript 1 (prior to hydrolysis): 0.44, 0.19 and 1.90 of PTE, CAU and PTA respectively. 7 All water originated by recent precipitation (precipitation-fed surface water bodies). Water body distance to P. aquilinum: ~2 m. No data of compound content in the plants is available. The waters are accessible for drinking by cattle and horses. 8 Precipitation data collected from the closest meteorological stations Ringsted(1,2), Gribskov(3), Helsingor(4) (dmi.dk/vejrarkiv) and Orkelljunga(5,6) (smhi.se/data/meteorologi/nederbord).

______

The case study showed that PTE and CAU of Pteridium and other Pteridaceae are important compounds from the perspective of chemical ecology. It provided analytical standards and methods for their further studies. Due to absence of robust water sampling strategy, the concept of compound leaching to water was not explained. Nevertheless, the study provided proof of concept of the compound leaching to water, tools for a parallel relevant study1, and information that can be used for further studies. A part of results also indicate the need of assessment and official categorisation of toxicity of PTE and CAU, as IARC since 1987 has only been classifying the bracken ferns and its constituent PTA. Besides the developed methods and compound stability studies, the following contribution to subject of Pteridium and its constituents PTA-like compounds have been made (Figure 21, bold): 2

Bracken carcinogenicity referred by other studies usually only to PTA

Elucidation of Elucidation Elucidation PTA [165] [166] of CAU [178] of PTE [180] 1983 1997 2010

1893 1965 1972 1987 1990 1992 2005 2011 2014-2016 2020

Bracken IARC classifies PTA and pterosin B in

carcinogenic to bracken possibly water bodies [70–72] farm animals carcinogenic to

[149] humans [156] Pterosins A, B and CAU carcinogenic to

G in meat [164] farm animals [120] Bracken toxic to Increased odds-ratio of (Schneider de Oliveira farm animals human gastric cancer et al. 20202). [152] and residence in PTA and pterosin bracken-infested areas B in soil [86] [161] Milk of bracken-fed Milk carcinogenicity CAU, PTE, pterosins A and cattle reported explained by its G in water bodies [206] carcinogenic [185] constituent PTA [186] (Manuscript 3). Figure 21. The chronological order of the first reports regarding the toxicity of bracken, PTA-like compounds and the medium to which the compounds can be carried over.

Chapter 3. Pyrrolizidine alkaloids from Petasites species (Case 2)

It was discussed in Chapter 1 that physicochemical properties of toxic alkaloids make them potential water contaminants. Alkaloids have also been reported to leach from the terrestrial plant ragwort (genus Senecio) [209,210]. Common butterbur (Petasites hybridus (L.) P.Gaertn., B.Meyer & Scherb) is a plant that typically produces much higher biomass per unit land area than common ragwort, and produces similar types of toxic alkaloids. Moreover, Petasites hybridus needs highly moist soil and proliferates alongside freshwater resources. This plant and the toxic alkaloids it produces will therefore be discussed in this chapter from the perspective of potential water contamination by the alkaloids.

Genus Petasites

Petasites (family Asteraceae) is a genus of flowering plants with 24 taxonomically accepted species, subspecies and variations. The most common are Petasites albus Gaertn., Petasites hybridus, Petasites japonicus (Siebold & Zucc.) Maxim., and Petasites tricholobus Franch. [125]. The genus can spread by both wind-dispersed seeds and vegetatively. It needs moist soils and proliferates in damp forests, marshes, the shores of streams, ponds and ditches [211,212]. It mostly prefers fertile humic soils and partial shade, but in very moist soils it can grow in full sun. Petasites is also an indicator of nutrient-rich, well-oxygenated soils [213,214].

Most species of Petasites are native to Asia or Southern Europe and have spread or were introduced to other continents. The expansion of the plant is conditioned by both its invasive nature and by deliberate cultivation. Petasites extracts for medicinal purposes have been used for more than 2000 years [215]. Petasites tricholobus has been used for treatment of palsy, hypertension, coughs and snake-bites in China. Petasites japonicus (alternative name – bog rhubarb) for treatment of allergy and asthma in Korea, and later in Europe [216]. Petasites hybridus was used for treatment of rashes, arthritis, kidney and bladder stones, and it was believed to be effective against plague [213,217,218].

Petasites hybridus

Petasites hybridus (Figure 22) is originally native to Southern Europe and Western Asia. Nowadays it is very common in most of Europe and can be encountered in North America as well [125,215,216] (Figure 23). The species was first introduced to the Northern Europe around 1350 as a medicinal plant by monks [217,219,220]. The plant gradually expanded from cultivation sites in monasteries, castles and estates to natural habitats. The species bloom from April to May, its leaves mature up to 70 cm width on 1 m high stems and vegetate until late autumn [217]. In Northern Europe the plant is recognised as invasive but no mitigation operations exist. Moreover, it is a part of local folklore and is traditionally planted around garden lakes and ponds. With the

61 aid of thick powerful rhizomes and large leaves that shade other vegetation, the plant can form large monocultures [217,218].

Figure 22. Petasites hybridus at its flowering stage in late April (left) and decaying stage in early November (right) (Denmark).

Figure 23. Registered distribution of Petasites hybridus according to the Global Biodiversity Information Facility [125]. The observation points are mainly registered by volunteers worldwide and accepted by the administration of the facility. Therefore, the locations with no registration do not indicate the absence of the species.

Medicinal and pharmaceutical industries have scientific and economic interest in the plant. Due to anti-spasmodic and anti-inflammatory effects of its constituents sesquiterpenes petasin and isopetasin, plant extracts have applications where relaxation of muscle and vascular spasms are desired. The biologically active sesquiterpenes are found in rhizomes, roots and leaves of the plant. The most well documented evidence-based therapeutic application is treatment of migraine headaches, allergic rhinitis and urinary tract spasms [216,217,221,222]. However, application of

62 the plant extracts for therapeutic purposes are documented to cause severe liver damage (acute hepatitis, acute liver failure, hepatopathy, choledocholitiasis) [221]. In addition to bioactive sesquiterpenes, the plant contains flavonoids, furanoeremophilanes, tannins and pyrrolizidine alkaloids [213,218]. The effect of liver damage is exhibited by pyrrolizidine alkaloids [216,217, 221,223,224].

Alkaloids

Alkaloids are biologically active secondary metabolites produced by bacteria and animals, but primarily by fungi and plants [46,225–229]. In plants, alkaloids play important roles for protection against microbes, insects, herbivores and as allelopathic chemicals (against other plant species). Many plant alkaloids can serve specific biological functions non-related to defence of the host organisms, whereas the functions of certain alkaloids are not yet fully understood. Many plant alkaloids express profound physiological effects on humans and animals [230,231]. Bioactive alkaloids pose scientific and economic interest for medicinal and pharmaceutical industries, and are therefore well-studied compounds [46]. Nowadays more than 10,000 alkaloids have been isolated and identified [232]. Most of isolated alkaloids in their pure form are non-volatile, colourless, crystalline solids and tend to express bitter taste [46]. Typically encountered heterocyclic rings in their chemical structures determine environmental stability, whereas ample hydrogen bonds result in high water solubility.

Alkaloid-containing plants were almost certainly the first remedies applied by humans as analgesics, sedatives, stimulants, poisons, protection agents and recreational drugs [46]. Nowadays, carefully controlled doses of alkaloids are successfully applied for killing human pathogens (e.g., chelerythrine), widening blood vessels (e.g., vincamine), relieving asthma symptoms (e.g., ephedrine), stimulating hearth rhythms (e.g., quinidine) and for multiple other therapeutic functions. Many alkaloids act on central nervous system (e.g., caffeine, codeine, morphine, nicotine) [46,233]. Some are abused due to properties able to cause elation, inebriation or hallucinations (e.g. cocaine, psilocybin) [234]. Plant extracts rich in atropine, coniine, ergotamine, and strychnine have been used for homicide since ancient times. Strychnine and anabasine were also applied as early insecticides. The reason that they were later abandoned in agriculture is that the currently applied synthetic pesticides are less toxic to humans (!) [235].

A uniform definition of alkaloids does not exist, but the broadest term in use is “naturally occurring chemical compounds that mostly contain basic nitrogen atoms” [236]. Due to an extremely large structural variety, the term alkaloid is a very broad and vague term. Alkaloids can be further classified by their biological origin, biogenetic pathways, physicochemical properties or molecular structures. For the purpose of chemical ecology, they will be further classified by the molecular structures. Two major groups of molecular structures are heterocyclic alkaloids (containing nitrogen atom in a heterocyclic ring) and non-heterocyclic alkaloids (containing nitrogen atom in

a non-heterocyclic ring) [237]. Heterocyclic alkaloids is a larger group and it encompasses the compounds that were categorized in Chapter 1 as potential water contaminants. The structural sub-groups of heterocyclic alkaloids are termed by the ring types that they encompass (e.g. tropane, pyrrolizidine).

Pyrrolizidine alkaloids

Plant alkaloids that currently pose public concern for food safety in European Union are opium alkaloids (categorized by origin), tropane alkaloids and pyrrolizidine alkaloids [238–246]. While opium alkaloids are one of the smallest problems in food safety, the pyrrolizidine alkaloids (PAs) are often found at unsafe levels [244,247–251]. Out of these three types of alkaloids that are mentioned throughout European food legislation, a study on physiochemical properties of toxins produced by European plants – not related to food safety – distinguish only the PAs as potential water contaminants (see Chapter 1) (Figure 24).

At least several hundreds of different PAs are known and they are regarded among the most common natural toxins produced by plants as defence chemicals against insects and herbivores [243,252–254]. It is estimated that 3% of all flowering plants (e.g. Asteraceae, Apocynaceae, Boraginaceae, Euphorbiaceae, Fabaceae, Gramineae, Leguminosae, Orchidaceae, Ranunculaceae, Scrophulariaceae) contain at least one PA [244,232,255]. The PA-content of plant material can vary from trace amounts up to 19 % of dry weight [244]. Plants with high PA contents are reported responsible for cattle death in many countries [227,256].

Alkaloids of European plants, with physicochemical properties making them potential water contaminants

[54] (discussed in Chapter 1)

terpenoid, steroidal, quinolizidine, isoquinoline, indole, amaryllidaceae, pyrrolizidine, tropane, opium

Alkaloids that are regulated in the EU in food, or the proposed legislation is in place [238–246]

Figure 24. Overlap of food-regulated alkaloids and alkaloids reported as potential water contaminants

Since PAs are exclusively produced by plants, this could imply that the source of PA poisoning is exclusively food or feed derived from such plants. Indeed, the highest PA-levels occur in edible products and mixtures that contain PA-producing plants. Reported outbreaks of human poisonings with PAs are mostly linked with consumption of herbal teas and remedies [244,247]. However, plant-based food such as crops or salads that contain no inherent PAs can be accidentally contaminated at harvesting by PA-producing weeds and their seeds [244]. At least one massive

64 outbreak of acute liver poisoning is reported due to PAs in wheat flour accidentally contaminated with a PA-producing plant [248].

However, besides the direct poisoning, PAs are also reported in non-plant food. These compounds can be carried in the food chain from plants to animal products like milk and dairy products, eggs and meat. A recent study of over 1000 samples of foods collected at the point of sale in different European countries reported the presence of one or more PAs in 91 % of the teas (on average 460 μg/kg in dry teas) and 60 % in plant-based food supplements (highly variable concentrations). Besides, the study reported relatively low levels of PAs in 6% of all milk samples (0.05–0.17 μg/l) and in 1% of all eggs (~0.1 μg/kg) [249]. Other studies reported traces of PAs in cheese, yoghurt and meat (offal) [250], and from traces to considerable concentrations in 94% of honeys (from 1 to 1000 μg/kg) [251]. The origin of high PA contents in honey and other bee products was verified to be due to high PA contents in flowering plants visited by the bees (e.g. genera Echium, Senecio, Eupatorium). Beekeepers are encouraged to select or manage locations for the production of honey based on the absence of these plants [251].

The mode of PA toxicity depends on the specific molecular structure. PAs with an unsaturated bond in 1,2 position of the two-ring structure (Figure 25) (hereafter called unsaturated PAs) express nucleophilic properties. Therefore, ingested unsaturated PAs metabolically oxidise and react with proteins and nucleic acids [257,258,259,260] resulting in hepatotoxic, mutagenic and carcinogenic effects [46,243,244,260–264]. Low level consumption of unsaturated PAs can cause cumulative health damage in the long-term [216]. The lowest known doses associated with acute short-term toxicity of unsaturated PAs in humans are reported 3000 μg PA per kg of body weight per day (lethal outcome of a boy exposed for 4 days) and 800–1700 μg PA per kg of body weight per day (non-lethal outcome of an acute hepatic veno-occlusive disease associated with liver cirrhosis and high mortality, of a girl exposed for a 2 week-period) [244]. It has been reported that daily intake of just approximately 7 μg of PAs via herbal tea during pregnancy had no toxic effect on the mother’s liver, but damaged the foetal liver at such extent that the new-born died after two days [216].

Figure 25. Examples of PAs encountered in food [249]. The unsaturated part of a two-ring structure expressing nucleophilic properties is encircled on the left-most example.

Due to the genotoxic and carcinogenic nature of unsaturated PAs, EFSA states that it would be inappropriate to establish minimal safe doses for their intake [265]. EFSA therefore apply lower and upper bounds of a mean dietary exposure based on the observed carcinogenicity, that spans from 0 to 0.048 μg of daily intake per kg of body weight for different age groups [265]. WHO recommends minimising exposure of unsaturated PAs to humans as much as possible [266], whereas the Federal Institute for Risk Assessment (Germany) and the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (UK) suggest a universal tolerable level for all age groups of 0.007 μg of unsaturated PA daily intake per kg of bodyweight [243].

Occurrence in water The first study measuring PAs as environmental contaminants was published very recently (2019), reporting individual unsaturated PAs in the range of 3–1350 μg/kg in soil and 4–270 μg/l in water of a small lake [209] (the total measured PAs are provided in Table 16, Study No. 1). The compounds were reported to leach directly from a terrestrial plant common ragwort (Senecio jacobaea L.). This plant is also reported to cause fatal poisonings to horses and is among the plants suggested to be avoided for beekeepers due to resulting alarming PA contaminations in honey [251,267]1

For the purpose of the present study, an area infested with Petasites hybridus was selected in Eastern Denmark, containing both surface water and ground water wells. By the time of the study the only fully validated method for the PA quantification in water existed [209]. The method (LC- MS/MS) was applicable to detect all of the PAs by means of specific fragmentation patterns of the compounds, and was selected as appropriate for the purpose (instrumental LOD 2–7 µg/l, LOQ 5– 9 µg/l; 1000 time SPE pre-concentration factor with 90% recovery [209], resulting in 1000 times lower LOD and LOQ of the total method). Each of the 21 PAs detected in Study No. 3 (Table 16) were quantified against exact equivalent external standards (purchased from Phytolab, Germany). The results are summarized in Table 16 (Study No. 3).

1 Alternative name of Senecio jacobaea L. is Jacobaea vulgaris (Gaertn.)

Table 16. Summary of all studies reporting PAs in water resources.

Study Country, description Range detected Detected PAs Remarks References No. of sampling site (Total PAs µg/l) Jacoline, Jacoline N- oxide, Riddelline, Erucifoline, Seneciphylline N-oxide, Denmark. Pond Hama and adjacent to dense Erucifoline-N-oxide, Two sampling 1 116–529 Strobel 2019 Senecio Jacobaea Riddelline N-oxide, campaigns [209] meadows Integerrimine N-oxide, Senecionine N-oxide, Jacobine N-oxide, Seneciphylline (Σ11) Retrorsine, Retrorsine N-oxide, Senecionine, Switzerland. Stream Senecionine N-oxide, One sampling Gunthardt et 2 in proximity to 0.13 Seneciphylline, campaign al. 2020 [210] Senecio jacobaea Seneciphylline N-oxide, Senkirkine (Σ11) Jacobine, Jacobine N- Five sampling oxide, Monocrotaline, campaigns of a Monocrotaline N- stream (4 sites), Stream: oxide, Senecionine, three sampling 0.05 – 0.52 Senecionine N-oxide, campaigns of 4 Senecivernine, wells at 2 different Wells (2.2–3.0 m): Senecivernine N-oxide, water horizons. 0.02 – 0.23 Senkirkine, Erucifoline, Denmark. Stream Retrorsine, Retrorsine Peak concentration and water wells 3 Control wells N-oxide, registered in a Manuscript 4 adjacent to Petasites (no adjacent plant): Seneciphylline, stream after hybridus 0.00 Echimidine, precipitation event. Echimidine-N-oxide, Deep groundwater Lasiocarpine, Europine, More of PAs were (~60 m): Europine N-oxide, detected in waters 0.00 Heliotrine, than in the Intermedine, precursor plant Intermedine N-oxide (due to PA (Σ21) transformation)

Out of the 21 unsaturated PAs reported in both water types (Study No. 3, Table 16), the first 9 in the list dominated in all sampling events. These compounds also matched with the PAs detected in leaves and rhizomes of a nearby Petasites hybridus. Additional 12 PAs were detected occasionally in relatively low concentrations and could not have been directly linked to the plant. However, analysis of water in a control shallow groundwater wells, remote from Petasites hybridus by approx. 150–200 m, revealed that the waters contained no PAs. It was therefore concluded that the main 9 PAs in both water types leached from Petasites hybridus directly, whereas the extra 12 PAs formed indirectly as a result of PA transition and decay. No PAs were detected in deep groundwater in the area (~60 m depth) that supposedly occurs under impermeable geological layers and originates from remote sources (Manuscript 4 [268]).

The above mentioned study confirmed leaching of PAs from plants to surface water, and for the first time linked it with an invasive Petasites hybridus, that is very commonly encountered nearby freshwater bodies. The reported surface waters can pose a risk to aquatic life, land and farm animals accessing the water, and potentially farm animal products. The study also illustrated that leaching of PAs to surface waters depends on precipitation. Moreover, the study for the first time reported plant toxins in groundwater wells. According to the acceptable levels of unsaturated PAs proposed by the British and German health authorities (0.007 μg of daily intake per kg of bodyweight) [243], 2 l of the well water reported in this study (0.23 μg/l) would violate the daily ingestion limit for an average healthy human being.

General conclusions

Analytical standards and methods necessary for studying the selected terpene glycosides in water and plants were created. The methods applied for measuring the compounds in Pteridium demonstrated that the average combined contents of CAU and PTE in the plants are comparable to the average contents of PTA. Besides, PTE and CAU express higher water solubility and the same aquatic stability as PTA. Subsequently, these compounds and their hydrolysis products were detected in surface waters adjacent to Pteridium aquilinum.

The estimated tolerable concentration of PTA in drinking-water is 0.002 μg/l. The highest previously reported concentration of PTA in a stream feeding an active drinking-water reservoir is 2.28 µg/l, exceeding the tolerable limit by 1100 times. It can be assumed that the levels of carcinogenicity of waters assessed by the previous studies based on PTA were likely underestimated as PTE and CAU were not included in the quantification. The highest concentration of the total PTE, CAU and PTA concentrations measured by the present study in water was 5.3 µg/l (drainage water in Spain, accessible for drinking to horses and cattle). A simultaneous side study showed that cattle ingesting plants containing CAU developed clinical symptoms that are normally associated to cattle ingesting plants that contain PTA. Applying assumption that tolerable ingestion levels of CAU and PTE are the same as of PTA, the reported studied water, if used for drinking, would exceed the estimated tolerable limit by up to 2650 times.

The other part of the study reported up to 0.53 μg/l of unsaturated PAs in surface waters and up to 0.23 μg/l in shallow groundwater wells (2.2–3.0 m depth). The measurements in the plants and in the control wells confirmed that the PAs originated from an adjacently growing invasive plant species Petasites hybridus. Taking into consideration the tolerable levels of unsaturated PA intake proposed by the German and British health authorities (0.007 μg of daily intake per kg of bodyweight for all age groups), 1 L of the surface water or 2 L of well water reported in this study would violate the safety limit for an average healthy human. Moreover, European Food Safety Authority proposes tolerable levels of unsaturated PAs that span from 0 to 0.048 μg of daily intake per kg of body weight for different age groups, and WHO recommends minimizing PA exposure to humans as much as possible.

Identification of PAs in shallow water wells indicate a true risk to users of broad range of water resources. Many shallow groundwater wells are exploited in Europe in rural areas or summerhouses remote from centralized water supplies [269]. Contamination of many of these wells with nitrate have previously demonstrated that their quality is particularly vulnerable to seepage water moving in the ground. Given that 3% of all flowering plants produce at least one PA, and that the PAs are washable from the plants by precipitation, the compounds are likely to enter the shallow water wells and pose direct health risks. It is also well identified that pesticides being spread on land surface often contaminate deep groundwater resources below impermeable

69 geological levels. The contamination is usually transported through poorly maintained or old deep groundwater wells that become non-hermetic. No PAs in deep groundwater were detected by the presented study, indicating proper maintenance of the particular wells. Nevertheless, identification of PAs in seepage water of the constructions of deep groundwater wells demonstrates previously unknown risk of deep groundwater contamination with PAs by intrusion of seepage water into deep groundwater through improperly maintained wells.

Two types of toxic plant compounds were detected in water bodies out of the two options carefully selected for investigation. Many more plant toxins exist that could be potential water contaminants than those experimentally analysed. Some biologically active plant compounds applied as pharmaceuticals or stimulants (e.g. morphine, caffeine) are frequently reported in water resources due to their presence in outlets of wastewater treatment plants or other wastewater sources. Their presence in water resources demonstrates that certain compounds of plant origin can be very persistent. Many plant toxins are water-soluble, environmentally persistent and toxic (such as heterocyclic alkaloids, steroids, triazoles and polyketides). It is suggested that many more biologically active compounds produced by plants may directly leach into water resources and be undetected by contemporary drinking-water analyses.

The reasons why the persistent plant toxins like PAs were not detected in water wells before might have been a lack of attempts, but also lack of sensitive analytical techniques. Discovery of the PAs in water can be comparable to the discovery of anthropogenic Persistent Mobile Organic Compounds (PMOCs). Many PMOCs have been detected in water only recently due to the lack of available analytical techniques. Currently, only the methods with mass spectrometric detection provide the prerequisites needed to analyse the terpene glycosides and the PAs at the reported concentrations in water, and little attempts to analyse water for these compounds are reported.

This study indicates the need of further research investigating the degree of water contamination with plant toxins. If plant toxins were detected in water reservoirs exploited for drinking, it is recommended that regulatory bodies manage vegetation around the resources to prevent waterborne human exposure.

References

[1] World Health Organisation, “Guidelines for drinking-water quality: fourth edition incorporating the first addendum (ISBN: 978-92-4-154995-0),” Geneva, 2017.

[2] Paracelsus, “Die dritte Defension wegen des Schreibens der neuen Rezepte" Septem Defensiones 1538,” Darmstadt, Werke Bd. 2, 1965.

[3] Evans S., Andrews, D., Stoiber, T., Naidenko, O., “PFAS Contamination of Drinking Water Far More Prevalent Than Previously Reported,” 22 January 2020. [Online]. Available: https://www.ewg.org/research/national-pfas-testing/. [Accessed 14 March 2020].

[4] United States Environmental Protection Agency, “Basic Information on PFAS,” [Online]. Available: https://www.epa.gov/pfas/basic-information-pfas. [Accessed 18 April 2020].

[5] Ministry of Environment of Denmark, “Administrative considerations and determination of limit values for perfluorinated alkyl acid compounds (PFAS compounds), incl. PFOA, PFOS and PFOSA in drinking water, as well as soil and groundwater for assessment of contaminated soils,” 2015. [Online]. Available: https://mst.dk/media/91517/pfas-administrative-graensevaerdier-27-april- 2015-final.pdf. [Accessed 14 March 2020].

[6] Andrews D., “EWG Proposes PFAS Standards That Fully Protect Children’s Health,” 6 May 2019. [Online]. Available: https://www.ewg.org/research/ewg-proposes-pfas-standards-fully-protect- children-s-health. [Accessed 14 March 2020].

[7] Schmidt U., Struck S., Gruening B., Hossbach J., Jaeger I.S., Parol R., Lindequist U., Teuscher E., Preissner R., “SuperToxic: a comprehensive database of toxic compounds,” Nucleic Acids Res., vol. 37, no. Database issue, pp. 295-299, 2009.

[8] Ames B.N., Profet M., Gold L.S., “Dietary pesticides (99.99% all natural),” Proc Natl Acad Sci U S A, vol. 87, no. 18, p. 7777–7781, 1990.

[9] MassBank Europe, “High Quality Mass Spectral Database,” [Online]. Available: https://massbank.eu/MassBank. [Accessed 03 03 2020].

[10] UN General Assembly, “Transforming our world: the 2030 Agenda for Sustainable Development (A/RES/70/1),” 21 October 2015. [Online]. Available: https://www.refworld.org/docid/57b6e3e44.html. [Accessed 26 March 2020].

[11] Su M.J., Coulter A.R., Sutherland S.K., Chang C.C., “The presynaptic neuromuscular blocking effect and phospholipase A2 activity of textilotoxin, a potent toxin isolated from the venom of the Australian brown snake, Pseudonaja textilis,” Toxicon, vol. 21, no. 1, pp. 143-151, 1983.

[12] Ushkaryov Y.A., Volynski K.E., Ashton A.C., “The multiple actions of black widow spider toxins and theirselective use in neurosecretion studies,” Toxicon, vol. 43, p. 527–542, 2004.

[13] Wink, M., “Mode of action and toxicology of plant toxins and poisonous plants,” in Wirbeltierforschung in der Kulturlandschaft , Mitt. Julius Kühn-Inst, 2009, pp. 93-112.

[14] Wink M., “Plant breeding: importance of plant secondary metabolites for protection against pathogens and herbivores,” Theor. Appl. Genet., vol. 75, p. 225–233, 1988.

[15] Teuscher E., Lindequist U., Biogene Gifte. Biologie – Chemie – Pharmakologie., Stuttgart: Gustav Fischer, 1994.

[16] Horowitz B.Z., “Botulinum toxin,” Crit Care Clin., vol. 21, no. 4, pp. 825-839, 2005.

[17] World Health Organisation, “Botulism,” 10 January 2018. [Online]. Available: https://www.who.int/en/news-room/fact-sheets/detail/botulism. [Accessed 5 April 2020].

[18] Yokoyama A., Murata M., Oshima Y., Iwashita T., Yasumoto T., “Some Chemical Properties of Maitotoxin, a Putative Calcium Channel Agonist Isolated from a Marine Dinoflagellate,” J. Biochem., vol. 104, no. 2, p. 184–187, 1988.

[19] Chai P.R., Hayes B.D., Erickson T.B., Boyer E.W., “Novichok agents: a historical, current, and toxicological perspective,” Toxicology Communications, vol. 2, no. 1, p. 45–48, 2018.

[20] Wada O., “Health Impact of Dioxins,” JMAJ , vol. 45, no. 7, p. 312–315, 2002.

[21] Tokuyama T., Daly J., Witkop B., “The structure of batrachotoxin, a steroidal alkaloid from the Colombian arrow poison frog, Phyllobates aurotaenia, and partial synthesis of batrachotoxin and its analogs and homologs,” J Am Chem Soc., vol. 91, no. 14, pp. 3931-3938, 1969.

[22] Nagai H., “Recent Progress in Jellyfish Toxin Study,” Journal of Health Science, vol. 49, no. 5, p. 337–340, 2003.

[23] European Food Safety Authority, “Ricin (from Ricinus communis) as undesirable substances in animal feed ‐ Scientific Opinion of the Panel on Contaminants in the Food Chain,” EFSA Journal, vol. 6, no. 9, p. 726, 2008.

[24] Ushkarev I.A., Grishin E.V., “Neurotoxin of the black widow spider and its interaction with receptors from the rat brain,” Bioorg. Khim., vol. 12, pp. 71-80, 1986.

[25] Fisher Scientific, “Material Safety Data Sheet of Tetrodotoxin,” [Online]. Available: https://fscimage.fishersci.com/msds/01139.htm. [Accessed 5 April 2020].

[26] Wong J.H., Ng T.B., “Toxins from Basidiomycete Fungi (Mushroom): Amatoxins, Phallotoxins, and Virotoxins,” in Handbook of Biologically Active Peptides, Elsevier, 2006, pp. 131-137.

[27] The University of Oxford Department of Chemistry, “Safety data for aflatoxin B1,” 19 August 2003. [Online]. Available: http://msds.chem.ox.ac.uk/AF/aflatoxin_B1.html. [Accessed 5 April 2020].

[28] Matsuzawa M., Graziano M.J., Casida J.E., “Endothal and cantharidin analogs: relation of structure to herbicidal activity and mammalian toxicity,” J. Agric. Food Chem., vol. 35, no. 5, pp. 823-829, 1987.

[29] DrugBank bioinformatics and cheminformatics database , “Fentanyl Identification,” [Online]. Available: https://www.drugbank.ca/drugs/DB00813. [Accessed 5 April 2020].

[30] ATSDR - Agency for Toxic Substances and Disease Registry, “Toxicological profile for white prosphorus,” U.S. Public Health Service in collaboration with U.S. Environmental Protection Agecy (EPA), 1997.

[31] Fu Y.J. Yin L.T., Wang W., Chai B.F., Liang A.H., “ Synthesis, expression and purification of a type of chlorotoxin-like peptide from the scorpion, Buthus martensii karsch, and its acute toxicity analysis,” Biotechnology Letters, vol. 27, no. 20, pp. 1597-1603, 2005.

[32] Jones M.M., Schoenheit J.E., Weaver A.D., “Pretreatment and heavy metal LD50 values,” Toxicology and Applied Pharmacology, vol. 49, no. 1, pp. 41-44, 1979.

[33] The University of Oxford Department of Chemistry, “Safety data for arsenic,” 7 December 2005. [Online]. Available: https://web.archive.org/web/20110114204809/http://msds.chem.ox.ac.uk/AR/arsenic.html. [Accessed 5 April 2020].

[34] Saxena P.N., Anand S., Saxena N., Bajaj P., “Effect of arsenic trioxide on renal functions and its modulation by Curcuma aromatica leaf extract in albino rat,” J Environ Biol, vol. 30, no. 4, pp. 527- 531, 2009.

[35] Internationally Peer Reviewed Chemical Safety Information (INCHEM) (WHO), “Diamorphine,” September 1993. [Online]. Available: http://www.inchem.org/documents/pims/pharm/pim261f.htm. [Accessed 5 April 2020].

[36] EXTOXNET Extension Toxicology Network. A Pesticide Information Project of Cooperative Extension Offices of Cornell University, Michigan State University, Oregon State University, and University of California, “Demeton-s-methyl,” September 1995. [Online]. Available: http://pmep.cce.cornell.edu/profiles/extoxnet/carbaryl-dicrotophos/demeton-s-methyl- ext.html#9. [Accessed 6 April 2020].

[37] Gill D.M., “Bacterial toxins: a table of lethal amounts,” Microbiol Rev., vol. 46, no. 1, p. 86–94, 1982.

[38] Bradberry S.M., Dickers K.J., Rice P., Griffiths G.D., Vale J.A., “Ricin Poisoning,” Toxicol Rev, p. 65– 70, 2003.

[39] James Martin Center for Nonproliferation Studies, “Chronology of Incidents Involving Ricin,” 29 February 2008. [Online]. Available: https://web.archive.org/web/20141216133309/http://cns.miis.edu/reports/ricin_chron.htm. [Accessed 16 March 2020].

[40] Schep L.J., Temple W.A., Butt G.A., Beasley M.D., “Ricin as a weapon of mass terror — Separating fact from fiction,” Environ. Int., vol. 35, no. 8, pp. 1267-1271, 2009.

[41] Chandra Sekhar J., Sandhya S., Vinod K.R., Banji D., Sudhakar K., Chaitanya R.S.N.A.K.K., “Plant toxins-useful and harmful effects,” Hygeia.J.D.Med., vol. 4, no. 1, pp. 79-90, 2012.

[42] van Wyk B.E., Wink M., Phytomedicines, Herbal Drugs, and Poisons, University of Chicago Press, 2015.

[43] van Wyk B.E., Wink M., Medicinal Plants of the World, Second edition, Pretonia: Briza Publications, 2017.

[44] “Solanine poisoning,” BMJ, vol. 2, no. 6203 ( https://doi.org/10.1136/bmj.2.6203.1458-a), p. 1458– 1459, 1979.

[45] Wink M., Hartmann T., “Sites of Enzymatic Synthesis of Quinolizidine Alkaloids and Their Accumulation in Lupinus polyphyllus,” Zeitschrift für Pflanzenphysiologie, vol. 102, no. 4, pp. 337- 344, 1981.

[46] Aniszewski, T., Alkaloids: chemistry, biology, ecology, and applications. Second edition., Elsevier Science, 2015.

[47] Retief F.P., Cilliers L., “Poisons, poisoning and the drug trade in ancient Rome,” Akroterion, vol. 45, no. 1, pp. 88-100, 2000.

[48] Jones, D.E., Poison Arrows: North American Indian Hunting and Warfare, University of Texas Press, 2009.

[49] Bisset N.G., “Arrow and dart poisons,” J Ethnopharmacol., vol. 25, no. 1, pp. 1-41, 1989.

[50] Wadley L., Sievers C., Bamford M., Goldberg P., Berna F., Miller C., “Middle Stone Age Bedding Construction and Settlement Patterns at Sibudu, South Africa,” Science, vol. 334, no. 6061, pp. 1388-1391, 2011.

[51] Haber A.I., Sustache J.R., Carr D.E., “A generalist and a specialist herbivore are differentially affected by inbreeding and trichomes in Mimulus guttatus,” Ecosphere, vol. 9, no. 2, p. e02130, 2018.

[52] Mithofer A., Maffei M.E., “General Mechanisms of Plant Defense and Plant Toxins,” in Gopalakrishnakone P., Carlini C., Ligabue-Braun R. (eds) Plant Toxins. Toxinology, Dordrecht, Springer, 2017, pp. 1-22.

[53] International Agency for Research on Cancer (IARC) - Summaries & Evaluations, “Bracken fern (Pteridium aquilinum) and some of its constituents,” updated 1998. [Online]. Available: http://www.inchem.org/documents/iarc/vol40/brackenfern.html. [Accessed 14 April 2020].

[54] Gunthardt B.F., Hollender J., Hungerbuhler K., Scheringer M., Bucheli T.D., “Comprehensive Toxic Plants−Phytotoxins Database and Its Application in Assessing Aquatic Micropollution Potential,” J. Agric. Food Chem., vol. 66, no. 29, pp. 7577-7588, 2018.

[55] Moreno J.A., Peinado R.A., Enological Chemistry, Cordoba: Elsevier Academic Press, 2012.

[56] Bhat S.V., Nagasampagi B.A., Sivakumar M., Chemistry of Natural Products, Springer, 2005 .

[57] “Potato Glycoalkaloids: Chemistry, Analysis, Safety, and Plant Physiology,” Crit Rev Plant Sci, vol. 16, no. 1, pp. 55-132, 1997.

[58] Cherney E.C., Baran P.S., “Terpenoid‐Alkaloids: Their Biosynthetic Twist of Fate and Total Synthesis,” Isr. J. Chem., vol. 51, no. 3-4, pp. 391-405, 2011.

[59] Patel S., Nag, M.K., Daharwal S.J., Singh M.R., Singh D., “Plant Toxins: An Overview,” Research J. Pharmacology and Pharmacodynamics, vol. 5, no. 5, pp. 283-288, 2013.

[60] Picardo M., Filatova D., Nunez O., Farre M., “Recent advances in the detection of natural toxins in freshwater environments,” Trends Anal. Chem., vol. 112, pp. 75-86, 2019.

[61] Gupta V.K., Sharma B., “Forensic Applications of Indian Traditional Toxic Plants and their Constituents,” orensic Research & Criminology International J., vol. 4, no. 1, pp. 1-6, 2017.

[62] van der Heijden, K., Younes M., Fishbein M., Miller S. (editors), International Food Safety Handbook: Science, International Regulation, and Control, CRC Press, 1999.

[63] Wink, M., “Evolution of secondary metabolites from an ecological and molecular phylogenetic perspective,” Phytochemistry, vol. 64, no. 1, pp. 3-19, 2003.

[64] Wittstock U., Gershenzon J., “Constitutive plant toxins and their role in defense against herbivores and pathogens,” Curr Opin Plant Biol., vol. 5, no. 4, pp. 300-307, 2002.

[65] Rates S.M.K., Betti A.H., Muller L.G., Nunes J.M., “Plant Toxins as Sources of Drugs,” in Gopalakrishnakone P., Carlini C., Ligabue-Braun R. (eds) Plant Toxins, Dordrecht, Springer, 2017, pp. 81-105.

[66] Ma L.Q., Komar K.M., Tu C., Zhang W., Cai Y., Kennelley E.D., “A fern that hyperaccumulates arsenic,” Nature, vol. 409, p. 579, 2001.

[67] Ghosh P, Rathinasabapathi B., Ma L.Q., “Arsenic-resistant bacteria solubilized arsenic in the growth media and increased growth of arsenic hyperaccumulator Pteris vittata L.,” Bioresour. Technol., vol. 102, no. 19, pp. 8756-8761, 2011.

[68] Yang C.Y., Ho Y.N., Makita R., Inoue C., Chien M.F., “A multifunctional rhizobacterial strain with wide application in differentferns facilitates arsenic phytoremediation,” Sci. Total Environ., vol. 712, no. 134504, pp. 1-11, 2020.

[69] Rasmussen, L.H., Pedersen, H.Æ., “Screening for Ptaquiloside in Ferns: Using Herbarium Speciments for Qualitative Mapping Purposes,” Phytochem Anal, vol. 28, no. 6, pp. 575-583, 2017.

[70] Clauson-Kaas, F., Jensen, P.H., Jacobsen, O.S., Juhler, R.K., Hansen, H.C., “The naturally occurring carcinogen ptaquiloside is present in groundwater below bracken vegetation,” Toxicol Environ Chem., vol. 33, pp. 1030-1034, 2014.

[71] O'Driscoll C., Ramwell C., Harhen B., Morrison L., Clauson-Kaas F., Hansen H.C.B., Campbell G., Sheahan J., Misstear B., Xiao L.W., “Ptaquiloside in Irish Bracken Ferns and Receiving Waters, with Implications for Land Managers,” Molecules, vol. 543, no. 21, 2016.

[72] Clauson-Kaas F., Ramwell C., Hansen, H.C.B., Strobel B.W., “Ptaquiloside from bracken in stream water at base flow and during storm events,” Water Res., vol. 106, pp. 155-162, 2016.

[73] Manners G.D., “Plant Toxins: The Essences of Diversity and a Challenge to Research,” in Advances in Experimental Medicine and Biology, volume 391, New York, Plenum Press, 1996, pp. 9-35.

[74] Lee J., Cho K.S., Jeon Y., Kim J.B., Lim y., Lee K., Lee I.S., “Characteristics and distribution of terpenesin South Korean forests,” Journal of Ecologyand Environment, vol. 41, no. 19, pp. 1-10, 2017.

[75] Maurya, A.K., Patel, R.C. & Frost, C.J., “Acute toxicity of the plant volatile indole depends on herbivore specialization,” J Pest Sci, vol. 93, p. 1107–1117, 2002.

[76] Schmitt M.H., Shuttleworth A., Shrader A.M., Ward D., “The role of volatile plant secondary metabolites as pre-ingestive cues and potential toxins dictating diet selection by African elephants,” Oikos, vol. 129, p. 24–34, 2019.

[77] Bais H.P., Weir T.L., Perry L.G., Gilroy S., Vivanco J.M., “The role of root exudates in rhizosphere interactions with plants and other organisms,” Annu Rev Plant Biol., vol. 57, pp. 233-266, 2006.

[78] Badri D.V., Vivanco J.M., “Regulation and function of root exudates,” Plant, Cell and Environment , vol. 32, p. 666–681, 2009.

[79] Korenblum E., Dong Y., Szymanski J., Panda S., Jozwiak A., Massalha H., Meir S., Rogachev I., Aharoni A., “Rhizosphere microbiome mediates systemic root metabolite exudation by root-to- root signaling,” PNAS, vol. 117, no. 7, pp. 3874-3883, 2020.

[80] Rodriguez, A.M., Derita, M.G., Borkosky, S.A., Socolsky, C., Bardon, A., Hernandez, M.A.,, “Bioactive farina of Notholaena sulphurea (Pteridaceae): morphology and histochemistry,” Flora Morphol. Distrib. Funct. Ecol. Plants, vol. 240, p. 144–151, 2018.

[81] Lange B.M., “The Evolution of Plant Secretory Structures and Emergence of Terpenoid Chemical Diversity,” Annu. Rev. Plant Biol., vol. 66, p. 139–159, 2015.

[82] Langenheim, J.H., Plant resins: chemistry, evolution, ecology and ethnobotany, Portland: Cambridge: Timber Press., 2003.

[83] Gopalakrishnakone P., Carlini C., Ligabue-Braun R. (eds), Plant Toxins. Toxinology., Dordrecht: Springer, 2017.

[84] Rasmussen L.H., Schmidt B., Sheffield E., “Ptaquiloside in bracken spores from Britain,” Chemosphere, vol. 90, no. 10, pp. 2539-2341, 2013.

[85] Metz G., Bocher D., Metz J., “IgE-mediated allergy to castor bean dust in a landscape gardener,” Contact Dermatitis, vol. 44, no. 6, p. 367, 2001.

[86] Rasmussen L.H., Lauren D., Hansen H.C.B., “Sorption, degradation and mobility of ptaquiloside, a carcinogenic Bracken (Pteridium sp.) constituent, in the soil environment,” Chemosphere, vol. 58, no. 6, pp. 823-835, 2005.

[87] Garcia-Jorgensen D.B., Hansen H.C.B., Abrahamsen, P., Diamantopoulos E., “A novel model concept for modelling the leaching of natural toxins: results for the case of ptaquiloside,” Environ Sci Process Impacts, vol. 22, pp. 1768-1779, 2020.

[88] Bucheli T.D., “Phytotoxins: environmental micropollutants of concern?,” Environ Sci Technol., vol. 44, no. 22, pp. 13027-13033, 2014.

[89] Food and Agriculture Organization of the United Nations, “Pesticides indicators,” 2020. [Online]. Available: http://www.fao.org/faostat/en/#data/EP/visualize. [Accessed 11 April 2020].

[90] Rasmussen, L. H., Donnelly, E., Strobel, B. W., Holm, P. E., Hansen, H. C. B., “Land management of bracken needs to account for bracken carcinogens – A case study from Britain,” J. Environ. Manage., vol. 151, pp. 258-266, 2015.

[91] Shahin, M., Smith, B.L., Prakash, A.S., “Bracken carcinogens in the human diet,” Mutat. Res., vol. 443, p. 69–79, 1999.

[92] Alonso-Amelot, M.E., Avendano, M., “Human carcinogenesis and bracken fern: a review of the evidence,” Curr. Med. Chem., p. 675–686, 2002.

[93] Alonso-Amelot, M.E., Avendano M., “Possible association between gastric cancer and bracken fern in Venezuela: An epidemiologic study,” Int. J. Cancer., vol. 9, pp. 252-259, 2001.

[94] Rasmussen L. H., “Ptaquiloside – an environmental hazard?,” Chemistry Department, The Royal Veterinary and Agricultural University, Denmark, Frederiksberg, 2003.

[95] “Council Directive 98/83/EC of 3 November 1998 on the quality of water intended for human consumption,” Official Journal of the European Union, vol. 330, pp. 32-54, 1998.

[96] Government of Canada. Document for Public Consultation, “Guidance on Natural Organic Matter in Drinking Water,” 15 March 2019. [Online]. Available: https://www.canada.ca/en/health- canada/programs/consultation-organic-matter-drinking-water/document.html#b21. [Accessed 3 May 2020].

[97] Villacorte, L.O., Algal blooms and Membrane Based Desalination Technology, Delft: CRC Press/Balkema, 2014.

[98] Schenzel J., Hungerbühler K., Bucheli T.D., “Mycotoxins in the Environment: II. Occurrence and Origin in SwissRiver Waters,” Environ. Sci. Technol., vol. 46, no. 24, pp. 13076-13084, 2012.

[99] Reemtsma T., Berger U., Arp H.P.H., Gallard H., Knepper T.P., Neumann M., Quintana J.B., de Voogt P., “Mind the Gap: Persistent and Mobile Organic Compounds—Water Contaminants That Slip Through,” Environ. Sci. Technol., vol. 50, no. 19, pp. 10308-10315, 2016.

[100] Angeles L.F., Aga D.S., “Catching the Elusive Persistent and Mobile Organic Compounds: Novel Sample Preparation and Advanced Analytical Techniques,” Trends in Environmental Analytical Chemistry, Available online 14 December 2019, https://doi.org/10.1016/j.teac.2019.e00078.

[101] Niemuth N.J., Klaper R.D., “Emerging wastewater contaminant metformin causes intersex and reduced fecundity in fish,” Chemosphere, vol. 135, pp. 38-45, 2015.

[102] Bunke D., Moritz, S., Brack, W., Herráez D.L., Posthuma L., Nuss M., “Developments in society and implications for emerging pollutants in the aquatic environment,” Environ. Sci. Eur., vol. 31, pp. 1- 17, 2019.

[103] Prasad M., Vithanage M., Kapley A. (Editors), Pharmaceuticals and Personal Care Products: Waste Management and Treatment Technology, 1st Edition, Emerging Contaminants and Micro Pollutants, Butterworth-Heinemann, 2019.

[104] Stuart M., Lapworth D., Crane E., Hart A., “Review of risk from potential emerging contaminants in UK groundwater,” Sci Total Environ, vol. 416, pp. 1-21, 2012.

[105] Turner R.D.R., Warne M.S., Dawes L.A., Thompson K., Will G.D., “Greywater irrigation as a source of organic micro-pollutants to shallow groundwater and nearby surface water,” Sci Total Environ, vol. 669, pp. 570-578, 2019.

[106] Boxall A.B.A., “The environmental side effects of medication,” EMBO Rep, vol. 5, no. 12, pp. 1110- 1116, 2004.

[107] Gao J., Li J., Jiang G., Yuan Z., Eaglesham G., Covaci A., Mueller J.F., Thai P.K., “Stability of alcohol and tobacco consumption biomarkers in a real rising main sewer,” Water Research, vol. 138, pp. 19-26, 2018.

[108] Lian L., Yan S., Yao B., Chan S., Song W., “Photochemical transformation of nicotine in wastewater effluent,” Environ. Sci. Technol., vol. 51, no. 20, pp. 11718-11730, 2017.

[109] Jensen, S., Johnels A. G., Olsson M., Otterlind G., “DDT and PCB in Marine Animals from Swedish Waters,” Nature, vol. 224, p. 247–250, 1969.

[110] Furlong E.T., Batt A.L., Glassmeyer S.T., Noriega M.C., Kolpin D.W., Mash H., Schenck K.M., “Nationwide reconnaissance of contaminants of emerging concern in source and treated drinking waters of the United States: Pharmaceuticals.,” Sci Total Environ, vol. 579, pp. 1629-1642, 2017.

[111] Narberhaus, I., Zintgraf, V. & Dobler, S., “Pyrrolizidine alkaloids on three trophic levels – evidence for toxic and deterrent effects on phytophages and predators,” Chemoecology, vol. 15, p. 121–125, 2005.

[112] Trigo, J.R., “Effects of pyrrolizidine alkaloids through different trophic levels,” Phytochemistry Reviews, vol. 10, p. 83–98, 2011.

[113] Harkes P., Verhoeven A., Sterken M.G., Snoek L.B., van den Elsen S.J.J. Mooijman P.J.W. Quist C.W., Vervoort M.T.W., Helder J., “The differential impact of a native and a non‐native ragwort species (Senecioneae) on the first and second trophic level of the rhizosphere food web,” Oikos, vol. 126, p. 1790–1803, 2017.

[114] Benelli G., Pavela R., Petrelli R., Cappellacci L., Santini G., Fiorini D., Sut S., Dall’Acqua S., Canale A., Maggi F., “The essential oil from industrial hemp (Cannabis sativa L.) by-products as an effective tool for insect pest management in organic crops,” Ind Crops Prod, vol. 122, pp. 308-315, 2018.

[115] Scalerandi E., Flores G.A., Palacio M., Defago M.T, Carpinella M.C., Valladares G., Bertoni A., Palacios S.M., “Understanding Synergistic Toxicity of Terpenes as Insecticides: Contribution of Metabolic Detoxification in Musca domestica,” Front. Plant Sci., vol. 9, no. 1579, pp. 1-9, 2018.

[116] Thomson J.A., “Towards a taxonomic revision of Pteridium (Dennstaedtiaceae),” Telopea (Syd.), vol. 10, no. 4, pp. 793-803, 2004.

[117] Potter D.M., Baird M.S., “Carcinogenic effects of ptaquiloside in bracken fern and related compounds,” Br J Cancer., vol. 87, no. 3, pp. 914-920, 2000.

[118] Saito K., Nagao T., Matoba M., Koyama K., Natori S., Murakami T., SaikI Y., “Chemical assay of ptaquiloside, the carcinogen of pteridium aquilinum, and the distribution of related compounds in the pteridaceae,” Phytochemistry, vol. 28, no. 6, pp. 1605-1611, 1989.

[119] O’Connor P.J., Alonso-Amelot M.E., Roberts S.A., Povey A.C., “The role of bracken fern illudanes in bracken fern-induced toxicities,” Mutat Res Rev Mutat Res., vol. 782, no. 108276, pp. 1-39, 2019.

[120] Schneider de Oliveira L.G., Boabaid F.M., Kisielius V., Rasmussen L.H., Buroni F., Lucas M., Schild C.O., Lopez F., Machado M., Riet-Correa F., “Hemorrhagic diathesis in cattle due to consumption of Adiantopsis chlorophylla (Swartz) Fée (Pteridaceae),” Toxicon: X, vol. 5, no. 100024, 2020.

[121] Saito K., Nagao T., Takatsuki S., Koyama K., Natori S., “The sesquiterpenoid carcinogen of bracken fern, and some analogues, from the pteridaceae,” Phytochemistry, vol. 29, no. 5, pp. 1475-1479, 1990.

[122] Smith B.L., Embling P.P., Lauren D.R., Agnew M.P., Ross A.D., Greentree P.L., “Carcinogen in rock fern (Chielanthes sieberi) from New Zealand and Australia,” Aust Vet J., vol. 66, no. 7, pp. 230-231, 1989.

[123] Page C.N., “The taxonomy and phytogeography of bracken—a review,” Bot J Linn Soc, vol. 73, no. 1-3, p. 1–34, 1976.

[124] “The Plant List,” An interinstitutional working list database for plant species, [Online]. Available: http://www.theplantlist.org/tpl1.1/search?q=Pteridium. [Accessed 1 March 2020].

[125] Global Biodiversity Information Facility, [Online]. Available: https://www.gbif.org/. [Accessed 18 03 2020].

[126] Der J.P., Thomson J.A., Stratford J.K., Wolf P.G., “Global chloroplast phylogeny and biogeography of bracken (Pteridium; Dennstaedtiaceae),” Am. J. Bot., vol. 95, no. 5, pp. 1041-1049, 2009.

[127] Marrs R.H., Watt A.S., “Biological Flora of the British Isles: Pteridium aquilinum (L.) Kuhn,” J. Ecol., vol. 94, no. 6, pp. 1272-1321, 2006.

[128] Holland J., Aplin P., “Super-resolution image analysis as a means of monitoring bracken (Pteridium aquilinum) distributions,” ISPRS J Photogramm Remote Sens, vol. 75, pp. 48-63, 2013.

[129] Matongera T.N., Mutanga O., Dube T., Sibanda M., “Detection and mapping the spatial distribution of bracken fern weeds using the Landsat 8 OLI new generation sensor,” Int J Appl Earth Obs Geoinf, vol. 57, pp. 93-103, 2017.

[130] Pakeman R.J., Marrs R.H., Howard D.C., Barr C.J., Fuller R.M., “The bracken problem in Great Britain: its present extent and future changes,” Appl Geogr, vol. 16, pp. 65-86, 1996.

[131] Aranha P., “The fate of ptaquiloside in dairy cattle,” University of Copenhagen, 2013.

[132] Whitehead S.J., Digby J., “The morphology of bracken (Pteridium aquilinum (L) Kuhn) in the North York Moors - a comparison of the mature stand and the interface with heather (Calluna vulgaris (L) Hull),” Ann. appl. Biol., vol. 131, pp. 103-131, 1997.

[133] Sheffield E., Wolf P.G., Haufler C.H., “How big is a bracken plant?,” Weed Res., vol. 29, no. 6, pp. 455-460, 1989.

[134] Wang H.H., Chen B.J., Hsu L.M., Cheng Y.M., Liou Y.J., Wang C.Y., “Allelopathic effects of bracken fern (Pteridium aquilinum L. Kuhn) in Taiwan,” Allelopathy J., vol. 27, no. 1, pp. 97-110, 2011.

[135] Gil da Costa, R.M., Bastos, M.M.S.M., Oliveira, P.A., Lopes, C., “Bracken-associated human and animal health hazards: chemical, biological and pathological evidence,” J Hazard Mater, vol. 1, pp. 203-204, 2012.

[136] Cooper-Driver G., Finch S., Swain T., Bernays E., “Seasonal variation in secondary plant compounds in relation to the palatability of Pteridium aquilinum,” Biochem. Syst. Ecol., vol. 5, no. 3, pp. 177- 183, 1977.

[137] Jones C.G., Firn R.D., “The role of phytoecdysteroids in bracken fern,Pteridium aquilinum (L.) Kuhn as a defense against phytophagous insect attack,” J. Chem. Ecol., vol. 4, p. 117–138, 1978.

[138] Fenwick G.R., “Bracken (Pteridium aquilinum)—toxic effects and toxic constituents,” J. Sci. Food Agric, vol. 46, no. 2, pp. 147-173, 1989.

[139] Mohammad R.H., Nur-E-Alam M., Lahmann M., Parveen I., Tizzard G.J., Coles S.J., Fowler M., Drake A.F., Heyes D., Thoss V., “Isolation and characterisation of 13 pterosins and pterosides from bracken (Pteridium aquilinum (L.) Kuhn) rhizome,” Phytochemistry, vol. 128, pp. 82-94, 2016.

[140] Stephen R., Gliessman F.L.S., “Allelopathy in a broad spectrum of environments as illustrated by bracken,” Bot. J. Linn. Soc., vol. 73, no. 1-3, pp. 95-104, 1976.

[141] Gava A., da Silva Neves D., Gava D., de Moura S.T., Schild A.L., Riet-Correa F., “Bracken fern (Pteridium aquilinum) poisoning in cattle in southern Brazil,” Vet Hum Toxicol, vol. 44, no. 6, pp. 362-365, 2002.

[142] Smith B.L., “Bracken fern (genus Pteridium) toxicity - a global problem,” in Poisonous plants and related toxins by T. Acamovic, C.S. Stewar and T.W. Pennycott (editors), CABI Publishing, 2004, pp. 227-240.

[143] Cunningham I.J., “Bracken fern poisoning,” New Zeal. Vet. J., vol. 1, no. 2, pp. 36-37, 1952.

[144] Barnett K.C., Watson W.A., “Bright blindness in sheep. A primary retinopathy due to feeding bracken (Pteris aquilina),” Res Vet Sci., vol. 11, no. 3, pp. 289-290, 1970.

[145] Alonso-Amelot M.E., Oliveros-Bastidas A., “Kinetics of the natural evolution of hydrogen cyanide in plants in neotropical Pteridium arachnoideum and its ecological significance,” J Chem Ecol., vol. 31, no. 2, pp. 315-331, 2005.

[146] Harwood D.G,. Palmer N.M., Wessels M.E., Woodger N.G., “Suspected bracken poisoning in pigs,” Vet Rec., vol. 160, no. 26, pp. 914-915, 2007.

[147] Harper H.A., “Carbohydrate metabolism in thiamine deficiency,” J. Biol. Chem., pp. 239-248, 1942.

[148] Cebra C.K., Cebra M.L., “Altered mentation caused by polioencephalomalacia, hypernatremia, and lead poisoning,” Vet Clin Food Anim , vol. 20, p. 287–302, 2004.

[149] Evans., I.A., Mason. J., “Carcinogenic Activity of Bracken,” Nature, vol. 208, p. 913–914, 1965.

[150] “The Radiomimetic Nature of Bracken Toxin,” Cancer Res., vol. 28, no. 11, p. 2252–2256, 1968.

[151] Pamukcu, A.M., Olson, C., Price, J.M., “Assay of Fractions of Bovine Urine for Carcinogenic Activity after Feeding Bracken Fern (Pteris aquilina),” Cancer Res., vol. 26, no. 1, pp. 1745-1753, 1966.

[152] Storrar D.M., “Cases of vegetable poisoning in cattle,” Journal of Comparative Pathology and Therapeutics, vol. 6, pp. 276-279, 1893.

[153] Yamada, K., Ojika, M., Kigoshi H., “Ptaquiloside, the major toxin of bracken, and related terpene glycosides: chemistry, biology and ecology,” Nat Prod Rep., vol. 24, p. 798–813, 2007.

[154] Vetter J., “A biological hazard of our age: bracken fern [Pteridium aquilinum (L.) Kuhn] - a review,” Acta Vet. Hung., vol. 57, no. 1, p. 183–196, 2009.

[155] International Agency for Research on Cancer (IARC), “Bracken fern (Pteridium aquilinum) and some of its constituents,” 1998. [Online]. Available: http://www.inchem.org/documents/iarc/vol40/brackenfern.html. [Accessed 12 January 2020].

[156] “Agents Classified by the IARC Monographs, Volumes 1–123,” 2018. [Online]. Available: https://monographs.iarc.fr/agents-classified-by-the-iarc/. [Accessed 12 January 2020].

[157] Hirono I., “Edible Plants Containing Naturally Occurring Carcinogens in Japan,” Jpn. J. Cancer Res., vol. 84, no. 10, p. 997–1006, 1993.

[158] Recouso R.C., Stocco dos Santos R.C., Freitas R., Santos R.C., de Freitas A.C., Brunner O., Beçak W., Lindsey C.J., “Clastogenic effect of bracken fern (Pteridium aquilinum v. arachnoideum) diet in peripheral lymphocytes of human consumers: preliminary data,” Vet Comp Oncol, vol. 1, no. 1, pp. 22-29, 2003.

[159] Chen Y., Zhao Y., Hu Y., Wang L., Ding Z., Liu Y., Wang J., “Isolation of 5-hydroxypyrrolidin-2-one and other constituents from the young fronds of Pteridium aquilinum,” J Nat Med., vol. 62, no. 3, pp. 358-359, 2008.

[160] Mahmood S., Smith B.L., Prakash A.S., “Bracken carcinogens in the human diet,” Mutat. Res., vol. 443, p. 69–79, 1999.

[161] Galpin, O.P., Whitaker, C.J., Whitaker, R., Kassab, J.Y., “Gastric cancer in Gwynedd. Possible links with bracken,” Br J Cancer., vol. 61, pp. 737-740, 1990.

[162] Virgilio, A., Sinisi, A., Russo, V., Gerardo, S., Santoro, A., Galeone, A., Taglialatela-Scafati, O., Roperto, F., “Ptaquiloside, the major carcinogen of bracken fern, in the pooled raw milk of healthy sheep and goats: an underestimated, global concern of food safety,” J Agric Food Chem., vol. 63, pp. 4886-4892, 2015.

[163] Alonso-Amelot M.E., Castillo U., Smith B.L., Lauren D.R., “Bracken ptaquiloside in milk,” Nature, vol. 382, p. 587, 1996.

[164] Fletcher M.T., Reichmann K.G., Brock I.J., McKenzie R.A., Blaney B.J., “Residue Potential of Norsesquiterpene Glycosides in Tissues of Cattle Fed Austral Bracken (Pteridium esculentum),” J. Agric. Food Chem., vol. 59, pp. 8518 - 8523, 2011.

[165] van der Hoeven, J.C, Lagerweij, W.J, Posthumus, M.A, van Veldhuizen, A., Holterman, HA , “Aquilide A, a new mutagenic compound isolated from bracken fern (Pteridium aquilinum (L.) Kuhn),” Carcinogenesis, vol. 4, no. 12, pp. 1587-1590, 1983.

[166] Niwa, H., Ojika, M., Wakamatsu, K., Yamada, K., Ohba, S., Saito, Y., Hirono, I., Matsushita, K., “Stereochemistry of ptaquiloside, a novel norsesquiterpene glucoside from bracken, Pteridium aquilinum var. latiusculum,” Tetrahedron Lett., vol. 24, no. 48, pp. 5371-5372, 1983.

[167] Matoba M., Saito E., Saito K., Koyama K., Natori S., Matsushima T., Takimoto M., “Assay of ptaquiloside, the carcinogenic principle of bracken, Pteridium aquilinum, by mutagenicity testing in Salmonella typhimurium,” Mutagenesis, vol. 2, p. 419–423, 1987.

[168] Hirono I., Ogino H., Fujimoto M., Yamada K., Yoshida Y., Ikagawa M., Okumura M., “Induction of Tumors in ACI Rats Given a Diet Containing Ptaquiloside, a Bracken Carcinogen,” J. Natl. Cancer Inst., vol. 79, p. 1143–1149, 1987.

[169] Gomes J., Magalhaes A., Michel V., Amado I.F., Aranha P., Ovesen R.G., Hansen H.C.B., Gartner F., Reis C.A., Touati E., “Pteridium aquilinum and its ptaquiloside toxin induce DNA damage response in gastric epithelial cells, a link with gastric carcinogenes,” Toxicol. Sci., vol. 126, pp. 60-71, 2012.

[170] Ojika M, Wakamatsu K, Niwa H, Yamada K, “Ptaquiloside, a potent carcinogen isolated from bracken fern Pteridium aquilinum var. latiusculum: structure elucidation based on chemical and spectral evidence, and reactions with amino acids, nucleosides, and nucleotides,” Tetrahedron, vol. 43, no. 22, pp. 5261-5274, 1987.

[171] Kushida T, Uesugi M, Sugiura Y, Kigoshi H, Tanaka H, Hirokawa J, Ojika M, Yamada K, “DNA damage by ptaquiloside, a potent bracken carcinogen: detection of selective strand breaks and identification of DNA cleavage products,” J. Am. Chem. Soc, vol. 116, no. 2, p. 479–486, 1994.

[172] Prakash A.S., Pereira T.N., Smith B.L., Shaw G., Seawright A.A., “Mechanism of bracken fern carcinogenesis: evidence for H-ras activation via initial adenine alkylation by ptaquiloside,” Nat Toxins., vol. 4, no. 5, pp. 221-227, 1996.

[173] Kigoshi H., Tanaka H., Hirokawa J., Mizuta K., Yamada K., “Synthesis of analogues of a bracken ultimate carcinogen and their DNA cleaving activities,” Tetrahedron Lett., vol. 33, no. 44, pp. 6647- 6650, 1992.

[174] Rasmussen L.H., Hansen H.C.B, “Growth of bracken in Denmark and the content of ptaquiloside in fronds, rhizomes and roots,” in Poisonous plants and related toxins. Stewart C.S., Pennycott T.W., Acamovic T., Wallingford, CABI Publishing, 2004, pp. 354-360.

[175] Engel P., Brandt K.K., Rasmussen L.H., Ovesen R.G., Sørensen J., “Microbial degradation and impact of Bracken toxin ptaquiloside on microbial communities in soil,” Chemosphere, vol. 67, no. 1, pp. 202-209, 2007.

[176] Gomes J., Magalhaes, A., Carvalho, A.S., Hernandez, G.E., Papp, S.L., Head, S.R., Michel, V., David, L., Gartner, F., Touati, E., Reis, C.A., “Glycophenotypic Alterations Induced by Pteridium aquilinum in Mice Gastric Mucosa: Synergistic Effect with Helicobacter pylori Infection,” PLOS ONE, vol. 7, no. 6, pp. 1-10, 2012.

[177] Fukuoka M., Kuroyanagi M., Yoshihira K., Natori S., “Chemical and Toxicological Studies on Bracken Fern, Pteridium aquilinum var. latiusculum. II. Structures of Pterosins, Sesquiterpenes having 1- Indanone Skeleton,” Chem. Pharm. Bull, vol. 26, no. 8, pp. 2365-2385, 1978.

[178] Castillo U.F., Wilkins A.L., Lauren D.R., Smith B.L., Towers N.R., Alonso-Amelot M.E., Jaimes- Espinoza R., “Isoptaquiloside and caudatoside, illudane-type sesquiterpene glucosides from Pteridium aquilinum var. caudatum,” Phytochemistry, vol. 44, no. 5, pp. 901-906, 1997.

[179] Castillo U.F., Ojika M., Alonso-Amelot M., Sakagami Y., “Ptaquiloside Z, a new toxic unstable sesquiterpene glucoside from the neotropical bracken fern Pteridium aquilinum var. caudatum,” Bioorg. Med. Chem., vol. 6, no. 11, p. 2229–2233, 1998.

[180] Fletcher M.T., Hayes P.Y., Somerville M.J., De Voss J.J., “Ptesculentoside, a novel norsesquiterpene glucoside from the Australian bracken fern Pteridium esculentum,” Tetrahedron Lett., vol. 51, pp. 1997-1999, 2010.

[181] Somvanshi R., Lauren D.R., Smith B.L., Dawra R.K., Sharma O.P., Sharma V.K., Singh A.K., Gangwar N.K., “Estimation of the fern toxin, ptaquiloside, in certain Indian ferns other than bracken,” Curr. Sci., vol. 91, no. 11, pp. 1547-1552, 2006.

[182] Fletcher M.T., Brock I.J., Reichmann K.G. McKenzie R.A. Blaney B.J., “Norsesquiterpene Glycosides in Bracken Ferns (Pteridium esculentum and Pteridium aquilinum subsp. Wightianum) from Eastern Australia: Reassessed Poisoning Risk to Animals,” J. Agric. Food Chem., vol. 59, pp. 5133- 5138, 2011.

[183] Wiberg K.B., “Cyclobutane—physical properties and theoretical studies,” in The Chemistry of Cyclobutanes. Rappoport J. Liebman F. (Editors), New Jersey, John Wiley & Sons, Ltd., 2005.

[184] Castillo U.F., Sakagami Y., Alonso-Amelot M., Ojika M., “Pteridanoside, the first protoilludane sesquiterpene glucoside as a toxic component of the neotropical bracken fern Pteridium aquilinum var. caudatum,” Tetrahedron, vol. 55, no. 42, pp. 12295-12300, 1999.

[185] Evans I.A., Jones R.S., Mainwaring-Burton R., “Passage of bracken fern toxicity into milk,” Nature, vol. 237, no. 5350, pp. 107-108, 1972.

[186] Alonso Amelot M.E. Castillo U., Sanchez M.D., DeJongh F., “Detection of ptaquiloside the potent carcinogen of bracken fern Pteridium aquilinum in bovine milk,” Am Chem Soc, vol. 203, no. 1-3, p. 89, 1992.

[187] Alonso-Amelot M.E., Castillo U., De Jongh F., “Passage of the bracken fern carcinogen ptaquiloside into bovine milk,” Lait, vol. 73, no. 3, pp. 323-332, 1993.

[188] F. Clauson-Kaas, H.C. Hansen, B.W. Strobel, “UPLC-MS/MS determination of ptaquiloside and pterosin B in preserved natural water,” Anal. Bioanal. Chem., vol. 408, p. 7981–7990, 2016.

[189] Caceres-Pena YC, Naya M, Calcagno-Pissarelli MP, Alonso-Amelot ME, “Influence of bracken fern (Pteridium caudatum L. Maxon) pre-treatment on extraction yield of illudane glycosides and pterosins,” Phytochem Anal, vol. 24, no. 4, pp. 290-295, 2012.

[190] Clauson-Kaas, F., Environmental fate and analysis of ptaquiloside from the bracken fern (PhD thesis), University of Copenhagen, 2016.

[191] Kisielius V., Lindqvist D.N., Thygesen M.B., Rodamer M., Hansen H.C.B., Rasmussen L.H., “Fast LC- MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns,” J. Chromatogr. B, vol. 1138, no. 121966, 2020.

[192] Rai S.K., Sharma R., Kumari A., Rasmussen L.H., Patil R.D., Bhar R., “Survey of Ferns and Clinico- pathological Studies On the Field Cases of Enzootic Bovine Haematuria in Himachal Pradesh, a North-western Himalayan State of India,” Toxicon, vol. 138, pp. 31-36, 2017.

[193] Aranha P.C.R., Hansen H.C.B., Rasmussen L.H., Strobel B.J., Friis C., “Determination of ptaquiloside and pterosin B derived from bracken (Pteridium aquilinum) in cattle plasma, urine and milk,” J Chromatogr B , Vols. 951-952, pp. 44-51, 2014.

[194] Jensen P.H., Jacobsen O.S., Hansen H.C.B., Juhler R.K., “Quantification of Ptaquiloside and Pterosin B in Soil and Groundwater Using Liquid Chromatography−Tandem Mass Spectrometry (LC−MS/MS),” J. Agric. Food Chem., vol. 56, no. 21, p. 9848–9854, 2008.

[195] Ovesen R.G., Rasmussen L.H., Hansen H.C.B., “Degradation kinetics of ptaquiloside in soil and soil solution,” Environ. Toxicol. Chem., vol. 27, no. 2, pp. 252-259, 2008.

[196] Ayala-Luis K.B., Hansen P.B., Rasmussen L.H., Hansen H.C., “Kinetics of ptaquiloside hydrolysis in aqueous solution,” Environ Toxicol Chem., vol. 25, no. 10, pp. 2623-2629, 2006.

[197] Kisielius V., Markussen B., Rasmussen., L.H., “Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium,” to be submitted in 2020 (See Manuscript 2 of this thesis).

[198] “Danmarks Miljøportal,” [Online]. Available: https://naturdata.miljoeportal.dk/. [Accessed 2020].

[199] Rasmussen, L.H., Jensen, L.S. & Hansen, H.C.B., “Distribution of the carcinogenic terpene ptaquiloside in Bracken fronds, rhizomes (Pteridium aquilinum), and litter in Denmark,” Journal of Chemical Ecology, vol. 29, pp. 771-778, 2003.

[200] Alonso-Amelot, M.E. Rodulfo-Baechler, S. Jaimes-Espinoza, R., “Comparative dynamics of ptaquiloside and pterosin B in the two varieties (caudatum and arachnoideum) of neotropical bracken fern (Pteridium aquilinum L. Kuhn),” Biochemical Systematics and Ecology, vol. 23, pp. 709-716, 1995.

[201] Zaccone, C., Cavoski, I., Costi, R., Sarais, G., Caboni, P., Traversa, A. and Miano, T. M., “Ptaquiloside in Pteridium aquilinum subsp. aquilinum and corresponding soils from the South of Italy: Influence of physical and chemical features of soils on its occurrence,” Science of The Total Environment, vol. 496, pp. 365-372, 2014.

[202] Smith, B.L., Embling, P.P., Agnew, M.P., Lauren, D.R. and Holland, P.T., “Carcinogenicity of Bracken Fern (Pteridium esculentum) in New Zealand,” New Zealand Veterinary Journal, vol. 36, pp. 56-58, 1988.

[203] Smith, B.L., Seawright, A.A., Ng, J.C., Hertle, A.T., Thomson, J.A. and Bostock, P.D., “Concentration of ptaquiloside, a major carcinogen in Bracken fern (Pteridium spp.), from eastern Australia and from a cultivated worldwide collection held in Sydney, Australia,” Natural Toxins, vol. 2, pp. 347- 353, 1994).

[204] Alonso-Amelot, M.E., Calcagno, M.P. and Pérez-Inojosa, M., “Growth and selective defensive potential in relation to altitude in neotropical Pteridium aquilinum var. caudatum,” in Bracken Fern: Toxicity, Biology and Control. Proceedings of the International Bracken Group Conference Manchester 1999, Aberystwyth: International Bracken Group, 1999, pp. 43-47.

[205] Sandersen J., Lindqvist D.N., Rasmussen H.H., “Degradation of the carcinogenic ptaquiloside from Bracken ferns in rain and surface water,” in The World Plant Toxin Forum, Winnipeg, Canada, 2016.

[206] Kisielius V., Drejer M., Skrbic N., Dornhoff J.K., Lindqvist D.N., Hansen H.C.B., Rasmussen L.H., “Occurrence and stability of ptesculentoside, caudatoside and ptaquiloside from bracken ferns in surface waters,” Tto be submitted in 2020. (See Manuscript 3 of this thesis).

[207] Rios‐Gutierrez, M., Domingo L.R., Alonso‐Amelot, M.E., “A DFT Study of the Conversion of Ptaquiloside, a Bracken Fern Carcinogen, to Pterosin B in Neutral and Acidic Aqueous Medium,” Biological Chemistry & Chemical Biology, vol. 2, p. 8178–8186, 2017.

[208] Rasmussen L.H., Kroghsbo S., Frisvad J.C., Hansen H.C.B., “Occurrence of the carcinogenic Bracken constituentptaquiloside in fronds, topsoils and organic soillayers in Denmark,” Chemosphere, vol. 51, p. 117–127, 2003.

[209] Hama J.R., Strobel B.W., “Pyrrolizidine alkaloids quantified in soil and water using UPLC-MS/MS,” RSC Adv., vol. 9, no. 52, pp. 30350-30357, 2019.

[210] Günthardt B.F., Schonsee C.D., Hollender J., Hungerbuhler K., Scheringer M., Bucheli T.D., “"Is There Anybody Else Out There?" - First Insights From a Suspect Screening for Phytotoxins in Surface Water,” Chimia, vol. 74, no. 3, pp. 129-135, 2020.

[211] Chen Y., Nordenstam B., Flora of China, Vols. 20-21, p. 461.

[212] eFloras.org, “Flora of North America,” [Online]. Available: http://www.efloras.org/flora_page.aspx?flora_id=1. [Accessed 18 April 2020].

[213] Ozarowski M., Przystanowicz J., Adamczak A., “Phytochemical, pharmacological and clinical studies of Petasites hybridus (L.) P. Gaertn., B. Mey. and Scherb. A reviev,” Herba Polonica, vol. 59, no. 4, pp. 108-128, 2013.

[214] Toman J., “A Taxonomic Survey of the Genera Petasites and Endocellion,” Folia Geobotanica & Phytotaxonomica, vol. 7, p. 381–406, 1972.

[215] Giles M., Ulbricht C., Khalsa K.P.S., DeFranco Kirkwood C., Park C., Basch E., “Butterbur,” J. Herb. Pharmacother., vol. 5, no. 3, pp. 119-143, 2009.

[216] Aydın A.A., Zerbes V., Parlar H., Letzel T., “The medical plant butterbur (Petasites): Analytical and physiological (re)view,” J Pharm Biomed Anal, vol. 75, p. 220– 229, 2013.

[217] Giversen I., Brimer L., Kristiansen B., Danmarks Vilde Lægeplanter, Gyldendal A/S, 2014.

[218] Tys J., Szopa A., Lalak J., Chmielewska M., Serefko A., Poleszak E., “A botanical and pharmacological description of petasites species,” Curr. Issues Pharm. Med. Sci., vol. 28, no. 3, pp. 151-154, 2015.

[219] Åsen, Å., “Plants of possible monastic origin, growing in the past or present, at medieval monastery grounds in Norway,” in Plants and Culture: seeds of the cultural heritage of Europe, Edipuglia, 2009, pp. 227-238.

[220] Solberg S.O., “More than just weeds. NordGen’s work with Cultural Relict Plants and Bernt Løjtnant’s inventories from Denmark,” Nordic Genetic Resource Center, Alnarp, Sweden, 2014.

[221] Anderson N., Borlak J., “Hepatobiliary Events in Migraine Therapy with Herbs-The Case of Petadolex, A Petasites Hybridus Extract,” J Clin Med, vol. 8, no. 5, p. 652, 2019.

[222] Kozlov V., Lavrenova G., Savlevich E., Bazarkina K., “Evidence-based phytotherapy in allergicrhinitis rhinitis,” Clinical phytoscience, vol. 4, no. 23, pp. 1-8, 2018.

[223] Avula B., Wang Y.H., Wang M., Smillie T.J., Khan I.A., “Simultaneous determination of sesquiterpenes and pyrrolizidine alkaloids from the rhizomes of Petasites hybridus (L.) GM et Sch and dietary supplements using UPLC-UV and HPLC-TOF-MS methods,” J Pharm Biomed Anal, vol. 70, pp. 53-63, 2012.

[224] Kalin P., “The common butterbur (Petasites hybridus) - Portrait of a medicinal herb,” Forsch Komplementarmed Klass Naturheilkd, vol. 10, pp. 41-44, 2003.

[225] Roberts M. F., Wink M., Alkaloids: Biochemistry, Ecology, and Medicinal Applications, Springer, 1998.

[226] International Agency for Research on Cancer (IARC), “IARC Monographs on the evaluation of carcinogenic risks to humans; Some traditional herbal medicines, somemycotoxins, naphthalene and styrene,” IARC Press, Lyon, 2002.

[227] Xia Q., He X., Ma L., Chen S., Fu P.P., “Pyrrolizidine Alkaloid Secondary Pyrrolic Metabolites Construct Multiple Activation Pathways Leading to DNA Adduct Formation and Potential Liver Tumor Initiation,” Chem. Res. Toxicol, vol. 31, no. 7, pp. 619-628, 2018.

[228] Yang L., Stockigt J., “Trends for diverse production strategies of plant medicinal alkaloids,” Nat Prod Rep, vol. 27, no. 10, pp. 1469-1479, 2010.

[229] Wang T., Frandsen H.L., Christiansson N.R., Rosendal S.E., Pedersen M., Smedsgaard J., “Pyrrolizidine alkaloids in honey: Quantification with and without standards,” Food Control, vol. 98, pp. 227-237, 2019.

[230] The Editors of Encyclopaedia Britannica, “Alkaloid. Chemical compound.,” in Encyclopaedia Britannica, Encyclopædia Britannica, inc. , 2018.

[231] Sayhan H., Beyaz S.G., Celiktas A., The Local Anesthetic and Pain Relief Activity of Alkaloids, IntechOpen, 2017.

[232] Zilinskas A., Alkaloidai. Mokomoji knyga (in English: Alkaloids. Educational book), Vilnius University.

[233] Dembitsky V.M., Gloriozova T.A., Poroikov V.V., “Naturally occurring plant isoquinoline N-oxide alkaloids: their pharmacological and SAR activities,” Phytomedicine, vol. 22, no. 1, pp. 183-202, 2015.

[234] Scholtz S., MacMorris L., Krogmann F., Auffarth G.U., “Poisons, Drugs and Medicine: On the Use of Atropine and Scopolamine in Medicine and Ophthalmology: An Historical Review of their Applications,” J Eye Stud Treat, vol. 1, pp. 51-58, 2019.

[235] Matolcsy G., Nadasy M., Andriska V., Pesticide Chemistry, Volume 32, Elsevier Science, 1989.

[236] Wieczorek P.P., Witkowska D., Jasicka-Misiak I., Poliwoda A., Oterman M., Zielinska K., “Chapter 5 - Bioactive Alkaloids of Hallucinogenic Mushrooms,” in Studies in Natural Products Chemistry, Elsevier, 2015, pp. 133-168.

[237] Taylor S.L., Hefle S.L., “Chapter 16 - Naturally Occurring Toxicants in Foods,” in Foodborne Diseases (3rd edition), Elsevier , 2017, pp. 327-344.

[238] European Commission, “Commission Regulation (EC) No 1881/2006 of 19 December 2006 setting maximum levels for certain contaminants in foodstuffs,” Official Journal of the European Union, vol. 49, no. L 364, pp. 5-24, 2006.

[239] European Commission, “Regulation (EC) No 1334/2008 of the European Parliament and of the Council of 16 December 2008 on flavourings and certain food ingredients with flavouring properties for use in and on foods,” Official Journal of the European Union, vol. 51, no. L 354, pp. 34-50, 2008.

[240] de Nijs M., Noordam M.Y., Mol H.G.J., “Short inventory of EU legislation on plant toxins in food,” Quality Assurance and Safety of Crops & Foods, vol. 9, no. 1, pp. 129-139, 2017.

[241] UK Food Standards Agency, “June 2019 Stakeholder Update on Rapidly Developing Policy on Food Contaminants,” 2019. [Online]. Available: https://www.food.gov.uk/print/pdf/node/2546. [Accessed 14 April 2020].

[242] Kempf M., Reinhard A., Beuerle T., “Pyrrolizidine alkaloids (PAs) in honey and pollen-legal regulation of PA levels in food and animal feed required,” Mol Nutr Food Res., vol. 54, no. 1, p. 158–168, 2010.

[243] European Medicines Agency, Committee on Herbal Medicinal Products, “Public statement on the use of herbal medicinal products containing toxic, unsaturated pyrrolizidine alkaloids (PAs),” London, 2014.

[244] European Food Safety Authority, “Scientific Opinion on Pyrrolizidine alkaloids in food and feed,” EFSA journal, vol. 9, no. 11, 2011.

[245] “Compilation of agreed monitoring recommendations as regards the presence of mycotoxins and plant toxins in food,” 2014. [Online]. Available: https://ec.europa.eu/food/sites/food/files/safety/docs/cs_monitoring_recommendations_en.pdf. [Accessed 10 April 2020].

[246] European Commission, “Summary report of the standing committee on plants, animals, food and feed held in Brusssels on 18 February 2016 - 19 February 2016 (Ares (2016) 956112),” Brussels, 2016.

[247] Johanns E.S., Sijben A.E., Peters P.W., “An epidemic of epileptic seizures after consumption of herbal tea,” Nederlands Tijdschrift Voor Geneeskunde, vol. 146, no. 17, pp. 813-816, 2002.

[248] Kakar F., Akbarian Z., Leslie T., Mustafa M.L., Watson J., van Egmond H.P., Omar M.F., Mofleh J., “An Outbreak of Hepatic Veno-Occlusive Disease in Western Afghanistan Associated with Exposure to Wheat Flour Contaminated with Pyrrolizidine Alkaloids,” J Toxicol., vol. 313280, 2010.

[249] European Food Safety Authority, “Occurrence of Pyrrolizidine Alkaloids in food,” EFSA supporting publication, Vols. EN-859, no. 2015, pp. 1-116, 2015.

[250] Ma C., Liu Y., Zhu L., Ji H., Song X., Guo H., Yi T., “Determination and regulation of hepatotoxic pyrrolizidine alkaloids in food: A critical review of recent research,” Food Chem Toxicol., vol. 119, pp. 50-56, 2018.

[251] Dubecke A., Beckh G., Lullmann C., “Pyrrolizidine alkaloids in honey and bee pollen,” Food Addit Contam, vol. 28, no. 3, p. 348–358, 2011.

[252] Prakash A.S., Pereira T.N., Reilly P.E., Seawright A.A., “Pyrrolizidine alkaloids in human diet,” Mutat Res, vol. 433, no. 1-2, pp. 53-67, 1999.

[253] Bolechova M., Caslavsky J., Pospichalova M., Kosubova P., “UPLC-MS/MS method for determination of selected pyrrolizidine alkaloids in feed,” Food Chem., vol. 170, pp. 265-270, 2015.

[254] van Egmond H.P., “Natural toxins: risks, regulations and the analytical situation in Europe,” Anal Bioanal Chem, vol. 378, no. 5, pp. 1152-1160, 2004.

[255] Smith L.W., Culvenor C.C., “Plant sources of hepatotoxic pyrrolizidine alkaloids,” J Nat Prod., vol. 44, no. 2, pp. 129-152, 1981.

[256] Moreira R., Pereira D.M., Valentão P., Andrade P., “Pyrrolizidine Alkaloids: Chemistry, Pharmacology, Toxicology and Food Safety,” Int J Mol Sci, vol. 19, no. 6, p. 1668, 2018.

[257] Molyneux R.J., Gardner D.L., Colegate S.M., Edgar J.A., “Pyrrolizidine alkaloid toxicity in livestock: a paradigm for human poisoning?,” Food Addit Contam Part A Chem Anal Control Expo Risk Assess, vol. 28, no. 3, pp. 293-307, 2011.

[258] Ebmeyer J., Braeuning A., Glatt H., These A., Hessel-Pras S., Lampen A., “Human CYP3A4-mediated toxification of the pyrrolizidine alkaloid lasiocarpine,” Food Chem Toxicol, vol. 130, pp. 79-88, 2019.

[259] He X., Xia Q., Wu Q., Tolleson W.H., Lin G., Fu P.P., “Primary and secondary pyrrolic metabolites of pyrrolizidine alkaloids form DNA adducts in human A549 cells,” Toxicol In Vitro, vol. 54, pp. 286- 294, 2019.

[260] Chen T., Mei N., Fu P.P., “Genotoxicity of pyrrolizidine alkaloids,” J Appl Toxicol., vol. 30, no. 3, pp. 183-196, 2010.

[261] Hoogenboom L.A.P., Mulder P.P.J., Zeilmaker M.J., van den Top H.J., Remmelink G.J., Brandon E.F.A., Klijnstra M., Meijer G.A.L., Schothorst R., Van Egmond H.P., “Carry-over of pyrrolizidine

alkaloids from feed to milk in dairy cows,” Food Addit Contam Part A Chem Anal Control Expo Risk Assess., vol. 28, no. 3, pp. 359-372, 2011.

[262] Huxtable R.J., “Activation and pumonary toxicity of pyrrolizidine alkaloids,” Pharmacol. Therapeut., vol. 47, p. 371–389, 990.

[263] McLean E.K., “The toxic actions of pyrrolizidine senecio-D alkaloids,” Pharmacol Ther., vol. 47, no. 3, pp. 371-389, 1990.

[264] Mori H., Kawai K., Ohbayashi F., Bunai Y., Yamada K., Hirono I., “Some toxic properties of a carcinogenic pyrrolizidine alkaloid, petasitenine,” J Toxicol Sci., vol. 9, no. 2, pp. 143-149, 1984.

[265] Europen Food Safety Authority, “Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions and food supplements,” 2017.

[266] International Programme on Chemical Safety (WHO), “Pyrrolizidine Alkaloids. Environmental Health Criteria 80,” Geneva, 1988.

[267] Pohlmann J., van Loon G., Lefere L., Vanschandevijl K., Nollet H., De Clercq D., Delesalle C., Deprez P. , “Hepatoencephalopathy caused by Senecio jacobaea intoxication in five horses,” Vlaams Diergeneeskundig Tijdschrift, vol. 74, no. 6, pp. 440-445, 2005.

[268] Kisielius V., Hama J.R., Skrbic N., Hansen H.C.B., Strobel B.W., Rasmussen L.H., “Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids,” Accepted by Sci. Rep. (Spetember 2020) (See Manuscript 4 of this thesis).

[269] The Danish Environmental Protection Agency, “"Natur & Vand?" (English: "Nature & Water?"),” [Online]. Available: https://mst.dk/natur-vand/. [Accessed September 2020].

Part III: Manuscripts

0. Introduction to manuscripts

This part contains 4 manuscripts prepared in the scope of this PhD thesis:

1. Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns. 2. Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium. 3. Occurrence and stability of ptesculentoside, caudatoside and ptaquiloside from bracken ferns in surface waters. 4. Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids. Manuscript 1 was prepared to provide a universal method for quantification of several reactive terpene glycosides in Pteridium. The LC-MS technique has been chosen as sensitive enough for the detection of the compounds in the plants. The manuscript contains a procedure for preparation of analytical standards, fully validated method for sample treatment and analysis, and a case study analysing the compounds in the plant samples from 6 continents of the world. The study demonstrated that ptaquiloside-like compounds are present in the plants in significant concentrations. It also argues that these compounds have been under-investigated by the previous studies referring to ptaquiloside of Pteridium only because of the lack of knowledge about their presence, and the absence of analytical standards and methods. The manuscript was published in January 2020 in the Journal of Chromatography B. In May 2020 it was among the top 3 most downloaded articles of the Journal over the past 90 days.

Manuscript 2 contains a technical report of spatial and temporal variation of ptesculentoside, caudatoside and ptaquiloside in the plants. The sampling campaign was performed in July 2018 in Denmark, Sweden and Finland (spatial variation), and in June-November of 2018 in Denmark (temporal variation). The study was performed to uncover any environmental parameters that influence high concentrations of the compounds in the plants. The determined environmental parameters imply an enhanced risk of pollution of water resources. Performed statistical analyses were performed to determine the dependency of the compound concentrations on geographical position and environmental factors (plant access to sunshine, species of adjacently growing trees). It illustrated parameters that statistically significantly determine the concentrations of the compounds and their relative distribution. This work has been technically finalised within the timeframe of the PhD project, but will be supplemented with a discussion and submitted to a journal after the submission of this thesis.

Manuscript 3 complements the method presented in Manuscript 1 with a sample preparation and preconcentration technique needed for the detection of terpene glycosides and their hydrolysis products in water. It includes a study of surface waters in Denmark, Sweden and Spain sampled in 2019 and provides the first report of ptesculentoside, caudatoside and their hydrolysis products in water bodies. In order to fill the knowledge gap about the stability of the ptaquiloside-like compounds, the study performed the stability analysis as a function of pH, temperature and microbial activity in laboratory incubation of lake waters. The results demonstrated that ptesculentoside, caudatoside and ptaquiloside degrade in nearly identical rates under all analysed aquatic conditions. The manuscript will be submitted to a journal with only minor additional data analysis.

Manuscript 4 was prepared to illustrate natural occurrence of other types of plant toxins in water bodies. The study monitored surface and groundwaters nearby invasive plant species Petasites hybridus for an array of unsaturated pyrrolizidine alkaloids. The monitoring was performed in Denmark in the vegetation season of 2019. In order to detect any of these compounds, a validated LC-MS/MS technique has been chosen, allowing an analysis of common characteristic fragments of all the compounds. The study detected in total 21 unsaturated pyrrolizidine alkaloids in both water types. It reports the first detection of plant toxins in groundwater wells and discusses implications of findings in the light of prevalence of the plants producing the same and/or similar compounds. The manuscript was submitted to Scientific Reports on March 5, and accepted with minor revision in September 10, 2020 (the editorial suggestions are not reflected in the Manuscript).

1. Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns

DOI: 10.1016/j.jchromb.2019.121966

Graphical abstract:

Journal of Chromatography B 1138 (2020) 121966

Contents lists available at ScienceDirect

Journal of Chromatography B

journal homepage: www.elsevier.com/locate/jchromb

Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and T corresponding pterosins in bracken ferns ⁎ Vaidotas Kisieliusa,b, , Dan Nybro Lindqvista, Mikkel Boas Thygesenc, Michael Rodamerd, Hans Christian Bruun Hansenb, Lars Holm Rasmussena a Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark b Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark c Department of Chemistry, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark d Agilent Technologies, Hewlett-Packard St. 8, 76337 Waldbronn, Germany

ARTICLE INFO ABSTRACT

Keywords: Ptaquiloside (PTA) is an illudane glycoside partly responsible for the carcinogenicity of bracken ferns (Pteridium Pteridium sp.). The PTA analogues ptesculentoside (PTE) and caudatoside (CAU) have similar biochemical reactivity. Phytotoxins However, both compounds are highly under-investigated due to the lack of analytical standards and appropriate Carcinogens methods. This study presents a robust method for preparation of analytical standards of PTE, CAU, PTA, the NMR analysis corresponding hydrolysis products: pterosins G, A and B, and an LC-MS based method for simultaneous quan- Compound isolation tification of the six compounds in bracken. The chromatographic separation of analytes takes 5min.Theob- Method validation served linear range of quantification was 20–500 µg/L for PTA and pterosin B, and 10–250 µg/L forthere- maining compounds (r > 0.999). The limits of detection were 0.08–0.26 µg/L for PTE, CAU and PTA and 0.01–0.03 µg/L for the pterosins, equivalent to 2.0–6.5 µg/g and 0.25–0.75 µg/g in dry weight, respectively. The method was applied on 18 samples of dried fern leaves from 6 continents. Results demonstrated high variation in concentrations of PTE, CAU and PTA with levels prior to hydrolysis up to 3,900, 2,200 and 2,100 µg/g respectively. This is the first analytical method for simultaneous and direct measurement of all six compounds. Its application demonstrated that bracken ferns contain significant amounts of PTE and CAU relative to PTA.

1. Introduction hydrolysis [14,15]. The International Agency for Research on Cancer under WHO evaluates the evidence of bracken carcinogenicity in ex- Bracken is a group of carcinogenic ferns comprising the genus perimental animals as sufficient and classifies the plant as possibly Pteridium (Pteridaceae) [1]. Pteridium is found at all continents except carcinogenic to humans [16,17]. Nevertheless, traditional food con- Antarctica and is regarded as the fifth most abundant plant in the world taining bracken can be found in many nations of the world [18,19]. [2–4]. Under favourable conditions bracken can proliferate at extreme Correlation between human consumption of bracken and upper gas- rates. In the United Kingdom Pteridium aquilinum L. (Kuhn) covers 1.7 trointestinal tract cancers have been identified in Japan, Brazil and million ha which corresponds to 7.3% of the total land surface area of Venezuela [20–22]. the country [5]. Increase in abundance and distribution of bracken is PTA was identified in 1983 simultaneously by a Dutch anda observed in many parts of the world and is predicted to continue due to Japanese research groups [23,24]. The structure was elucidated by 1H changing climate, deforestation and other changes in land management NMR using PTA extracted from Pteridium aquilinum and Pteridium [6,7]. aquilinum subsp. latiusculum (Desvaux), respectively. Later on, several Bracken is the only plant that is well-documented to naturally cause PTA-like compounds were isolated from other species of bracken: iso- cancer in animals [8]. Several studies link bracken carcinogenicity to ptaquiloside (PTA isomer) and caudatoside (CAU) from Pteridium cau- the illudane glycoside ptaquiloside (PTA) [9–12]. PTA is found in datum (L. Maxon) [25], ptaquiloside Z from Pteridium caudatum [26] variable concentrations in the ferns reaching levels of 45,000 µg/g [13]. and hydroxyl-ptaquiloside ptesculentoside (PTE) from Pteridium escu- PTA is a pre-carcinogen, where the genotoxic effect is caused bya lentum (G. Forst.) [27] (concentrations unspecified). More PTA analo- highly reactive conjugated dienone formed from PTA as result of gues are known in Hypolepis punctata and Dennstaedtia hirsta

⁎ Corresponding author at: Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark. E-mail address: [email protected] (V. Kisielius). https://doi.org/10.1016/j.jchromb.2019.121966 Received 27 September 2019; Received in revised form 29 December 2019; Accepted 30 December 2019 Available online 03 January 2020 1570-0232/ © 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/BY-NC-ND/4.0/). V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

Table 1 Structures of analytes in decreasing order of polarity with numbering of carbon atoms of PTA. The cyclopropane functional group causing formation of dienone agents comprises atoms No. 7, 12 and 13.

Compound PTE CAU PTA Pterosin G Pterosin A Pterosin B

Common molecular structure

R1 H CH3 H H CH3 H

R2 CH2OH CH2OH CH3 CH2OH CH2OH CH3

Molecular Formula C20H30O9 C21H32O9 C20H30O8 C14H18O3 C15H20O3 C14H18O2 CAS number – – 87625-62-5 35964-50-2 35910-16-8 34175-96-7 Estimated LogP (Source: PubChem) −2 −1.4 −1.3 1.9 2.2 2.5

(Pteridaceae), like hypolosides A, B and C [28]. However, these have 2. Materials and methods never been reported in bracken. PTE, CAU and PTA are believed to comprise the majority of illudane glycosides in bracken [29] (Table 1). 2.1. Solvents and chemicals The content of PTA is known to be quite variable due to environmental and genetic factors. Large temporal and spatial variation is encountered LC-MS grade acetonitrile was obtained from Merck Millipore and PTA hot-spots are found in some geographical regions. The hot- (LiChrosolv hypergrade for LC-MS, Germany). LC-MS grade methanol spots are correlated with bracken induced chronic diseases among farm was obtained from Honeywell (LC-MS Chromasolv, Germany). Water animals. Similarly, proximity to bracken is a potential risk factor for for preparative HPLC and LC-MS (electrical resistivity 18.2 MΩcm, TOC human gastric and oesophageal cancers [30–34]. less than 2 µg/L) was dispensed from Satorius Arium Pro Ultrapure Quantification of PTA in larger surveys is based on aqueous ex- water system (Germany). Acids and bases (sodium hydroxide, tri- traction of PTA from dried bracken. However, sample preparation and fluoroacetic, formic and hydrochloric acids) were all of analytical grade extraction can cause transformation of PTA into pterosin B as result of from Sigma-Aldrich (Denmark). Polyamide was obtained from Fluka hydrolysis under alkaline or acid conditions, exposure to heat and/or Analytical Sigma-Aldrich co (Polyamide for column chromatography 6, enzymes [35]. As a result, PTA quantification methods comprise either Germany). individual measurements of PTA and pterosin B (direct method) [36–38] or measurement of pterosin B after deliberate total conversion 2.2. Analytical instruments of PTA (conversion method) [39–43]. Similar conditions are expected for PTE and CAU requiring quantification of equivalent pterosins Gand For preparation of analytical standards Shimadzu LC 20-AD semi- A(Table 1). The only way to avoid formation of pterosins is to use hot preparative LC system equipped with Shimadzu SPD-10A UV–VIS de- water or hot methanol extraction of freshly harvested plant material tector and Hitachi D 2000 Chromato-Integrator were used. 1H NMR [35]. This is however inapplicable for high-throughput screening sur- spectra were recorded with Bruker Avance 300 MHz NMR spectro- veys. meter. The method of chromatographic separation and quantification of The analytical methods for quantification of PTA and associated analytes was developed in an analytical Agilent 1260 Infinity HPLC pterosin B are based on HPLC-UVVIS, LC-MS, LC-MS/MS and GC-MS. system equipped with an Agilent 6130 Single Quadrupole mass spec- The LOD range from 0.014 to 0.6 µg/L for PTA and 0.008 to 500 µg/L trometer. UV absorption profiles of the analytes were obtained with an for pterosin B. A single GC-MS based method exists for the simultaneous Agilent 1260 DAD VL detector. quantification of PTE, CAU and PTA (LOD not specified) [29]. It is based on the conversion of the illudane glycosides into pterosins as the 2.3. Collection of bracken samples former cannot be analysed by GC. The disadvantage of the conversion approach is that it does not differentiate the pterosins originally present The method was applied on 18 samples of dried grinded fern leafs. in analytical samples from the pterosins intentionally generated in The species and geographic locations are provided in Table 9. The sample treatment. The issue can be handled by sequential analysis of collection time, sampling, preservation and storage methods applied sample fractions containing pterosins and fractions containing illudane are provided in Table S1 of Supplementary Material (SM). glycosides converted into pterosins. However, this is time-consuming and a direct method is desired. 2.4. Preparation of analytical standards Bracken carcinogenicity is amplified by the number of PTA analogues potentially being present in the ferns [44]. Most studies evaluating No certified reference materials of PTE, CAU and PTA exist and bracken toxicity do not quantify or discuss PTA analogues. Hence, the therefore analytical standards have to be prepared from the compounds purpose of this study was to develop a fast, direct and versatile method extracted, isolated and purified from bracken. For preparation of ana- for high-throughput screening of bracken for content of PTE, CAU, PTA lytical standards, a record of bracken containing sufficient amounts and and pterosins G, A and B. In addition, a novel method for purification of proportion of PTE, CAU and PTA to be extracted was found in the lit- the analytes is presented. erature (Pteridium esculentum in Bribie Island, South East Queensland, Australia [29]). Recently emerged furled bracken fronds were collected

2 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

Table 2 v12.0.3 resulted in a purity of PTA of 94% based on integration of the H Observed UV absorption profiles max1(λ > λmax2 > λmax3) and literature- atom (H-5) attached to C-5 (5.80 ppm) and 90% based on the H atom based molar extinction coefficients of pterosins at 217 nm. (H-14) attached to C-14 (1.32 ppm) (Table 1) relative to the internal 1 λmax1 (nm) λmax2 (nm) λmax3 (nm) ε at λmax1 (abs/mol/L) standard. H NMR (D2O, 20 mM KH2PO4, 300 MHz): δ 5.80 (s, 1H; H- [29,47] 5), 4.77 (overlapped by HDO; H-1′), 3.97 and 3.77 (2dd, J = 12.5 and 2.0 Hz and J = 12.5 Hz and 5.3 Hz respectively, 2H; H-6′), 3.56–3.41 PTE 202 none none n/a (m, 3H; H-3′ and H-4′ and H-5′), 3.32 (dd, J = 9.0 and 8.0 Hz, 1H; H- CAU 202 none none n/a PTA 202 none none n/a 2′), 2.78 (d, J = 1.4 Hz, 1H; H-9), 2.57–2.44 and 2.05–1.97 (2 m, 3H; Pterosin G 217 263 306 26,915 H-2 and H-3), 1.58 (d, J = 1.3 Hz, 3H; H-11), 1.32 (s, 3H; H-14), 1.13 Pterosin A 217 263 306 34,674 (d, J = 6.7 Hz, 3H; H-10) , 1.01–0.87 and 0.79–0.64 (2 m, 4H; H-12 Pterosin B 217 260 304 37,154 and H-13) (Fig. S2 of SM). For representation of the purity of PTA the average (92%) was used. Half of the volume of obtained aqueous solutions of PTE and CAU in the reported location in 2017, freeze-dried, grinded to fine powder were allocated for production of pterosins G and A. An aqueous solution and kindly provided by Prof Mary T. Fletcher (The University of of known concentration of PTA was prepared for production of pterosin Queensland, Australia). For isolation of the compounds 20 g of the B. The glycosides were converted by mixing the aqueous solutions with powder were mixed with 200 mL deionized water in a glass bottle, 75 μL of 1 M NaOH per 1 mL of solution, submerging for 15 min. in shaken at 100 rpm for 30 min. on a shaking table and centrifuged for 35 °C water bath, allowing to cool to room temperature and mixing 20 min. at 2500 g at 1 °C (Sigma Zentrifugen 4 K15). The supernatant with 75 μL of 2.5 M trifluoroacetic acid per 1 mL of solution [46]. The was filtered through a 2 µm pore size filter paper (Whatman No.41, 100% conversion rates were confirmed in LC-MS by observed dis- Sigma-Aldrich, Germany) under vacuum. Two cylindrical glass funnels appearance of the glycosides and occurrence of pure pterosins. The (Ø = 1.5 cm; l = 20 cm, BIO-RAD, Econo-Column, USA) were dry concentration of the obtained solution of pterosin B was determined packed with approximately 10 g of polyamide. The aqueous extract was applying 1:1 M conversion ratio from the original solution of PTA. The passed through the columns at an approximate speed of 3 mL/min to concentrations of the obtained solutions of pterosins G and A were retain hydrophobic impurities and pterosins [45]. Extra 20 mL of determined comparing absorption of pterosins G, A and B in 217 nm UV deionized water was added to each column to fully elute the illudane with molar extinction coefficients ε (Table 2). The concentrations of glycosides. The bottles and flasks used throughout the process were solutions of PTE and CAU were back-calculated applying 1:1 M con- wrapped in aluminium foil and kept in ice bath to prevent photolysis version ratios of these compounds to pterosins G and A (Fig. 1). In order and thermal degradation. to prevent hydrolysis and degradation of the obtained analytical stan- The analytes were retrieved from the eluate by solid phase extrac- dards, the aliquots were preserved in 50% v/v MeOH and stored at tion. OASIS HLB 6 cc 200 mg cartridges (Waters co, Ireland) were −80 °C. conditioned with 12 mL of deionized water, loaded with 30 mL of the eluate, washed with 5 mL of deionized water and eluted with 5 mL of 40% methanol at an approximate speed of 3 mL/min. The pre- 2.5. LC-MS method development concentrated eluate was further purified by semi-preparative HPLC system using a reversed-phase Agilent ZORBAX Eclipse XDB-C18 5 µm 2.5.1. Chromatographic separation 9.4 × 250 mm column (C18 5 µm 9.4 × 15 mm guard column) at A high-throughput reversed-phase column was selected for fast ambient temperature (injection volume 5 mL, mobile phase 40% me- analysis of the hydrophilic compounds (Agilent InfinityLab Poroshell thanol, flowrate 4 mL/min, λ 220 nm). The retention times ofthe 120, 2.7 μm EC-C18, 3.0 × 50 mm column; 2.7 μm EC-C18 3.0 × 5 mm compounds at 23 °C were 7.7 min for PTE, 14.5 min for CAU and guard column). The chromatographic separation was developed to- 19.9 min for PTA (Fig. S1 of SM). The injection was repeated 8 times to wards full resolution of all analytes and fast elution of the latest eluting accumulate approx. 50–80 mL of each fraction. The compound iden- analyte in different gradient steepness of methanol and acetonitrile. The tities were confirmed in LC-MS and the methanol was removed from the peak resolution values were monitored in Agilent OpenLab CDS eluates using a rotary evaporator (BUCHI Rotavapor R-210) at 35 °C ChemStation C.01.07. water bath temperature. Attempts to isolate PTE, CAU and PTA demonstrated that pure crystals of the glycosides instantly absorb ambient humidity and the crystals transform into sticky transparent gels. Concentration of the dried least hydroscopic glycoside PTA was determined as the reference to calculate concentrations of the rest of the analytes. The aqueous solution of PTA was frozen to −80 °C and freeze-dried at −55 °C (freeze-drier: Scanvac Coolsafe 55-4) in a 30 mL amber bottle covered with perforated aluminium foil. After approximately 36 h the pressure in the freeze-dryer vacuum compartment was gradually increased supplying dried nitrogen gas (nitrogen generator: Peak Scientific NM32LA). The bottle with dry PTA was immediately closed and transported for weighing and handling in a glove bag (Atmosbag 39″ × 48″, Sigma-Aldrich, Germany) inflated with dried nitrogen gas and loaded with silica gel beads to absorb humidity. Several minutes after a humidity meter in the glove bag showed relative humidity lower than lowest detectable (20%), the NMR tube was filled with 2.9 mg PTA, 950 µL D2O (99.9 atom % D, Sigma-Aldrich, Germany), 25 µL 20 mM KH2PO4 and 25 µL 20 mM 3-(trimethylsilyl propionic-2,2,3,3-d4 acid, sodium salt solution (internal standard) (99.9% pure; ≥98.0 atom % D, Sigma-Aldrich, Germany). The data of NMR analysis processed with Mestrelab Research S.L. analytical chemistry software MestReNova Fig. 1. Flowchart of preparation of analytical standards.

3 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

Table 3 Mass fragments in chronological order of retention times. The masses monitored in SIM are underlined.

Peak No. Compound Monoisotopic mass (g/mol) m/z 200–460 (relative abundance ≥ 10%, excluding isotopes)

+ + + + 1 PTE 414.2 217.1 [M-glucose-H2O + H] , 235.1 [M-glucose + H] , 437.2 [M + Na] , 453.1 [M + K] + + + 2 CAU 428.2 231.1 [M-glucose-H2O + H] , 249.1 [M-glucose + H] , 451.2 [M + Na] 3 pterosin G 234.1 235.1 [M + H]+, 257.1 [M + Na]+ + + + + 4 PTA 398.2 201.1 [M-glucose-H2O + H] , 219.1 [M-glucose + H] , 421.2 [M + Na] , 437.1 [M + K] 5 pterosin A 248.1 249.1 [M + H]+, 271.1 [M + Na]+ 7 pterosin B 218.1 219.1 [M + H]+

Table 4 Fractional factorial experimental design.

Parameter Factor − Factor + Method of Rai et al. [30] for PTA and pterosin B

A: Extraction duration (min.) 20 70 60 B: Solvent composition 100% water 20% v/v MeOH 100% water

C: pH (buffer) of extraction 4.5 (0.1 M NH4AC) 6 (0.1 M NH4AC) Approx. 5 (not buffered) D: Sample/solvent 40 mg/40 mL 120 mg/40 mL 100 mg/40 mL

Method 1 2 3 4 5 6 7 8 9 10 11 12 Parameter A + + − + + + − − − + − − B − + + − + + + − − − + − C − − + + − + + + − − − − D + − − + + − + + + − − −

2.5.2. Optimisation of mass detector from the samples stored for 3 and 12 weeks were compared. The eluents were selected comparing the strengths of the signals of + + + + + analytes forming with adducts of H , Na ,K , Li and NH4 in dif- 2.8. Extraction of illudane glycosides and pterosins from bracken ferent combinations of LC mobile phase compositions. Electrospray ionisation (ESI) parameters were optimised in full factorial experi- 2.8.1. Extraction method development mental design wherein the effects on signal strengths by 4 factors at3 Sample preparation was optimised based on a method for extraction 4 levels were compared in 81 (3 ) possible combinations (cone voltage: of PTA and pterosin B from dry bracken powder by Rai et al. [30]. The 1,000–3,000–5,000 V; drying gas flow: 7–10–13 L/min; drying gas extracting solvent was made adjusting 0.1 M ammonium acetate buffer temperature: 150–250–350 °C; nebulizer pressure: 20–40–60 psi). The with HCl to pH 4.5 or 6.0 (pH Meter: MeterLab PHM220 Lab). The ions underlined in Table 3 were monitored in the mass spectrometer extraction was performed by mixing the dry plant powder with 40 mL operated in single ion mode (SIM). of the extracting solvent in a Nunc 50 mL conical polypropylene centrifuge tubes (Thermo Fisher Scientific, Korea) for 20 min at 75rpm 2.6. Validation of the analytical method (mixer: ELMI Intelli-Mixer RM-2L). The influence of 4 factors at 2 levels on the extraction efficiency of the 6 targeted compounds were com- To delineate the linearity range, 3 identical sets of 7 vials with pared in a fractional factorial experimental design (Table 4). different concentrations for each analytical standard were in- A method extracting the highest amount of each of the 6 targeted dependently made (in total 21 vials for each compound) and analysed compounds per mass of the dry bracken powder was identified. in a randomised order. The average SIM areas measured from the 3 Efficiency of the given method for the particular compound was vials containing the same concentrations of a given analyte were used equated to 100% (the maximum yield). The concentrations of each for construction of the calibration curves. The linearities of PTE, CAU, compounds extracted by the other methods were expressed as relative pterosins G and A were explored in the 10–250 µg/L range, whereas the concentrations (x % of the maximum yield). The method yielding the linearities of PTA and pterosin B were explored in the 20–500 µg/L highest average relative concentrations of all 6 compounds was selected range. LOD was calculated as 3.3 times the SDintercept/slope of the ca- as optimal. libration curves. LOQ was calculated as 10 times the SDintercept/slope of the calibration curves. Precision of the instrument was calculated using 2.8.2. Precision and accuracy of the extraction method relative standard deviation (RSD) for replicate injections (n = 7) of The precision of the extraction method was determined as RSD for analytical standards injected from the same vial. Precision of the the measured concentrations of all compounds obtained performing 7 standards was calculated using RSD for single injections of 3 different parallel extractions from the same sample. The accuracy of the ex- independently made vials of identical concentrations (n = 3). The traction method was tested performing parallel extractions by replacing precision of instrument and standards were expressed as the range from the extracting solvent with 25%, 50% and 100% of the previously made the lowest to the highest RSD observed. extract of the same plant sample. The accuracy of the method was determined as the recovery percentage of all targeted compounds in the 2.7. Robustness of analytical standards during storage obtained more concentrated extraction solutions.

Due to instability of the illudane glycosides [35] robustness of the 2.9. Matrix effects analytical standards during storage were tested. The identical triplicate sets of standards used for making the calibration curves were dis- The matrix effects of the plant extracts were tested injecting them tributed to store one set of each compound at −20 °C, +5 °C and ap- into LC-MS as pure extracts and diluted 2, 5, 10, 15, 20, 35 and 50 prox. +23 °C (room temperature). The slopes, linearities and pairwise times. Dilution was made with 50% v/v MeOH in a filtering vial t-tests (95% confidence interval) of the calibration curves produced (Syringeless filter device Mini-Uniprep, GE Healthcare Life Sciences,

4 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

UK). Suspended plant particles were removed by the 0.2 µm pore size Table 5 membrane in the vial, and the measurements were performed injecting Validation data of the LC-MS method. Compounds are listed in the descending 20 µL of the solution into LC-MS directly from the vial. The curves of order of the slopes of calibration curves. the measured concentrations versus the dilution factors were plotted Compound Calibration LOD LOQ Precision of the Precision of and the ranges of linearity were explored. The matrix analyses were range (µg/L) (µg/L) (µg/L) instrument (%; standards (%; made on plant samples that naturally contained the 6 targeted com- n = 7) n = 3) pounds and represented different subspecies, origin and drying PtrA 10–250 0.02 0.05 1.16–1.81 1.27–5.34 methods: freeze-dried Pteridium esculentum from Australia (sample No. PtrB 20–500 0.03 0.09 0.52–0.81 0.46–2.09 12) and plant-press dried Pteridium aquilinum from Sweden (sample No. PtrG 10–250 0.01 0.03 0.36–2.70 1.46–3.26 4). PTA 20–500 0.22 0.68 1.07–3.16 0.09–7.24 PTE 10–250 0.08 0.25 1.38–5.79 3.32–7.22 CAU 10–250 0.26 0.78 3.08–4.59 4.68–8.61 2.10. Statistical analysis

Data analysis (linear regression, descriptive statistics, etc.) were revealed the following set of parameters as the most optimal: cone performed with Microsoft Excel 2016. Effects of parameters applied voltage: 3,000 V; drying gas flow: 13 L/min; drying gas temperature: optimising methods of analyte extraction and MS ionisation were ex- 350 °C; nebulizer pressure: 40 psi. (Effects of all parameters on ESI amined with data analysis software Minitab 18 and graphical re- efficiency analysed and graphically presented in Figs. S4 and S5 of SM). presentation software Plotly. The optimised ESI parameters yielded 2.2, 2.4 and 2.2 times higher responses of PTE, CAU and PTA respectively than application of the set 3. Results of ESI parameters reported in PTA study with the same instrument by Rai et al. [30]. 3.1. Preparation of analytical standards The correlation coefficients (r) of the calibration curves gave ex- cellent linearities with values > 0.999 for all 6 compounds over the Previously reported methods of PTA isolation depend on its sorption concentration ranges investigated (10–250 or 20–500 µg/L) (Fig. S6 of to XAD-2 polymeric absorbent [8,38,42,43,48,49]. The developed SM). LOD, LOQ and the ranges of precision of instrument and standards method compared with the most recent method of PTA isolation [38] are provided in Table 5. showed that application of disposable OASIS HLB cartridges requires less labour and results in 19% loss of the mass of PTA on the absorbent 3.3. Robustness of analytical standards during storage in contrast to 72% loss in XAD-2. The glycosides kept for 3 weeks at room temperature hydrolysed, 3.2. LC-MS method validation calibration curves did not follow linear patterns and data are not shown. The parameters of calibration curves of analytical standards The analytes were separated in the analytical HPLC system ther- kept in −20 °C and + 5 °C for 3 and 12 weeks are provided in Tables 6 mostated at 35 °C at 1 mL/min flow with a mobile phase comprising and 7. water (eluent A) and acetonitrile (eluent B) both with 0.1% v/v formic The test demonstrated that analytical standards of the illudane acid in the following gradient elution: 0–1 min 10% B; 3 min 35% B; glycosides can be kept in a fridge or a temperature-controlled auto- 4–4.5 min 95% B; 4.6–5 min 10% B. The chromatographic separation sampler (5 °C) for at least 3 weeks. However, after 12 weeks the cali- monitored with UV (injection volume 20 µL, λ 220 nm) is displayed in bration curves of the stored samples expressed poor linearity (≤0.991) Fig. 2. and recovery in comparison to the curves stored for the same period in The chromatographic separation yielded full resolution of the least −20 °C (≤86.4). The analytical standards of pterosins could be stored resolved peaks (No. 3 and 4). Nevertheless, in case of overlap the at 5 °C for at least 12 weeks. temperature of the column compartment could be increased to dis- tribute resolution of the peaks No. 3–5 more evenly. It has been de- 3.4. Sample preparation monstrated that the increase of temperature to 60 °C has little or no effect on quantification of the analytes (Table S2 of SM). Sample preparation method No. 3 providing the highest average Full factorial experimental design of ESI parameter analyses extraction of compounds was identified as the optimal method

Fig. 2. Chromatogram of the compounds measured at 220 nm. 1: PTE (5,300 µg/L); 2: CAU (4,200 µg/L); 3: pterosin G (4,500 µg/L); 4: PTA (3,400 µg/L); 5: pterosin A (400 µg/L); 6: pterosin G methyl analogue (can form with methanol residues in incomplete conversion of PTE to pterosin G [50]); 7: pterosin B (1,300 µg/L).

5 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

Table 6 35% of acetonitrile in only 5 min. gradient elution. Two times higher Comparison of calibration curves of analytical standards of illudane glycosides prices of acetonitrile than methanol (European market) make ex- stored in different temperatures in time. ploitation of the newly developed LC-MS method 1.5 times cheaper. Compound −20 °C 5 °C Pairwise t-test Application of 0.5 mM sodium acetate in the eluent of the previously reported PTA and pterosin B quantification method resulted in pre- Slope Intercept r Slope Intercept r Recovery % cipitation of sodium salts in the mass spectrometer and required regular cleaning of chromatographic column with organic solvents. The method 3 weeks PTE 127 482 0.999 123 261 0.997 96.4 ± 3.79 reported in this study does not require additional maintenance neither CAU 63 605 0.993 60 350 0.992 96.9 ± 2.95 compromises durability of the equipment. PTA 197 4,530 0.995 188 5,310 0.992 108 ± 12.6 The LODs of the LC-MS method were 0.22 µg/L for PTA and 12 weeks 0.03 µg/L for pterosin B, whereas the ranges of previously reported PTE 522 446 0.999 364 2,250 0.955 66.9 ± 4.22 methods span from 0.014 to 0.6 µg/L for PTA and 0.008 to 500 µg/L for CAU 309 212 0.999 204 1,280 0.937 71.9 ± 6.46 pterosin B. A single method for quantification of PTE and CAU in GC- PTA 629 6,620 0.999 481 12,200 0.991 86.4 ± 6.6 MS after conversion of the compounds exists, but the LODs are not specified [29]. The LODs of PTA and pterosin B in the LC-MS method (Table 8). No significant effect of a single parameter on the extraction developed in this study do not exceed LODs of the methods applying efficiency was observed (Effects of all parameters on extraction effi- high-resolution mass spectrometry. However, the presented method ciency are presented in Fig. S7 of SM). applies more affordable single quadruple mass spectrometry andis The precision of the extraction method was 5.5, 9.0, 6.5, 5.6, 3.8, developed for compound quantification in plants where concentrations 3.4 and the accuracy was 1.0, 2.4, 1.7, 0.8, 1.3, 2.0 for PTE, CAU, PTA, of the targeted analytes proved to significantly exceed the LODs. The pterosins G, A and B, respectively. The test of the matrix effects de- method reported in this study would quantify but not differentiate monstrated that plant extracts cause ion enhancement of the 6 analytes. between isomers (for example, isoptaquiloside) and would not detect However, the quantification of the analytes was linear after dilution of PTA analogues other than PTE and CAU. extracts by 20 or more times (Figs. S8 and S9 of SM). A dilution factor The full analytical method allows fast and feasible screening of of 25 times was applied resulting in LOQ in the plant samples of bracken ferns for wide range of targeted compounds. Its application on 6.3–19.5 µg/g for glycosides and 0.8–2.3 µg/g for pterosins. samples from six continents of the world revealed high distribution of PTA in the range similar to reported by other studies (0–2,100 µg/g in dry weight). However, the results demonstrated significant concentra- 3.5. Method application on fern samples tions of PTA analogues. PTE was identified in 7 out of 17 samples of bracken. The species in which PTE was originally identified (Australian The method was applied for measuring contents of the analytes in P. esculentum) [27] had the highest proportion of PTE (nearly 80% of the samples of ferns from various regions of the world. In total 18 the mass of the three glycosides prior to hydrolysis). Nevertheless, samples of ferns, 17 of which represented the genus Pteridium were quantification in other species of European, African and South-Amer- extracted in duplicates and analysed in a randomised order. The ican bracken revealed that PTE comprised up to nearly 20% of the mass average coefficients of variation between the measured contents of of the three glycosides. CAU was identified in 15 samples, in 6 of which CAU, PtrA, PTE, PtrG, PTA and PtrB were 0.17, 0.09, 0.01, 0.12, 0.10 its concentrations exceeded PTA. The single PTA neutral bracken and 0.17 respectively. The average measured contents are provided in sample (European P. aquilinum subsp. Latiusculum) contained CAU. Table 9. None of the illudane glycosides were restricted to particular subspecies Illudane glycosides or corresponding pterosins were detected in all or origin of bracken. A more extensive empirical study would be re- samples of Pteridium and not detected in the single sample of other fern quired to link biosynthesis of the compounds to factors like genetics or (No. 3). The concentrations of glycosides in the samples prior to hy- environmental conditions. drolysis were obtained summing the concentrations of quantified gly- The observed relative concentrations of PTA analogues strongly cosides with concentrations of quantified corresponding pterosins and suggest that in most cases the conventional quantification of PTA does applying 1:1 M conversion ratios (Table S3 of SM). Relative distribu- not represent the content of the total reactive illudane glycosides. tions of glycoside concentrations prior to hydrolysis are presented in Estimated LogP values and observed order of chromatographic elution Fig. 3. demonstrate that the PTA analogues are more soluble than PTA. The illustrated common occurrence together with high solubility suggest 4. Discussion significant risk of leaching of the compounds from bracken to soils and water. Previous studies reported PTA in milk of animals grazing cattle The most recently reported LC-MS method measuring PTA and (up to 3.1 µg/L) [31,32,51,52] as well as in surface waters and shallow pterosin B in the same instrumental setup used 53% methanol in a ground waters adjacent to bracken ferns, with peak concentrations 10 min. isocratic elution [30] whereas the new method uses on average

Table 7 Comparison of calibration curves of analytical standards of pterosins stored in different temperatures in time.

Compound −20 °C 5 °C Pairwise t-test

Slope Intercept r Slope Intercept r Recovery %

3 weeks Pterosin G 343 209 0.996 295 87.6 0.991 86.6 ± 12.6 Pterosin A 698 2,160 0.999 643 1,640 0.994 93.6 ± 3.58 Pterosin B 643 7,050 0.998 647 7,490 0.998 104 ± 3.69

12 weeks Pterosin G 2,480 2,820 0.999 2,571 15,300 0.998 112 ± 4.84 Pterosin A 4,390 8,280 0.999 4,280 11,000 0.999 101 ± 1.25 Pterosin B 2,740 29,800 0.998 2,666 21,800 0.999 99.4 ± 2.64

6 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

Table 8 Comparison of relative concentrations of the compounds from different extraction methods.

Method 1 2 3 4 5 6 7 8 9 10 11 12

Relative concentration PTE 85.3 92.6 98.6 93.0 81.8 81.2 87.7 95.4 86.8 99.7 96.9 100 (x % of the maximum yield) CAU 95.0 93.2 94.5 100 84.5 73.5 88.5 98.9 90.2 89.9 87.6 91.9 PTA 85.6 89.2 100 71.4 63.3 72.2 64.5 73.0 64.9 84.8 87.5 86.4 pterosin G 84.4 96.5 99.8 80.7 74.9 84.8 76.8 78.5 76.7 100 95.6 98.1 pterosin A 96.0 95.4 100 89.9 167* 83.7 86.5 88.3 87.2 94.8 93.5 93.2 pterosin B 93.9 100 79.5 57.1 52.7 58.6 50.5 46.4 58.1 53.3 67.3 53.2 Average 90.1 94.5 95.4 82.0 87.4 75.7 75.8 80.1 77.3 87.1 88.1 87.1

* The data point too distant from other data points disregarded as an outlier. following precipitation of up to 4 µg/L [53–55]. The present study non-plant environmental samples such as soils, water and food would particularly warrants re-evaluation of the total reactive illudane gly- require combination with sample extraction and preconcentration cosides as environmental contaminants and provides applicable method techniques. for preparation of analytical standards and chromatographic separation of analytes. However, the isotope internal standards are not available and the production of analytical standards proved to be complicated CRediT authorship contribution statement due to the hygroscopic nature of the compounds. To foster research of the illudane glycosides, commercial availability of analytical standards Vaidotas Kisielius: Conceptualization, Investigation, Writing - and reference materials would be of advantage. original draft. Dan Nybro Lindqvist: Conceptualization, Investigation, Besides application in research of illudane glycoside fate, toxicity Supervision. Mikkel Boas Thygesen: Investigation, Writing - review & and carcinogenesis, the method could serve studying yet indefinite editing. Michael Rodamer: Supervision, Resources. Hans Christian taxonomy of the genus [1], advancing understanding on its spreading Bruun Hansen: Supervision, Writing - review & editing. Lars Holm strategies and potential impacts on biodiversity and ecosystem services. Rasmussen: Conceptualization, Supervision, Data curation, Writing - Furthermore, the method could serve for enhancing emerging appli- review & editing. cations of pterosins for pharmaceutical purposes [56–58]. The developed quantification method of the compounds in the samples is directly applicable only for plant material. However, the LC-MS part of the Declaration of Competing Interest method could as well be adapted and validated for rapid screening of illudane glycosides in non-plant environmental samples. The formerly The authors declare that they have no known competing financial reported upper concentrations of PTA in water and milk exceed LODs of interests or personal relationships that could have appeared to influ- the LC-MS method presented in this study. However, its application on ence the work reported in this paper.

Table 9 Measured content of analytes in diverse fern samples.

Sample No. Species Country of origin Location Measured µg/g (dry weight) (WGS84 coordinations) CAU PtrA PTE PtrG PTA PtrB

1 P. aquilinum Denmark 55.947095, 9.759093 75b 133a 28c 1,370a 345a 2 P. aquilinum Denmark 55.123898, 14.917845 23b 2b 5a 9a 3 Dryopteris cristata (L) A. Gray Denmark Ordrup Næs peninsula (55.84, 11.37) 4 P. aquilinum Sweden 56.29849, 13.370333 63a 71a 177a 146a 1,530a 207a 5 P. aquilinum Sweden 57.092993, 14.736379 traceb 7a tracec 33a 12a 6 P. aquilinum subsp. latiusculum Finland 62.203272, 30.086088 16b 24a 7 P. aquilinum Lithuania 55.459371, 26.190076 551a 61a 6a 4a 8 P. aquilinum UK 54.50408, 38b 8b 42a 35a −0.83638 9 P. aquilinum UK 54.502781, 28b 21b 286a 94a −0.841872 10 P. aquilinum UK 54.388075, 26b 15b 213a 57a −0.692965 11 P. aquilinum India Kullu district traceb 5b 12 P. esculentum Australia −27.023612, 153.125651 215a 124a 2386a 856a 792a 110a 13 P. aquilinum Tanzania Moshi Rural district 205a 402a 53c 788a 732a (−3.293, 37.5) 14 P. aquilinum Tanzania Siha district 3c 34a 36a (−3.164, 37.121) 15 P. arachnoideum Brazil Wanderley municipality 1302a 18c 26a 16 P. arachnoideum Brazil Wanderley municipality 16c 22c 88a 17 P. aquilinum USA 46.558961, 239a 814a 77a 141a −90.662296 18 P. aquilinum USA 46.39491, 524a 327a 57a 29a −89.14131

The levels of certainty of determination of the compounds were characterized by the following levels: a Distinctive mass fragments. b Retention time and detection of corresponding pterosin (parent glycoside) in the same sample. c Retention time without emerging distinctive mass fragments due to co-elution of other substances from the plants.

7 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

bracken (Pteridium aquilinum) distributions, ISPRS J. Photogramm. Remote Sens. 75 (2013) 48–63. [7] T.N. Matongera, O. Mutanga, T. Dube, M. Sibanda, Detection and mapping the spatial distribution of bracken fern weeds using the Landsat 8 OLI new generation sensor, Int. J. Appl. Earth Obs. Geoinf. 57 (2017) 93–103. [8] K. Yamada, M. Ojika, H. Kigoshi, Ptaquiloside, the major toxin of bracken, and related terpene glycosides: chemistry, biology and ecology, Nat. Prod. Rep. 24 (2007) 798–813. [9] I. Hirono, H. Ogino, M. Fujimoto, K. Yamada, Y. Yoshida, M. Ikagawa, M. Okumura, Induction of tumors in ACI rats given a diet containing ptaquiloside, a bracken carcinogen, J. Natl. Cancer Inst. 79 (1987) 1143–1149. [10] J. Gomes, A. Magalhaes, V. Michel, I.F. Amado, P. Aranha, R.G. Ovesen, H.C.B. Hansen, F. Gartner, C.A. Reis, E. Touati, Pteridium aquilinum and its ptaquiloside toxin induce DNA damage response in gastric epithelial cells, a link with gastric carcinogenesis, Toxicol. Sci. 126 (2012) 60–71. [11] M. Matoba, E. Saito, K. Saito, K. Koyama, S. Natori, T. Matsushima, M. Takimoto, Assay of ptaquiloside, the carcinogenic principle of bracken, Pteridium aquilinum, by mutagenicity testing in Salmonella typhimurium, Mutagenesis 2 (1987) 419–423. [12] D.M. Potter, M.S. Baird, Carcinogenic effects of ptaquiloside in bracken fern and related compounds, Br. J. Cancer. 83 (2000) 914–920. [13] P. Engel, K.K. Brandt, L.H. Rasmussen, R.G. Ovesen, J. Sørensen, Microbial degradation and impact of Bracken toxin ptaquiloside on microbial communities in soil, Chemosphere 67 (2007) 202–209. [14] M. Ojika, K. Wakamatsu, H. Niwa, K. Yamada, Ptaquiloside, a potent carcinogen isolated from bracken fern Pteridium aquilinum var. latiusculum: structure eluci- Fig. 3. Ternary distribution of concentrations of glycosides prior to hydrolysis dation based on chemical and spectral evidence, and reactions with amino acids, nucleosides, and nucleotides, Tetrahedron. 43 (1987) 5261–5274. in 17 positive fern samples. [15] T. Kushida, M. Uesugi, Y. Sugiura, H. Kigoshi, H. Tanaka, J. Hirokawa, M. Ojika, K. Yamada, DNA damage by ptaquiloside, a potent bracken carcinogen: detection of Acknowledgements selective strand breaks and identification of DNA cleavage products, J. Am. Chem. Soc. 116 (1994) 479–486. [16] International Agency for Research on Cancer (IARC) - Summaries & Evaluations, The authors are grateful to EU framework programme Horizon 2020 Bracken fern (Pteridium aquilinum) and some of its constituents. http://www. and providers of analytical samples: Mary T. Fletcher (The University of inchem.org/documents/iarc/vol40/brackenfern.html, 1998. (accessed 20 January 2019). Queensland, Australia), Henrik Ærenlund Pedersen (Natural History [17] Agents Classified by the IARC Monographs, Volumes 1–123. https://monographs. Museum of Denmark), Rinku Sharma (Indian Veterinary Research iarc.fr/list-of-classifications-volumes/, 2018. (Accessed 20 January 2019). Institute), Anya Burton (International Agency for Research on Cancer), [18] I. Hirono, Edible Plants Containing Naturally Occurring Carcinogens in Japan, Jpn. J. Cancer Res. 84 (1993) 997–1006. Blandina Theophil Mmbaga (Kilimanjaro Clinical Research Institute), [19] R.C. Recouso, R.C. Stocco dos Santos, R. Freitas, R.C. Santos, A.C. de Freitas, Paulo Vargas Peixoto (Federal Rural University of Rio de Janeiro, O. Brunner, W. Beçak, C.J. Lindsey, Clastogenic effect of bracken fern (Pteridium Brazil), Paulo César dos Reis Aranha (University of Copenhagen), aquilinum v. arachnoideum) diet in peripheral lymphocytes of human consumers: preliminary data, Vet. Comp. Oncol. 1 (2003) 22–29. Latisha Coffin and Owen Holly Maroney (Great Lakes Indian Fishand [20] S. Mahmood, B.L. Smith, A.S. Prakash, Bracken carcinogens in the human diet, Wildlife Commission, USA). The authors would also like to thank Mutat. Res. 443 (1999) 69–79. Nicoline Wildenrath Brandtoft Hansen (intern at University of [21] M.E. Alonso-Amelot, M. Avendano, Human carcinogenesis and bracken fern: a review of the evidence, Curr. Med. Chem. 9 (2002) 675–686. Copenhagen), laboratory technician Jimena Bertorello Martinez and [22] M.E. Alonso-Amelot, M. Avendano, Possible association between gastric cancer and laboratory technician students Jonas Halgren Mikkelsen and Michael bracken fern in Venezuela: An epidemiologic study, Int. J. Cancer. 91 (2001) Josefsen (University College Copenhagen), LC and LC-MS application 252–259. specialists Florian Rieck and Lilla Guricza (Agilent technologies, [23] J.C. van der Hoeven, W.J. Lagerweij, M.A. Posthumus, A. van Veldhuizen, H.A. Holterman, Aquilide A, a new mutagenic compound isolated from bracken fern Germany), honourable constant supporters Lina Kisielienė, Monika (Pteridium aquilinum (L.) Kuhn), Carcinogenesis 4 (1983) 1587–1590. Kisieliūtė and Kotryna Kisieliūtė. [24] H. Niwa, M. Ojika, K. Wakamatsu, K. Yamada, S. Ohba, Y. Saito, I. Hirono, K. Matsushita, Stereochemistry of ptaquiloside, a novel norsesquiterpene glucoside from bracken, Pteridium aquilinum var. latiusculum, Tetrahedron. Lett. 24 (1983) Funding 5371–5372. [25] U.F. Castillo, A.L. Wilkins, D.R. Lauren, B.L. Smith, N.R. Towers, M.E. Alonso- This project was supported by the European Union’s Horizon 2020 Amelot, R. Jaimes-Espinoza, Isoptaquiloside and caudatoside, illudane-type sesquiterpene glucosides from Pteridium aquilinum var. caudatum, Phytochemistry 44 research and innovation programme under the Marie Sklodowska-Curie (1997) 901–906. grant agreement No. 722493. [26] U.F. Castillo, M. Ojika, M. Alonso-Amelot, Y. Sakagami, Ptaquiloside Z, a new toxic unstable sesquiterpene glucoside from the neotropical bracken fern Pteridium aquilinum var. caudatum, Bioorg. Med. Chem. 6 (1998) 2229–2233. Appendix A. Supplementary material [27] M.T. Fletcher, P.Y. Hayes, M.J. Somerville, J.J. De Voss, Ptesculentoside, a novel norsesquiterpene glucoside from the Australian bracken fern Pteridium esculentum, Supplementary data to this article can be found online at https:// Tetrahedron. Lett. 51 (1997) 1997–1999. [28] K. Saito, T. Nagao, S. Takatsuki, K. Koyama, S. Natori, The sesquiterpenoid carci- doi.org/10.1016/j.jchromb.2019.121966. nogen of bracken fern, and some analogues, from the pteridaceae, Phytochemistry 29 (1990) 1475–1479. References [29] M.T. Fletcher, I.J. Brock, K.G. Reichmann, R.A. McKenzie, B.J. Blaney, Norsesquiterpene Glycosides in Bracken Ferns (Pteridium esculentum and Pteridium aquilinum subsp. Wightianum) from Eastern Australia: Reassessed [1] J.A. Thomson, Towards a taxonomic revision of Pteridium (Dennstaedtiaceae), Poisoning Risk to Animals, J. Agric. Food Chem. 59 (2011) 5133–5138. Telopea (Syd.) 10 (2004) 793–803. [30] S.K. Rai, R. Sharma, A. Kumari, L.H. Rasmussen, R.D. Patil, R. Bhar, Survey of ferns [2] J. Vetter, A biological hazard of our age: bracken fern [Pteridium aquilinum (L.) and clinico-pathological studies on the field cases of Enzootic bovine haematuria in Kuhn] - a review, Acta Vet. Hung. 57 (2009) 183–196. Himachal Pradesh, a north-western Himalayan state of India, Toxicon 138 (2017) [3] R.H. Marrs, A.S. Watt, Biological Flora of the British Isles: Pteridium aquilinum (L.) 31–36. Kuhn, J. Ecol. 94 (2006) 1272–1321. [31] R.M. Gil da Costa, M.M.S.M. Bastos, P.A. Oliveira, C. Lopes, Bracken-associated [4] C.N. Page, The taxonomy and phytogeography of bracken - a review, Bot. J. Linn. human and animal health hazards: chemical, biological and pathological evidence, Soc. 73 (1976) 1–34. J. Hazard. Mater. 1 (2012) 203–204. [5] R.J. Pakeman, R.H. Marrs, D.C. Howard, C.J. Barr, R.M. Fuller, The bracken pro- [32] A. Virgilio, A. Sinisi, V. Russo, S. Gerardo, A. Santoro, A. Galeone, O. Taglialatela- blem in Great Britain: its present extent and future changes, Appl Geogr. 16 (1996) Scafati, F. Roperto, Ptaquiloside, the major carcinogen of bracken fern, in the 65–86. pooled raw milk of healthy sheep and goats: an underestimated, global concern of [6] J. Holland, P. Aplin, Super-resolution image analysis as a means of monitoring food safety, J. Agric. Food Chem. 63 (2015) 4886–4892.

8 V. Kisielius, et al. Journal of Chromatography B 1138 (2020) 121966

[33] J. Gomes, A. Magalhães, V. Michel, I.F. Amado, P. Aranha, R.G. Ovesen, herbarium speciments for qualitative mapping purposes, Phytochem. Anal. 28 H.C. Hansen, F. Gärtner, C.A. Reis, E. Touati, Pteridium aquilinum and its ptaqui- (2017) 575–583. loside toxin induce DNA damage response in gastric epithelial cells, a link with [47] M. Fukuoka, M. Kuroyanagi, K. Yoshihira, S. Natori, Chemical and toxicological gastric carcinogenesis, Toxicol. Sci. 126 (2012) 60–71. studies on bracken fern, Pteridium aquilinum var. latiusculum. II. Structure of [34] O.P. Galpin, C.J. Whitaker, R. Whitaker, J.Y. Kassab, Gastric cancer in Gwynedd. pterosins, sesquiterpenes having 1-indanone skeleton, Chem. Pharm. Bull. 26 Possible links with bracken, Br. J. Cancer. 61 (1990) 737–740. (1978) 2365–2385. [35] Y.C. Caceres-Pena, M. Naya, M.P. Calcagno-Pissarelli, M.E. Alonso-Amelot, [48] P.B. Oelrichs, J.C. Ng, J. Bartley, Purification of ptaquiloside, a carcinogen from Influence of bracken fern (Pteridium caudatum L. Maxon) pre-treatment onex- Pteridium aquilinum, Phytochemistry 40 (1995) 53–56. traction yield of illudane glycosides and pterosins, Phytochem. Anal. 24 (2012) [49] R.M.G. da Costa, P. Coelho, R. Sousa, M.M.S.M. Bastos, B. Porto, J.P. Teixeira, 290–295. I. Malheiro, C. Lopes, Multiple genotoxic activities of ptaquiloside in human lym- [36] P.H. Jensen, O.S. Jacobsen, H.C. Hansen, R.K. Juhler, Quantification of ptaquiloside phocytes: Aneugenesis, clastogenesis and induction of sister chromatid exchange, and pterosin B in soil and groundwater using liquid chromatography-tandem mass Mutat. Res. Genet. Toxicol. Environ. Mutagen. 747 (2012) 77–81. spectrometry (LC-MS/MS), J. Agric. Food Chem. 56 (2008) 9848–9854. [50] M.T. Fletcher, K.G. Reichmann, I.J. Brock, R.A. McKenzie, B.J. Blaney, Residue [37] P.C.R. Aranha, H.C.B. Hansen, L.H. Rasmussen, B.W. Strobel, C. Friis, potential of norsesquiterpene glycosides in tissues of cattle fed austral bracken Determination of ptaquiloside and pterosin B derived from bracken (Pteridium (Pteridium esculentum), J. Agric. Food Chem 59 (2011) 8518–8523. aquilinum) in cattle plasma, urine and milk, J. Chromatogr. B Analyt. Technol. [51] M.E. Alonso-Amelot, U. Castillo, B.L. Smith, D.R. Lauren, Bracken ptaquiloside in Biomed. Life Sci. 951–952 (2014) 44–51. milk, Nature 382 (1996) 587. [38] F. Clauson-Kaas, H.C. Hansen, B.W. Strobel, UPLC-MS/MS determination of pta- [52] F. Bonadies, G. Berardi, R. Nicoletti, F.S. Romolo, F. Giovanni, A. Marabelli, quiloside and pterosin B in preserved natural water, Anal. Bioanal. Chem. 408 C. Santoro, A. Raso, A. Tagarelli, F. Roperto, V. Russo, S. Roperto, A new, very (2016) 7981–7990. sensitive method of assessment of ptaquiloside, the major bracken carcinogen in the [39] M.P. Agnew, D.R. Lauren, Determination of ptaquiloside in bracken fern (Pteridium milk of farm animals, Food Chem. 124 (2011) 660–665. esculentum), J. Chromatogr. A. 538 (1991) 462–468. [53] F. Clauson-Kaas, C. Ramwell, H.C.B. Hansen, B.W. Strobel, Ptaquiloside from [40] P.W. Burkhalter, P.M.J. Groux, U. Candrian, P. Hubner, J. Luthy, Isolation, de- bracken in stream water at base flow and during storm events, Water Res. 106 termination and degradation of ptaquiloside - A bracken fern (Pteridium aquilinum) (2016) 155–162. carcinogen, J. Nat. Toxins. 5 (1996) 141–159. [54] C. O'Driscoll, C. Ramwell, B. Harhen, L. Morrison, F. Clauson-Kaas, H.C.B. Hansen, [41] M.E. Alonso-Amelot, U. Castillo, B.L. Smith, D.R. Lauren, Excretion, through milk, G. Campbell, J. Sheahan, B. Misstear, L.W. Xiao, Ptaquiloside in Irish Bracken ferns of ptaquiloside in bracken-fed cows.A quantitative assessment, Lait. 78 (1998) and receiving waters, with implications for land managers, Molecules 543 413–423. (2016) 21. [42] L.H. Rasmussen, H.C.B. Hansen, D. Lauren, Sorption, degradation and mobility of [55] F. Clauson-Kaas, P.H. Jensen, O.S. Jacobsen, R.K. Juhler, H.C.B. Hansen, The ptaquiloside, a carcinogenic Bracken (Pteridium sp.) constituent, in the soil en- naturally occurring carcinogen ptaquiloside is present in groundwater below vironment, Chemosphere 58 (2005) 823–835. bracken vegetation, Toxicol. Environ. Chem. 33 (2014) 1030–1034. [43] L.H. Rasmussen, S. Kroghsbo, J.C. Frisvad, H.C. Hansen, Occurrence of the carci- [56] F.L. Hsu, C.F. Huang, Y.W. Chen, Y.P. Yen, C.T. Wu, B.J. Uang, R.S. Yang, S.H. Liu, nogenic Bracken constituent ptaquiloside in fronds, topsoils and organic soil layers Antidiabetic effects of pterosin A, a small-molecular-weight natural product, on in Denmark, Chemosphere 51 (2003) 117–127. diabetic mouse models, Diabetes 62 (2013) 628–638. [44] A. Oliveros-Bastidas, M.P. Calcagno-Pissarelli, M. Naya, J.L. Avila-Nunez, [57] Y. Yahara, H. Takemori, M. Okada, A. Kosai, A. Yamashita, T. Kobayashi, K. Fujita, M.E. Alonso-Amelot, Human gastric cancer, Helicobacter pylori and bracken car- Y. Itoh, M. Nakamura, H. Fuchino, N. Kawahara, N. Fukui, A. Watanabe, T. Kimura, cinogens: A connecting hypothesis, Med. Hypotheses. 88 (2016) 99–191. N. Tsumaki, Pterosin B prevents chondrocyte hypertrophy and osteoarthritis in [45] L.H. Rasmussen, E. Donnelly, B.W. Strobel, P.E. Holm, H.C.B. Hansen, Land man- mice by inhibiting Sik3, Nat. Commun. 7 (2016) 1–12. agement of bracken needs to account for bracken carcinogens - A case study from [58] H. Sheridan, N. Frankish, R. Farrell, Smooth muscle relaxant activity of pterosin Z Britain, J. Environ. Manage. 151 (2015) 258–266. and related compounds, Planta Med. 65 (1999) 271–272. [46] L.H. Rasmussen, H.Æ. Pedersen, Screening for ptaquiloside in ferns: using

9 Supplementary Material

Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns

Vaidotas Kisielius, Dan Nybro Lindqvist, Mikkel Boas Thygesen, Michael Rodamer, Hans Christian Bruun Hansen, Lars Holm Rasmussen

102

Information on tested plant samples:

Table S1 Origin of dried fern powder used in the study

Sample Sample information Previous analysis of the sample no. Upper 40 cm part of fully matured frond collected for the purpose of this study in the summer of 2018, dried for 4-5 PTE, CAU, PTA with a conversion 1 months at ambient temperature in an ordinary plant press, method of a sample from the same grinded to fine powder and stored at -20 °C for 2-3 months. location [182]

Upper 40 cm part of fully matured fronds collected for the purpose of this study in the summer of 2018, dried for 4-5 2, months at ambient temperature in an ordinary plant press, n/a 4-10 grinded to fine powder and stored at -20 °C for 2-3 months.

Fully matured frond collected in 2014, dried for 1-2 months at ambient temperature in an ordinary plant press, grinded to fine powder and stored at ˗20°C for 4 years. Provided by 3 PTA and pterosin B [46] Assoc. Prof. Henrik A. Pedersen (Natural History Museum of Denmark).

Fully matured frond collected in 2017, shade-dried at room temperature under a fan for 2 days, grinded to fine powder, 11 sent by post and stored at -20 °C for 1.5 years. Provided by Dr PTA and pterosin B [30] Rinku Sharma (Indian Veterinary Research Institute).

Recently emerged furled fronds (croziers) collected in April 2017, freeze-dried, grinded to fine powder, sent by post and PTE, CAU, PTA with a conversion 12 stored at ˗20°C for 1.5 years. Provided by Prof. Mary T. method of a sample from the same Fletcher (The University of Queensland, Australia). location [29]

Fully matured fronds collected in October 2016, dried at room temperature in paper bags, grinded to fine powder, sent by post and stored at -20 °C for 2 years. Provided by Dr Anya 13, 14 Burton (International Agency for Research on Cancer) and Dr No Blandina Theophil Mmbaga (Kilimanjaro Clinical Research Institute).

Dry powder of matured fronds provided by Prof. Paulo V. Peixoto (Federal Rural University of Rio de Janeiro, Brazil) and 15, 16 Paulo César dos Reis Aranha (University of Copenhagen). No Stored at room temperature for several years.

Recently emerged fronds collected in May 2017, grinded to fine powder, sent by post and stored at -20 °C for 1.5 years. 17, 18 No Provided by Latisha Coffin and Owen H. Maroney (Great Lakes Indian Fish and Wildlife Commission, USA).

103

Preparation of analytical standards:

Fig. S1 Elution of preconcentrated solution of bracken fern, injected to preparative HPLC system. Report printed on thermal paper by Hitachi D 2000 Chromato-Integrator. The peaks of eluting PTE, CAU and PTA are emphasised.

104

1H NMR:

Fig. S2 1H NMR spectrum of PTA and an internal standard with absorption units on y-axis. Equating the peak area at 0.00 ppm (nuclide #1 of the internal standard) to 1.00, the peak at 5.80 ppm (nuclide # 5 of the analyte) corresponds to relative area 1.52. The peak at 1.32 ppm (nuclide # 14 of the analyte) corresponds to relative area 4.36.

Fig. S3 Molecular structure of 3-(trimethylsilyl)propionic-2,2,3,3-d4 acid, sodium salt (internal standard) used in 1H NMR, with numbering of nuclides.

105

The calculation for the purity of the analyte in 1H NMR:

퐴퐼푥 푁푁𝑖 푚푥 massx = ∙ ∙ 푛𝑖 ∙ 푀푊푥 ∙ 푃𝑖 ; 푝푢푟푖푡푦x = 퐴퐼𝑖 푁푁푥 푚푠푎푚푝푙푒 where: subscripts x and i relate to the analyte and internal standard, respectively; RI is the relevant integral for a signal; NN is the corresponding number of nuclides (protons); n is number of moles (mol); MW is the molar mass (g/mol); m is the weighed or calculated mass of material (mg); Pi is the purity of internal standard (decimal number).

The internal standard (nuclide #1) corresponds to NNi = 9. Ptaquiloside at a singlet peak 1.32 ppm (nuclide # 14) corresponds to NNx = 3. Ptaquiloside at a singlet peak at 5.80 ppm (nuclide # 5) corresponds to NNx = 1.

The purity calculations proceeded as follows: 4.36 9 g mass = ∙ ∙ 5.0 ∙ 10-4 mol ∙ 398.45 ∙ 0.999 = 2.603 mg (1.32) 1.0 3 mol 2,603 mg purity = = 0.90 ≈ 90 % (1.32) 2.9 mg 1.52 9 g mass = ∙ ∙ 5.0 ∙ 10-4 mol ∙ 398.45 ∙ 0.999 = 2.723 mg (5.80) 1.0 1 mol 2.723 mg purity = = 0.94 ≈ 94 % (5.80) 2.9 mg 90 + 94 Average purity = = 92 % 2

Effect on quantification of analytes by HPLC column temperature:

Table S2 Effect of HPLC column temperature on quantification of the three illudane glycosides and comparison of retention times.

PTE CAU PTA Peak area (mAU*s x106) at 35 °C (duplicate measurement) 2.30; 2.30 1.30; 1.30 0.98; 0.99 Peak area (mAU*s x106) at 60 °C (duplicate measurement) 1.99; 2.10 1.20; 1.16 1.00; 1.00 Peak height (mAU*s x106) at 35 °C (duplicate measurement) 0.31; 0.31 0.29; 0.29 0.23; 0.23 Peak height (mAU*s x106) at 60 °C (duplicate measurement) 0.20; 0.22 0.26; 0.25 0.23; 0.24 Retention time (min.) at 35 °C 2.8 3.4 3.6 Retention time (min.) at 60 °C 2.3 3.2 3.5 Retention time (min.) at 35 °C replacing acetonitrile with methanol 3.4 3.8 3.9* * Resolution was not achieved between PTA and pterosin G

106

Effect on ionisation efficiency of analytes by ESI parameters:

Fig. S4 Signal response in 81 ESI methods (Graphical analysis software Plotly).

107

The results in Fig. S4 demonstrate the average signal strength (Average response) of PTE, CAU and PTA in the mass spectrometer applying different ESI methods. Each data point represents method with individual set of parameters: drying gas temperature, nebulizer pressure, drying gas flow (graph axes) and cone voltage (graph titles). The colour chart scales individual to each graph represent the average signal strength. The graphical representation shows the trend of increased ionization efficiency towards the middle cone voltage (3,000 V) and maximum drying gas temperature (350 °C).

Fig. S5 Comparison of the effects of ESI parameters (factors) on ionisation efficiency in a Pareto plot. The single parameter of drying gas temperature is beyond the standardized effect value (1.997) and has significant effect on ESI (Statistical analysis software Minitab 18). Figure S5 demonstrates that the efficiency of ESI significantly depends only on drying gas temperature (factor C).

108

Linearities of calibration curves over investigated concentration ranges

Fig. S6 Calibration curves of the standard solutions of analytes. Each calibration curve was made from 7 data points. Each data point was derived as a mean average of the measured peak areas from single injections of 3 independently made vials (n=3). The error bars representing the standard errors between the individual measurements of each data point are too small to be visually discernible. The numerical RSD values between the individual measurements are presented in Table 5 of the main text.

109

Effect on extraction efficiency of analytes by solvent and mixing duration:

Fig. S7 Comparison of the effects of parameters (factors) on the extraction efficiency of analytes in a Pareto plot. None of the parameters are beyond the standardized effect value (16.41) (Statistical analysis software Minitab 18). Figure S7 demonstrates that none of the single parameters or a combination of parameters have significant effect on the extraction efficiency. Hence, the results of individually performed extraction events are highly comparable, even in cases when not identical parameters (e.g. extraction time) are applied.

110

Matrix effects:

Fig. S8 Matrix effects of plant extract of freeze-dried Pteridium esculentum (Australia).

Fig. S9 Matrix effects of plant extract of plant-press dried Pteridium aquilinum (Sweden).

Figures S8 and S9 demonstrate that quantification of all analytes follows linear patterns after dilution of plant extract by 20 and more times. The region of linear quantification represents elimination of matrix effects.

111

UV absorption profiles of analytes:

Fig. S10 UV absorption profiles at 190 – 400 nm of PTE (top), CAU (middle) and PTA (bottom). The maximum absorption wavelength for all glycosides is 202 nm.

112

Fig. S11 UV absorption profiles at 190 – 400 nm of pterosins G (top), A (middle) and B (bottom). The maximum absorption wavelength for all pterosins is 217 nm.

113

Estimation of glycoside concentrations in analytical samples prior to hydrolysis:

Table S3. Calculated concentrations of glycosides prior to hydrolysis

Calculated original µg/g DW sample in plant sample No. PTE CAU PTA Concentrations of glycosides prior to hydrolysis were obtained 1 50 305 2003 summing the concentrations of quantified glycosides with 2 0 26 22 concentrations of quantified corresponding pterosins and 3 0 0 0 applying 1:1 molar conversion ratios.

4 436 185 1903 푀푃푇퐸 푃푇퐸푝푟𝑖표푟 푡표 ℎ푦푑푟표푙푦푠𝑖푠 = 푃푇퐸푑푒푡푒푐푡푒푑 + 푃푡푟퐺푑푒푡푒푐푡푒푑 ∙ 5 0 13 56 푀푃푡푟퐺 6 0 58 0 푀퐶퐴푈 7 0 656 13 퐶퐴푈푝푟𝑖표푟 푡표 ℎ푦푑푟표푙푦푠𝑖푠 = 퐶퐴푈푑푒푡푒푐푡푒푑 + 푃푡푟퐴푑푒푡푒푐푡푒푑 ∙ 푀푃푡푟퐴 8 0 51 106 푀푃푇퐴 9 0 64 457 푃푇퐴 = 푃푇퐴 + 푃푡푟퐵 ∙ 푝푟𝑖표푟 푡표 ℎ푦푑푟표푙푦푠𝑖푠 푑푒푡푒푐푡푒푑 푑푒푡푒푐푡푒푑 푀 10 0 52 316 푃푡푟퐵 11 0 0 9 12 3900 429 992

13 94 898 2125 14 6 0 100 15 31 2246 47

16 39 27 161 17 0 1643 335 18 0 1088 110

Mass spectra of analytes:

(Next 6 pages)

114

Fig. S12 Mass spectra of analytical standard of PTE

115

Fig. S13 Mass spectra of analytical standard of CAU

116

Fig. S14 Mass spectra of analytical standard of PTA

117

Fig. S15 Mass spectra of analytical standard of PtrG

118

Fig. S16 Mass spectra of analytical standard of PtrA

119

Fig. S17 Mass spectra of analytical standard of PtrB

120

2. Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium

Technical report

Vaidotas Kisielius1,2, Bo Markussen3 Lars Holm Rasmussen1

Purpose

Manuscript 1 developed a new analytical method to quantify illudane glycosides ptesculentoside (PTE), caudatoside (CAU) and ptaquiloside (PTA) in bracken ferns (genus Pteridium). The method applied on 18 samples from 6 continents of the world demonstrated that the two previously underinvestigated compounds PTE and CAU can be present in the plants in relatively high concentrations, also exceeding PTA. Moreover, PTE and CAU are more water-soluble compounds than PTA. Nevertheless, no systematic investigation of concentrations of these compounds in the plants was performed. This study has been prepared to demonstrate the spatial and temporal variation of the compounds in the plants and to determine any environmental risk factors that influence higher concentrations, that in turn would imply higher risk to adjacent water resources.

Materials and methods Sample collection

Spatial variation of illudane glycosides Worldwide common Pteridium aquilinum L. (Kuhn) and northern Hemisphere specific Pteridium aquilinum var. latiusculum (Desv.) (hereafter both referred to as bracken) were sampled once in 66 sites along a 1500 km NE-SW gradient in Northern Europe during July 12–31, 2018 (Figure 1). Three parts of the bracken fronds were taken in each site: 40 cm tip of a fully matured frond (“Triplicate No. 1”), the whole fully matured lowest pinna of the same frond (“Triplicate No. 2”) and 40 cm tip of a fully matured frond remote from the Triplicate No. 1 by 10–20 m (“Triplicate No. 3”) (see spreadsheet in the Pages 128-132). In the statistical analysis the bracken tips are referred to as upper, and the lowest pinnae as lower parts of the fronds. The WGS84 coordinates of the sampling locations were determined by an off-road navigation app for Android Denmark Topo Maps (©2018 ATLOGIS Geoinformatics GmbH & Co. KG). The subspecies

1 Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark

2 Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark

3 Department of Mathematical Sciences, University of Copenhagen, Universitetsparken 5, 2100 Copenhagen, Denmark

121 were identified by a taxonomic criteria provided in Den nye nordiske flora [1]. The heights of the sampled fronds from the surface of the ground were measured, information about the bracken stand access to the direct sunshine recorded, adjacently growing tree species registered and the images of the bracken stands and the surrounding environments taken.

Figure 1. The sites of the spatial variation study sampled in Northern Europe in July 12-31, 2018. (Background map © deviantart.com).

Temporal variation of illudane glycosides Two locations encircling the sites of the spatial variation study (Figure 1: No 4 (Grib Skov) and No 12 (Humleore)) were studied for the temporal variation of the illudane glycosides throughout the vegetation season of 2018. The locations were selected as large forests extensively infested with Pteridium aquilinum within the distances from the research laboratory allowing to process the samples on the sampling day. Besides, the both locations were selected as containing surface water bodies and active groundwater extraction wells in order to potentially facilitate future studies assessing the risks of the illudane glycosides to water. Each location was divided into 7 subsites (Figure 2). Every subsite was sampled the 1st week of each month from June to November by cutting from the surface of the ground 2 whole fully matured average-height fronds remote from one another by 20–40 m and merging them in a plastic bag as one sample.

122

Figure 2. The sites of the temporal variation study in the locations Grib Skov (A) and Humleore (B), sampled the first week of every month from June till November, 2018. Background map © Google maps.

Sample preservation

The samples of the spatial variation study were pressed in an ordinary cardboard-based plant presses within several minutes after collection (blotting paper for plant press 28x43 cm, Frederiksen Scientific, Denmark). The samples No 1–27 and 28–66 were carried in a vehicle until the middle and until the end of the expedition period, respectively. Unloaded pressed samples were stored for 4–5 months in approx. 22 °C well-ventilated room. To maintain the reference of the subspecies, high resolution images of the bottom side of each dried pinnae (sample 2) were taken under a neutral light including the size and the colour references (camera Nikon Coolpix A 900, optical zoom 4K 35X, light source for optical microscopy Olympus FLQ 75, certified white balance reference card WhiBal G7 WB7-PC). The samples of the temporal variation study were preserved at the same day of each sampling event. The fronds were cut lengthwise at the centre of the stems, chopped crosswise into 3-5cm pieces, each sample placed in a separate open paper bag and dried for 72 hours in 45 °C fan- ventilated oven manually rotating the contents in each bag 1–2 times per day [2]. The dried material was transferred into plastic bags and kept in -20 °C for 4–8 months until grinding. The volume of the samples of the temporal variation study was reduced by vigorously mixing and scooping a full 180 ml glass of the sample content. The reduced samples of the temporal variation study and the whole samples of the spatial variation study were individually grinded to fine powder

123

(Kenwood AT320A blade-chopper ran by 1000W Kenwood Chef Kitchenmach engine). After each grinding the blades and the plant touching parts of the equipment were manually cleaned with DI water following 70% ethanol, wiped with disposable tissues and dried under gentle air flow. The 100 % efficiency of the cleaning procedure was verified by grinding dry leaves of green tea in the equipment and not detecting illudane glycosides or pterosins in them with the LCMS. All grinded samples were kept for 3–6 moths in a plastic zip-bags in -20 °C prior to analysis.

Chemical analysis

All samples were prepared and analysed for illudane glycosides PTE, CAU, PTA and corresponding pterosins G, A and B on Agilent 1260 Infinity HPLC System equipped with Agilent 6130 Single Quadrupole mass spectrometer by the method described in Kisielius at. al. [3]. The contents of illudane glycosides in the plant samples prior to hydrolysis were referred to as the sums of the concerned quantified glycosides and their corresponding pterosins applying 1:1 M conversion ratios [3]. The results are provided in the spreadsheets in pages 128-135.

Statistical analysis

The sites of the spatial variation study were classified based on the field-recorded data of the adjacently growing tree species in categories of dropping leaves and others (deciduous, coniferous, both types and no trees) and by the light transparency of their canopies (light, dark, both types and no trees) (Supplementary material). As longitude and latitude of the sampling locations were collinear (푅2 = 0.84) these variables were replaced by their first principal component representing geographic gradient (PC1). The variables of the datasets investigating the studies of the spatial and temporal variation are listed in Table 1 and Table 2, respectively.

Table 1. Variables in the statistical analysis of the study of spatial variation

Variable Value (proportion) / Range [min ; max] Usage PTE 0 (75%) / [0.01828 ; 0.50146] response CAU 0 (5%) / [0.02236 ; 5.45902] response PTA 0 (2%) / [0.00313 ; 10.5496] response longitude [85,065 ; 3,100,000] not used latitude [5,504,385 ; 6,319,842] not used PC1 [-1.4263 ; 3.1381] fixed effect country Denmark (45%), Sweden (30%), Finland fixed effect (24%) subspecies aquilinum (85%), var. latiusculum (15%) fixed effect sunshine No (76%), Yes (24%) fixed effect deciduous No (24%), Yes (76%) fixed effect coniferous No (53%), Yes (47%) fixed effect

124

light No (79%), Yes (21%) fixed effect dark No (52%), Yes (48%) fixed effect part lower (33%), upper (66%) fixed effect site 66 different locations random effect

Table 2. Variables in the statistical analysis of the study of temporal variation.

Variable Value (proportion) / Range [min ; max] Usage PTE 0 (67%) / [0.00983 ; 0.18451] response CAU 0 (4%) / [0.06050 ; 0.96170] response PTA 0 (7%) / [0.00106 ; 4.21487] response month Jun (16%), …, Nov (16%) fixed effect month number 6 (16%), …, 11 (16%) fixed effect location G (50%), H (50%) fixed effect (†) sample G1 (7%), …, G7 (7%), H1 (7%), …, H7 (7%) random effect (†)

The analyses were performed by investigating the two datasets in four separate parts in statistical computing software R. The first three parts were marginal analyses of the contents of PTE, CAU and PTA, respectively, and the fourth part was the relative distribution of the three compounds. The marginal analyses were performed in a two-step procedure. First a logistic regression on the response being larger than zero, and then a normal regression on the logarithm of the strictly positive responses [4]. The normal regressions were validated by residual and normal quantile plots, which suggested that the logarithm of the content gave the better modelling. The relative contents between PTE, CAU and PTA were analysed in a Dirichlet regression using the alternative parameterization of Maier [5] with PTA content as the reference value. The Dirichlet package does not allow for random effects. Thus, in the analysis of the spatial variation the hypotheses were evaluated by permutation tests based on the negative log likelihood statistics. As PC1, country, subspecies, sunshine, deciduous, coniferous, light and dark were nested within site, the permutations for these effects were done between samples. As all levels for part were measured for all samples, the permutations for the effect of part were done within samples. To obtain parsimonious models backward model reduction was done via significance tests on a 5 percent significance level, where the initial models are given by including all the fixed effects listed in Table 1 and Table 2. Significance tests and estimates were reported from the reduced models. In the study of temporal variation sample was used as a fixed effect in the Dirichlet regression, which does not allow for random effects. Since sample was nested in location (†), location was not used in the Dirichlet regression (Table 2).

125

Results Sample analysis

The spreadsheet below provides the results of the analysed contents of pterosins and glycosides in the samples of spatial variation study. All analyses were performed and the contents of glycosides prior to hydrolysis (Σ) were obtained based on [3] (empty values represent no detection; 0.00 represent ˂0.005).

mg/g dry weight Site Triplicate Σ Σ Σ PTE PtrG CAU PtrA PTA PtrB No. No. PTE CAU PTA

1 0.15 0.13 0.37 1.43 0.52 2.39 1 2 0.12 0.13 0.34 0.25 0.63 1.39 3 0.31 0.15 0.56 1.52 0.40 2.26

1 0.24 0.03 0.28 0.16 0.07 0.28 2 2 0.54 0.06 0.64 0.92 0.22 1.33 3 0.06 0.10 0.14 0.03 0.20 1 0.11 0.06 0.22 0.65 0.66 1.80 3.65 1.75 6.85 3 2 0.19 0.09 0.34 1.22 0.93 2.82 6.12 2.43 10.55 3 0.08 0.01 0.10 0.40 0.21 0.76 1.47 0.40 2.20

1 0.02 0.04 0.05 0.07 0.17 4 2 0.01 0.02 0.02 0.03 3 0.04 0.08

1 5 2 0.01 0.03 0.07 3 0.01 0.04 0.09

1 0.30 0.06 0.39 0.56 0.13 0.80 6 2 0.28 0.08 0.41 0.44 0.14 0.71 3 0.42 0.09 0.58 0.19 0.09 0.35

1 0.28 0.08 0.41 0.93 0.31 1.49 7 2 0.27 0.07 0.39 0.76 0.35 1.41 3 0.48 0.15 0.73 1.47 0.31 2.02

1 0.02 0.04 0.20 0.07 0.32 0.71 0.19 1.06 8 2 0.02 0.03 0.46 0.10 0.63 1.05 0.41 1.79 3 0.02 0.02 0.06 0.48 0.25 0.91 2.00 0.53 2.97 1 0.11 0.05 0.20 0.32 0.12 0.53 2.77 0.65 3.95 9 2 0.11 0.06 0.22 0.41 0.14 0.65 3.62 0.98 5.41 3 0.06 0.01 0.07 0.30 0.11 0.49 1.78 0.38 2.47

1 0.26 0.08 0.39 1.50 0.48 2.38 10 2 0.21 0.08 0.35 1.68 0.60 2.77 3 0.06 0.02 0.09 0.11 0.08 0.25 0.89 0.29 1.41

1 0.09 0.16 11 2 0.49 0.16 0.78 3 0.07 0.12 0.30 0.13 0.54

126

1 0.36 0.09 0.52 1.29 0.35 1.92 12 2 0.49 0.11 0.68 2.13 0.37 2.81 3 0.85 0.26 1.30 1.73 0.68 2.97

1 0.13 0.23 1.48 0.62 2.61 13 2 0.09 0.15 1.82 0.49 2.71 3 0.18 0.12 0.40 1.78 0.22 2.18

1 0.14 0.14 0.38 1.35 0.47 2.21 14 2 0.14 0.12 0.34 1.30 0.64 2.47 3 0.28 0.18 0.60 0.99 0.36 1.64

1 0.73 1.27 2.25 2.14 6.15 15 2 0.41 0.46 1.21 2.85 1.57 5.72 3 0.37 0.26 0.83 1.13 0.61 2.25 1 0.07 0.03 0.13 0.26 0.04 0.33 0.90 0.24 1.34 16 2 0.05 0.02 0.09 0.21 0.04 0.28 1.12 0.27 1.62 3 0.09 0.08 0.24 0.31 0.12 0.52 2.15 0.60 3.25

1 0.03 0.05 0.11 0.07 0.23 17 2 0.25 0.09 0.41 1.78 0.36 2.44 3 0.07 0.11 0.14 0.05 0.22

1 0.08 0.14 1.00 0.39 1.71 18 2 0.09 0.15 1.18 0.71 2.48 3 0.08 0.14 0.23 0.08 0.38

1 0.17 0.29 0.67 0.28 1.18 19 2 0.60 0.19 0.93 2.22 0.47 3.07 3 0.19 0.04 0.27 0.01 0.01 0.03 1 0.10 0.04 0.18 0.18 0.13 0.40 2.09 0.49 2.99 20 2 0.06 0.10 0.46 0.23 0.88 3 0.02 0.00 0.02 0.24 0.09 0.40 0.98 0.14 1.24 1 0.06 0.06 0.16 0.93 0.32 1.48 2.29 0.73 3.63 21 2 0.32 0.08 0.45 1.22 0.54 2.15 3.50 1.22 5.73 3 0.09 0.04 0.16 1.13 0.47 1.94 3.77 0.67 4.99

1 0.27 0.46 2.38 0.54 3.37 22 2 0.73 0.46 1.52 2.79 1.11 4.82 3 0.25 0.07 0.38 0.33 0.08 0.47

1 0.11 0.04 0.17 0.04 0.07 0.48 0.29 1.00 23 2 0.05 0.03 0.11 0.11 0.06 0.21 0.80 0.36 1.45 3 0.13 0.09 0.29 0.40 0.37 1.04 2.15 1.07 4.10

1 0.35 0.09 0.51 2.26 0.38 2.96 24 2 0.23 0.07 0.35 1.55 0.44 2.36 3 0.33 0.12 0.54 2.32 0.55 3.32 1 0.08 0.09 0.24 0.33 0.20 0.68 2.69 0.61 3.81 25 2 0.20 0.12 0.41 0.64 0.25 1.08 4.18 0.22 4.58 3 0.04 0.02 0.07 0.21 0.15 0.48 1.79 0.29 2.32

1 0.25 0.07 0.38 0.95 0.20 1.32 26 2 0.02 0.04 0.44 0.10 0.63 1.54 0.33 2.15 3 0.32 0.10 0.49 0.85 0.12 1.07

127

1 0.45 0.08 0.59 0.96 0.16 1.25 27 2 0.62 0.12 0.84 2.02 0.34 2.64 3 0.39 0.11 0.58 0.87 0.15 1.15

1 0.24 0.06 0.34 1.13 0.29 1.65 28 2 0.37 0.08 0.51 1.67 0.54 2.67 3 0.05 0.03 0.10 0.31 0.12 0.51 1.85 0.49 2.74

1 0.23 0.06 0.34 29 2 0.29 0.10 0.47 3 0.12 0.03 0.16

1 0.21 0.14 0.46 0.82 0.26 1.30 30 2 0.27 0.06 0.38 0.56 0.15 0.84 3 0.25 0.14 0.50 0.76 0.15 1.03

1 0.04 0.06 0.61 0.16 0.90 31 2 0.17 0.06 0.26 1.40 0.36 2.07 3 0.08 0.13 0.92 0.24 1.36

1 0.14 0.10 0.31 0.60 0.41 1.35 32 2 0.15 0.07 0.26 0.55 0.33 1.16 3 0.23 0.14 0.47 0.59 0.24 1.04

1 0.43 0.14 0.67 0.04 0.03 0.10 33 2 0.65 0.28 1.13 0.02 0.05 0.11 3 0.36 0.15 0.61 0.02 0.04

1 0.47 0.15 0.73 0.96 0.22 1.36 34 2 0.96 0.25 1.39 1.95 0.37 2.63 3 0.35 0.06 0.45 0.66 0.13 0.90

1 0.15 0.04 0.22 0.20 0.05 0.30 35 2 0.15 0.05 0.25 0.28 0.12 0.49 3 0.06 0.10 0.15 0.01 0.17

1 0.19 0.05 0.28 2.02 0.27 2.51 36 2 0.03 0.01 0.06 3 0.24 0.07 0.36 1.42 0.15 1.68 1 0.07 0.04 0.14 0.31 0.07 0.43 1.11 0.21 1.50 37 2 0.07 0.07 0.19 0.40 0.12 0.61 1.69 0.51 2.62 3 0.38 0.08 0.51 0.81 0.12 1.03

1 0.76 0.15 1.02 0.03 0.05 0.12 38 2 1.13 0.23 1.53 0.02 0.05 0.11 3 1.00 0.24 1.41 0.03 0.05 1 0.10 0.03 0.15 0.64 0.15 0.90 1.67 0.30 2.22 39 2 0.18 0.03 0.23 0.45 0.13 0.68 1.17 0.30 1.72 3 0.07 0.01 0.10 0.88 0.23 1.29 2.17 0.40 2.90

1 0.17 0.04 0.23 0.25 0.11 0.45 40 2 0.14 0.02 0.17 0.11 0.12 0.34 3 0.17 0.07 0.28 0.32 0.07 0.45

1 0.13 0.03 0.18 0.08 0.06 0.18 41 2 0.05 0.02 0.08 0.13 0.04 0.20 3 0.15 0.06 0.25 0.09 0.02 0.12

128

1 0.32 0.06 0.41 1.40 0.40 2.13 42 2 0.20 0.03 0.25 0.43 0.17 0.75 3 0.86 0.21 1.23 1.04 0.24 1.47

1 1.87 0.94 3.49 0.11 0.06 0.21 43 2 1.84 1.02 3.60 0.10 0.07 0.23 3 0.94 0.46 1.73 0.01 0.02

1 0.17 0.07 0.29 0.72 0.30 1.26 44 2 0.11 0.02 0.14 0.31 0.14 0.57 3 0.25 0.07 0.37 0.01 0.02

1 0.34 0.04 0.40 0.47 0.09 0.64 45 2 0.22 0.03 0.26 0.28 0.11 0.48 3 0.37 0.08 0.50 0.77 0.10 0.96

1 0.16 0.04 0.23 0.38 0.13 0.60 46 2 0.22 0.03 0.26 0.16 0.06 0.27 3 0.05 0.08 0.12 0.07 0.24

1 0.21 0.02 0.24 0.38 0.08 0.53 47 2 0.25 0.02 0.28 0.24 0.08 0.39 3 0.46 0.07 0.58 0.43 0.12 0.64

1 0.32 0.04 0.38 0.31 0.14 0.57 48 2 0.25 0.03 0.30 0.22 0.12 0.44 3 1.49 0.15 1.75 0.00 0.01

1 0.01 0.02 0.29 0.06 0.40 49 2 0.08 0.01 0.11 3 0.10 0.05 0.19 0.42 0.03 0.49 1 0.13 0.01 0.15 0.15 0.02 0.18 0.32 0.06 0.43 50 2 0.21 0.01 0.22 0.12 0.05 0.21 3 0.20 0.06 0.30 0.05 0.02 0.09 1 0.08 0.03 0.14 0.21 0.06 0.31 0.85 0.23 1.26 51 2 0.08 0.03 0.13 0.08 0.05 0.16 0.30 0.18 0.62 3 0.17 0.03 0.22 0.49 0.14 0.73 1.31 0.20 1.67 1 0.09 0.04 0.15 0.27 0.04 0.33 0.83 0.17 1.15 52 2 0.08 0.05 0.18 0.49 0.09 0.65 1.06 0.30 1.61 3 0.11 0.01 0.12 0.22 0.07 0.34 0.71 0.16 1.00

1 0.23 0.03 0.27 0.25 0.07 0.39 53 2 0.20 0.04 0.27 0.23 0.10 0.40 3 0.09 0.04 0.15 0.23 0.08 0.36 0.94 0.21 1.31

1 0.16 0.04 0.24 0.55 0.16 0.85 54 2 0.82 0.21 1.19 2.69 0.51 3.62 3 0.19 0.07 0.32 0.59 0.15 0.87

1 0.48 0.21 0.84 0.03 0.05 55 2 0.50 0.22 0.88 0.03 0.05 3 0.82 0.46 1.61 0.03 0.05

1 0.69 0.23 1.08 0.02 0.04 56 2 0.83 0.28 1.32 0.04 0.08 3 2.16 0.50 3.02 0.03 0.05

129

1 0.90 0.15 1.15 0.01 0.04 0.08 57 2 1.02 0.15 1.29 0.02 0.03 0.08 3 1.75 0.30 2.27 0.04 0.06 1 0.12 0.04 0.19 0.62 0.15 0.88 1.58 0.48 2.46 58 2 0.11 0.04 0.19 0.64 0.16 0.92 1.21 0.44 2.01 3 0.39 0.06 0.50 0.86 0.27 1.32 2.54 0.72 3.86

1 0.11 0.08 0.25 0.04 0.07 59 2 0.13 0.09 0.27 0.04 0.07 3 1.14 0.45 1.91 0.04 0.07 0.16 1 0.13 0.03 0.18 0.62 0.13 0.85 1.08 0.24 1.50 60 2 0.06 0.03 0.11 0.44 0.11 0.63 0.62 0.20 0.99 3 0.04 0.03 0.08 3.78 0.98 5.46 0.20 0.08 0.36

1 0.14 0.02 0.18 0.20 0.08 0.35 61 2 0.11 0.03 0.16 0.46 0.16 0.75 3 0.34 0.08 0.48 0.49 0.17 0.79

1 0.17 0.04 0.24 0.49 0.22 0.89 62 2 0.05 0.02 0.09 0.51 0.11 0.71 3 0.18 0.07 0.30 0.81 0.10 0.99

1 0.54 0.26 0.99 0.06 0.05 0.14 63 2 0.25 0.16 0.52 0.04 0.03 0.10 3 1.44 0.19 1.76 0.09 0.04 0.15

1 1.28 0.19 1.61 0.02 0.02 0.07 64 2 2.15 0.38 2.81 0.03 0.06 3 1.42 0.32 1.97 0.01 0.02 1 0.14 0.04 0.20 0.38 0.12 0.58 1.39 0.31 1.96 65 2 0.08 0.03 0.12 0.19 0.05 0.28 0.51 0.17 0.82 3 0.12 0.04 0.19 0.67 0.22 1.06 2.18 0.42 2.94 1 0.02 0.01 0.03 0.11 0.01 0.13 0.16 0.06 0.27 66 2 0.02 0.02 0.38 0.02 0.40 3 0.16 0.06 0.26 0.41 0.07 0.55 Σ 4.73 1.91 8.11 78.49 27.81 126.48 176.82 53.08 273.71 % 2% 31% 67%

The chart with the data above shows that PTE, CAU and PTA in all samples comprised 2%, 31% and 67% of the total amount of the compounds, respectively. The graphs below show their geographical distribution with the dash lines separating the countries (non-uniform y axes).

130

Denmark Sweden Finland

131

The spreadsheet below provides the results of the analysed contents of pterosins and glycosides in the samples of temporal variation study. All analyses were performed and the contents of glycosides prior to hydrolysis (Σ) were obtained based on [3] (empty values represent no detection; 0.00 represent ˂0.005). mg/g dry weight Σ Σ Σ PTE PtrG CAU PtrA PTA PtrB Location Month PTE CAU PTA Grib Skov 1 JUN 0.07 0.01 0.08 0.14 0.09 0.29 0.64 0.15 0.91

JUL 0.17 0.06 0.28 0.14 0.03 0.20

AUG 0.18 0.06 0.29 0.59 0.02 0.64

SEP 0.05 0.08 0.00 0.00

OCT 0.04 0.06 0.01 0.02 NOV 0.12 0.20 0.09 0.16 Grib Skov 2 JUN 0.19 0.09 0.34 0.26 0.14 0.52

JUL 0.11 0.05 0.20 0.15 0.02 0.18

AUG 0.08 0.05 0.17 0.00 0.00

SEP 0.04 0.07

OCT NOV 0.04 0.08 Grib Skov 3 JUN 0.08 0.08 0.26 0.14 0.51 1.13 0.27 1.63

JUL 0.18 0.06 0.27 0.75 0.07 0.87

AUG 0.04 0.04 0.80 0.10 0.96 1.76 0.16 2.05

SEP 0.02 0.02 0.27 0.07 0.38 0.82 0.19 1.16

OCT 0.08 0.14 0.59 0.05 0.68 NOV 0.22 0.37 0.48 0.15 0.75 Grib Skov 4 JUN 0.04 0.04 0.17 0.10 0.35 0.69 0.24 1.12

JUL 0.25 0.06 0.35 0.69 0.04 0.76

AUG 0.19 0.07 0.31 0.45 0.01 0.46

SEP

OCT 0.10 0.17 0.16 0.00 0.16 NOV 0.14 0.24 0.22 0.06 0.33 Grib Skov 5 JUN 0.05 0.05 0.24 0.07 0.36 1.07 0.09 1.22

JUL 0.13 0.07 0.25 0.38 0.03 0.44

AUG 0.20 0.07 0.32 1.19 0.06 1.30

SEP

OCT 0.06 0.03 0.10 0.15 0.00 0.16 NOV 0.10 0.07 0.21 0.20 0.02 0.24 Grib Skov 6 JUN 0.12 0.12 0.26 0.07 0.38 0.90 0.09 1.06

JUL 0.02 0.02 0.25 0.06 0.36 0.60 0.04 0.68

AUG 0.24 0.05 0.32 0.51 0.01 0.54

SEP 0.06 0.11 0.08 0.01 0.10

OCT 0.04 0.08 0.08 0.13 0.31 NOV 0.14 0.24 0.20 0.13 0.43 Grib Skov 7 JUN 0.13 0.01 0.14 0.29 0.11 0.48 2.03 0.26 2.50 JUL 0.09 0.00 0.10 0.21 0.07 0.33 1.42 0.09 1.59 AUG 0.08 0.01 0.10 0.32 0.11 0.52 1.41 0.46 2.26 SEP 0.04 0.00 0.05 0.14 0.05 0.22 0.74 0.02 0.77 OCT 0.00 0.01 0.01 0.14 0.05 0.23 0.61 0.03 0.66 NOV 0.00 0.02 0.04 0.00 0.10 0.18 0.43 0.10 0.60 132

mg/g dry weight Σ Σ Σ PTE PtrG CAU PtrA PTA PtrB Location Month PTE CAU PTA Humleore 1 JUN 0.05 0.05 0.49 0.14 0.73 1.73 0.33 2.34

JUL 0.02 0.02 0.20 0.06 0.29 0.41 0.02 0.45

AUG 0.28 0.06 0.38 0.89 0.04 0.96

SEP 0.17 0.05 0.26 0.29 0.01 0.31

OCT 0.15 0.06 0.26 0.17 0.01 0.20 NOV 0.07 0.13 0.03 0.01 0.05 Humleore 2 JUN 0.01 0.01 0.13 0.07 0.25 0.31 0.16 0.60

JUL 0.04 0.04 0.21 0.05 0.29 0.90 0.06 1.01

AUG 0.15 0.05 0.23 0.52 0.02 0.57

SEP 0.11 0.05 0.19 0.05 0.04 0.13

OCT 0.10 0.04 0.17 0.03 0.01 0.04 NOV 0.13 0.06 0.24 0.05 0.02 0.09 Humleore 3 JUN 0.01 0.01 0.10 0.07 0.22 0.32 0.23 0.73 JUL 0.03 0.00 0.03 0.15 0.06 0.24 0.46 0.08 0.61

AUG 0.02 0.02 0.12 0.05 0.21 0.45 0.01 0.47

SEP 0.04 0.07 0.04 0.01 0.06

OCT 0.04 0.07 0.12 0.02 0.16 NOV 0.06 0.11 0.04 0.07 Humleore 4 JUN 0.11 0.03 0.16 0.29 0.10 0.47 1.15 0.24 1.59

JUL 0.05 0.05 0.17 0.07 0.29 0.76 0.09 0.93

AUG 0.12 0.05 0.20 0.49 0.02 0.53

SEP 0.04 0.07 0.11 0.02 0.14

OCT 0.08 0.14 0.26 0.05 0.35 NOV 0.14 0.23 0.10 0.19 Humleore 5 JUN 0.10 0.10 0.19 0.08 0.33 0.99 0.16 1.28

JUL 0.02 0.02 0.26 0.06 0.37 0.47 0.04 0.54

AUG 0.01 0.01 0.20 0.06 0.29 0.36 0.05 0.46

SEP 0.01 0.01 0.09 0.04 0.16 0.25 0.01 0.27

OCT 0.11 0.05 0.20 NOV 0.10 0.17 0.04 0.07 Humleore 6 JUN 0.16 0.01 0.18 0.35 0.22 0.73 2.66 0.85 4.21

JUL 0.35 0.09 0.50 1.27 0.19 1.61

AUG 0.40 0.08 0.53 1.54 0.11 1.74

SEP 0.21 0.05 0.29 0.84 0.06 0.95

OCT 0.20 0.06 0.30 0.38 0.01 0.40 NOV 0.22 0.08 0.35 0.75 0.06 0.85 Humleore 7 JUN 0.08 0.05 0.17 0.08 0.03 0.14

JUL 0.09 0.05 0.17 0.23 0.01 0.25

AUG 0.19 0.06 0.30 0.64 0.02 0.68

SEP 0.00 0.05 0.08 0.07 0.01 0.08

OCT 0.00 0.05 0.09 0.09 0.00 0.10 NOV 0.00 0.07 0.13 0.00 0.05 0.10

133

Statistical analysis

Spatial variation

PTE content

Each site was investigated on 2 upper and 1 lower parts of the fronds. The cross tabulation of the raw data (Table 3) shows that except for 8 sites either all of the three samples contained PTE (13 sites) or no samples contained PTE (45 sites).

Table 3. Cross tabulation of raw data of presence of PTE

Locations with 0 1 2 PTE upper upper upper 0 lower 45 4 1 1 lower 1 2 13

Since the main variation was between the sites, it has been chosen to compare the 21 sites with PTE in at least one sample against the 45 sites without PTE. The associated logistic regression selected only the estimated odds ratio of PTE occurrence in plants. Statistics of the odds ratio of PTE occurrence in plants was not further investigated, because PTE was detected only in very little contents (close to LOD), and there was not possible to differentiate with certainty between the absence of PTE in the plants and the presence of PTE in the plants

CAU content

The cross tabulation of the raw data (Table 5) shows that 63 of the 66 sites contained CAU in both upper samples and only 2 sites contained no CAU in any of the three samples. Hence the logistic regression for containing CAU in at least one of the three samples did not select any variables.

Table 5. Cross tabulation of raw data for presence of CAU

Leaves with PTE 0 upper 1 upper 2 upper 0 lower 2 1 2 1 lower 0 0 61

The backward model reduction selected PC1 (p=0.0446) in the normal regression for ln(CAU) for the 188 samples with strictly positive CAU. The estimated slope against PC1 was 0.1450 (95% CI: [0.0036 ; 0.2865]). This implies that when moving 1 standard deviation in the north-east direction from Denmark to Finland the estimated relative increase of the population content of CAU equals:

(exp(0.1450) − 1) ∗ 100% = 15.6%

134

The variation for ln(CAU) between the sites was estimated at 0.5285. and the variation between the samples within the sites was estimated at 0.3316. This implies that 61 percent of the total variation not explained by the geographical position is due to variation between the sites.

PTA content

Only 3 samples did not contain observable content of PTA. therefore it was not reasonable to perform the logistic regression. The normal regression for ln(PTA) for the 195 samples with PTA selected country (p=0.0002). The estimated ratios of PTA content between Denmark, Sweden and Finland are provided in Table 6.

Table 6. The estimated ratios of PTA content between countries

95% confidence p-value for Ratio Estimate interval ratio=1 Denmark/Sweden 3.12 [1.41 ; 6.92] 0.0031 Denmark/Finland 3.60 [1.53 ; 8.44] 0.0018 Sweden/Finland 1.15 [0.46 ; 2.91] 0.9279

The PTA contents in the samples were not significantly different in Sweden and Finland, and more than 3 times as large in Denmark. The variation for ln(PTA) between the sites was estimated at 1.1013, and the variation between the samples within the sites was estimated at 0.6496. This implies that 63 percent of the total variation not explained by country was due to variation between the sites.

Relative distribution of PTE, CAU and PTA

PTA was used as the reference for the composition analysis. As the reference values should be strictly positive, in the compositional analysis the few zero measurements for PTA (3 out of 198) were changed to half of the minimal value of the non-zero measurements (0.00313/2=0.001565). The backward model reduction selected country (p=0.0002) and part (p=0.0096) for the relative distribution of PTE, CAU and PTA. The corresponding ternary plots of the observed compositions is visualized in Figure 3.

135

Figure 3. Relative composition of PTE, CAU and PTA in the 198 measured samples separated by the selected effects. DK, SE and FI represent Denmark, Sweden and Finland, respectively; lower and upper – lower and upper parts of the fronds.

The marginal analyses of PTE, CAU and PTA showed that the random effect of site is important as there is a large variation between samples that cannot be explained by the fixed effects. Thus, as the Dirichlet-package in R did not allow for inclusion of a random effect of site, care should be taken when interpreting parameter estimates in the used Dirichlet regression.

Temporal variation of PTE, CAU and PTA

The temporal variation of the illudane glycosides is displayed in Boxplot on a logarithmic scale of the measurements of strictly positive PTE, CAU and PTA. The numbers below the boxplots designate the number of measurements (out of 14 samples) with zero content for the three compounds.

Figure 4. the temporal variation of PTE, CAU and PTA 136

PTE content

PTE was detected only in very little contents (close to LOD). Statistics of temporal PTE variation was not further investigated because it was not possible to differentiate with certainty between the absence of PTE in the plants and the presence of PTE in the plants

CAU content

Only 3 measurements did not contain observable CAU content, therefore it was not reasonable to perform the logistic regression. The backward model reduction selected month (p<0.0001) in the normal regression for ln(CAU) for the 81 observations with strictly positive CAU. The estimated content of CAU is given in

Table 8, where months not sharing a group letter are significantly different on the 5 percent significance level.

Table 8. The estimated content of CAU

95% confidence Group Month Estimate interval letter JUN 0.370 [0.288 ; 0.476] a JUL 0.290 [0.225 ; 0.373] a AUG 0.322 [0.250 ; 0.414] a SEP 0.138 [0.106 ; 0.180] b. c OCT 0.132 [0.102 ; 0.171] c NOV 0.189 [0.147 ; 0.242] b

The variation for ln(CAU) between the samples was estimated at 0.1179, and the variation between observations within the samples was estimated at 0.0952. This implies that 55 percent of the total variation not explained by month was due to the variation between the samples.

PTA content

Only 6 measurements did not contain observable PTA therefore it did not make sense to perform the logistic regression. The backward model reduction selected month (p<0.0001) in the normal regression for ln(PTA) for the 78 observations with strictly positive PTA. The estimated content of PTA is given in Table 9. where months not sharing a group letter are significantly different on the 5 percent significance level.

137

Table 9. The estimated content of PTA

95% confidence Group Month Estimate interval letter JUN 1.092 [0.547 ; 2.180] a JUL 0.593 [0.297 ; 1.183] a AUG 0.546 [0.274 ; 1.090] a SEP 0.129 [0.061 ; 0.270] b OCT 0.152 [0.074 ; 0.312] b NOV 0.176 [0.087 ; 0.357] b

The variation for ln(PTA) between the samples was estimated at 0.8111, and the variation between observations within samples was estimated at 0.8069. This implies that 50 percent of the total variation not explained by month is due to variation between samples.

138

Main conclusions

The study concludes that among the three investigated compounds in the Northern European Pteridium the most commonly encountered was PTA followed by CAU. Analysis of the spatial variation revealed that among all 198 plant samples only 18.7% contained PTE, 98.0% contained CAU and 98.5% contained PTA. Regarding the total amount of the compounds in all samples, PTE comprised 2.0%, CAU: 31.0% and PTA: 67.0% of the three compounds. No statistically significant difference of the contents of compounds between the subspecies was demonstrated.

The study investigated environmental and geographical factors determining the contents of these compounds in the plants. No dependency of the contents of PTA and CAU were observed based on fixed environmental factors. However, It has been demonstrated that the contents of PTA in the plants were on average 3.12 times higher in Denmark than in Sweden (p=0.0031) and 3.6 times higher in Denmark than in Finland (p=0.0018). The contents between the plants sampled in Sweden and Finland did not vary (p=0.9279).

Even though the PTA content was significantly higher in Denmark than in the other two countries (this can be also observed without the statistical analysis), no statistically significant continuous gradual geographical change of the PTA content was demonstrated. Conversely, the content of CAU increased gradually towards the north-east direction. By moving 1 standard deviation in the north- east from Denmark to Finland the estimated relative increase of the CAU content was equal to 15.6% (p=0.0446). PTE was detected only rarely and in very low contents (up to 0.5 mg/g in dry weight) and its contents were not linked to any fixed environmental factors.

Acknowledgements

The authors are grateful to the forest ranger Bjarne Persson of Grib Skov and a forest owner Ulrich of Humleore for provided permissions for sampling activities. The authors would also like to thank the laboratory technician students of University College Copenhagen Jonas Halgren Mikkelsen and Michael Josefsen for assistance in processing samples, and a laboratory technician Jimena Bertorello Martinez for general technical support.

Funding

This project was supported by the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 722493.

139

References used in the technical report Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium

[1] Mossberg B.. Stenberg L.. Den nye nordiske flora. 2nd edition. Copenhagen: Gyldendal A/S. 2014. [2] Rasmussen L.H.. Pedersen H.A.. "Screening for Ptaquiloside in Ferns: Using Herbarium Specimens for Qualitative Mapping Purposes." Phytochem Anal. vol. 28. no. 6. pp. 575-583. 2017. [3] Kisielius V.. Lindqvist D.N.. Thygesen M.B.. Rodamer M.. Hansen H.C.B.. Rasmussen L.H.. "Fast LC-MS quantification of ptesculentoside. caudatoside. ptaquiloside and corresponding pterosins in bracken ferns." J. Chromatogr. B. vol. doi: 10.1016/j.jchromb.2019.121966. Epub 2020 Jan 3. [4] Thiele J.. Markussen B.. "Potential of GLMM in modelling invasive spread." CAB Reviews. vol. 7. no. 16. pp. 1-10. 2012. [5] Maier. M.J.. "DirichletReg: Dirichlet Regression for Compositional Data in R." Department of Statistics and Mathematics. Vienna University of Economics and Business. Vienna. Research Report Series/. http://epub.wu.ac.at/4077/).

140

Supplementary material

Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium

The spreadsheet below provides recorded information used for statistical analysis about the sampling locations.

sub-species Sample (Aquilinum/ direct No latitude longitude Country Latiusculum) sunshine Tree type 1 Tree type 2 1 5580678 1247600 DK L - deciduous Dark 2 5592783 1234852 DK A + deciduous Dark 3 5603362 1256166 DK A + deciduous Light 4 5601947 1231138 DK A - deciduous Light 5 5598664 1208600 DK A + deciduous Dark 6 5591060 1198022 DK A - deciduous Light, Dark 7 5582563 1223877 DK A - deciduous Dark 8 5534658 1178511 DK A - coniferous Dark 9 5505719 1174527 DK A - coniferous Dark 10 5504385 1226230 DK A - deciduous Dark 11 5514045 1202613 DK A - deciduous Dark 12 5547483 1190649 DK A - deciduous Dark 13 5543796 1152479 DK A - deciduous Dark 14 5562964 1153478 DK A - deciduous Light 15 5530091 1082634 DK A + deciduous, coniferous Light, Dark 16 5553400 1052932 DK A + deciduous, coniferous Light, Dark 17 5513785 1034402 DK A + coniferous Dark 18 5551496 969795 DK A - deciduous Light, Dark 19 5612474 990716 DK A + coniferous Dark 20 5642987 1051195 DK A - deciduous Dark 21 5657370 965237 DK A - deciduous Dark 22 5678236 975107 DK A - deciduous Dark 23 5726235 1019518 DK A - coniferous Dark 24 5672611 953380 DK A - deciduous Light 25 5654088 850665 DK A - deciduous Light, Dark 26 5625477 851052 DK A - deciduous Light 27 5554447 863914 DK A - deciduous Light 28 5514371 1471485 DK A - deciduous Light 29 5526328 1476237 DK A - deciduous Light 30 5510014 1510326 DK A + deciduous Dark 31 5553219 1326748 SE A - deciduous Dark 32 5566968 1403258 SE A - deciduous Dark 33 5565715 1426418 SE A - deciduous Dark 141

34 5558880 1372492 SE A - deciduous Dark 35 5588317 1366398 SE A - deciduous Dark 36 5595572 1325764 SE A + deciduous, coniferous Dark 37 5615649 1334669 SE A + coniferous Dark 38 5651645 1413639 SE A + - - 39 5738137 1572451 SE A - deciduous, coniferous Dark 40 5778995 1576087 SE A + deciduous, coniferous Light, Dark 41 5853731 1603987 SE A + coniferous Dark 42 5876834 1699811 SE L - deciduous, coniferous Light, Dark 43 5943621 1807476 SE A - deciduous, coniferous Light, Dark 44 5972428 1903942 SE A + - - 45 5963529 1663297 SE L - deciduous, coniferous Light, Dark 46 5931375 1529605 SE A + - - 47 5867140 1460341 SE L - deciduous Light 48 5834846 1432337 SE L - coniferous Dark 49 5760570 1393799 SE L - coniferous Dark 50 5685314 1302913 SE L - deciduous, coniferous Light, Dark 51 6024749 1948384 FI L - deciduous, coniferous Light, Dark 52 6007292 2014584 FI A - deciduous, coniferous Light 53 6036900 1991617 FI A - deciduous, coniferous Light, Dark 54 6063566 2203978 FI A + coniferous Dark 55 6123323 2270801 FI A - deciduous, coniferous Light, Dark 56 6191875 2533720 FI L - deciduous Light 57 6254933 2697207 FI A - coniferous Dark 58 6319842 2825248 FI A - deciduous, coniferous Light, Dark 59 6275316 3100000 FI A - deciduous Light 60 6256208 3057539 FI A - deciduous, coniferous Light, Dark 61 6242426 3038000 FI A - deciduous Light 62 6195601 2845833 FI A - coniferous Dark 63 6173810 2752081 FI A - coniferous Dark 64 6133358 2616437 FI L - deciduous, coniferous Light 65 6106630 2580758 FI A - deciduous, coniferous Light, Dark 66 6071496 2447065 FI A - deciduous, coniferous Light, Dark

142

3. Occurrence and stability of ptesculentoside, caudatoside and ptaquiloside from bracken ferns in surface waters

Manuscript

Vaidotas Kisielius13,14, Mikkel Drejer1 Natasa Skrbic1,15, Jimmy Kjellerup Dornhoff1, Dan Nybro Lindqvist1, Hans Christian Bruun Hansen1, Lars Holm Rasmussen1

Abstract

The natural carcinogen ptaquiloside (PTA) is a common contaminant in water bodies adjacent to bracken ferns (genus Pteridium). Besides PTA the genus produces structurally nearly identical and seemingly carcinogenic compounds: ptesculentoside (PTE) and caudatoside (CAU). This study was conducted to examine whether PTE and CAU also occur in water bodies and to investigate their aquatic stability. Surface waters adjacent to Pteridium were sampled in Denmark, Sweden and Spain, and analyzed for PTE, CAU, PTA and their corresponding hydrolysis products – pterosins G, A and B. The compounds were detected in 11 out of 14 water bodies at concentrations up to 21.2, 13.7 and 3.4 ng/l of PTA, pterosins G and A respectively in streams and lakes, and up to 5300, 6.5, 2280 and 7.3 ng/l of PTE, CAU, pterosins G and A, respectively in drainage waters. The stability of PTE, CAU and PTA was tested in laboratory incubation of lake waters as a function of pH, temperature and microbial activity. The compounds degraded in a first-order reaction: fast under slightly basic pH and at relatively high temperature (half-lives of ~2 days at pH 7.4 and 15 ˚C) and significantly slower under moderately acidic to neutral pH and at low temperature (half-lives of ~12 days at pH 5.2-6.5, T 5 ˚C). The PTE, CAU and PTA were similarly stable under all conditions. This study presents the first report of natural occurrence of PTE and CAU in water bodies. The results imply that the previous assessments of water toxicity adjacent to bracken based solely on PTA were likely underestimated as PTE and CAU were not included.

Keywords:

Water quality, pteridium, phytotoxins, carcinogens, pterosins, solid-phase extraction

13 Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark

14 Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark

15 Greater Copenhagen Utility HOFOR, Ørestads Blvd. 35, 2300 Copenhagen, Denmark

143

Introduction

Bracken (generic name Pteridium, family Dennstaedtiaceae) is the fifth most abundant plant genus in the world [1]. Its carcinogenicity is usually related to the illudane glycoside ptaquiloside (PTA) [2] [3–5]. An electrophilic cyclopropyl group in an active form of ptaquiloside (conjugated dienone, Figure 1) readily reacts with nucleophiles such as water, alkaloids, amines, and DNA [6,7]. By alkylating DNA it can cause mutations [4,8–11]. Many studies link bracken-containing food with human upper gastrointestinal tract cancers [12–14]. Even if the plant is not consumed, longer term residence in areas with dense bracken growth has been reported to correlate with increased risk of dying from gastric cancer [9,15]. The risk is assumed to be caused due to PTA found dairy products of domestic farm animals that occasionally graze the plant [15,16].

The increased human carcinogenicity in bracken infested areas may also be related to plant leachates seeping to drinking water reservoirs [17–19]. PTA is a very polar compound easily released from the surface of plants by precipitation and is almost not absorbed by sediments and soils [20,21]. As a result, PTA has been detected in shallow groundwater nearby bracken stands in Denmark (up to 90 ng/l) [17], in surface drinking water abstraction sites in Ireland (up to 670 ng/l) [18] and in a stream feeding an active drinking water reservoir in the United Kingdom (up to 2200 ng/l) [19]. Other studies demonstrated that water containing PTA caused in vitro DNA damage of human gastric epithelial cells and in vivo gastric carcinogenesis of mice [4,22].

Several compounds with chemical structures nearly identical to PTA have been elucidated in Dennstaedtiaceae plant family. Among them the majority in Pteridium is believed to comprise ptesculentoside (PTE) [23] and caudatoside (CAU) [24] (hereafter called PTA analogues (Figure 1) [25]. The PTA analogues contain the cyclopropyl group that is ultimately responsible for DNA alkylation and genetic mutations caused by PTA. A recent study demonstrated that cattle grazing ferns that contain no PTA (Adiantopsis chlorophylla (Swartz) Fée (Pteridaceae)) developed mutagenic symptoms similar to those normally caused by PTA, due to CAU [26].

Reports quantifying PTA analogues in plants are rare, however revealing significant concentrations of PTE and CAU in comparison to PTA [25,27,28]. In various Pteridium species distributed over 6 continents of the world, the relative average contents of the three illudane glycosides were 22%, 37% and 41% of PTE, CAU and PTA respectively (0.009 – 5.3 mg/g of the total Illudane glycosides in dry weight of the plants) [27]. A systematic sampling of Pteridium aquilinum and Pteridium aquilinum var. latiusculum (Desv.) across a 1500 km geographic gradient in northern Europe demonstrated average contents of 2%, 31% and 67% for PTE, CAU and PTA, respectively [28]. Other studies demonstrated comparable relative average compositions with an exception of PTE comprising 40-73% of the illudane glycosides in Australian Pteridium esculentum [25,27].

Studies of PTA stability reveal that the compound hydrolyzes in water into the less toxic and non- carcinogenic product pterosin B (Figure 1). Rapid first-order hydrolysis of PTA take place in both basic and strongly acidic conditions. However, the compound is rather stable under neutral to

144

moderately acidic conditions (pH 4-7) and at low temperatures [6,8,20,29–31]. In aqueous buffered solutions of pH 2.9 to 8.9 at 22 ˚C its half-lives (t1/2) range from seconds to hours at pH > 7 and from hours to days at pH < 4 [29]. Assuming that PTA degradation occurs due to hydrolysis only, its half- life at pH 4.5 at 8 ˚C was estimated to be 3.7 months [29]. At pH 2, 9, 10 and 11 (25 ˚C) the half-lives of PTA were only 174, 40, 6 and 1 min, respectively [6]. One study reported a 7 day half-life at pH 4.0 at 37 ˚C [30]. Sensitivity of PTA to higher than room temperatures is also known from observed decrease of the compound content in oven-dried plants [32,33]. The kinetic properties imply that PTA leaching and endurance in water shall be facilitated by slightly acidic soils and waters, and cold and rainy weather.

The load of the total PTA content in mature bracken stands has been estimated to 0.1 – 5.9 kg of PTA per ha [21,34]. The abovementioned results of PTA analogues imply that the loads of the total reactive illudane glycosides can be nearly double this load. Even though PTA has been demonstrated to leach from bracken to water, to the best of our knowledge PTA analogues have not been monitored in water before. Despite the slightly lower overall average contents of PTE and CAU than PTA, the former compounds may have more implications to water quality due to their higher mobility than PTA [25,27].

We hypothesize that PTA analogues are overlooked water contaminants. To test the hypothesis we sampled surface waters from locations adjacent to Pteridium aquilinum in Denmark, Sweden and Spain. The waters were analyzed for concentrations of PTE, CAU, PTA. The aqueous hydrolysis products of these compounds – pterosins G, A and B were monitored as well as they report on former presence of the illudane glycosides [27]. Moreover, even though the pterosins are not carcinogenic, their biological activity indicate their potential adverse health effects [35–38]. To improve LODs of the applied analytical method, a 100 fold sample pre-concentration technique was developed. A full factorial experimental design was applied to determine the reaction rates of the glycosides in natural surface waters from different origins and different laboratory-controlled values of pH, temperatures, and presence/absence of microbial activity.

145

Figure 1. Hydrolysis of PTA in aqueous solution (A) and the structures of PTA analogues and corresponding pterosins (B). The electrophilic cyclopropyl groups are encircled.

Materials and methods

Solvents and chemicals

Acids, bases and buffers (formic, hydrochloric, acetic and trifluoroacetic acids, sodium hydroxide, ammonium acetate) were all of analytical grade from Sigma-Aldrich (Denmark). Polyamide for column chromatography was from Fluka Analytical, Sigma-Aldrich Co (Germany). LC-MS grade acetonitrile was from Merck Millipore (Germany), LC-MS grade methanol from Honeywell (Germany). LC-MS water (electrical resistivity 18.2 MΩcm, TOC < 2000 ng/L) was prepared by Satorius Arium Pro Ultrapure dispenser (Germany).

Analytical method

Reference aqueous standards of PTE, CAU, PTA and the corresponding pterosins from plant material were prepared from bracken ferns and the compounds analyzed using an Agilent 1260 Infinity HPLC System equipped with Agilent 6130 Single Quadrupole mass spectrometer by the method described in Kisielius et al. [27]. The method does not detect PTA analogues other than PTE and CAU, but supposedly quantifies (though not differentiates) epimers of the targeted compounds [24]). The 146

instrumental limits of detection and quantification of illudane glycosides were 80–260 ng/l and 250- 780 ng/l, respectively, and 10–30 ng/l and 30–90 ng/l for the pterosins [27]. To make the analytical method applicable for low concentrations in water it was supplemented with a 100 fold pre- concentration by solid-phase extraction (SPE).

Solid-phase extraction

The SPE cartridges OASIS MAX 20 cc Vac RC 60 mg (Waters co, Ireland) were successively conditioned by 2 ml of MeOH and 2 ml of DI water, loaded with 20 ml of the water sample, washed with 2ml DI water and run dry for 20 s. The compounds were eluted into 2 ml volume analytical HPLC autosampler vials in two stages: first with 0.75ml and 0.5 ml MeOH, allowing the cartridges to run dry for 20 s after each stage. Approximate speed of conditioning, loading, washing and eluting was 3 ml/min. The contents of the vials were dried with a gentle air flow in 30 °C heating block (Mikrolab Aarhus Supertherm) and re-dissolved with 0.2 ml mixture of 40 % (v/v) MeOH and 0.1 M ammonium acetate adjusted to pH 5.0 with glacial acetic acid.

The efficiency of the combined SPE and evaporation procedure was determined for 20 ml of DI water samples spiked with 0.2 ml of 0.1 ng/ml of the compounds. Water containing each single compound was processed in separate SPE cartridges in duplicates and analyzed by LC-MS. The resulting rates of the compound recoveries are provided in Table 1. Recovery of the most polar compound PTE can be increased by lowering the volumes of loading and/or washing, whereas recovery of the least polar PtrB – by increasing the volume of elution.

Table 1. Recovery of the compounds resulting from 100 times pre-concentration of aqueous samples by SPE and evaporation. Recoveries are listed according to the elution time from the most to the least polar compound [27]. The SDs are based on duplicate independently performed measurements.

Compound Recovery (%) SD PTE 74.2 3.3 CAU 91.2 3.8 PtrG 91.8 5.1 PTA 99.2 1.8 PtrA 96.2 1.4 PtrB 71.5 1.9

Mass-spectrometric data

As part of this study mass spectral data for PTE, CAU, PTA and the corresponding pterosins have been included in European MassBank database allowing free data access, exchange and share [39].

147

Field sample collection and analysis

Samples of surface waters (lakes, rivers, creeks, ditches) for analysis of PTE, CAU, PTA and the corresponding pterosins were taken during the bracken growing period by drawing water from 1 cm depth at approx. 0.5 m distance from the shore into a 50 ml sterile syringe followed by filtering 40 ml through a 0.45 µm pore size sterile cellulose acetate filter (Q-Max Frisenette Syringe Filters) into sterile conical 50 ml polypropylene centrifuge tube (Thermo Fisher Scientific, Korea). The samples were buffered by adding 1 ml of 0.3 M ammonium acetate adjusted with glacial acetic acid to pH 5 optimal for stabilizing PTA against hydrolysis [19]. All samples were taken in triplicates at one or more sampling events from 14 surface water bodies adjacent to populations of Pteridium: 4 in Denmark, 2 in Sweden and 8 in Spain (Table 3). The filtered and buffered samples were immediately put on ice and within 8 hours frozen in -20 ˚C. The samples from Spain were packed with dry ice and delivered by express mail for analysis in Denmark.

The concentrations of the compounds are reported as the average concentrations measured in triplicate samples. If a compound was quantified in only 2 of the triplicate samples (close to the approximate limit of quantification of the full method), the concentration was reported as the average of the 2 positive samples. If a compound was quantified in only 1 of the triplicate samples, the concentration was not quantified and reported as trace.

Stability of illudane glycosides

Sampling of natural waters

The spiking experiment for determination of compound stability required waters containing none of the glycosides and pterosins investigated. Therefore, two lakes were selected in areas with no adjacent Pteridium in Eastern Denmark: a peat-brown acidic water of Bøllemosen (WGS84: 55.8269, 12.5639, hereafter called lake A) and a clear weakly alkaline water of Skodsborg Dam (WGS84: 55.8177, 12.5659, hereafter called lake B). The waters were sampled in 10 December, 2019 at approx. 2 m distance from the shore in approx. 50 cm depth into 1 L polypropylene bottles. Prior to use, the bottles were washed with 10 % HCl, rinsed 5 times with DI water and 1 time with the water of the same lake. An estimate of the colony forming units (CFU) [40] determined 64 CFUs in 1 ml of A and an indefinite number between 121 and 340 in 1 ml of B, indicating that microorganisms in both waters were present, though largely dormant. The physicochemical parameters of water A and B were: pH 5.2 and 7.4 (VWR pHenomenal MU 6100 H), turbidity 2.6 and 0.3 FNU (HACH TU5200 laser meter), non-purgeable organic carbon (NPOC) 38.1 and 13.3 mg/L (Shimadzu TOC-VCPN), electric conductivity 65 and 210 µS/cm (VWR pHenomenal MU 6100 H, electrode VWR CO 11), respectively.

148

Experimental design

The stability of illudane glycosides in waters was tested on 4 selected physicochemical and microbiological water parameters in a full factorial experimental design. The design comprised 2 discrete values of each parameter assuming all possible combinations, allowing study of the effects of interactions between the parameters (Table 2). In total 16 (24) possible combinations of the parameters were tested (Figure 2)

Table 2. Values of parameters tested in a full factorial experimental design to determine the compound degradation rates

Parameter Values Water origin A / B Temperature 5°C / 15°C pH controlled pH 6.5 / original pH Sterilization sterilized / not sterilized

The waters were constantly kept at 5 °C and 15 °C in ventilated incubators (Sanyo MIR-253). The pH values of the waters were kept either at natural pH or adjusted to approximately pH 6.5 using 1 M HCl or 1 M NaOH. The waters were kept as sampled or sterilized by filtering through a 0.2µm pore size sterile cellulose acetate filters (Q-Max Frisenette Syringe Filters). The waters were distributed in 95 ml portions into 100 ml loosely closed Blue Cap glass bottles and kept for a 6 day equilibration period prior to spiking with PTA, CAU and PTE. The pH values were subsequently measured at 23, 72 and 504 hours after spiking the compounds (MeterLab PHM220).

Spiking with PTA, CAU and PTE

A fine powder of dried Pteridium esculentum leaves from Bribie Island, Australia, containing 2400, 220 and 800 µg/g of PTE, CAU and PTA respectively was kindly provided by Prof Mary T. Fletcher (The University of Queensland) [27]. An aqueous extract containing the three illudane glycosides was prepared from 3.2 g of the powder and 160 ml of DI water (in 4 conical 50 ml polypropylene centrifuge tubes Thermo Fisher Scientific, Korea), shaking for 20 min at 75 rpm (ELMI Intelli-Mixer RM-2L), centrifuging for 20 min at 2500 g at 1 °C (Sigma Zentrifugen 4 K15) and filtering the supernatant through a 2 μm pore size filter paper (Whatman No. 41, Sigma-Aldrich, Germany) under vacuum [27].

The extract was cleaned from pterosins and hydrophobic impurities by passing it through polyamide packed in two cylindrical glass funnels (ø = 1.5 cm; l = 20 cm, BIO-RAD, Econo-Column, USA) (approx. 8 g of polyamide per funnel) at an approximate speed of 3 ml/min and adding extra 8 ml of DI water to each funnel to fully elute the illudane glycosides [34]. The collecting flasks were kept on ice to prevent thermal degradation. The Blue Cap bottles with waters representing each test (Table 2)

149

were spiked with 5 ml of the obtained aqueous plant extract and vigorously shaken resulting in initial concentration in each bottle of 2.40, 0.22 and 0.80 mg/l of PTE, CAU and PTA, respectively.

Monitoring of illudane glycoside degradation

The concentration of the analytes were monitored in the flasks by sampling 0.2 ml water immediately after spiking (initial concentrations) and after 0.75, 2, 4, 6, 23, 72, 96 and 504 h. The samples were transferred to HPLC autosampler vials, and stabilized against microbial degradation and hydrolysis by adding 0.2 ml mixture of 40 % (v/v) MeOH and 0.1 M ammonium acetate adjusted to pH 5 with glacial acetic acid. The samples were filtered through 0.2 μm pore size membranes integrated in the vial caps (Syringeless filter device Mini-Uniprep, GE Healthcare Life Sciences, UK) and stored in -20 °C prior to analysis. All samples were taken in duplicates. The bottles were manually shaken before each sampling but kept loosely closed in the incubators to maintain aerobic conditions.

Determination of compound degradation rates

The concentrations of PTE, CAU and PTA in samples at time t are expressed in percentages of their initial concentrations. The compound degradation rates were calculated by Equation 1. 푑(퐴) = −푘(푡) 푑(푡)

Equation 1. “A” is the average concentration of the analyte at time “t” in duplicate samples. “–k(t)” is compound degradation rate constant.

The average compound degradation rates were determined in tests that shared the same value of one parameter (e.g. 8 tests performed at 5 °C). The average compound degradation rates in the tests representing the opposite values of the same parameter (e.g. between the 8 tests in 5 °C and 8 tests in 15 °C) were compared in pairwise t-tests with 95% confidence intervals (Microsoft Excel 2016). Comparison of the average degradation rates of the three compounds throughout the tests was made setting the null-hypothesis that the average rates were equal and testing it in a One- Factor Analysis of Variance (statistical model ANOVA, Microsoft Excel 2016).

Results

Water analysis

In 11 out of the 14 monitored water bodies at least one illudane glycoside or pterosin was detected. The detected concentrations are listed in Table 3.

150

Table 3. Concentrations of illudane glycosides and corresponding pterosins in natural waters. The empty cells represent no detection (˂LOD). Different values of the same compound in one site represent quantifications at different sampling events. Figures in brackets represent the number of sampling events when the trace concentrations were measured. Distance Number of (number of sampling events with positive detection): detected range ng/l Site Location Water Country WGS84 coordinates from sampling ID name type PTE PtrG CAU PtrA PTA PtrB bracken (m) events 1 Humleore Denmark Stream 55.47542, 11.90639 30 9 (1): trace (5): trace – 3.4 (1): 5.9 2 Humleore Denmark Pond 55.47487, 11.90836 10 9 (1): 13.7 (1): trace (1): 21.2 3 Gribskov Denmark Lake 55.99644, 12.3516 5 2 (2): trace – 2.1 4 Helsingor Denmark Lake 56.03350, 12.56145 5 2 5 Asljunga Sweden Stream 56.3063, 13.37894 100 2 (1): trace 6 Asljunga Sweden Lake 56.30315, 13.3743 25 2 7 43.22904, -5.65816 ~2 2 (1): 12.6 (1): 6.5 (2): 1.3 – 2.5 8 43.22994, -5.67083 1 (1): 5300 (1): 1230 9 43.22915, -5.6754 1 (1): 26.8 10 Drainage 43.2272, -5.65708 1 (1): 3.3 (1): 1.4 Asturias Spain 11 water* 43.22934, -5.67141 1 (1): 399 (1): 1190 12 43.22749, -5.67983 1 (1): 403 (1): 2280 (1): 7.3 13 43.22739, -5.68088 1 14 43.23083, -5.67636 1 (1): 399 (1): 1000 *Seepage water in drainage ditches and pits, accessible for drinking to horses and beef cattle

In general, the frequency of occurrence of PTE and CAU was not lower than the frequency of PTA detection. The concentrations in samples from Denmark and Sweden were relatively low and did not correlate to the distance to the bracken populations. The high concentrations of PTE in samples from Spain clearly correlated with high concentrations of pterosin G, which could be expected since pterosin G originates from PTE [27].

151

Stability of illudane glycosides

Stability over 96 hours Results of the 16 tests show that in 10 of them the compounds degraded before the last measurement at 21 days (504 h) after the spiking. In all tests the compounds were present until the preceding measurement (at 96 h) and their degradation rates could be best fitted by linear regression corresponding to zero-order kinetics. Therefore the rates were determined as the slopes of the straight lines over the period of 0 to 96 h. A single exception was PTA in (natural pH lake B, T 15 ˚C, not sterilized) (Figure 2), that had degraded before 96 h and its degradation rate was determined instead for the period 0 to 72 h. The degradation rates of each compound in all combinations of conditions are provided in Figure 2. The slopes of the lines the compound degradation rates in Figure 2. demonstrate that individual compounds in all degraded in very similar rates.

Under natural conditions (natural pH, T 15 ˚C, not sterilized) the compounds in waters A and B degraded at significantly different rates. In lake water A approximately 80% of the added glycosides remained after 96 h, whereas in lake water B half of the compounds had degraded already after approximately 48 h demonstrating the strong influence of the water matrix on degradation. Between a pair of tests representing the same waters and conditions but with the same pH (pH 6.5, T 15 ˚C, not sterilized) the compounds degraded in very similar rates and approximately 80% of the original compounds remained after 96 hours. This clearly indicates that the rapid degradation rates at natural pH of the lake B water were caused due to its relatively high pH (pH 7.4).

In the identical conditions but cold water [natural pH lake B, T 5 ˚C, not sterilized] approximately 80% of the compounds remained after 96 hours. This demonstrates that not only the pH, but the combination of pH and temperature had a strong impact on degradation rates. At identical conditions but sterilized water [natural pH, T 15 ˚C, sterilized] the compounds degraded at similar rates as in non-sterilized water. The absence of effect due to sterilization confirms that the rapid compound degradation in water with higher pH is due to abiotic hydrolysis.

152

Figure 2. Measured pH and degradation rates (± 95% confidence intervals) of the three illudane glycosides in different combinations of parameters of two surface waters.

Results of the pairwise t-test on the effects of individual parameters are provided in Table 4. The relatively large 95% confidence intervals originated because in the majority of the 16 experiments the compounds degraded slowly. This resulted in average uncertainty of the measurements being close to the differences of the measured values.

Table 4. Effect of individual parameters on the stability of the glycosides. The rates are provided as average values given in Figure 2 of all compounds from all experiments sharing the same value of a given parameter.

Average k 95% confidence Pairwise Parameter Value over 96 h intervals t-tests Sterilized 0.164 ±0.108 Sterilization p = 0.001 Not sterilized 0.229 ±0.103 5 °C 0.079 ±0.034 Temperature p ˂ 0.001 15 °C 0.314 ±0.128 Not adjusted of A (pH 7.4) 0.499 ±0.208 pH p ˂ 0.001 Not adjusted of B (pH 5.2) 0.078 ±0.053 Water source A with adjusted pH 0.106 ±0.060 p = 0.895 (reconciled pH) B with adjusted pH 0.103 ±0.058 PTE (average) 0.179 ±0.116 CAU (average) 0.187 ±0.140 PTA (average) 0.231 ±0.151 Different glycosides Null-hypothesis p = 0.009 PTE against CAU (averages) p = 0.783 PTA against PTE (averages) p = 0.001 PTA against CAU (averages) p = 0.004

The results show that three of the studied parameters had statistically significant effects on the compound degradation rates (p ˂ 0.05). The variables with highest impact on the rates were temperature and pH. The compounds degraded differently in the waters of different lakes only due to the difference in pH. The combination of other natural physicochemical parameters of waters (significantly different NPOC, turbidity and electric conductivity of the natural waters provided in the Materials and methods section) did not have effect on the rates (water origin with reconciled pH: p =0.895 >> 0.05).

153

Controlled pH (~6.5): Lake A Lake B

Sterilized

C ˚ ˚

T=5

Notsterilized

Sterilized

C ˚ ˚

T=15

Notsterilized

Figure 2. Measured pH and degradation trends of all compounds over 96 (continues next page). 154

Natural pH:

Lake A: pH~5.2 Lake B: pH~7.4

Sterilized

C ˚ ˚

T=5

Notsterilized

Sterilized

C ˚ ˚

T=15

Notsterilized

155

The average degradation rates of individual compounds (PTE, CAU, PTA) had large confidence intervals overlapping with the average rates of the other compounds (Table 2, 95% confidence intervals of different glycosides). We tested the null-hypothesis that the degradation rates of the 3 compounds throughout the 16 tests were the same. It was rejected by p ˂ 0.05 implying statistically significant difference between the degradation rates of individual glycosides. Three additional pairwise t-tests were performed for each combination between the pairs of compounds to determine the difference. No statistically significant difference between the rates of PTE and CAU was found (p 0.783 > 0.05), implying that the rate of PTA was different from the rates of PTE and CAU. The p values between PTA and the other compounds (p ˂ 0.05) together with values of average rates (k) indicate that over the 96 h period PTA degraded approx. 20 % faster than the other two compounds.

Figure 3. Plot of co-dependency of water properties on degradation rates of PTA, PTE and CAU. Parallel lines reflect no co-dependency of parameters tested (graph A: sterilization has the same effect on compound degradation rates regardless the water source), whereas increasing angle between the lines indicates increasing co-dependency of parameters.

The interaction effects of all pairs of parameters are plotted in Figure 3. Parallel lines (graph A) reflect that parameters did not interdepend. Sterilization of samples reduced the rates by the same extent regardless of the water origin. The non-parallel lines (e.g. graph E) indicate interdependency between the parameters. Regardless of sterilization, in 5 °C the compounds degraded evenly. In 15 °C the average rates depended on sterilization, indicating the presence of microbial degradation. Graph F indicates that the compound degradation rates in waters with different natural pH were very different, but if the pH was brought to the same level, the rates became equal.

156

Stability over 21 days

In 6 tests representing the most stable conditions (Figure 4), the compounds were present in water after 21 days. Except for a single test representing 15 °C and a single test representing not sterilized conditions, the compounds were relatively stable. The most stable conditions represented the combination of sterilization, low temperature and pH 5.2 (94, 95, 94 % of PTE, CAU and PTA remained, respectively). The stability was nearly the same in the test at pH 6.5 (92, 94, 93), but differed in test at pH 7.4 (54, 40, 64). The set of conditions of pH 5.2, 5 ˚C and sterilization were the most optimal for the compound preservation. The experiment also indicates that the methods for PTA preservation in waters by sterile filtration, low temperature and pH 5.0-5.5 developed by former studies [19] [29] [41] apply for the other two glycosides.

157

Controlled pH (~6.5) Natural pH (A: 5.2; B:7.4)

Sterilized

˚ ˚ T=5

All compounds tested in any other experiment representing not sterilized waters degraded before the termination of 21 day period (Figure 2).

Notsterilized

All compounds tested in any other

experiment representing 15˚C degraded

C ˚ ˚

before termination of the 21 day period T=15

(Error! Reference source not found. 2). Sterilized

Figure 4. Measured pH and degradation trends of all compounds over 21 days in the 6 most stable combinations of parameters of surface waters.

158

Discussion

This study detected PTA analogues PTE and CAU and their corresponding aqueous hydrolysis products pterosins G and A in water bodies for the first time. In general, the frequency of occurrence of PTE and CAU was not lower than of PTA. The concentrations of total illudane glycosides in the samples from Spain correlated with the speed of the water flow. Higher concentrations of PTE (the most polar illudane glycoside) and its corresponding hydrolysis product pterosin G were observed in the samples of stagnant waters.

The test of aqueous stability of the illudane glycosides was performed in natural waters with distinctive physicochemical parameters (different content of organic carbon, turbidity, pH and electric conductivity) in different laboratory-controlled values of pH, temperature and microbial activity. Results demonstrated that regardless of the combinations of conditions the degradation rates of the compounds were very similar. The compounds degraded faster under slightly basic pH and higher temperatures and were significantly more stable under moderately acidic to neutral pH (5.2-6.5) and at low temperature (5 ˚C). The most stable conditions resulted in approx. 95% recovery of all compounds after 21 days. This implies that stabilization developed by other studies for PTA (pH 5.0-5.5, sterilization and -20 ˚C) [19,29,41] shall be applicable for the other two illudane glycosides.

The compound degradation rates over the first 96 hours were best fitted by linear regression (zero- order functions). Other PTA studies have demonstrated that degradation follows first-order reaction [29,30]. This could be due to a rapid microbial degradation of PTA in other studies, whereas in this study the used waters were sampled in December and the CFU count revealed that the metabolically active microorganisms were present, though largely dormant. Besides the different best-fitting models, the rates among the previous studies analysing PTA stability in water with respect to pH and temperature conditions were also slightly different. This is presumably due to different used buffer solutions between the experiments. In this study the buffer solutions were not used, but pH was monitored throughout the experiment. The overall observation that PTA is the most stable in moderately acidic to neutral pH and cold temperatures demonstrated in this study match with all previous studies.

The illudane glycosides in water pose potential risk to drinking water resources, aquatic life, land animals accessing the water and potentially farm animal products. To the best of our knowledge the waters reported in this study from Denmark and Sweden were not utilized. However, the analyzed waters from the locations in Spain were occasionally orally admitted by horses and meat cattle. Provided that bracken leachates enter surface waters predominantly due to precipitation [18,19], acidic surface waters adjacent to Pteridium at moderately frequent precipitation and low temperature can potentially contain the illudane glycosides throughout the season of bracken vegetation. 159

Conclusion

This study presents the first report of natural occurrence of PTE and CAU in water bodies. The stability of PTE and CAU were nearly identical to PTA across all studied water conditions implying that the compounds have the same persistence in water after they are released from the plant. The overall results imply that the previous assessments of water toxicity adjacent to bracken based solely on PTA were likely underestimated due to the absence of analysis of PTE and CAU. Even though PTE and CAU express similar aqueous stability, these compounds presumably have greater potential of leaching from the plant and not absorbing to sediments and soils (not studied and not discussed above).

Funding This project was funded by the European Union’s Horizon 2020 research and innovation program under Marie Sklodowska-Curie grant agreement No. 722493.

Acknowledgements The authors are grateful to Prof Elena María Fernández González (University of Oviedo, Spain) for sampling waters in Asturias region in Spain, Prof Mary T. Fletcher (University of Queensland, Australia) for providing plant material needed for compound extraction, Dr. Werner Brack and Dr. Tobias Schulze (Helmholtz Centre for Environmental Research, Germany) for prepared detailed mass spectral data for European MassBank. Anita Schjødt Sandager, Sorivan Chhem Kieth (University of Copenhagen, Denmark) and Greater Copenhagen Utility HOFOR (Denmark) for physicochemical analysis of waters.

160

References used in the manuscript Occurrence and stability of ptesculentoside, caudatoside and ptaquiloside from bracken ferns in surface waters

[1] Thomson JA, "Towards a taxonomic revision of Pteridium (Dennstaedtiaceae)," Telopea (Syd.), vol. 10, no. 4, pp. 793-803, 2004. [2] Matoba, M., Saito, E., Saito, K., Koyama, K., Natori, S., Matsushima, T., Takimoto, M., "Assay of ptaquiloside, the carcinogenic principle of bracken, Pteridium aquilinum, by mutagenicity testing in Salmonella typhimurium," Mutagenesis, vol. 2, p. 419–423, 1987. [3] Hirono, I., Ogino, ., Fujimoto, ., Yamada, ., Yoshida, Y., Ikagawa, M., Okumura, M., "Induction of Tumors in ACI Rats Given a Diet Containing Ptaquiloside, a Bracken Carcinogen," J. Natl. Cancer Inst., vol. 79, p. 1143–1149, 1987. [4] Gomes, J., Magalhaes, A., Michel, V., Amado, I.F., Aranha, P., Ovesen, R.G., Hansen, H.C.B., Gartner, F., Reis, C.A., Touati, E., "Pteridium aquilinum and its ptaquiloside toxin induce DNA damage response in gastric epithelial cells, a link with gastric carcinogenes," Toxicol. Sci., vol. 126, pp. 60-71, 2012. [5] Potter, D.M., Baird, M.S., "Carcinogenic effects of ptaquiloside in bracken fern and related compounds," Br J Cancer., vol. 87, no. 3, pp. 914-920, 2000. [6] Ojika M, Wakamatsu K, Niwa H, Yamada K, "Ptaquiloside, a potent carcinogen isolated from bracken fern Pteridium aquilinum var. latiusculum: structure elucidation based on chemical and spectral evidence, and reactions with amino acids, nucleosides, and nucleotides," Tetrahedron, vol. 43, no. 22, pp. 5261-5274, 1987. [7] Kushida T, Uesugi M, Sugiura Y, Kigoshi H, Tanaka H, Hirokawa J, Ojika M, Yamada K, "DNA damage by ptaquiloside, a potent bracken carcinogen: detection of selective strand breaks and identification of DNA cleavage products," J. Am. Chem. Soc, vol. 116, no. 2, p. 479–486, 1994. [8] Yamada, K., Ojika, M., Kigoshi H., "Ptaquiloside, the major toxin of bracken, and related terpene glycosides: chemistry, biology and ecology," Nat Prod Rep., vol. 24, p. 798–813, 2007. [9] Gil da Costa, R.M., Bastos, M.M.S.M., Oliveira, P.A., Lopes, C., "Bracken-associated human and animal health hazards: chemical, biological and pathological evidence," J Hazard Mater, vol. 1, pp. 203-204, 2012. [10] Prakash, A.S., Pereira, T.N., Smith, B.L., Shaw, G., Seawright, A.A., "Mechanism of bracken fern carcinogenesis: evidence for H-ras activation via initial adenine alkylation by ptaquiloside," Nat Toxins., vol. 4, no. 5, pp. 221-227, 1996.

161

[11] Kigoshi, H., Tanaka, H., Hirokawa, J., Mizuta, K., Yamada, K., "Synthesis of analogues of a bracken ultimate carcinogen and their DNA cleaving activities," Tetrahedron Lett., vol. 33, no. 44, pp. 6647-6650, 1992. [12] Shahin, M., Smith, B.L., Prakash, A.S., "Bracken carcinogens in the human diet," Mutat. Res., vol. 443, p. 69–79, 1999. [13] Alonso-Amelot, M.E., Avendano, M., "Human carcinogenesis and bracken fern: a review of the evidence," Curr. Med. Chem., p. 675–686, 2002. [14] Alonso-Amelot, M.E., Avendano M., "Possible association between gastric cancer and bracken fern in Venezuela: An epidemiologic study," Int. J. Cancer., vol. 9, pp. 252-259, 2001. [15]Galpin, O.P., Whitaker, C.J., Whitaker, R., Kassab, J.Y., "Gastric cancer in Gwynedd. Possible links with bracken," Br J Cancer., vol. 61, pp. 737-740, 1990. [16]Virgilio, A., Sinisi, A., Russo, V., Gerardo, S., Santoro, A., Galeone, A., Taglialatela-Scafati, O., Roperto, F., "Ptaquiloside, the major carcinogen of bracken fern, in the pooled raw milk of healthy sheep and goats: an underestimated, global concern of food safety," J Agric Food Chem., vol. 63, pp. 4886-4892, 2015. [17] Clauson-Kaas, F., Jensen, P.H., Jacobsen, O.S., Juhler, R.K., Hansen, H.C., "The naturally occurring carcinogen ptaquiloside is present in groundwater below bracken vegetation," Toxicol Environ Chem., vol. 33, pp. 1030-1034, 2014. [18] O'Driscoll C, Ramwell C, Harhen B, Morrison L, Clauson-Kaas F, Hansen HCB, Campbell G, Sheahan J, Misstear B, Xiao LW, "Ptaquiloside in Irish Bracken Ferns and Receiving Waters, with Implications for Land Managers," Molecules, vol. 543, no. 21, 2016. [19] Clauson-Kaas F., Ramwell C., Hansen, H.C.B., Strobel B.W., "Ptaquiloside from bracken in stream water at base flow and during storm events," Water Res., vol. 106, pp. 155-162, 2016. [20] Rasmussen, L.H., Hansen, H.C., Lauren, D., "Sorption, degradation and mobility of ptaquiloside, a carcinogenic Bracken (Pteridium sp.) constituent, in the soil environment," Chemosphere, vol. 58, no. 6, pp. 823-835, 2005. [21] Rasmussen, L.H., Kroghsbo, S., Frisvad, J.C., Hansen, H.C.B., "Occurrence of the carcinogenic Bracken constituent ptaquiloside in fronds, topsoils and organic soil layers in Denmark," Chemosphere, vol. 51, no. 2, pp. 117-127, 2003. [22] Gomes J., Magalhaes, A., Carvalho, A.S., Hernandez, G.E., Papp, S.L., Head, S.R., Michel, V., David, L., Gartner, F., Touati, E., Reis, C.A., "Glycophenotypic Alterations Induced by Pteridium aquilinum in Mice Gastric Mucosa: Synergistic Effect with Helicobacter pylori Infection," PLOS ONE, vol. 7, no. 6, pp. 1-10, 2012.

162

[23] Fletcher MT, Hayes PY, Somerville MJ, De Voss JJ, "Ptesculentoside, a novel norsesquiterpene glucoside from the Australian bracken fern Pteridium esculentum," Tetrahedron Lett., vol. 51, pp. 1997-1999, 2010. [24] Castillo UF, Wilkins AL, Lauren DR, Smith BL, Towers NR, Alonso-Amelot ME, Jaimes-Espinoza R, "Isoptaquiloside and caudatoside, illudane-type sesquiterpene glucosides from Pteridium aquilinum var. caudatum," Phytochemistry, vol. 44, no. 5, pp. 901-906, 1997. [25] Fletcher, M.T., Brock, I.J., Reichmann K.G., McKenzie R.A., Blaney B.J., "Norsesquiterpene Glycosides in Bracken Ferns (Pteridium esculentum and Pteridium aquilinum subsp. Wightianum) from Eastern Australia: Reassessed Poisoning Risk to Animals," J. Agric. Food Chem., vol. 59, pp. 5133-5138, 2011. [26] de Oliveira, L.G.S., Boabaid, F.M., Kisielius, V., Rasmussen, L.H., Buroni, F., Lucas, M., Schild, C.O., Lopez, F., Machado, M., Riet-Correa, F., "Hemorrhagic diathesis in cattle due to consumption of Adiantopsis chlorophylla (Swartz) Fée (Pteridaceae)," Toxicon: X, vol. 5, no. 100024, 2020. [27] Kisielius V., Lindqvist D.N., Thygesen M.B., Rodamer M., Hansen H.C.B., Rasmussen L.H., "Fast LC-MS quantification of ptesculentoside, caudatoside, ptaquiloside and corresponding pterosins in bracken ferns," J. Chromatogr. B, vol. 1138, no. 121966, 2020. [28] Kisielius, V., Markussen, B., Rasmussen., L.H., "Spatial and temporal variation of naturally occurring toxic illudane glycosides in Northern European Pteridium," submitted to J. Chem. Ecol, 2020. [29] Ayala-Luis K.B., Hansen P.B., Rasmussen L.H., Hansen H.C., "Kinetics of ptaquiloside hydrolysis in aqueous solution," Environ Toxicol Chem., vol. 25, no. 10, pp. 2623-2629, 2006. [30] Saito, K., Nagao, T., Matoba M., Koyama K., Natori, S., Murakami, T., SaikI Y., "Chemical assay of ptaquiloside, the carcinogen of pteridium aquilinum, and the distribution of related compounds in the pteridaceae," Phytochemistry, vol. 28, no. 6, p. 1989, 1989. [31] Ovesen, R.G., Rasmussen, L.H., Hansen, H.C.B., "Degradation kinetics of ptaquiloside in soil and soil solution," Environ. Toxicol. Chem., vol. 27, no. 2, pp. 252-259, 2008. [32] Caceres-Pena YC, Naya M, Calcagno-Pissarelli MP, Alonso-Amelot ME, "Influence of bracken fern (Pteridium caudatum L. Maxon) pre-treatment on extraction yield of illudane glycosides and pterosins," Phytochem Anal, vol. 24, no. 4, pp. 290-295, 2012. [33] Rasmussen, L.H., Pedersen, H.Æ., "Screening for Ptaquiloside in Ferns: Using Herbarium Speciments for Qualitative Mapping Purposes," Phytochem Anal, vol. 28, no. 6, pp. 575-583, 2017.

163

[34] Rasmussen LH, Donnelly E, Strobel BW, Holm PE, Hansen HCB, "Land management of bracken needs to account for bracken carcinogens - A case study from Britain," J. Environ. Manage., vol. 151, pp. 258-266, 2015. [35] Saito M., Umeda M., Enomoto M., Hatanaka, Y., Natori S., Yoshihira K., Fukuoka M., Kuroyanagi M., "Cytotoxicity and carcinogenicity of pterosins and pterosides, 1-indanone derivatives from bracken (Pteridium aquilinum)," Experientia, vol. 31, p. 829–831, 1975. [36] Hsu, F.L., Huang, C.F., Chen, Y.W., Yen, Y.P., Wu, C.T., Uang, B.J., Yang, R.S., Liu, S.H., "Antidiabetic Effects of Pterosin A, a Small-Molecular-Weight Natural Product, on Diabetic Mouse Models," Diabetes, vol. 62, no. 2, pp. 628-638, 2013. [37] Yahara Y, Takemori H, Okada M, Kosai A, Yamashita A, Kobayashi T, Fujita K, Itoh Y, Nakamura M, Fuchino H, Kawahara N, Fukui N, Watanabe A, Kimura T, Tsumaki N, "Pterosin B prevents chondrocyte hypertrophy and osteoarthritis in mice by inhibiting Sik3," Nat Commun, vol. 7, no. 10959, pp. 1-12, 2016. [38] Sheridan H, Frankish N, Farrell R, "Smooth Muscle Relaxant Activity of Pterosin Z and Related Compounds," Planta Med., vol. 65, no. 3, pp. 271-272, 1999. [39] "European MassBank," [Online]. Available: https://massbank.eu/MassBank. [Accessed 03 03 2020]. [40] "Danish standards on water survey sampling, transport and storage of samples for microbiological studies," Danish Standards Council, 1983. [41] Clauson-Kaas F., Hansen H.C.B., Strobel B.W., "UPLC-MS/MS determination of ptaquiloside and pterosin B in preserved natural water," Anal Bioanal Chem., vol. 408, p. 7981–7990, 2016. [42] Vetter J, "A biological hazard of our age: bracken fern [Pteridium aquilinum (L.) Kuhn] - a review," Acta Vet. Hung., vol. 57, no. 1, p. 183–196, 2009. [43] Marrs RH, Watt AS, "Biological Flora of the British Isles: Pteridium aquilinum (L.) Kuhn," J. Ecol., vol. 94, no. 6, pp. 1272-1321, 2006. [44] Page CN, "The taxonomy and phytogeography of bracken - a review," Bot. J. Linn. Soc., vol. 73, no. 1-3, p. 1–34, 1976.

164

Supplementary material

Table 1. Measured physicochemical parameters of sampled water bodies

conductivity Site ID pH (μS/cm) 1 6.1 159 2 7.2 180 3 6.4 123 4 6.3 120 5 6.9 77 6 7.9 413 7-14 No measurements

165

Controlled pH (~6.5): Lake A Lake B

Sterilized

C ˚ ˚

T=5

Notsterilized

Sterilized

C ˚ ˚

T=15

Notsterilized

Figure 1. Formation of pterosins G and B in waters of illudane glycoside degradation experiments (continues next page) 166

Natural pH: Lake A: pH~5.2 Lake B: pH~7.4

Sterilized

C ˚ ˚

T=5

Notsterilized

Sterilized

C ˚ ˚

T=15

Notsterilized

Figure 1. Formation of pterosins G and B in waters of illudane glycoside degradation experiments

167

4. Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids

Vaidotas Kisielius*,†*✉, Jawameer R. Hama2*, Natasa Skrbic1,‡, Hans Christian Bruun Hansen2, Bjarne W. Strobel2, Lars Holm Rasmussen1

Abstract

Pyrrolizidine alkaloids (PAs) are persistent mutagenic and carcinogenic compounds produced by many common plant species. These compounds are regulated in food and medicinal products to ensure consumer health and safety. However, there is little awareness that PAs can contaminate water resources. Therefore, no regulations exist to limit PAs in drinking water. This study measured a PA base concentration of ~70 ng/l in stream water adjacent to an invasive PA-producing plant Petasites hybridus. After intense rain the PA concentration increased 10 fold. In addition, PAs measured up to 230 ng/l in seepage water from groundwater wells. The dominant PAs in both water types corresponded to the most abundant PA types in the plants. According to acceptable levels proposed by health authorities, drinking 2 l of the well water reported in this study would violate the daily PA ingestion limit. The study presents the first discovery of persistent plant toxins in well water and their associated risks.

Keywords

Butterbur, Petasites, Water quality, Phytotoxins, Carcinogens

* Department of Technology, University College Copenhagen, Sigurdsgade 26, 2200 Copenhagen, Denmark

† Department of Plant and Environmental Sciences, University of Copenhagen, Thorvaldsensvej 40, 1871 Frederiksberg, Denmark

‡ Greater Copenhagen Utility HOFOR, Ørestads Blvd. 35, 2300 Copenhagen, Denmark

* Shared First Authorship

✉ [email protected] (V. Kisielius)

168

Graphical abstract

Introduction

A popular fairytale The Ugly Duckling written in Danish by Hans Christian Andersen in 1843 narrates that the ducklings were born under the leaves of a butterbur: the plant such tall that small children could stand upright under the largest ones [1]. The customized name of this regional plant was not translated into foreign languages. However, the English name butterbur also has a local customary background. It is supposed to have originated from the large leaves that were used to wrap butter during hot weather [2]. Invasive plants, such as the butterbur, and their adverse roles are little recognized by society, especially when they are inseparable parts of the culture or folklore.

Common butterbur (Petasites hybridus (G. Gaertn., B. Mey. & Scherb.)) is a flowering plant found in damp nutrient-rich soil and around the shores of fresh water bodies in northern Europe and parts of North America [3–4]. With the aid of powerful rhizomes it can form large populations; its leaves reach up to 70 cm in width on 1 m stems that shade other vegetation and thereby cause soil erosion [5]. In these regions the species is regarded as invasive since it is originally native to southern Europe and western Asia and was introduced to northern Europe around 1350 as a medicinal plant by monks [5–7]. The butterbur expanded from cultivation sites in monasteries, castles and estates to natural habitats across northern and western Europe. Nowadays, it is also marketed and deliberately planted for ornamental purposes, e.g. around local community fire safety ponds or in garden lakes as portrayed in H.C. Andersen’s The Ugly Duckling. 169

The bioactive sesquiterpene constituents in butterbur - petasin and isopetasin - express antispasmodic properties and offer a variety of medicinal applications where the relaxation of muscle and vascular spasms is desired. Historically, the plant’s leaves and rhizomes were also used to treat or prevent plague, rashes, arthritis, kidney and bladder stones. Its most documented evidence-based modern therapeutic application is the treatment of migraine headaches and allergic rhinitis [4,5,8,9]. However, commercialization of unrefined butterbur extracts is prohibited due to numerous cases of intensive liver damage caused by high concentrations of unsaturated pyrrolizidine alkaloids [4,5,8,10,11].

Alkaloids are a large class of naturally occurring diverse organic compounds structurally determined by a heterocyclic ring structure containing at least one nitrogen atom. They are produced as secondary metabolites primarily by fungi and plants, but also by bacteria and animals [12–17]. Many alkaloids are highly toxic and some contain properties desired for therapeutic or recreational purposes [18]. Atropine and coniine rich extracts have been used for homicide since ancient Rome. Strychnine and anabasine were used as insecticides before the development of a wide range of synthetic pesticides that are comparatively less toxic to humans [19]. For centuries alkaloids have been used as pharmaceuticals (morphine, codeine) and stimulants (cocaine, nicotine) [13,20]. Health authorities regulate specific alkaloid groups to ensure safety of food, tea, medicine, honey, supplements and other products [14,21–23].

Heterocyclic rings and ample hydrogen bonds determine alkaloid stability and contribute to their high water solubility. The polarity and persistence of alkaloids make them compatible with persistent and mobile organic compounds (PMOCs). PMOCs are emerging contaminants that are rarely detected in the environment due to the lack of available analytical techniques [24]. PMOCs include surfactants, polar industrial chemicals, pharmaceuticals and personal care products [25,26]. Alkaloids applied as pharmaceuticals or stimulants are also commonly found in waste water and sewage sludge, whereas groundwater surveys reveal alkaloid caffeine among the most frequently encountered compounds [20,27–31].

Pyrrolizidine alkaloids (PA) comprise several hundred compounds based on the heterocyclic two- ring structure necine [13,32,33] (Figure 1). PAs are among the most common natural toxins produced by plants as defence chemicals against insects and herbivores [34–37]. The PA composition in PA-producing plants is not necessarily constant and can vary due to climatic and environmental conditions, the age and the part of the plant, as well as discrete genotypes and chemotypes [22,38,39]. The mode of PA toxicity depends on the molecular structure. PAs with an unsaturated bond in 1,2 position of the necine unit (Figure 1) (hereafter called unsaturated PAs) express nucleophilic properties. When ingested, unsaturated PAs metabolically oxidize and react with proteins and nucleic acids [40–43] resulting in hepatotoxic, mutagenic and carcinogenic effects [13,34,38,43–47].

170

Oxidized PAs (PA N-oxides, Figure 1) are present in plants in nearly equal quantities as free base PAs [48]. The toxicity of PA N-oxides has been demonstrated but most are suggested to be less toxic than the PAs from which they originate [39,48]. However, metabolic reduction of N-oxides to their corresponding free base PAs has been reported to take place in the gut of animals [39,49]. Studies on human liver microsomal system confirm that the results from animal studies are also relevant to humans [50]. N-oxides are more polar than free base PAs, implying potentially greater aquatic emissions from plants and a higher mobility in soils and sediments.

Plants with high concentrations of unsaturated PAs are responsible for the death of cattle in many countries [15,51]. The World Health Organization (WHO) recommends minimizing exposure to humans as much as possible [52], whereas the Federal Institute for Risk Assessment (Germany) and the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (UK) suggest that tolerable levels of unsaturated PAs are 7 ng of daily intake per kg of bodyweight [34]. Based on the observed carcinogenicity the European Food Safety Authority (EFSA) applies a lower and upper bounds to a mean dietary exposure that spans from 0 to 48 ng of daily intake per kg of body weight for different age groups [21].

It is estimated that 3% of all flowering plants contain at least one unsaturated PA [53]. In regions where these plants are prevalent there is an emerging concern of PAs highly exceeding safety limits in honey [17] [21,54–56]. Despite their abundance in flora, the degree of natural PA emissions and their fate in the environment remains largely unknown. A recent study measured up to 1350 μg/kg of PAs in soil and up to 270 μg/l in lake water [57]. The study associated the compounds with adjacent densely growing common ragwort (Jacobaea vulgaris (Gaertn.)).

We hypothesize that PAs can naturally leach from butterbur into water and make it unsafe to drink. The stability of PAs and frequently reported cases of groundwater contamination by structurally similar caffeine imply that PAs can leach into groundwater. Since there are no regulations that require PA monitoring, these and other toxic alkaloids may be overlooked as emerging environmental contaminants.

Stream water was sampled over a 4-month period from maturation to decay of adjacently growing butterbur. To identify any PA washoff effects, stream sampling events occurred during base flow conditions, shortly after intense precipitation and during drought conditions. In addition, groundwater and seepage water inside groundwater wells were sampled from two wells with adjacently growing butterbur and two wells without. The three water types, the leaves and the rhizomes of the plant were analysed. The concentrations of individual and total PAs were examined and compared.

171

Figure 1. Basic structural groups of pyrrolizidine alkaloids

Results

A site with a channelized surface water stream and a system of groundwater abstraction was were selected in east Denmark (Figure 2). The shores of the stream and the land surface surrounding two of the wells (G2 and G3) were infested with butterbur. Every well consisted of a vertical pipe providing access to deep groundwater (estimated water level of approximately 60 m depth) and a surrounding casing supported by concrete rings that naturally filled with seepage water from the soil (observed water level 2.2–3.0 m depth) (Figure 3). The waters from different depths are not mixed and the deep groundwater is pumped for municipal water supply.

172

Figure 2. A map of the monitoring site with distribution of butterbur (Petasites hybridus)

Figure 3. Butterbur on both shores of the sampled stream (A) and surrounding the groundwater well G3 (B). Open G3 well with seepage water in ~2.5 m depth (C).

173

A qualitative test revealed 9 cyclic diester unsaturated PAs in butterbur leaves sprouting near well G2 in April 2019: jacobine, jacobine N-oxide, monocrotaline, monocrotaline N-oxide, senecionine, senecionine N-oxide, senecivernine, senecivernine N-oxide and senkirkine (hereafter referred to as Type 1 PAs). Subsequent quantitative analyses of matured butterbur leaves and rhizomes sampled near wells G2 and G3 on the dates listed in Table 2 of the supplementary material (SM) revealed the same PA compositions with concentrations provided in Figure 4.

A total of 21 PAs were detected in the stream water and in the seepage water of wells G2 and G3. They comprised all the type 1 PAs, together with lower concentrations of 12 unsaturated PAs that were not detected in the butterbur (hereafter referred to as Type 2 PAs). Type 2 PAs comprised cyclic diesters (erucifoline, retrorsine, retrorsine N-oxide, seneciphylline), open chain diesters (echimidine, echimidine-N-oxide, lasiocarpine) and monoesters (europine, europine N-oxide, heliotrine, intermedine, intermedine N-oxide). The concentrations of individual type 1 PAs measured in the stream water and in the seepage water in wells G2 and G3 are presented in Figure 4. No PAs were detected in the deep groundwater, in seepage waters of the control wells G1 and G4, or in blank water sampled as the field controls.

Plant (ng/g) Stream water (ng/l) Seepage water (ng/l)

Sampling event 1 1 6 6 6 6 1 2 4 6 7 3 3 5 5 7 7 Sampling location G2 G2 G2 G2 G3 G3 Surface water monitoring site G2 G3 G2 G3 G2 G3 Leaves / Rhizomes L R L R L R Compound

Jacobine

oxide

Jacobine N

Monocrotaline

oxide

Monocrotaline N

174

Plant (ng/g) Stream water (ng/l) Seepage water (ng/l)

Sampling event 1 1 6 6 6 6 1 2 4 6 7 3 3 5 5 7 7 Sampling location G2 G2 G2 G2 G3 G3 Surface water monitoring site G2 G3 G2 G3 G2 G3 Leaves / Rhizomes L R L R L R Compound

Senecionine

oxide

Senecionine N

Senecivernine

not detected

oxide

Senecivernine N

Senkirkine

Figure 4. Average concentrations of PAs detected in butterbur plants, stream and seepage waters ± SDs (in alphabetical order). The graphs of PAs in the plants have different scales on the y-axes. The empty values represent no detection.

PAs were detected in the stream water and in the seepage water in wells G2 and G3 during all sampling events. In all sampled matrices the concentrations of N-oxides were similar to the concentrations of the equivalent free base compounds. The most frequently found and abundant PAs in the various sampled water corresponded to the most abundant PAs in the plants (senkirkine, senecionine, and senecionine N-oxide).

Besides the established lower toxicity of N-oxides and other PA metabolites, there is no regulatory framework or literature on the toxicity of individual unsaturated PAs. Given that the toxicity is controlled by the necine ring, smaller PA molecules should be more toxic per mass concentration. No other PA metabolites were detected and N-oxides are known to be subject to back- 175

transformation. Taking into account that the molar masses of the PA compounds were very similar, further assessment of toxicity applies to total PAs. This evaluation is also compatible with the thresholds imposed by the authorities that regulate the unsaturated PAs in orally consumed products.

The base stream flow (sampling events No. 1, 6 and 7 (Table 2 of the SM)) had average total PA concentrations in the surface water monitoring site from 49 to 91 ng/l (Figure 5). Intense rapid rain between sampling events No. 1 and 2 (13 mm recorded by the Danish Meteorological Institute) resulted in a 10 fold increase in the total PA concentration, implying simple and quick toxin washoff and release from the plants. The segmentation of the stream into stagnant water patches due to summer drought conditions (sampling event No. 4) resulted in PA concentrations approximately 3 times higher than at the base flow.

Total PA concentrations in the well seepage waters were highest during sampling event No. 3 (Figure 5) which corresponded with the largest biomass of adjacent butterbur. The greater plant coverage of the well (well G3, Figure 2) was associated with higher PA concentrations in the well water. There were fewer plants around the wells at subsequent sampling events due to regular biomass cutting and removal by a maintenance company to ensure access to the wells. This biomass removal possibly caused reduced PA concentrations.

Figure 5. Concentrations of total PAs in the stream at the surface water monitoring site (A) and in seepage waters in the wells (B) ± SDs of the total PA concentrations in triplicate samples.

The concentrations of individual type 2 PAs did not exceed 10 ng/l in both water types, except in the stream water under non-base flow conditions (108 ng/l of retrorsine and 27 ng/l of seneciphylline in sampling event No. 2, and 56 ng/l of erucifoline and 46 ng/l of retrorsine N-oxide in sampling event No. 4). Type 1 PAs were only comprised of cyclic diesters, whereas type 2 PAs were comprised of cyclic diesters, open-chain diesters and monoesters (Table 1 of the SM). Open-

176

chain diester and monoester PAs have not been reported in butterbur in this or in previous studies [4].

Since other known PA producing plants were not observed in the area and the control wells contained neither PA types, this indicates that all type 2 PAs indirectly originated from the butterbur. The highest concentrations of type 2 PAs in the seepage water were observed in sampling event No. 7 which corresponded to the highest proportion of naturally wilted and decayed butterbur and the longest cumulative retention time of the toxins in the wells. This implies that type 2 PAs were formed in decaying biomass or in the water from the type 1 PAs [35,41,43,58].

Discussion

This research reports PAs from butterbur in surface water for the first time. Together with a recent study on common ragwort [57], it suggests PAs are newly identified environmental contaminants. The surface waters adjacent to PA-producing plants can contain substantial PA concentrations and pose a risk to aquatic life, land animals accessing the water and possibly to farm animals. Furthermore, this study illustrates PA leaching from PA-producing plants to groundwater wells via seepage water. Open bottomed shallow groundwater wells are common in remote one-household settlements with no centralized water supply in both developing and developed countries. These wells are frequently subject to pollution of organic matter, nitrate and pesticides that seep from farming areas and noncentralized sanitation. To the best of our knowledge PAs in well water have not been examined before.

According to the tolerable levels proposed by the Federal Institute for Risk Assessment (Germany) and the Committee on Toxicity of Chemicals in Food, Consumer Products and the Environment (UK) [34], 1 l of surface water (523 ng PAs/l) or 2 l of well water (233 ng PAs/l) reported in this study would violate the daily PA ingestion limit for a 70 kg healthy human. The stream and shallow well waters reported in this study caused no harm to human health only because these waters were not utilized for drinking. The analysed exploited deep groundwater contained no PAs because this water was pumped from beneath impermeable geological layers with no contact with leachates from the butterbur. However, the groundwater pipes immersed in contaminated seepage water (Figure 3 C) pose a potential risk to groundwater. Poorly drilled or maintained deep groundwater wells are subject to the downward transmission of water carrying microbiological and hazardous anthropogenic contaminants [59–61]. Microbiological and anthropogenic chemical pollution is methodically monitored and regulated in drinking water by law. The Directive of the European Parliament and of the Council on the protection of groundwater against pollution and deterioration (2006/118/EC) imposes EU member states to terminate exploitation of water from groundwater wells if active substance of any pesticide, including its relevant metabolites, degradation and reaction products exceeds the 100 177

ng/l threshold [62]. Threshold violations have been reported in Italy, the Netherlands, Belgium, Sweden, Finland and Denmark, leading to the closure of many drinking water wells [59].

Plant-produced toxic compounds can locally outcompete anthropogenic chemicals in quantity because of their continuous production and comparatively high concentrations in plants [63]. Invasive and foreign ornamental species tend to contain highly toxic natural compounds. Inedible parts of agricultural crops (e.g., lupin, potato) can produce natural toxins that are released to larger land areas. To some extent, the natural toxins are regulated in food. Plant toxins not entering the human food chain are generally regarded as not harmful to humans. However, their aquatic toxicity and related ingestion by humans remain largely unknown. In addition to direct harm, ingested plant toxins can interact with other contaminants in humans creating synergies and cumulative adverse health effects [64].

Anthropogenic activities directly or indirectly stimulate the spread of many invasive species. Studies show that among toxic invasive plants, producing PAs and other alkaloids are the most common [65]. For example, the genus Cytisus comprises about 50 species of invasive PA-producing flowering plants. Many other plant-produced toxic alkaloids can be stable in water, like alkaloids entailing toxic persistent heterocyclic triazole rings.

The PA releasing butterbur tends to invade moist and disturbed lands like shaded sides of roads, bridge embarkments, the shores of artificial surface water bodies, edges of ditches, channelized streams and water wells. According to the Global Biodiversity Information Facility, the species is omnipresent in most of western and northern Europe, common in the rest of the Europe and occasionally found across North America [66]. No mitigation operations of these species exist. In northern Europe climate change is leading to an increase in precipitation that together with various land management activities will further stimulate the spread of butterbur, leading to an increased risk of PA contaminated water.

The absence of regulations on natural toxins in water has resulted in a lack of suitable methodologies for their monitoring and identification. This study illustrates the need for the advancement of procedures to monitor persistent plant toxins and indicates the need for water cleaning technologies to facilitate their removal.

Methods

Sample collection

Butterbur plants were sampled by cutting 3 whole average-height leaves from the ground surface and digging 10-15 cm segments of 3 rhizomes. Each sample was collected approximately 5 m from one another. The sampled biomass was frozen within 8 hours at -20° C prior to analyses. The 178

stream water was sampled at the surface water monitoring site during the dates listed in Table 2 of the SM. Intense precipitation resulted in a sudden flow increase, whereas dry periods from July to September resulted in no water movement or dry stream bed. Supplementary stream water samples were irregularly taken at 3 additional surface water sampling sites (Figure 2).

The seepage water was sampled by manually scooping it from the top 50 cm in 1 l bottles with an extension stick. Deep groundwater was sampled using a pump. The samples of all water types were taken in triplicates of 1 l in polypropylene bottles and frozen within 8 hours at -20° C prior to analyses. To avoid any possible cross-contamination, no plant samples were taken during groundwater sampling events (Table 2 of the SM). A PA-free blank water was sampled, processed and analysed among all surface and groundwater samples as a field control.

Analytical

All plant and water samples were prepared by the method described in Hama and Strobel (2019) [57]. In short, three plants from each sampling point were finely chopped into 2 mm pieces, vigorously mixed, 0.1 g was measured into 10 ml amber glass tubes, 10 ml of MeOH was added and the mixture was sonicated for 15 minutes. Subsequently, the supernatants were transferred to 50 ml centrifuge tubes. Extraction cycles with additional MeOH and sonication from the remaining plant materials were repeated twice, resulting in a total of 30 ml of supernatant per sample. The supernatants were centrifuged for 10 min at 2100 g and dried under gentle nitrogen flow in a heating block at 40 °C. The dried extracts were dissolved in 1 ml 40% (V/V) MeCN and filtered through 0.2 μm polytetrafluoroethylene (PTFE) membrane filters prior to analysis.

The water samples were passed through a 2.5 μm filter paper (Whatman quantitative-Grade 42) under a vacuum and acidified to a pH of 3.0 with 0.1 M formic acid. Solid phase extraction cartridges Oasis MCX (6 cc, 150 mg 30 µm particle) were conditioned with 5 ml MeOH followed by

5 ml H2O, and then 1.0 l of the acidified water samples were loaded at a flow rate of 10 ml/min. The loaded cartridges were washed with 5 ml 0.065 mM formic acid and successively eluted with 5 ml MeOH and with 10 ml 3:1 (V/V) mixture of A: MeOH and B: 10% ammonia solution. A gentle nitrogen flow dried the eluates in a heating block at 40 °C. The dried extracts were dissolved in 1.0 ml 40% (V/V) MeCN and filtered through a 0.2 μm PTFE membrane prior to analysis.

The PAs were identified and quantified from individual fragmentation patterns with UPLC-MS/MS as described in Hama and Strobel (2019) [57]. The approximate instrumental limits of detection and quantification of PAs were 2000-7000 and 5000-9000 ng/l, respectively. The solid phase extraction provided a 1000 times concentration of the water samples with a 90% recovery rate. Field and laboratory blanks consisted of certified laboratory grade organic-free water (Milli-Q). All concentrations of PAs in water blanks were below the limits of detection. The average recovery

179

rate of surrogate (caffeine) was 94% ±11 (n=3). The concentrations of all the free base PAs and N- oxides were quantified against certified external standards purchased from Phytolab (Germany).

The concentrations of individual PAs were reported as the average concentrations quantified in triplicate samples. In cases when the individual PAs were quantified in only 2 of the triplicate samples (close to the approximate 5-9 ng/l limit of quantification of the full method), the concentrations of individual PAs were reported as the average of 2 positive samples. If individual PAs were quantified in only 1 of the triplicate samples, the concentrations were not quantified and reported as trace. Total PA concentrations (Figure 5) were reported as the averages of the sums of all PAs quantified in triplicate samples.

Acknowledgements

The authors would like to thank Steen Dam Larsen from Greater Copenhagen Utility HOFOR for providing access to groundwater wells and for technical support.

Funding

This project was funded by the European Union’s Horizon 2020 research and innovation program under Marie Sklodowska-Curie grant agreement No. 722493.

Author contributions

V.K. and L.H.R. conceived and directed the study, V.K. and J.R.H. performed analysis, V.K. and N.S. performed sampling, H.C.B.H and B.W.S supervised, V.K. wrote the article with feedback from J.R.H, H.C.B.H, B.W.S and L.H.R.

Competing interests

The authors declare no competing interests

180

References used in the manuscript Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids

[1] Andersen H.C. Den grimme ælling. Nye Eventyr. Første Bind. Første Samling (C.A. Reitzels Forlag, 1843).

[2] Brøndegaard V. J. Folk og flora: dansk etnobotanik (Rosenkilde og Bagger, 1987).

[3] Giles, M., Ulbricht, C., Khalsa, K.P.S., DeFranco, Kirkwood C., Park, C., Basch, E. Butterbur. J. Herb. Pharmacother. 5, 119-143 (2009).

[4] Aydın, A.A., Zerbes, V., Parlar, H., Letzel T. The medical plant butterbur (Petasites): Analytical and physiological (re)view. J Pharm Biomed Anal. 75, 220– 229 (2013).

[5] Giversen, I., Brimer, L., Kristiansen, B. Danmarks Vilde Lægeplanter (Gyldendal A/S, 2014).

[6] Asen, A. Plants of possible monastic origin, growing in the past or present, at medieval monastery grounds in Norway. Plants and Culture: seeds of the cultural heritage of Europe. 227- 238 (Edipuglia, 2009).

[7] Solberg, S.O. More than just weeds. NordGen’s work with Cultural Relict Plants and Bernt Løjtnant’s inventories from Denmark. (Nordic Genetic Resource Center, Alnarp, Sweden, 2014).

[8] Anderson, N & Borlak J. Hepatobiliary Events in Migraine Therapy with Herbs - The Case of Petadolex, A Petasites Hybridus Extract. J Clin Med. 8, 652 (2019).

[9] Kozlov, V., Lavrenova, G., Savlevich, E., Bazarkina, K. Evidence-based phytotherapy in allergicrhinitis rhinitis. Clinical phytoscience 4, 1-8 (2018).

[10] Avula, B., Wang, Y.H., Wang, M., Smillie, T.J., Khan, I.A. Simultaneous determination of sesquiterpenes and pyrrolizidine alkaloids from the rhizomes of Petasites hybridus (L.) GM et Sch and dietary supplements using UPLC-UV and HPLC-TOF-MS methods. J Pharm Biomed Anal. 70, 53-63 (2012).

[11] Kalin, P. The common butterbur (Petasites hybridus) - Portrait of a medicinal herb. Forsch Komplementarmed Klass Naturheilkd. 10, 41-44 (2003).

[12] Roberts, M. F. & Wink, M. Alkaloids: Biochemistry, Ecology, and Medicinal Applications (Springer, 1998).

181

[13] Aniszewski, T. Alkaloids: chemistry, biology, ecology, and applications. Second edition. (Elsevier Science, 2015).

[14] International Agency for Research on Cancer. IARC Monographs on the evaluation of carcinogenic risks to humans; Some traditional herbal medicines, somemycotoxins, naphthalene and styrene. (IARC Press, Lyon, 2002).

[15] Xia, Q., He, X., Ma, L., Chen, S., Fu, P.P., Pyrrolizidine Alkaloid Secondary Pyrrolic Metabolites Construct Multiple Activation Pathways Leading to DNA Adduct Formation and Potential Liver Tumor Initiation. Chem. Res. Toxicol. 31, 619-628 (2018).

[16] Yang, L., Stockigt, J. Trends for diverse production strategies of plant medicinal alkaloids. Nat Prod Rep. 27, 1469-1479 (2010).

[17] Wang, T., Frandsen, H.L., Christiansson, N.R., Rosendal, S.E., Pedersen, M., Smedsgaard J. Pyrrolizidine alkaloids in honey: Quantification with and without standards. Food Control. 98, 227-237 (2019).

[18] Scholtz, S., MacMorris, L., Krogmann, F., Auffarth, G.U. Poisons, Drugs and Medicine: On the Use of Atropine and Scopolamine in Medicine and Ophthalmology: An Historical Review of their Applications. J Eye Stud Treat. 1, 51-58 (2019).

[19] Matolcsy, G., Nadasy, M., Andriska, V. Pesticide Chemistry. 32 (Elsevier Science, 1989).

[20] Dembitsky, V.M., Gloriozova, T.A., Poroikov, V.V. Naturally occurring plant isoquinoline N- oxide alkaloids: their pharmacological and SAR activities. Phytomedicine 22, 183-202 (2015).

[21] Europen Food Safety Authority. Risks for human health related to the presence of pyrrolizidine alkaloids in honey, tea, herbal infusions and food supplements. EFSA Journal. 15 (2017).

[22] European Medicines Agency. Public statement on contamination of herbal medicinal products/traditional herbal medicinal products with pyrrolizidine alkaloids - transitional recommendations for risk management and quality control. (EMA Committee on Herbal Medicinal Products, London, 2016).

[23] Lopez-Pacheco, I.Y., Silva-Nunez, A., Salinas-Salazar, C., Arevalo-Gallegos, A., Lizarazo- Holguin, L.A., Barcelo, D., Iqbal, H.M.N., Parra-Saldivar, R. Anthropogenic contaminants of high concern: Existence in water resources and their adverse effects. Sci Total Environ. 690, 1068- 1088 (2019).

[24] Angeles, L.F. & Aga, D.S. Catching the Elusive Persistent and Mobile Organic Compounds: Novel Sample Preparation and Advanced Analytical Techniques. Trends Environmen Anal. https://doi.org/10.1016/j.teac.2019.e00078 (2019). 182

[25] Reemtsma, T., Berger, U., Arp, H.P.H., Gallard, H., Knepper, T.P., Neumann, M., Quintana, J.B., de Voogt, P. Mind the Gap: Persistent and Mobile Organic Compounds - Water Contaminants That Slip Through. Environ. Sci. Technol. 50, 10308-10315 (2016).

[26] Furlong, E.T., Batt, A.L., Glassmeyer, S.T., Noriega, M.C., Kolpin, D.W., Mash, H., Schenck, K.M. Nationwide reconnaissance of contaminants of emerging concern in source and treated drinking waters of the United States: Pharmaceuticals. Sci Total Environ. 579, 1629-1642 (2017).

[27] Boxall, A.B.A. The environmental side effects of medication. EMBO Rep. 5, 1110-1116 (2004).

[28] Gao, J., Li, J., Jiang, G., Yuan, Z., Eaglesham, G., Covaci, A., Mueller, J.F., Thai, P.K. Stability of alcohol and tobacco consumption biomarkers in a real rising main sewer. Water Res. 138, 19-26 (2018).

[29] Lian, L., Yan, S., Yao, B., Chan, S., Song, W. Photochemical transformation of nicotine in wastewater effluent. Environ. Sci. Technol. 51, 11718-11730 (2017).

[30] Stuart, M., Lapworth, D., Crane, E., Hart, A. Review of risk from potential emerging contaminants in UK groundwater. Sci Total Environ. 416, 1-21 (2012).

[31] Turner, R.D.R., Warne, M.S., Dawes, L.A., Thompson, K., Will, G.D. Greywater irrigation as a source of organic micro-pollutants to shallow groundwater and nearby surface water. Sci Total Environ. 669, 570-578 (2019).

[32] Robertson, J. & Stevens, K. Pyrrolizidine alkaloids. Nat. Prod. Rep. 31, 1721-1788 (2014).

[33] Rosemann, G.M., Botha, C. J., Eloff, J.N. Distinguishing between toxic and non-toxic pyrrolizidine alkaloids and quantification by liquid chromatography–mass spectrometry. Phytochem. Lett. 8 126-131 (2014).

[34] European Medicines Agency. Public statement on the use of herbal medicinal products containing toxic, unsaturated pyrrolizidine alkaloids (PAs). (EMA Committee on Herbal Medicinal Products, London, 2014).

[35] Prakash, A.S., Pereira, T.N., Reilly, P.E., Seawright, A.A. Pyrrolizidine alkaloids in human diet. Mutat Res. 433 53-67 (1999).

[36] Bolechova, M., Caslavsky, J., Pospichalova, M., Kosubova, P. UPLC-MS/MS method for determination of selected pyrrolizidine alkaloids in feed. Food Chem. 170, 265-270 (2015).

[37] van Egmond, H.P. Natural toxins: risks, regulations and the analytical situation in Europe. Anal Bioanal Chem. 378, 1152-1160 (2004).

183

[38] Hoogenboom, L.A.P., Mulder, P.P.J., Zeilmaker, M.J., van den Top, H.J., Remmelink, G.J., Brandon, E.F.A., Klijnstra, M., Meijer, G.A.L., Schothorst, R., Van Egmond, H.P. Carry-over of pyrrolizidine alkaloids from feed to milk in dairy cows. Food Addit Contam Part A Chem Anal Control Expo Risk Assess. 28, 359-372 (2011).

[39] Mattocks, A.R. Chemistry and toxicology of pyrrolizidine alkaloids (Academic Pr, 1986).

[40] Molyneux, R.J., Gardner D.L., Colegate S.M., Edgar J.A. Pyrrolizidine alkaloid toxicity in livestock: a paradigm for human poisoning? Food Addit Contam Part A Chem Anal Control Expo Risk Assess. 28, 293-307 (2011).

[41] Ebmeyer, J., Braeuning, A., Glatt, H., These, A., Hessel-Pras, S., Lampen, A. Human CYP3A4- mediated toxification of the pyrrolizidine alkaloid lasiocarpine. Food Chem Toxicol. 130, 79-88 (2019).

[42] He, X., Xia, Q., Wu, Q., Tolleson, W.H., Lin, G., Fu, P.P. Primary and secondary pyrrolic metabolites of pyrrolizidine alkaloids form DNA adducts in human A549 cells. Toxicol In Vitro. 54, 286-294 (2019).

[43] Chen, T., Mei, N., Fu, P.P. Genotoxicity of pyrrolizidine alkaloids. J Appl Toxicol. 30, 183-196 (2010).

[44] Huxtable, R.J. Activation and pumonary toxicity of pyrrolizidine alkaloids. Pharmacol. Therapeut. 47, 371–389 (1990).

[45] McLean, E.K., The toxic actions of pyrrolizidine senecio-D alkaloids. Pharmacol Ther. 47, 371- 389 (1990).

[46] Mori, H., Kawai, K., Ohbayashi, F., Bunai, Y., Yamada, K., Hirono, I. Some toxic properties of a carcinogenic pyrrolizidine alkaloid, petasitenine. J Toxicol Sci. 9, 143-149 (1984).

[47] European Food Safety Authority. Scientific Opinion on Pyrrolizidine alkaloids in food and feed. EFSA journal. 9, (2011).

[48] Wang, Y.P., Yan, J., Fu, P.P., Chou, M.W. Human liver microsomal reduction of pyrrolizidine alkaloid N-oxides to form the corresponding carcinogenic parent alkaloid. Toxicol Lett., 155, 411- 420 (2005).

[49] Chou, M.W., Wang, Y.P., Yan, J., Yang, Y.C., Beger, R.D., Williams, L.D., Doerge, D.R., Fu, P.P. Riddelliine N-oxide is a phytochemical and mammalian metabolite with genotoxic activity that is comparable to the parent pyrrolizidine alkaloid riddelliine. Toxicol Lett. 145, 239-247 (2003).

184

[50] Xia, Q., Chou, M.W., Kadlubar, F.F., Chan, P.C., Fu, P.P. Human liver microsomal metabolism and DNA adduct formation of the tumorigenic pyrrolizidine alkaloid, riddelliine. Chem Res Toxicol. 16, 66-73 (2003).

[51] Moreira, R., Pereira, D.M., Valentao P., Andrade, P. Pyrrolizidine Alkaloids: Chemistry, Pharmacology, Toxicology and Food Safety. Int J Mol Sci. 19, 1668 (2018).

[52] International Programme on Chemical Safety. Pyrrolizidine Alkaloids. Environmental Health Criteria 80. (WHO, Geneva, 1988).

[53] Smith, L.W. & Culvenor, C.C. Plant sources of hepatotoxic pyrrolizidine alkaloids. J Nat Prod. 44, 129-152 (1981).

[54] Edgar, J.A., Roeder, E., Molyneux, R.J. Honey from plants containing pyrrolizidine alkaloids: a potential threat to health. J Agric Food Chem. 50, 2719-2730 (2002).

[55] Kempf, M., Reinhard, A., Beuerle, T. Pyrrolizidine alkaloids (PAs) in honey and pollen-legal regulation of PA levels in food and animal feed required. Mol Nutr Food Res. 54, 158-168 (2010).

[56] Dubecke, A., Beckh, G., Lullmann, C. Pyrrolizidine alkaloids in honey and bee pollen. Food Addit Contam Part A Chem Anal Control Expo Risk Assess. 28, 348-358 (2011).

[57] Hama, J.R. & Strobel B.W. Pyrrolizidine alkaloids quantified in soil and water using UPLC- MS/MS. RSC Adv. 9, 30350-30357 (2019).

[58] Ruan, J., Yang, M., Fu, P., Ye, Y., Lin, G. Metabolic activation of pyrrolizidine alkaloids: insights into the structural and enzymatic basis. Chem Res Toxicol. 27, 1030-1039 (2014).

[59] Ellegaard-Jensen, L., Horemans, B., Raes, B., Aamand, J., Hansen, L.H. Groundwater contamination with 2,6-dichlorobenzamide (BAM) and perspectives for its microbial removal. Appl Microbiol Biotechnol. 101, 5235-5245 (2017).

[60] Stuart, M., Lapwortha, D., Cranea, E., Hart, A. Review of risk from potential emerging contaminants in UK groundwater. Sci Total Environ. 416, 1-21 (2012).

[61] Lapworth, D.J., Baran, N., Stuart, M.E., Ward, R.S. Emerging organic contaminants in groundwater: A review of sources, fate and occurrence. Environmental Pollution. 163, 287-303 (2012).

[62] Directive 2006/118/EC of the European Parliament and of the Council of 12 December 2006 on the protection of groundwater against pollution and deterioration. https://eur- lex.europa.eu/legal-content/EN/TXT/?uri=celex:32006L0118. [Accessed 20 January 2020].

[63] Bucheli, T.D. Phytotoxins: environmental micropollutants of concern? Environ Sci Technol. 48, 13027-13033 (2014).

185

[64] Aichinger, G., Pantazi, F., Marko, D. Combinatory estrogenic effects of bisphenol A in mixtures with alternariol and zearalenone in human endometrial cells. Toxicol. Lett. 319, 242-249 (2020).

[65] Gunthardt, B.F., Hollender J., Hungerbuhler K., Scheringer M., Bucheli T.D. Comprehensive Toxic Plants−Phytotoxins Database and Its Application in Assessing Aquatic Micropollution Potential. J. Agric. Food Chem. 66, 7577-7588 (2018).

[66] Global Biodiversity Information Facility. https://www.gbif.org/species/9490132. [Accessed 06 February 2020].

186

Supplementary material Invasive plant contaminates stream and seepage water in groundwater wells with hepatotoxic pyrrolizidine alkaloids

Vaidotas Kisielius, Jawameer R. Hama, Natasa Skrbic, Hans Christian Bruun Hansen, Bjarne W. Strobel, Lars Holm Rasmussen

Table 1. Average concentrations of type 2 PAs detected in water bodies ± SDs (in alphabetical order). The empty values represent no detection.

Stream water (ng/l) Seepage water (ng/l) Sampling event 1 2 4 6 7 3 3 5 5 7 7 Sampling location Surface water monitoring site G2 G3 G2 G3 G2 G3 Compound

other sites: up to 4 ng/l

trace

trace Echimidine

9 ± 9

oxide

Echimidine N

other sites: up to 13 ng/l

56 ± 4 4 ± 1 3 ± 2

12 Erucifoline

other sites: trace

trace

Europine

other sites: trace

oxide

Europine N

other sites: up to 10 ng\l

10 ± trace trace 9 ± 7

5 Heliotrine

187

Stream water (ng/l) Seepage water (ng/l) Sampling event 1 2 4 6 7 3 3 5 5 7 7 Sampling location Surface water monitoring site G2 G3 G2 G3 G2 G3 Compound

other sites: trace

Intermedine

other sites: up to 14 ng\l

oxide

Intermedine N

other sites: up to 24 ng\l

trace trace

9 ± 3 Lasiocarpine

other sites: up to 9 ng\l

108 2 ± 1 9 4 ± 2

± 13 Retrorsine

other sites: trace

46 ± trace

oxide

Retrorsine N

other sites: up to 9 ng\l

27 ± 5 ± 1 8 ± 1 trace trace

3 Seneciphylline

Table 2. The dates of sampling events. The marks indicate taken samples.

Sampling Date Taken samples (x) event (2019) Plant Stream water Well water 1 June 14 x x (base flow) 2 June 16 x (intense flow) 3 July 3 x (no water) x 4 August 16 x (no flow) 5 September 12 x (no flow) x 6 October 1 x x (base flow) 7 October 8 x (base flow) x

188

About the author Vaidotas was born in 1987 and raised in Vilnius, the capital of Lithuania. In childhood he intensively practiced music but eventually settled at Vilnius Gediminas Technical University (VGTU). With BSc in Environmental Engineering he obtained a specialty of Water Supply and Management. Shortly before graduation he received a scholarship from Wetsus, European Centre of Excellence for Sustainable Water Technology, for a 2 year MSc program at Wageningen University (Netherlands). There he obtained MSc in Biotechnology. At the end of the MSc studies Vaidotas returned to Lithuania for a position at Ministry of Environment. During a 6 month period in 2013 of the rotating Lithuanian Presidency to the Council of the European Union (Brussels) he acted there as a deputy chair of the Working Party on the Environment. After the 4.5 year period as a civil servant, shortly before expiration of an Early Stage Researcher status, he applied and was offered a PhD position at University of Copenhagen with an employment at University College Copenhagen. This thesis was written in close collaboration together with both institutions. Vaidotas is married to a fellow student from VGTU. The family currently lives in Copenhagen and raises two charming daughters Monika and Kotryna.

Sampling location No. 10. Goathland, UK. 06.2018 (Journal of Chromatography B 1138 (2020) 121966) Photo credit: Lars Holm Rasmusse189n