<<

Research Collection

Doctoral Thesis

Assessment of the risk potential of reactive chemicals with multiple modes of toxic action

Author(s): Harder, Angela

Publication Date: 2002

Permanent Link: https://doi.org/10.3929/ethz-a-004522771

Rights / License: In Copyright - Non-Commercial Use Permitted

This page was generated automatically upon download from the ETH Zurich Research Collection. For more information please consult the Terms of use.

ETH Library

Assessment of the Risk Potential of Reactive Chemicals with Multiple Modes of Toxic Action

Angela Harder

Diss. ETH No.14966

Diss. ETH No. 14966

Assessment of the Risk Potential of Reactive Chemicals with Multiple Modes of Toxic Action

A dissertation submitted to the Swiss Federal Institute of Technology Zurich for the degree of Doctor of Natural Sciences

presented by Angela Harder

Dipl. Biotechnol., Technical University Braunschweig born on July 20, 1968 Citizen of the Federal Republic of Germany

accepted on the recommendations of Prof. Dr. René P. Schwarzenbach, examiner Prof. Dr. Joop L.M. Hermens, co-examiner Dr. Paolo Landini, co-examiner Dr. Beate I. Escher, co-examiner

Zurich, 2002

iii

Table of Contents

Summary...... vii Zusammenfassung...... ix

1 General Introduction...... 1 1.1 Electrophiles and their Industrial Significance ...... 1 1.2 Emissions and Exposure...... 2 1.3 Human Health Risks ...... 3 1.4 Environmental Risk Assessment ...... 3 1.5 Molecular Based Concepts for Ecotoxicological Effect Assessment...... 5 1.6 Reactions with Nucleophiles ...... 6 1.7 Objectives and Structure of this Thesis ...... 8 1.8 Literature Cited...... 9

2 Evaluation of Bacterial Biosensors for Toxicity Assessment and Mode of Action Classification of Reactive Chemicals ... 13 2.1 Introduction...... 14 2.2 Design of Bacterial Biosensor Set ...... 15 2.3 Experimental Section ...... 18 2.3.1 Chemicals ...... 18 2.3.2 Quantification of Growth Inhibition...... 19 2.3.3 Determination of Glutathione Influence on Growth Inhibition ...... 20 2.3.4 Determination of the Influence of DNA Repair on Decrease of Colony-Forming Units 20 2.3.5 Induction of DNA Repair Processes ...... 21 2.3.6 Determination of Mutation Rates ...... 21 2.3.7 Determination of Cell Vitality, Glutathione Depletion, and DNA Strand Breaks ...... 21 2.4 Results and Discussion...... 23 2.4.1 Toxicity of Reactive Chemicals in E. coli and Comparability of Different Endpoints .... 23 2.4.2 Evaluation of Suitable Biosensors for Assessing the Contribution of the Reaction with Glutathione to the Toxicity in E. coli...... 25 2.4.3 Evaluation of Suitable Biosensors for Assessing the Contribution of the Reaction with DNA to the Toxicity in E. coli ...... 28 2.4.4 Mode of Action Classification...... 31 2.4.5 Transferability of Modes of Action Classification to Higher Organisms...... 32 2.5 Literature Cited...... 33

iv

3 Nucleophilic Substitution Reactions of Organochlorines and Epoxides with Biological Nucleophiles: Classification

According to the Reaction Order ...... 37 3.1 Introduction...... 38 3.2 Experimental Section ...... 40 3.2.1 Chemicals ...... 40 3.2.2 Measurement of Hydrolysis Rate Constants ...... 40 3.2.3 Measurement of Reaction Rate Constants with Glutathione...... 41 3.2.4 Measurement of Reaction Rate Constants with 2´-Deoxyguanosine...... 42 3.2.5 Derivation of (Pseudo) First-Order Hydrolysis Rate Constants...... 42 3.2.6 Derivation of Second-Order Reaction Rate Constants with Glutathione and 2’-Deoxyguanosine ...... 42 3.3 Results and Discussion...... 45

3.3.1 Classification of Reactions according to SN1 and SN2...... 45 3.3.1.1 Organochlorines...... 46 3.3.1.2 Epoxides ...... 47 3.3.2 Comparison of Experimental Rate Constants with Literature Data ...... 51 3.4 Literature Cited...... 53

4 Methods and Tools for the Prediction of the Toxicity of Electrophilic Chemicals...... 57 4.1 Introduction...... 58 4.2 Methods and Data Sets...... 59 4.2.1 Determination of Octanol-Water Partition Coefficients...... 59 4.2.2 Data Sets ...... 61 4.3 Results and Discussion...... 61 4.3.1 Prediction of Bacterial Toxicity Using Reaction Rate Constants with Biological Nucleophiles...... 61 4.3.2 Comparison of Bacterial Toxicity with Toxicity in Algae, Daphnia, and Fish...... 65 4.3.3 Hazard Assessment...... 70 4.4 Literature Cited...... 71

5 Outlook ...... 75 5.1 Literature Cited...... 77

Acknowledgments ...... 79 Curriculum Vitae ...... 81

v

List of Figures Fig. 1.1 Lethality of Electrophiles for the Fish Pimephales promelas compared to their Theoretical Baseline Toxicity and the Environmental Hazardous Concentration...... 4

Fig. 1.2 Species Sensitivity Distribution for Baseline Chemicals and the Reactive Chemicals Epichlorohydrin, Benzyl Chloride, and Acrolein ...... 5

Fig. 1.3 Time Scale of Events from Uptake of Electrophiles to Possible Toxic Effects and Continuous Scale of Hard and Soft Tendency of Nucleophiles and Electrophiles Leading to Different Proportions of Biomarkers and Toxic Effects ...... 6

Fig. 1.4 Synthesis of Glutathione and Chemical and Enzymatically Catalyzed Reactions with Oxidizing and Alkylating Agents and the Likely Subsequent Toxicological Effects for a Prokaryote...... 7

Fig. 1.5 Schematic Structure of the Thesis and Association between Results of the Chapters...... 8

Fig. 2.1 Toxicokinetic and Toxicodynamic Effects from Single Cells to Populations and Related Biosensor Set for the Evaluation of Single and Combined Toxicodynamic Effects ...... 16

Fig. 2.2 Structures of the Investigated Organochlorines, Epoxides, Acrylates, and Acrylic Compounds...... 19

Fig. 2.3 Effect of Styrene Oxide, 2-Hydroxyethyl Acrylate, and 2,3-Dichloropropene on Growth of MJF276 and MJF335 ...... 27

Fig. 2.4 Effect of Styrene Oxide, 2-Hydroxyethyl Acrylate, and 2,3-Dichloropropene on Colony Forming Units of MV1161 and MV4108 ...... 30

Fig. 3.1 Nucleophilic Attack of Glutathione to the Organochlorine Benzyl Chloride and of 2´-Deoxyguanosine to the Epoxide Styrene Oxide ...... 39

Fig. 3.2 Deduction of Substitution Mechanisms from Known Reaction Mechanism...... 45

Fig. 3.3 Dependence of Hydrolysis Rate Constants upon σ+ and Reaction Rate Constants with Glutathione upon σ ...... 47

Fig. 3.4 Rate Determining Step of SN1 reaction and SN2 Reaction of Epoxides ...... 48

Fig. 3.5 Correlation of Hydrolysis Rate Constants and Reaction Rate Constants with 2´-Deoxyguanosine to the Molecular Volume and Taft σ* Constant ...... 50

Fig. 4.1 Plot of cytotoxicity values of E. coli of Specifically Reacting Compounds versus Logarithmic Reaction Rate Constants with Glutathione and with 2´-Deoxyguanosine ...... 63

vi

Fig. 4.2 Plot of cytotoxicity values of E. coli of Unspecifically Reacting Compounds versus Logarithmic Reaction Rate Constants with Glutathione and with 2´-Deoxyguanosine ...... 64

Fig. 4.3 Linear Correlation between Growth Inhibition of E. coli and Photosystem II Inhibition of Scenedesmus vacuolatus...... 65

Fig. 4.4 Linear Correlation between Growth Inhibition of E. coli and Lethality of Daphnia magna ...... 66

Fig. 4.5 Linear Correlation between Growth Inhibition of E. coli and Lethality of the Fish Pimephales promelas and Poecilia reticulata ...... 67

Fig. 4.6 Comparison of Sensitivity of Endpoints for Electrophiles in Algae and Bacteria ...... 68

Fig. 5.1 Comparison of Expression and Translational Profiles with Mode of Toxic Action Classification by Bacterial Biosensors for Identification of the Mode of Toxic Action in Higher Organisms...... 76 List of Tables Table 2.1 Description of E. coli Strains ...... 18

Table 2.2 EC50 Values Indicating Toxicity of Different E. coli Strains, EC50 Values of GSH Depletion, and Toxicity Ratios of GSH and DNA Repair Sufficient and Deficient Strains ...... 24

Table 2.3 Growth Difference of DNA Repair Sufficient and Deficient Strain, Induction of DNA Repair, DNA Fragmentation, and Mutation Rates ...... 29

Table 3.1 Suggested Dominant Reaction Mechanism, Pseudo First-Order Hydrolysis Rate, and Reaction Rate Constant with 2´-Deoxyguanosine and Reduced Glutathione of Electrophiles Investigated in this Study ...... 44

Table 3.2 σ and σ+ Hammett Constants used for Classification of the Reaction Rates with Structure-Activity Correlations ...... 46

Table 3.3 Molecular Volume and σ* Taft Substituent Constants used for Classification of Reaction Rate with Structure-Activity Correlations ...... 49

Table 3.4 Reaction Rate Constants Reported in Literature for Some Electrophiles in Aqueous Solution...... 51

Table 4.1 Toxicity Values for E. coli, Algae, Daphnia, and Fish, Mode of Toxic Action Classification, Octanol-Water Partition Coefficients, and Reaction Rate Constants with Glutathione and 2´-Deoxyguanosine ...... 62

Table 4.2 Enhancement of Aliphatic and Aromatic Nucleophilic Substitution and Michael Addition by Glutathione S-transferases from Rat Liver ...... 69

vii

Summary

Reactive organic chemicals comprise a large number of compounds with a variety of reactive moieties. This thesis focuses on three groups of electrophilic chemicals: reactive organochlorines, epoxides, and compounds with an activated double bond. Compounds belonging to those groups are widely used as intermediates in organic synthesis or as monomers in polymer synthesis. Many of them are high production volume chemicals, produced or imported in quantities over 1000 t per year in the European Union.

The environmental hazard of reactive chemicals is caused by their excess toxicity. They are usually neither persistent nor bioaccumulative. Compared to their theoretical baseline toxicity they are as much as 10 to 10 000 times more toxic. Apart from their high acute toxicity potential, continuous exposure to low concentrations of reactive chemicals may result in chronic effects such as mutations and cancer.

Both acute and chronic toxicity of reactive chemicals are based on an initial reaction with biomolecules. Nucleophilic sites in peptides, proteins, and DNA are most susceptible target sites for electrophilic chemicals. Molecular reactions include alkylation of cysteines in peptides and proteins, and alkylation of nitrogen and oxygen groups in DNA. An important peptide containing cysteine is glutathione, which is present in millimolar concentration in nearly all organisms. The reaction with glutathione is an important detoxification mechanism, as it prevents alkylation of nucleophilic sites in other biomolecules. However, the following decrease of glutathione concentration not only exposes other nucleophilic targets, but also increases the susceptibility to other toxic effects, such as internal oxidative stress. Alkylation of nucleophilic sites in DNA is the primary reaction triggering mutagenesis and subsequent carcinogenesis. Since all described reactions are based on formation of covalent bonds, toxic effects of electrophiles are most likely to be irreversible.

Apart from the chemical reactivity the toxicity of an electrophile is determined by its target site occupation, which is modulated by uptake, distribution, and metabolism. For the reduction of influences on the target site occupation, we studied the toxic effects of a series of selected electrophiles in the prokaryote Escherichia coli. The working hypothesis of this thesis was that electrophiles react specifically with either DNA or glutathione and proteins and that the toxicity of electrophilic chemicals in E. coli is directly linked to the respective chemical reactivity with these biological nucleophiles.

Evaluated toxic effects were determined in several E. coli strains and included cytotoxicity, glutathione depletion, DNA damage, induction of DNA repair, and growth differences of two pairs of strains that varied in susceptibility towards electrophilic chemicals due to lack of glutathione viii synthesis or DNA repair systems. These pairs of strains were found to be the most suitable biosensors for detection of mechanisms that underlay the observable cytotoxicity of the tested electrophiles. The observed growth differences allowed the distinction of three reactive modes of action: DNA damage, glutathione depletion related toxicity, and unspecific reactivity. Most examined epoxides caused primarily DNA damage and all compounds with an activated double bond were found to cause toxicity in direct relation to glutathione depletion. All reactive organochlorines and some epoxides were however classified as “unspecifically reactive” because their reaction was initiated by reactions with both types of biological nucleophiles.

While electrophiles that were assigned towards glutathione depletion related toxicity might cause severe acute effects, DNA damaging chemicals have a high inherent ability to cause mutations and cancer. Chemicals with multiple modes of toxic action possess both risk potentials. For these chemicals it is likely that the mode of toxic action may change with concentration and exposure time. Thus, acute and chronic effect assessment can not be simplified by application of statistical acute-chronic ratios, but must consider different underlying mechanisms of toxicity.

In order to evaluate how close the toxicity of chemicals assigned to one reactive mode of toxic action is related to the chemical reaction with either glutathione or DNA, reaction rate constants of nucleophilic substitution with glutathione and 2´-deoxyguanosine, representing DNA, were determined. For chemicals with a specific reactive mode of toxic action, the respective reaction rate constants served well to set up mode of toxic action specific quantitative structure- activity relationships for bacteria. However reaction rate constants could not describe the toxicity of “unspecifically reactive” compounds, reacting with both important biological targets DNA and glutathione.

Since assignment to a mode of toxic action is neither stringently correlated to the electrophilic moiety nor to the magnitude of the reaction rate constant with biological nucleophiles, knowledge of the reaction rate constant is not sufficient to apply QSARs as predictive tools in risk assessment. However, a selection of four E. coli biosensors proved to be sufficient to clearly classify the reactive mode of toxic action and to provide cytotoxicity values as descriptors for prediction of acute toxicity in algae, daphnia, and fish.

Hence, we recommend to base risk and hazard assessment of reactive chemicals on the proposed set of biosensors and to use predictive models based on E. coli cytotoxicity values instead of chemical reaction rate constants, and other physico-chemical descriptors. ix

Zusammenfassung

Reaktivchemikalien umfassen eine grosse Anzahl von Substanzen mit verschiedenen reaktiven funktionellen Gruppen. Die vorliegende Arbeit befasst sich mit elektrophilen Reaktivchemikalien und konzentriert sich auf reaktive Chlorverbindungen, Epoxide und Substanzen mit endständigen aktivierten Doppelbindungen. Substanzen, die diesen Gruppen zugeordnet werden, werden häufig als Zwischenprodukte oder als Monomere für z.B. Kunststoffe verwendet. Nur wenige Elektrophile werden direkt als Produkte eingesetzt. Als Basischemikalien werden sie in der EU in grossen Mengen produziert und importiert, wobei die Jahresmengen häufig 1000 Tonnen pro Jahr übersteigen. Sie gehören damit zur Gruppe der sogenannten High Production Volume Chemicals.

Aufgrund ihrer hohen Toxizität besitzen reaktive Chemikalien ein hohes Gefährdungspotenzial für die Umwelt. Die meisten sind in der Umwelt weder persistent noch bioakkumulierend. Verglichen mit ihrer theoretischen Basistoxizität sind sie 10 bis 10i000 mal toxischer. Neben sehr starken akut-toxischen Effekten, ist jedoch auch mit chronischen Effekten, wie Mutationen und Krebs, zu rechnen.

Die akute und chronische Toxizität von reaktiven Chemikalien wird durch ihre Reaktion mit Biomolekülen initiiert, wobei Elektrophile bevorzugt mit nukleophilen Gruppen in Peptiden, Proteinen und DNA reagieren. Nukleophile Gruppen in Biomolekülen umfassen zum Beispiel die Sulfhydrylgruppe in der Aminosäure Cystein, oder Stickstoff- und Sauerstoffgruppen in DNA Basen. Das Tripeptide Glutathion ist eines der wichtigsten zellulären Peptide, weil es mit Cystein eine sehr reaktive Sulfhydrylgruppe enthält. In nahezu allen Organismen kommt Glutathion in millimolaren Konzentrationen vor und verhindert durch seine Reaktivität, dass andere Biomoleküle angegriffen werden können. Sofern die Konzentration an Glutathion jedoch unter einen kritischen Level gesunken ist, werden andere Biomoleküle nicht nur angreifbar, der Organismus oder die Zelle ist ausserdem auch empfindlicher gegenüber anderen Prozessen, die mit Glutathion in Zusammenhang stehen, wie zum Beispiel oxidativem Stress. Die Reaktion von Elektrophilen mit DNA kann zu Mutationen und möglicherweise Krebs führen. Alle beschriebenen Reaktionen beruhen auf der Bildung von kovalenten Bindungen, darum sind toxische Effekte von elektrophilen Chemikalien zumeist irreversibel.

Neben der chemischen Reaktivität ist die Toxizität eines Elektrophils dadurch bestimmt, wie viel von ihm letztlich an den Wirkort gelangt. Dies ist wiederum beeinflusst durch die Aufnahme in den Organismus, die Verteilung, und mögliche metabolische Umwandlungen. Um die Einflüsse auf die Wirkortkonzentration möglichst klein zu halten, wurden toxische Effekte von Elektrophilen in dem Prokaryonten Escherichia coli untersucht. Die Arbeitshypothese der vorliegenden Arbeit x war, dass Elektrophile in E. coli spezifisch entweder mit DNA oder Proteinen und Glutathion reagieren, und dass die Toxizität direkt mit der entsprechenden chemischen Reaktion zwischen Elektrophil und Nukleophil in Zusammenhang steht.

Die Bewertung der Toxizität in verschiedenen E. coli Stämmen beruhte auf der Bestimmung der Zytotoxizität, der Abnahme der Glutathion Konzentration, der Bestimmung von DNA Strangbrüchen und DNA Reparaturenzymen, sowie auf der Bestimmung des Wachstumsunterschiedes zwischen jeweils genetisch identischen Paaren von Stämmen, die sich jedoch entweder in ihrer Fähigkeit Glutathion zu synthetisieren oder DNA Schäden zu reparieren, voneinander unterschieden. Mit Hilfe dieser Stämmepaare war es möglich, drei verschiedene Klassen von Wirkmechanismen zu unterscheiden. Die Klassen wurde stichwortartig mit DNA Schäden, Glutathion-Abnahme verursachte Toxizität und unspezifische Reaktivität charakterisiert. Die meisten der untersuchten Epoxide erzeugten DNA Schäden, wohingegen Substanzen mit endständigen Doppelbindungen aufgrund ihrer Reaktion mit Glutathion und nachfolgenden Effekten toxisch wirkten. Reaktive chlororganische Verbindungen und einige Epoxide wurden als unspezifisch reaktiv eingestuft, da ihre Toxizität durch die Reaktionen mit beiden Biomolekülen initiiert wurde.

Während Elektrophile, die aufgrund ihrer Reaktion mit Glutathion toxisch wirken, sehr schwere akute Effekte bewirken können, ist für DNA schädigende Substanzen das Risiko gegeben, Mutationen und Krebs auszulösen. Die Substanzen, die sowohl mit Glutathion als auch mit DNA reagieren können, weisen beide Gefährdungspotenziale auf, und je nach Konzentration und Expositionszeit wird der eine oder andere Wirkmechanismus entscheidend sein. Von daher kann die Bewertung möglicher akuter und chronischer Wirkungen nicht darauf reduziert werden, statistische Verhältnisse zwischen akuter und chronischer Toxizität anzuwenden, sondern muss darauf beruhen, herauszufinden, welcher Wirkmechanismus für die beobachtete Toxizität verantwortlich ist.

Um zu beurteilen, ob die Toxizität von spezifisch reaktiven Chemikalien mit der entsprechenden chemischen Reaktion zwischen Elektrophil und Glutathion bzw. DNA zusammenhängt, wurden entsprechende Reaktionsgeschwindigkeitskonstanten mit Glutathion und 2´-Desoxyguanosin, stellvertretend für DNA, bestimmt. Für spezifisch reaktive Substanzen konnten mit entsprechenden Konstanten quantitative Struktur-Aktivitätsbeziehungen (QSAR) zur Toxizität abgeleitet werden. Dies war jedoch für unspezifisch reaktive Substanzen nicht möglich.

Da die Zuordnung von Elektrophilen zu einer Klasse von Wirkmechanismen nicht zwingend mit der reaktiven funktionellen Gruppe des Elektrophils verknüpft ist und auch die Höhe der entsprechenden Reaktionsgeschwindigkeitskonstanten nicht direkt mit dem Wirkmechanismus korreliert ist, ist es für die Anwendung von QSARs als Vorhersageinstrument für die Toxizität von reaktiven Chemikalien nicht ausreichend, die Reaktionsgeschwindigkeitskonstante zu kennen. xi

Mit den hier vorgestellten vier E. coli Biosensoren, war es jedoch gleichzeitig möglich, den Wirkmechanismus zu erkennen und Zytotoxizitätswerte zu bestimmen, von denen aus eine Vorhersage der Toxizität in Algen, Wasserflöhen und Fischen möglich war.

Ausgehend von diesen Resultaten empfehlen wir, die Bewertung des Gefährdungspotenzials und des Risikos von Reaktivchemikalien eher auf der Zuordnung des Wirkmechanismus mit Hilfe des vorgeschlagenen mikrobiellen Testsytems und den experimentell bestimmten mikrobiellen Zytotoxizitätswerten beruhen zu lassen, als auf der Anwendung von chemischen Reaktionsgeschwindigkeitskonstanten oder anderen physiko-chemischen Deskriptoren.

xii

1 General Introduction

Economic development in Europe has been driven to a considerable extent by the progress and innovation of the chemical industry. Today the European Union accounts for about 30% of the world production of chemicals. The development has led to an ever increasing number and quantity of chemicals. Presently about 100 000 chemicals are traded commercially in the European Union (EU) (Stanners and Bourdeau 1995). However, toxicological and ecotoxicological information about the enormous number of chemicals is rare. A minimum set of toxicological data is available for only about 25% of the imported or produced chemicals, and until 1997 for only 0.4% a risk assessment had been conducted (Gee 1998).

The slow process of acquiring information required for risk assessment of chemicals calls for fast, reliable experimental test systems and simple predictive tools, like quantitative structure- activity relationships (QSARs). Prior to the application of QSARs in effect assessment, a classification according to modes of toxic action is necessary. A fast screening based on simple test systems and mode of action classification for identification of hazards, would allow focusing on environmentally problematic chemicals. A fast identification of the most hazardous chemicals permits to set up priorities for further testing, and gives reasons for precautionary actions to reduce probable risks, e.g., by restriction of the use of certain chemicals (EC 2001; Gee 1998). The White Paper on the Strategy for a future Chemicals Policy of the Commission of the European Communities (EC 2001) therefore explicitly states the importance to promote the development and evaluation of quantitative structure-activity relationships and in-vitro test systems that can be used for hazard and effect assessment of chemicals.

This thesis will present the development of an in-vitro test system for the evaluation of hazards and mode of action classification of reactive chemicals. The general introduction illustrates the industrial importance of reactive chemicals, their use pattern, and subsequent emission scenarios. Risks for human health are presented and problems of a classical environmental risk assessment for reactive chemicals are discussed. The conceptual foundation of an alternative hazard and effect assessment for reactive chemicals based on an in-vitro test system is presented.

1.1 Electrophiles and their Industrial Significance Reactive organic chemicals comprise a large number of compounds with a variety of reactive moieties. This thesis focuses on electrophiles, i.e. those reactive chemicals that possess sites with electron deficiencies, and thus are attacked by nucleophilic chemicals. Electrophiles may be either grouped by their reactive moiety, e.g. aldehydes, cyanates, epoxides, or by their reaction mechanism with nucleophiles. An excellent overview of electrophilic moieties and reaction 2 mechanisms was given by Hermens (Hermens 1990). In this thesis chemical and ecotoxicological aspects of three groups of electrophiles with different reactive moieties were examined: epoxides, reactive organochlorines (either allylic chlorides or benzylic chlorides), and compounds with an activated double bond, e.g., acrylates. The first two groups react according to nucleophilic substitution, the latter one according to Michael addition. For this introduction one compound of each group is chosen as example: the epoxide epichlorohydrin, the organochlorine benzyl chloride, and acrolein, representing compounds with an activated double bond. All three chemicals are high production volume chemicals, which are produced or imported in quantities over 1000 t/a in the European Union (European Chemicals Bureau 2002). Typical for most reactive compounds, including the chosen examples, is their use as intermediates or as monomers for polymer synthesis. Most of the produced epichlorohydrin is reacted with bisphenolvA forming epoxy resins; a second important use is the oxidation to synthetic glycerin. Minor applications are the production of additives in paper and textile industry and the production of flocculants for wastewater treatment (BUA 1992). The major proportion of benzyl chloride is used for production of benzyl alcohol and synthesis of benzyl phthalates, e.g., butyl benzyl phthalate. To a lesser extent benzyl chloride is used for the production of textile chemicals, emulsifiers, and disinfectants (BUA 1996). More than 90% of acrolein is used directly for the production of acrylic acids and acrylonitrile, which are further processed to polymers. Acrolein is also used for production of the amino acid methionine and for synthesis of glutardialdehyde. A direct application of acrolein in the EU is not known. In some countries, e.g., in the US, Canada, and Australia, it is used as an effective biocide in irrigation canals, public waterways and cooling water tanks (BUA 1995).

1.2 Emissions and Exposure Due to the application pattern of reactive chemicals exposure is mostly restricted to the production site, diffusive losses from unpolymerized monomers, or accidental releases, e.g., resulting after transport accidents. The problem of accidental loss, although occurring rarely, should not be underestimated. Many basic chemicals, including monomers or important intermediates, are transported over long distances to industrial sites specialized in production of, e.g., special adsorbers or textile aids. On Swiss railroads 10 million tons of hazardous materials are transported annually, representing a quarter of all transported goods in Switzerland. 25% of those hazardous materials are basic chemicals (Stoll 2000). Whereas exposure to chemicals on work sites or exposure to trace amounts of chemicals in drinking water, underlie regulatory restrictions, accidental releases can not be avoided. Transport accidents reported for epichlorohydrin include release of 2.5 tons from a freighter in the river Elbe near Brunsbuettel, Germany, in 1989, and release of 30 to 40 tons after a train collision in Bad Muender, Germany in 2002 (Hamburger Abendblatt 2002; Stern 2002). The train collision caused a fire and a subsequent explosion and besides epichlorohydrin the combustion products hydrogen chloride 3 and phosgene caused intoxication symptoms in about 600 people (Tagesschau 2002). The derailment of a train transporting epichlorohydrin in the central station of Lausanne, Switzerland, in 1994, would have caused even worse consequences if derailed tanks had started leaking. Like epichlorohydrin, acrolein can form explosive vapors, as demonstrated by a vast explosion of a production site in Taft, USA, in 1982, that initiated the evacuation of 20 000 people (UNEP 2002).

1.3 Human Health Risks Regular human exposure to reactive chemicals is mostly restricted to industrial production sites and biomarker studies in industrial hygiene monitor the risks and effects of occupational exposure. Epichlorohydrin was found to cause significantly elevated levels of chromosomal aberrations in lymphocytes of exposed workers (BUA 1992). Retrospective studies showed elevated levels of lung cancer and central nervous system neoplasms (Giri 1997), but as the analysis of living habits such as smoking was not examined, results were ambiguous. Workers that were in contact with benzyl chloride reported irritation of the respiratory tract. Retrospective studies indicate a higher incidence of cancer of respiratory organs (BUA 1996). For acrolein no effect studies in industry are reported. The greatest risk to exposure to acrolein is posed by cigarette smoke, automobile exhaust emissions, and ingestive uptake caused by thermal destruction of PVC food wraps (Nunoshiba and Yamamoto 1999). Based on regulatory classification (EC 1992) epichlorohydrin is categorized as a possible carcinogen and as a chemical that causes burns and sensitization. Benzyl chloride is categorized as a possible carcinogen and irritating chemical; acrolein causes burns.

1.4 Environmental Risk Assessment Hazards of reactive chemicals are often underestimated. Compounds are classified as dangerous for the environment if either toxicity endpoints of algae (72 h), daphnia (48 h), or fish (96 h) are below 1 mg/l or the octanol-water partition coefficient is higher than 1000 and compounds are not readily degradable (EC 1992). Reactive compounds are often not recognized as dangerous to the environment, despite their inherent toxicity. This underestimation of hazards originates from two reasons. Reactive compounds with low octanol-water partition coefficient may be up to 4 orders of magnitude more toxic than their theoretical baseline toxicity (Figure 1.1), which reflects partitioning into membranes and disturbance of their integrity. For these compounds a limit of 1 mg/l is obviously much too high. The second problem arises from using toxicity data that are determined in static experiments. Because reactive compounds usually have rather high hydrolysis rates (Chapter 3), the nominal concentration in toxicity experiments is much higher than the actual concentration. Thus, it is possible that when taking hydrolysis into account, epichlorohydrin and benzyl chloride would similar to acrolein, be classified as dangerous to the environment (Figure 1.1). 4

Figure 1.1: Lethality of -1 electrophiles for the fish baseline toxicity Pimephales promelas (96 h (M) -2 LC50) compared to their theoretical baseline toxicity -3 (Veith et al. 1983) and the epichlorohydrin environmental hazardous -4 concentration of 1 mg/l. benzyl chloride Results from static experi- -5 ments (EPA 2002) are de- Pimephales promelas acrolein 50 50 picted by open circles, -6 results from dynamic experi- log LC log 1 mg/l limit ments (BUA 1995) are symbolized by closed -7 squares. -0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0

log KOW

The present standard environmental risk assessment procedure for the assessment of chemicals in the EU (EC 1996) is particularly unsuited for reactive chemicals. The environmental risk is expressed by risk quotients derived from the predicted environmental concentrations (PEC) and predicted no effect concentrations (PNEC), for which no adverse environmental effects are expected. The derivation of PEC is based on predicted emissions resulting from typical production, use, and disposal of a chemical; the extrapolated concentrations are therefore steady state concentrations. PNEC values are most often extrapolated from single organism tests taking safety factors for extrapolation into account. For reactive chemicals the risk quotient (PEC/PNEC) is usually smaller than one, indicating no environmental risk. However, this procedure neglects that the predominant exposure scenario for reactive chemicals is the accidental release (Harder 2000), and the safety factor of 1000, applied to extrapolate from single organism tests to probable effects on ecosystems, may be insufficient to account for large differences of organism susceptibility as outlined below.

Toxicity values for different organisms of epichlorohydrin, benzyl chloride, and acrolein (BUA 1995; EPA 2002) were normalized to their theoretical baseline toxicity. The normal distribution of these toxic ratios (TR) represents the species sensitivity distribution of the electrophiles (Figure 1.2). For comparison the species sensitivity distribution of typical narcotic chemicals is shown. Two characteristics are striking: The broad species sensitivity distribution of a given electrophile itself spanning several orders of magnitude and the range of TR-values between different electrophilic chemicals. Because of the broad species sensitivity distribution, extrapolations from single organism tests, may underestimate the differences of susceptibility that should be accounted for by safety factors. 5

Figure 1.2: Species baseline toxicants sensitivity distribution for baseline chemi- cals and the reactive reactive chemicals chemicals epichloro- BCl hydrin (EPI) benzyl

distribution EPI ACR chloride (BCl), and acrolein (ACR). species sensitivity -101234567 EC (baseline) log TR = log 50 EC50 (experimental)

1.5 Molecular Based Concepts for Ecotoxicological Effect Assessment Explanations for the large range of toxicity for different reactive chemicals and the species sensitivity distribution can only be found if the underlying mechanism of toxicity is understood. The initial molecular interactions of chemicals with the biological target sites membranes, proteins and peptides, as well as DNA, are supposed to trigger a series of events which result in an observable characteristic set of physiological and behavioral signs, called mode of toxic action (Escher and Hermens 2002). Initial molecular interactions include van der Waals interactions, polarizability, hydrogen bonding forces, ionic interactions and formation of covalent bonds (McKinney 1996). A systematic cataloguing of principally possible modes of toxic action based on target sites and mechanism was given by Escher and Hermens (2002).

Several predictive concepts have been developed for the assignment of a chemical to specific modes of toxic action. Depending on the complexity of the methods and tests used, these concepts were able to discriminate up to six different modes of toxic action. Based on the absence of certain functional groups, chemicals were assigned to baseline toxicity (Verhaar et al. 1992). A more detailed classification system is based on the combination of behavioral, physiological, and biochemical responses of fish (McKim et al. 1987). Among the modes of action characterized were: baseline toxicity, electrophilic/proelectrophilic reactivity, uncoupling of oxidative phosphorylation, acetylcholinesterase inhibition, respiratory inhibition, and central nervous system seizure. The relation of structural fragments of tested chemicals with those modes of action was used to set up a computer-based expert system for prediction of modes of toxic action (Russom et al. 1997). Using test-sets of chemicals with known mode of toxic action and a set of in-vivo and in-vitro tests, definite patterns of response were found (Nendza et al. 1995; Wenzel et al. 1997). Chemicals were classified towards baseline toxicity, uncoupling of oxidative phosphorylation, inhibition of photosynthesis, inhibition of acetylcholinesterase, and towards the group of reactive chemicals. Application of the latter concepts gave the result that certain chemicals could not be classified according to one mode of toxic action but rather follow multiple modes of toxic action. 6

While baseline toxicity is caused by partitioning of chemicals into membranes, which in turn is determined by molecular van der Waals interactions and hydrogen bonding forces, the molecular mechanism of electrophilic chemicals triggering toxicity is their covalent reaction with nucleophilic target sites in peptides and proteins and in DNA. The toxicity of reactive chemicals is supposed to be both dependent on chemical reactivity of the electrophile and the target site occupation (Escher and Hermens 2002). The concept is supported by results of toxicity studies with fish. Toxicity of epoxides (Deneer et al. 1988) could be described in terms of chemical reactivity and hydrophobicity, for organochlorines (Hermens et al. 1985) a term of chemical reactivity was sufficient. A critical target occupation was observed for the toxicity of acrylates reacting with the nucleophilic tripeptide glutathione (Freidig et al. 1999).

1.6 Reactions with Nucleophiles Proteins, peptides and DNA are the susceptible cellular target sites for electrophiles. Reactions with both types of biomolecules could lead to severe toxicological effects. It was postulated that electrophiles react rather unspecifically with nucleophilic sites in biomolecules (Hermens 1990). However examples from human monitoring studies showed that electrophiles might have preferred nucleophilic targets, and DNA or peptides and proteins react to a different extent with electrophiles (van Welie et al. 1992). Based on the concept of hard acids and bases, introduced by Edwards and Pearson (1962) it was postulated that hard electrophiles (acids) tend to react with the hard nucleophile (base) DNA whereas soft electrophiles (acids) tend to react with soft nucleophiles (bases) like the tripeptide glutathione (GSH) and amino acids in proteins (Figure 1.3). Therefore, for this study, a set of compounds including soft electrophiles with an activated double bond and hard electrophiles like epoxides was chosen.

Figure 1.3: Time scale of events from uptake of cytosol target site biomarker effect electrophiles to possible electrophiles like “hard acid“ “hard base”: toxic effects (left to organic cations, DNA adduct carcinogenicity DNA right) and continuous epoxides mutagenicity scale of hard and soft tendency of nucleo- philes and electrophiles leading to different pro- ??? ? portions of biomarkers and toxic effects “soft base”: electrophiles toxicity (adapted from van protein and protein and like acrylates, related to Welie et al. (1992)). “soft acid“ GSH GSH adduct quinones GSH depletion

Reaction with either peptides and proteins or with DNA leads to different toxicological effects. Conjugation to glutathione is one of the most important detoxification pathways (Timbrell 2000). Concentrations of glutathione in cells of different species range between 1 and 10 mM (Ohlenschläger 1991). Possible effects of the reaction with glutathione are depicted in Figure 1.4. 7

Decrease of cellular concentrations of GSH is either caused by oxidation to a glutathione dimer (GSSG) or alkylation to a glutathione conjugate (GSR). Oxidation to GSSG may be caused by reactive radical species or reactive oxygen species (Carmel-Harel and Storz 2000; Comporti et al. 1991; Ketterer 1998), whereas electrophiles, illustrated by an organohalide RX, form a substituted glutathione (GSR) (Ketterer 1982), which may either be transformed by higher organisms to mercapturic acids (van Welie et al. 1992), or exported out of the cell, which is likely for prokaryotes (Ishikawa 1992). At that point of time, where the new synthesis of GSH can not back up the oxidation or alkylation anymore, the decreased GSH content leads to diminished defense against oxidative stress and the destruction of ion homeostasis [K+ for bacteria (Ferguson et al. 1995) and Ca2+ (Reed 1990) for higher organisms] and redox homeostasis (Carmel-Harel and Storz 2000), the final effect being cell lysis.

Export pump Diminished defense against internal reactive oxygen species: lipid peroxida- tion, increased membrane permeability ATP ADP +Pi GSSG Thioltrans- ferase Feedback-Inhibition RH R* L-Glutamate γ-Glutamylcysteine- Glutathione- synthetase synthetase GSH-Reductase L-γ-glutamyl- L-cysteine GSH Decreased protection of L-Cysteine L-Glycin RX HX protein-thiols, missing reduction of protein- disulfides: loss of impor- RX tant enzymatic functions + + K Glutathione S- K + Likely export Increased permeability: transferase + K GSR loss of K homeostasis HX out of cell ATP ADP +Pi and proton gradient activated by GSH adducts K+channel Export pump

Figure 1.4: Synthesis of GSH and chemical and enzymatically catalyzed reactions with oxidizing and alkylating agents and the likely subsequent toxicological effects for a prokaryote.

DNA provides different nucleophilic targets sites. Endocyclic and exocyclic oxygens and nitrogens of DNA bases as well as the phosphodiester bridge are possible reaction partners for electrophiles. Alkylation of the phosphodiester bridge may result in instability and following strand breaks (Kolman et al. 1997). Alkylation of DNA bases however may cause mutations and potentially subsequent carcinogenesis (Uziel et al. 1992). Prior to the impairing manifestation of the alkylation, DNA adducts might be recognized and repaired. Cells possess an array of different 8

DNA repair enzymes, which are best classified by their type of repair: reversal of DNA damage, base excision repair, and nucleotide excision repair (Friedberg et al. 1995).

1.7 Objectives and Structure of this Thesis In light of the enormous number of chemicals, including a high number of reactive chemicals (Katritzky 2001; Nendza 2000; Russom et al. 1997; Verhaar et al. 1994), for which appropriate toxicological information is missing, it was the aim of this thesis to contribute to the development of simple and reliable methods for the hazard assessment and mode of action classification for reactive chemicals. The explicit objectives of this thesis were:

• Development of methods for identification of the molecular mechanisms of reactive chemicals that are responsible for observed cytotoxicity.

• Evaluation if toxicity of reactive chemicals based on specific molecular mechanisms can be described in terms of chemical reactivity with the nucleophiles glutathione and DNA.

Figure 1.5 gives an overview of the structure of the thesis.

Figure 1.5: Schematic structure Chapter 2 Chapter 3 of the thesis and association between results of the chapters. Biological Effects Chemical Reactivity

cytotoxicity reaction mechanism ? GSH depletion related reactivity towards toxicity, DNA damage DNA and GSH

understanding of suitable molecular underlying mechanism descriptors? Chapter 4 Prediction of Toxicity and Hazard Assessment

Chapter 5 Outlook

In Chapter 2 the cytotoxicity of epoxides, reactive organochlorines, and compounds with an activated double bond is evaluated. Several in-vitro test systems were tested for their suitability to identify the molecular mechanism(s) of reactive chemicals responsible for overall cytotoxicity. Biosensors eligible for classification towards the relevant mode of toxic action are suggested.

In Chapter 3, reaction mechanisms of epoxides and a series of reactive organochlorines with water, 2´-deoxyguanosine, and reduced glutathione were characterized in respect of reaction mechanism, and rate constant. 9

The assessment of the ecotoxicological risk potential of examined electrophiles is made in Chapter 4. The suitability of different tools for prediction of toxic effects of reactive chemicals is critically evaluated.

Chapter 5 gives an outlook how biosensors for electrophilic chemicals can be applied in risk assessment and proposes a strategy for identification of mode of toxic action in higher organisms.

1.8 Literature Cited BUA (1992). Epichlorhydrin. VCH Verlagsgesellschaft mbH, Weinheim, Germany. BUA (1995). Acrolein. S. Hirzel Wissenschaftliche Verlagsgesellschaft, Stuttgart, Germany. BUA (1996). Benzylchlorid. S. Hirzel Wissenschaftliche Verlagsgesellschaft, Stuttgart, Germany. Carmel-Harel, O., and Storz, G. (2000). Roles of the glutathione- and thioredoxin-dependent reduction systems in the Escherichia coli and Saccharomyces cervisiae responses to oxidative stress. Annu. Rev. Microbiol. 54, 439-461. Comporti, M., Maellaro, E., Del Bello, B., and Casini, A. F. (1991). Glutathione depletion: its effects on other antioxidant systems and hepatocellular damage. Xenobiotica 21, 1067-1076. Deneer, J. W., Sinnige, T. L., Seinen, W., and Hermens, J. L. M. (1988). A quantitative structure- activity relationship of some epoxy compounds to the guppy. Aquat. Tox. 13, 195-204. EC (1992). Council Directive 92/32/EEC of 30 April 1992 amending for the 7th time Directive 67/548/EEC on approximation of the laws, regulations and administrative provisions relating to the classification, packaging and labeling of dangerous substances. Off. J. Eur. Communities, L154/1. EC (1996). Technical Guidance Document in Support of Commission Directive 93/67/EEC on Risk Assessment for New Notified Substances and Commission Regulation (EC) No 1488/94 on Risk Assessment for Existing Substances. Office for Official Publications of the European Communities, Luxembourg. EC (2001). White Paper on the Strategy for a Future Chemicals Policy, COM (2001) 88 final. Commission of the European Communities, Brussels, Belgium. Edwards, J. O., and Pearson, R. G. (1962). The factors determining nucleophilic reactivities. J. Am. Chem. Soc. 84, 16-24. EPA (2002). http://www.epa.gov/ecotox/, accessed on 09/26/2002. Escher, B. I., and Hermens, J. L. M. (2002). Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201-4217. European Chemicals Bureau (2002). http://ecb.jrc.it/existing_chemicals, accessed on 10/30/2002. Ferguson, G. P., McLaggan, D., and Booth, I. R. (1995). Potassium channel activation by glutathione S-conjugates in Escherichia coli: protection against methylglyoxal is mediated by cytoplasmic acidification. Mol. Microb. 17, 1025-1033. Freidig, A., Verhaar, H. J. M., and Hermens, J. L. M. (1999). Comparing the potency of chemicals with multiple modes of toxic action in aquatic toxicity: acute toxicity due to narcosis versus reactive toxicity of acrylic compounds. Environ. Sci. Technol. 33, 3038-3043. Friedberg, E. C., Walker, G. C., and Siede, W. (1995). DNA repair and mutagenesis. ASM Press, Washington DC, USA. Gee, D., ed. (1998). Chemicals in the European Environment: Low Doses, High Stakes? European Environmental Agency, United Nations Environmental Programme, Copenhagen, Denmark. 10

Giri, A. K. (1997). Genetic toxicology of epichlorohydrin: A review. Mutation Res. Rev. 386, 25-38. Hamburger Abendblatt. (2002). Giftunfall: Die Angst vor Folgen. 10/09/2002. Harder, A. (2000). Vergleich deterministischer und probabilistischer Risikobewertung reaktiver Chemikalien. Nachdiplomkurs Risiko und Sicherheit. Eidg. Technische Hochschule (ETH), Zürich, Switzerland. Hermens, J. L. M. (1990). Electrophiles and acute toxicity to fish. Environ. Health Persp. 87, 219- 225. Hermens, J. L. M., Busser, F., Leeuwach, P., and Musch, A. (1985). Quantitative correlations studies between the acute lethal toxicity of 15 organic halides to the guppy (Poecilla reticulata) and chemical reactivity towards 4-nitrobenzylpyridine. Toxicol. Environ. Chem. 9, 219-236. Ishikawa, T. (1992). The ATP-dependent glutathione S-conjugate export pump. Trends Biochem. Sci. 17, 463-468. Katritzky, A. R., Tatham, D.B., Maran, U. (2001). Theoretical descriptors for the correlation of aquatic toxicity of environmental pollutants by quantitative structure-toxicity relationships. J. Chem. Inf. Comput. Sci. 41, 1162-1176. Ketterer, B. (1982). The role of nonenzymatic reactions of glutathione in xenobiotic metabolism. Drug Metabolism Reviews 13, 161-187. Ketterer, B. (1998). Glutathione S-transferases and prevention of cellular free radical damage. Free Radical Research 28, 647-658. Kolman, A., Spivak, I., Naslund, M., Dusinska, M., and Cedervall, B. (1997). Propylene oxide and epichlorohydrin induce DNA strand breaks in human diploid fibroblasts. Environ. Mol. Mutagen. 30, 40-46. McKim, J. M., Bradbury, S. P., and Niemi, G. J. (1987). Fish acute toxicity syndromes and their use in the QSAR approach to hazard assessment. Environ. Health Persp. 71, 171-186. McKinney, J. D. (1996). Reactivity parameters in structure-activity relationship-based risk assessment of chemicals. Environ. Health Persp. 104, 810-816. Nendza, M., Wenzel, A., and Wienen, G. (1995). Classification of contaminants by mode of action based on in vitro assays. SAR and QSAR in Environmental Research 4, 39-50. Nendza, M. M., M. (2000). Discriminating toxicant classes by mode of action: 2. physico-chemical descriptors. Quant. Struct.-Act. Relat. 19, 581-598. Nunoshiba, T., and Yamamoto, K. (1999). Role of glutathione on acrolein-induced cytotoxicity and mutagenicity in Escherichia coli. Mutat. Res. 442, 1-8. Ohlenschläger, G. (1991). Das Glutathionsystem: Ordnungs- und informationserhaltende Grundregulation lebender Systeme. Verlag für Medizin, Dr. Ewald Fischer GmbH, Heidelberg, Germany. Reed, D. J. (1990). Glutathione−toxicological implications. Annu. Rev. Pharmacol. Toxicol. 30, 603-631. Russom, C. L., Bradbury, S. P., Broderius, S. J., Hammermeister, D. E., and Drummond, R. A. (1997). Predicting modes of toxic action from chemical structure: Acute toxicity in the fathead minnow (Pimephales promelas). Environ. Toxicol. Chem. 16, 948-967. Stanners, D., and Bourdeau, P., eds. (1995). Europe´s Environment - The Dobris Assessment. Office for Official Publications of the European Communities, Luxembourg. Stern (2002). Giftgaswolke nach Kollision zweier Züge. http://www.stern.de/politik/news/tagesthema/artikel/?id=293015, accessed on 09/26/2002. 11

Stoll, H.-P. (2000). Risikoanalyse Bahntransport. Nachdiplomkurs Risiko und Sicherheit, oral presentation, Eidg. Techn. Hochschule (ETH) Zürich, Switzerland, 3/22/2000. Tagesschau (2002). Bad Muender: Bahn ignorierte angeblich Bremsprobleme. http://www.tagesschau.de/aktuell/meldungen/0,2044,OID1127296_TYP4,00.html, accessed on 09/26/2002. Timbrell, J. (2000). Principles of Biochemical Toxicology. Taylor & Francis Ltd., London, Great Britain. UNEP (2002). UNEP Disasters Database. http://www.unepie.org/pc/apell/disasters/lists/disastercat.html, accessed on 10/31/2002. Uziel, M., Munro, N. B., Katz, D. S., Vo Dinh, T., Zeighami, E. A., Waters, M. D., and Griffith, J. D. (1992). DNA adduct formation by 12 chemicals with populations potentially suitable for molecular epidemiological studies. Mutat. Res. 277, 35-90. van Welie, R. T., van Dijck, R. G., Vermeulen, N. P., and van Sittert, N. J. (1992). Mercapturic acids, protein adducts, and DNA adducts as biomarkers of electrophilic chemicals. Crit. Rev. Tox. 22, 271-306. Veith, G. D., Call, D. J., and Brooke, L. T. (1983). Structure toxicity relationships for the fathead minnow (Pimephales promelas) - narcotic industrial chemicals. Can. J. Fish Aquat. Sci. 40, 743- 748. Verhaar, H. J. M., van Leeuwen, C. J., Bol, J., and Hermens, J. L. M. (1994). Application of QSARs in risk management of existing chemicals. SAR and QSAR in Environ. Res. 2, 39-58. Verhaar, H. J. M., van Leeuwen, C. J., and Hermens, J. L. M. (1992). Classifying environmental pollutants. 1: Structure-activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471-491. Wenzel, A., Nendza, M., Hartmann, P., and Kanne, R. (1997). Testbattery for the assessment of aquatic toxicity. Chemosphere 35, 307-322.

12

2 Evaluation of Bacterial Biosensors for Toxicity Assessment and Mode of Toxic Action Classification of Reactive Chemicals

Abstract

The toxicity of electrophiles, including reactive organochlorines, epoxides, and compounds with an activated double bond was investigated. A set of different biosensors of genetically modified Escherichia coli strains was set up to quantify cytotoxicity and specific reactivity towards the important biological nucleophiles DNA and glutathione (GSH). The significance of GSH for detoxification was assessed by cellular GSH depletion as well as by growth inhibition of a GSH deficient strain. Tests for DNA damage comprised the measurement of induction of DNA repair systems, DNA fragmentation, and growth inhibition of a strain deficient in major DNA repair mechanisms. The most suitable biosensors for detection of mechanisms that underlie the observable cytotoxicity of the tested electrophiles were two sets of strains either lacking GSH or DNA repair in combination with their corresponding parent strains. Comparison of toxicity observed in those strains suggests three distinguishable modes of toxic action for electrophilic chemicals: “DNA damage”, “GSH depletion related toxicity”, and “unspecific reactivity”. The class of chemicals causing “DNA damage” includes the epoxides 1,2-epoxybutane, (2,3-epoxypropyl) benzene, and styrene oxide. The class of chemicals with “GSH depletion related toxicity” includes compounds with an activated double bond, like acrylates and acrolein. All reactive organochlorines and some epoxides were classified as “unspecifically reactive”, because their toxicity is initiated by reactions with both biological nucleophiles. The work presented in this chapter is a contribution for an alternative hazard and effect assessment of organic pollutants based on mode of toxic action classification.

14

2.1 Introduction

Predictive ecotoxicological risk assessment of chemicals relies on the correct assignment of a chemical towards the relevant mode(s) of toxic action. A widely used approach for rough classification of chemicals into groups of different modes of toxic action is the approach of Verhaar et al., who assigned chemicals to one of four general classes based on the presence of certain functional groups (Verhaar et al. 1992). Chemicals were divided in narcotics, polar narcotics, reactive chemicals, and specifically acting chemicals with receptor mediated toxicity. More detailed classification systems, based on behavioral and physiological responses of fish (McKim et al. 1987; Russom et al. 1997) or on a set of in-vitro tests (Wenzel et al. 1997), discriminate different types of narcotics, different types of specifically acting compounds, and reactive chemicals, which are described as either electrophilic/proelectrophilic or SH-alkylating compounds, respectively. Out of 225 arbitrarily chosen industrial organic chemicals more than 70% were classified as narcotics, and as much as 20% were classified as reactive organic chemicals (Russom et al. 1997), indicating the industrial relevance of this chemical group.

Reactive chemicals comprise substances with a large number of different reactive moieties, e.g., epoxides, isocyanates, aldehydes, or acrylates (Verhaar et al. 1992). Many of these chemicals are used as intermediates in chemical synthesis or as monomers for polymerization. The economic significance is both expressed by their number and their production volume: a large number of reactive chemicals are high production volume chemicals with a yearly production volume over 1000 t (European Chemicals Bureau 2002a). The production, the transport, and the use of reactive chemicals lead to a continuous emission, and particularly the transport process provides a risk for accidental releases of high amounts.

Compared to baseline toxicity, i.e. the toxicity that compounds would (theoretically) have if they act only by narcosis, reactive chemicals are 10 to 10000 times more toxic (see Figure 1.2 in Chapter 1). In the case of electrophiles, the enhanced toxicity of reactive chemicals is presumably caused by reactions with biological nucleophiles – such as proteins and DNA – either directly or after metabolic activation (Hermens 1990). Consequences of those initial molecular target interactions are toxic effects such as cell death or mutagenesis. Nevertheless, not all electrophiles react with both, DNA and proteins/peptides, to the same extent, e.g., for ethyl acrylate no DNA adducts were found (McCarthy et al. 1994). A tendency to react preferentially with certain biological nucleophiles was observed in human health studies of occupational exposure towards reactive organic chemicals (van Welie et al. 1992). In these studies different products derived from exposure to reactive chemicals were used as biomarkers to detect human health risks. Typical biomarkers were mercapturic acids resulting from reactions with glutathione (GSH), adducts with amino acids in the main serum proteins hemoglobin and albumin, and DNA 15 adducts, which result from alkylation of DNA bases. Van Welie et al. (1992) found that exposure to ethylene oxide and 1,2-dibromoethane results in ratios of 1:10 and 1:107 of DNA to GSH adducts, respectively. Based the hard and soft acid and base (HSAB) concept, first introduced by Edwards and Pearson (1962), they proposed that proportions between GSH and DNA adducts are related to the hardness of the electrophiles. Accordingly, hard electrophiles, e.g., ethylene oxide, have a higher tendency to react with the hard nucleophile DNA than soft electrophiles like 1,2-dibromoethane, and consequently the proportion of GSH adducts is higher for soft electrophiles than for hard electrophiles.

The aim of the study presented in this chapter was to evaluate if the toxicity of electrophiles is determined by reaction with both GSH and DNA or specific reaction with either GSH or DNA. With respect to the HSAB concept a set of reactive chemicals with different reactive moieties, covering the spectrum from soft acrylates to hard epoxides was chosen. Another important goal of this work was to find biosensors that can be used for reliable identification of reactive chemicals, in addition to a set of biosensors indicating other toxic mechanisms (Escher et al. 1997). Combined with the prediction of internal effect concentrations, this mechanism-based mode of toxic action classification would facilitate the process of ecotoxicological risk assessment and set it on a more scientifically sound basis (Escher and Hermens 2002).

2.2 Design of Bacterial Biosensor Set

The biosensor set that was tested for the evaluation of reactive chemicals allows the determination of effects step after step, after the initial molecular interaction between the electrophile and cellular nucleophiles (Figure 2.1). Included are the direct measurement of target interaction, the determination of biochemical responses for compensation of deleterious effects, the determination of physiological effects, and quantification of population effects. Comparison of the growth of reference strains to their mutants, either lacking GSH or major DNA repair mechanisms, offers the opportunity to evaluate the relevance of the target interaction and the biochemical response on population level.

A detailed description of the genetic particularities of the used strains is given in Table 2.1. Characteristics of the genotype that are crucial for the purposes of this investigation are detailed below.

16

Exposure to chemical

Biotransformation e.g., by dehalogenases, glutathione S-transferases, epoxide hydrolases

Target interaction e.g., with glutathione, Isolation of glutathione and structural proteins, key DNA fragments from DPD2794 enzymes, membranes, DNA, RNA

Biochemical response Determination of induction e.g., production of stress proteins, of DNA repair systems induction of repair and with PQ37 and MV3766 defense mechanisms

Physiological effects Determination of mutation and deficient (MJF335) strains of glutathione sufficient (MJF276) e.g., cell lysis, mutation rates of CC102 Determination of growth difference

Determination of growth Population effects inhibition of CC102 and deficient (MV4108) strains of DNA repair sufficient (MV1161) e.g., density, alteration Determination of reduction Determination of growth difference of genetic structure of light output of DPD2794

Figure 2.1: Toxicokinetic and toxicodynamic effects from single cells to populations (left hand side) and related biosensor set (right hand side) for the evaluation of single and combined toxicodynamic effects.

The E. coli strain CC102 was used to measure growth inhibition as a general toxicity marker and mutagenicity. The detection of mutagenicity is based on the reversion of a point mutation in the gene of ß-galactosidase (lacZ), which results from the transition from guanosine to adenine. Compared to strains bearing other point mutations in the lacZ gene, strain CC102 was shown to have the highest sensitivity against alkylating chemicals (Cupples and Miller 1989), however CC102 does not display any deficiency of DNA repair enzymes.

The (de)toxifying effects of GSH on population growth were evaluated using the strains MJF276 and MJF335. These two strains differ in the capability to produce GSH, which is synthesized in a two-step process. In the first step γ-glutamylcysteine synthetase combines glutamate with cysteine; in the second step GSH synthetase links glycine (Penninckx and Elskens 1993) to the dipeptide to produce GSH. While mutants that are deficient in GSH synthetase (gshB) still possess a (de)toxifying dipeptide, γ-glutamycysteine synthetase mutants (gshA) totally lack the capability of (de)toxification with cysteine containing peptides. MJF335 is a gshA mutant. Both strains, MJF276 and MJF335, are additionally deficient in two potassium ion channels, which are regulated by GSH adducts (Evans et al. 2000). The efflux of potassium, and thereby destabilization of cytoplasmatic pH, was avoided using a medium with high potassium concentration (see Experimental Section).

The strains MV1161 and MV4108 were used to evaluate the importance of DNA as a target for reactive chemicals and the role played by DNA repair counteracting their toxicity. Alkylative damage of DNA can be repaired by several DNA repair pathways, which either recognize specific 17

DNA adducts or have a broad adduct specificity. The repair processes include specific transfer of small adducts, which is typical for the adaptive response (Ada), excision of specific alkylated bases by glycosylases (e.g., TagA, AlkA), and the unspecific excision of nucleotide sequences around alkylated bases, which is typical for the SOS response (e.g., RecA, UvrA). MV4108 contains a number of mutations in described DNA repair enzymes and is therefore extremely sensitive towards DNA damaging chemicals. An extensive overview of DNA repair mechanisms in prokaryotes can be found in Friedberg et al. (1995).

PQ37 and MV3766 were used to detect the induction of DNA repair processes. The measurement of the induction of DNA repair processes in these strains is based on the induction of the lacZ reporter gene, encoding the enzyme ß-galactosidase. In PQ37 the lacZ gene is coupled to the promoter of the sfi gene belonging to the SOS response; in MV3766 lacZ is coupled to the promoter of the ada gene belonging to the adaptive response. Thus, induction of lacZ in PQ37 is triggered by single strand breaks in DNA (for details of SOS response see, e.g., Koch and Woodgate (1998)), while in MV3766 lacZ is induced by alkylated phosphodiesters in DNA (for details of adaptive response see, e.g., Volkert (1988)).

For detection of direct target interaction either with GSH or DNA the E. coli strain DPD2794 was used. DPD2794 carries a 17.55 kbp multicopy plasmid in which the lux-genes of Vibrio fischeri are coupled to the promoter of the recA-gene (Vollmer et al. 1997). The product of the recA-gene is the control element of SOS response in bacteria. RecA binds to single-stranded DNA, which may result as a consequence of direct chemical damage to DNA or from DNA repair activities. Upon binding to single-stranded DNA, RecA is activated and triggers the proteolysis of the LexA repressor. This in turn enables the expression of an array of proteins participating in DNA repair processes. Because of the high copy number of this plasmid, the number of LexA molecules present in the cell is insufficient to bind all recA promoters and to repress its expression. Thus, even without DNA damage, high background levels of luciferase were observed (for details of the mechanism see Rupani et al. (1996)). Taking advantage of this phenomenon and using the reduction of light emission, reflecting a reduced energy state, an unspecific toxicity endpoint for cell vitality was defined. Severe DNA damage by reactive chemicals was determined by plasmid DNA fragmentation. GSH depletion was determined by quantification of isolated cellular GSH.

18

Table 2.1: Description of E. coli strains. strain chromosomal and episomal genotype reference or source

- - CC102 ara ∆(lac proB), F with lacI , lacZ (Cupples and Miller 1989) MJF276 kdpABC thi rha lacI lacZ kup KefB KefC::Tn10; F- (Evans et al. 2000) MJF335 as MJF276, plus gshA::Tn10(Kanr) (Evans et al. 2000) MV1161 argE his ∆(gpt-proA) leu ara galK lacY mtl xyl thi rpsL (Volkert et al. 1986) supE tsx rfa; F- MV4108 as MV1161, plus ∆(srlC-recA)::Tn10 uvrA ∆(ada- gift from M. Volkert, MIT, alkB::Camr) alkA tagA zhb::Tn5 thr rfb mgl kdgK Boston, MA, USA MV3766 as MV1161, plus lacZ::Tetr alkB::Tn10(lacZ Camr) P. Landini, EAWAG, Duebendorf, Switzerland PQ37 sfi::mud(Ap lac) cts lac∆U169 uvrA galE galY phoC rfa (Quilliardet and Hofnung thr leu his pyrD thi trp::Muc srl::Tn10; F- 1985) DPD2794 galK lac rpsL; pUCD615 (Ampr, Kanr) with (Vollmer et al. 1997) recA´::luxCDABE, F-

Abbreviations used for description of the genotype are explained, e.g., on web page http://www.ecocyc.com (Encyclopedia of Escherichia coli Genes and Metabolism, ECOCYC 2002).

2.3 Experimental Section

2.3.1 Chemicals The set of electrophiles (for structures see Figure 2.2) comprised benzyl chloride (BCl, CAS 100-44-7), 3-methylbenzyl chloride (3MBCl, CAS 620-19-9), 4-nitrobenzyl chloride (NBCl, CAS 100-14-1), 2,3-dichloro-1-propene (DClP, CAS 78-88-6), trans-1,4-dichloro-2-butene (DClB, CAS 110-57-6), styrene oxide (SOX, CAS 96-09-3), (2,3-epoxypropyl) benzene (EPOX, CAS 4436-24- 2), 2-(4-nitro-phenyl)-oxirane (NOX, CAS 6388-74-5), 1,2-epoxybutane (EOX, CAS 106-88-7), epichlorohydrin (EPI, CAS 106-89-8), (1S, 2S)-(–)-1-phenylpropylene oxide (PPOX, CAS 4518- 66-5), 2-methyl-2-vinyloxirane (MVIN, CAS 1838-94-4), acrolein (ACR, CAS 107-02-8), ethyl acrylate (EA, CAS 140-88-5), 2-hydroxyethyl acrylate (HEA, CAS 818-61-1), isobutyl acrylate (IBA, CAS 106-63-8), acrylonitrile (ACN, CAS 107-13-1) and acrylamide (ACA, CAS 79-06-1). BCl, NBCl, EPI, EOX, ACR, EA, HEA, IBA, and ACA were purchased from Fluka Chemie AG, Buchs, Switzerland. 3MBCl, DClP, and SOX were obtained from Sigma-Aldrich Chemie AG, Steinheim, Germany. DClB, NOX, EPOX, PPOX, and MVIN were bought from Aldrich Chem. Co. Inc, Milwaukee, USA. ACN was bought from Riedel de Haën, Seelze, Germany. All chemicals were of highest purity available (≥95%) and used as received, ACA was received and used as a 40% aqueous solution. 19

reactive Figure 2.2: Structures of the investigated Cl organo- Cl organochlorines, epoxides, acrylates, and chlorines Cl acrylic compounds. R DClP BCl: R = H 3MBCl: R = m-CH Cl 3 Cl NBCl: R = p-NO2 DClB

epoxides O O R

SOX: R = phenyl PPOX EPOX: R = benzyl O NOX: R = p-nitrophenyl

EOX: R = C2H5 EPI: R = CH2Cl MVIN activated O double bonds N ACN R

ACR: R = H O

EA: R = O-C2H5 HEA: R = O-C2H4-OH H2N ACA IBA: R = O-sec-C4H9

2.3.2 Quantification of Growth Inhibition All experiments with strain CC102 were performed in minimal medium (MM, pH 7.0) which consisted of 33 mM KH2PO4, 60 mM K2HPO4, 7.6 mM (NH4)2SO4, 1.7 mM Na3-citrate, 1 mM

MgSO4, 0.1 ‰ (w/v) vitamin B1, and 11 mM glucose as a sole carbon source. Cells were grown aerobically at 30°C on a shaking incubator to early exponential phase (cell densities between 7·107 and 1·108 cells/ml). Thereafter geometrical dilution series of most chemicals were directly made with cell suspension in glass tubes closed with teflon-coated screw-caps. The solid electrophiles NBCl and NOX were dissolved in cyclohexane and different quantities of stock solution were pipetted into the glass tubes. After evaporation of cyclcohexane, crystallized NBCl and NOX were quickly redissolved with cell suspension. Different volumes of aqueous ACA solution were directly added to aliquots of cell suspension, differences of the resulting sample volume were corrected by adding water. Seven to eight treated samples and controls in duplicate were then incubated for a time t of 3 h at 30°C on a shaking incubator. The gas volume in the closed tubes was sufficient for aerobic growth during the 3 h incubation time. Experiments were performed at least in duplicate. Growth was monitored by light scattering at 600 nm (OD600) and growth related to control was calculated according to Equation 2.1.

OD600,t(sample)− OD600,t=0(control) % growth of control = ⋅ 100 (2.1) OD600,t(control) − OD600,t=0 (control)

The concentrations resulting in 50% inhibition of growth (EC50) were derived from a logistic fit of the concentration-effect curves (Equation 2.2), using the software Prism (GraphPad Software, San Diego, CA), which computed the best fit for experimental data of all parallels under the

20 prerequisites of fixed minimum at 0% and fixed maximum at 100% growth. Adjustable parameters were the slope m and the EC50. EC50 values of chemicals with high air-water partition coefficient (DClP, DClB, EA, IBA) were corrected for the loss of the chemical to the gas phase.

100 % effect = (2.2) 1+ 10m⋅(log EC 50 −log conc)

2.3.3 Determination of Glutathione Influence on Growth Inhibition The determination and calculation of growth inhibition of MJF276 and MJF335 was performed identically to CC102 (see above) but, to obtain reliable differences in growth, the incubation time with chemicals was increased to 6 hours. Pre-cultures of MJF276 contained 25 mg/l tetracycline and pre-cultures of MJF335 were grown with 25 mg/l kanamycine. The influence of GSH on growth inhibition of the used chemicals was characterized by the toxic ratio (TR) of EC50 values of

MJF276 and MJF335 (Equation 2.3), TRGSH. Confidence limits of the TR-values were derived from error propagation from the standard deviation of the EC50 values (Motulski 1995).

EC50 MJF276 TRGSH = (2. 3) EC50 MJF335

2.3.4 Determination of the Influence of DNA Repair on Decrease of Colony- Forming Units MV1161 and MV4108 at cell densities between 2 and 2.5·108 cells/ml were incubated with chemicals for 45 min, identical to the procedures described for CC102 (see above). MM was supplemented with 2.5% Luria-Bertani medium (LB) (10 g/l tryptone, 5 g/l yeast extract, and 10 g/l NaCl); pre-cultures of MV4108 were grown with 25 mg/l tetracycline. Likely due to its numerous mutations the strain MV4108 showed a tendency to form filaments when incubated with chemicals. This made it difficult to set up a linear correlation between cell number and optical density. Instead of measuring the light scattering at 600 nm it was necessary to evaluate the effect of chemicals on the growth on MV1161 and MV4108 by counting colony-forming units. After the incubation with the chemicals, cells were spun down by centrifugation. The supernatant was removed and the cells were resuspended in phosphate buffer of the same composition as the used MM. Dilution series of cells were made between 1:102 and 1:105. Aliquots of the dilution series of MV1161 were plated in duplicate on LB plates (LB medium plus 15 g/l agar), and incubated for one day at 30°C before counting. MV4108 was plated in duplicate on LB plates that contained 25 mg/l tetracycline, incubated at 37°C, and counted after two days of incubation. Experiments were performed only once. The percentage of the decrease of colony forming units of treated samples was calculated by division with the average number of colony forming units of the controls. EC50 values were derived from the concentration-response curves with Equation 2.2. The influence of DNA repair on growth inhibition of used chemicals was characterized by the toxic ratio of EC50 values of MV1161 and MV4108 (Equation 2.4), TRDNA. 21

EC50 MV1161 TRDNA = (2.4) EC50 MV4108

2.3.5 Induction of DNA Repair Processes Both PQ37 and MV3766, were grown in MM with 2.5% LB on a shaking incubator to cell densities between 7·107 and 1·108 cells/ml. Pre-cultures of PQ37 were grown with addition of 25img/l tetracycline, pre-cultures of MV3766 were supplemented with 25 mg/l chloramphenicol. The measurement of ß-galactosidase induction was performed according to the method described by Miller (1972) with the following modifications: isopropyl-ß-D-thiogalactoside was omitted from the culture and cells were lysed with 1 drop of 0.1% sodiumdodecylsulfate (SDS) and 2 drops of chloroform instead of toluene. Experiments were performed once in two series with overlapping concentration in order to find the concentration with maximal induction. Only samples with growth higher than 50% of control (Equation 2.1) were used for further analysis. As positive controls 2-methyl-2-vinyloxirane was used for the SOS response and methyl methane- sulfonate was used for the adaptive response.

2.3.6 Determination of Mutation Rates CC102 was grown in MM at 30°C on a shaking incubator to cell densities of 5 to 7·108 cells/ml. Geometrical dilution series of acrylates and acrylic compounds were directly made with cell suspension in gas tight glass tubes (procedure for ACA see above). Seven to eight treated samples and controls in duplicate were then incubated for 3 hours at 30°C on a shaking incubator. Growth was monitored by light scattering at 600 nm (OD600) and growth related to control was calculated according to Equation 2.1. After incubation, aliquots of cell suspension were diluted 1:105 in phosphate buffer of the same composition as the used MM. Aliquots of the dilution series were plated in duplicate on LB plates, incubated for one day at 37°C, and colonies formed counted thereafter. For determination of mutants 2 ml aliquots of the intoxicated samples were spun down and washed once with MM. The supernatant was removed leaving approximately 100 µl that were directly plated on MM-lac plates, containing 5.8 mM lactose instead of glucose as a carbon source. The number of revertants was counted after two days of incubation at 37°C. Experiments were performed only once. The mutation rate was determined by division of the number of revertants per ml cell culture on MM-lac plates with the number of colony forming units per ml determined on LB plates. Only samples with growth higher than 50% of control according to Equation 2.1 were used.

2.3.7 Determination of Cell Vitality, Glutathione Depletion, and DNA Strand Breaks In order to yield high amounts of DNA and GSH E. coli strain DPD2794 was grown to late exponential phase with cell densities between 6·108 and 7·108 cells/ml in MM supplemented with 25 mg/l kanamycine at 30°C on a shaking incubator. After reaching the necessary cell density,

22 cells were incubated with chemicals as described for strain CC102 for 45 min at 30°C. Thereafter the effect of the chemicals on cell vitality, GSH, and DNA was evaluated as follows. All experiments were performed at least in duplicate.

For determination of cell vitality 100 µl aliquots of cell suspension were placed on an opaque 96 wells plate, and the light output was measured for 5 seconds at 30°C for each well in a luminometer (MicroLumat LB96P, Berthold GmbH & Co KG, Bad Wildbad, Germany). The percentage of light output of a sample was derived by division with the average light output of controls. The EC50 values based on % light output of control were computed with Prism analogously to Equation 2.2, using only data between 0 and 100%.

For GSH isolation aliquots of 1.5 ml cell suspension were spun down. The supernatant was removed and the cells were resuspended in 200 µl SDS solution (2.5%, 60 mg/l Na2-EDTA) to lyse the cells for 30 s under vigorous shaking on a vortex. Addition of 200 µl of 5% trichloroacetic acid (TCA) and further vortexing for 5 min caused precipitation of proteins. The precipitated protein was spun down and 100 µl of supernatant was diluted with 500 µl water and stored at -20°C until HPLC analysis, which is described in Chapter 3. In difference to the there described method, the GSH stock solution was prepared with 0.2% SDS and 0.4% TCA. The percentage of the GSH content of a sample was determined by division with the average GSH content of controls. The EC50 values based on % GSH content of control were computed with Prism analogously to Equation 2.2, using only data between 0 and 100%.

DNA smaller than 20 kbp was isolated from 1.5 ml cell suspension using a plasmid miniprep kit (Genomed GmbH, Bad Oeyenhausen, Germany). The method employed a modified alkaline/SDS method to prepare a clear lysate of E. coli cells. After neutralization, the lysate was directly applied onto a DNA adsorbing matrix. With ethanolic washing buffer RNA, proteins, and other impurities were removed, and the purified DNA was eluted from the matrix with 10 mM tris(hydroxymethyl)-aminomethane / HCl of pH 8.0. The samples were kept at -20°C until electrophoresis, which was performed in a 1% agarose gel at 5 V/cm. Stained with 0.3 mg/l ethidium bromide the DNA was visualized on a UV-transilluminator, photographed, and printed (BioPhotoronics Corporation Gel Print 2000 I). DNA fragments of plasmid and genomic DNA appeared as a stained smear of fragments smaller than 20 kbp. 23

2.4 Results and Discussion

2.4.1 Toxicity of Reactive Chemicals in E. coli and Comparability of Different Endpoints Acute toxicity in bacteria is rather undefined since, unlike for higher organisms, cell death is not distinguishable from the inhibition of the ability to reproduce. There are a number of possible endpoints that have been used to measure toxicity in bacteria: growth inhibition, reduced colony forming units, cell morphology, enzyme leakage, or reduced reductive capacity. Here, the relationship between the following toxicity endpoints on population level was examined: Growth inhibition based on optical density measurement (for strains CC102, MJF276), reduction of colony forming units (for strain MV1161), and reduced light production (for strain DPD2794). The results are summarized in Table 2.2. As designated in the footnotes a, b, and d, all determination methods gave results that were proportional to each other. The slopes of these correlations were close to one, confirming equal relative sensitivity, which indicates the potential interchangeability of the used toxicity parameters. Differences of the intercept of toxicity correlations based on growth inhibition or colony forming units are caused by different incubation times with the electrophiles: for the same exposure concentration, effects rose with increasing incubation times. Comparing the toxicity based on colony forming units with toxicity based on reduction of light production, both for the same incubation time, the latter method proved to be much more sensitive, because it already reflects the toxic impact on the energy state prior to severe cell damage. However, the applicability of this more sensitive method is hampered by some outliers, pointing to the problems of the method: decrease of light production (SOX, NOX) might additionally be caused by inhibition of enzymes necessary for light production, and apparently high effect concentrations (EPI) might result from a very high induction of the SOS response interfering with the inhibitory effect.

Table 2.2: EC50 values indicating toxicity of different E. coli strains, EC50 values of GSH depletion, and toxicity ratios of GSH and DNA repair sufficient and deficient strains.

EC50 (mM) EC50 (mM) EC50 (mM) EC50 (mM) c EC50,GSH (mM) EC50 (mM) EC50 (mM) c e electrophile a b TR d TR class CC102 DPD2794 MJF276 MJF335 GSH DPD2794 MV1161 MV4108 DNA ACR 1.65·10-2 9.42·10-3 1.04·10-2 f 2.13·10-3 f 4.9 (2.1-11.3) 6.91·10-2 g n.d. h n.d. - GSH NBCl 0.117 1.37·10-2 6.40·10-2 1.44·10-2 4.4 (1.8-11.1) 0.279 0.58 8.5·10-2 7 (1-68) u.r. 3MBCl 0.372 0.153 0.449 0.282 1.6 (0.6-4.1) 1.02 g 0.56 4.4·10-2 13 (1-213) u.r. DClB 0.381 0.271 0.161 4.60·10-2 3.5 (1.2-10.2) no depletion 1.7 6.8·10-2 25 (7-87) u.r. BCl 0.466 0.137 0.391 0.146 2.7 (1.1-6.6) 1.56 g 1.0 0.13 8 (0-531) u.r. DClP 0.511 0.203 0.225 8.75·10-2 2.6 (1.2-5.7) 3.50 g 2.9 0.48 6 (0-111) u.r. NOX 0.545 4.94·10-2 0.234 8.56·10-2 2.7 (1.2-6.4) 0.234 2.0 i 0.16 13 (3-45) u.r. IBA 1.45 0.514 1.21 0.433 2.8 (1.1-7.4) 2.85 n.d. n.d. - GSH HEA 1.72 0.220 0.704 0.153 4.6 (1.9-11.4) 0.882 8.2 > 5.9 j < 1.4 (0-16) GSH EA 1.79 0.653 1.00 0.186 5.4 (2.4-12.1) 1.68 n.d. n.d. - GSH EPOX 2.83 0.516 1.12 1.15 1.0 (0.4-2.4) 7.18 10 i 0.34 29 (8-103) DNA PPOX 3.25 0.598 1.57 2.18 0.7 (0.4-1.4) no depletion 5.4 2.1 3 (0-22) n.c. EPI 3.41 11.0 4.79 2.03 2.4 (1.0-5.6) 16.1 34 0.36 94 (19-471) u.r. SOX 3.65 0.188 2.10 1.81 1.2 (0.6-2.2) no depletion 13 i 0.26 50 (25-105) DNA ACN 5.43 4.24 2.38 0.570 4.2 (1.9-9.4) 6.52 21 > 30 j < 0.7 (0-3) GSH MVIN 32.2 21.5 31.9 14.6 2.2 (0.9-5.4) 95.2 78 0.27 289 (34-2586) u.r. EOX 46.2 24.4 26.2 24.8 1.1 (0.4-2.9) 114.7 177 i 1.4 f 126 (17-987) DNA ACA 77.9 16.1 32.1 7.53 4.3 (1.8-9.9) 44.0 390 1146 0.3 (0-2) GSH a 2 Toxicity of CC102 and DPD2794 is linearly correlated: log EC50(DPD2794) = 1.031 · log EC50(CC102) - 0.503, r = 0.836. b 2 Toxicity of CC102 and MJF276 is linearly correlated: log EC50(MJF276) = 0.995 · log EC50(CC102) - 0.229, r = 0.964. c Toxicity ratios given with 95% confidence intervals based on the propagation of errors of EC50 values of MJF276, MJF335 and MV1161, MV4108, respectively. d 2 Toxicity of CC102 and MV1161 is linearly correlated: log EC50(MV1161) = 1.016 · log EC50(CC102) + 0.557, r = 0.925. e Mode of toxic action classification: GSH: glutathione depletion related toxicity, DNA: DNA damage, u.r.: unspecific reactivity; n.c.: not classified. f For an optimal fit for values around 50% effect, the calculation of EC50 was modified from Equation 2.2 by using maximum and minimum as additional adjustable parameters. g Depletion occurred at concentrations, at which cell lysis was observed in growth inhibition tests of either CC102 or MJF276. The potential loss due to cell lysis, however, was calculated to be significantly lower than the observed depletion. h n.d.: not determined. i Due to limited water solubility (NOX, EPOX, SOX) or very high toxic level (EOX) too few data points were experimentally obtained to apply sigmoidal fit. Data were therefore calculated with the equation given in footnote d. j Given as a borderline value is the highest tested concentration. 25

Effect concentrations of compounds with reactive organochlorines, epoxides, and compounds with an activated double bond span four orders of magnitude. There is no obvious correlation between reactive group and toxicity, as none of the group exhibits a clearly higher or lower toxicity. Nevertheless, comparing the toxicity of similar compounds with different reactive moieties, e.g., SOX with BCl or NOX with NBCl, organochlorines exhibit a clearly higher toxicity. Comparison of toxicity values within subgroups of chemicals with the same reactive moiety shows the effect of electron-withdrawing substituent. For ACA, EA and IBA, and ACR, toxicity rises with increasing aliphatic inductive effect of the : -CONH2 < -COOR < -COH (Hansch et al. 1995). However, the high inductive effect of -CN of ACN stands in contrast to its relatively low toxicity. A further example is the pair EPI/EOX, reflecting the influence of the negative inductive effect of chlorine. The parallel increase of toxicity and inductive effect might indicate a direct correlation to chemical reactivity. Freidig et al. (1999) measured the reactivity of acrylates, methacrylates, ACR, ACN, and ACA with glutathione and observed a direct correlation of toxicity of the fish Pimephales promelas (96 h LC50) with the corresponding reaction rate constants. Comparing NBCl with BCl and NOX with SOX, the nitro compounds are far more toxic. This might be accounted to change from dominant nucleophilic substitution of first order for BCl and SOX to dominant nucleophilic substitution of second order for NBCl and NOX (see Chapter 3). The high nucleophilic reactivity of NBCl might be a reason why NBCl is frequently used as standard substrate for characterization of glutathione S-transferases (Iizuka et al. 1989; Vuilleumier 1997).

2.4.2 Evaluation of Suitable Biosensors for Assessing the Contribution of the Reaction with Glutathione to the Toxicity in E. coli If the toxicity of a chemical is related to the reaction with GSH, a significant GSH depletion as well as a reasonable growth difference between GSH producing and not producing E. coli strains would be expected. So far, cellular GSH depletion has been used as an indicator for the reaction of a chemical with GSH (D`Souza et al. 1988; Freidig et al. 1999) and concurrently observed toxicity is explained by subsequent effects, such as oxidative stress (Comporti et al. 1991), the destruction of ion homeostasis (Ferguson et al. 1995; Reed 1990) and redox homeostasis (Carmel-Harel and Storz 2000).

Cytosolic GSH concentrations of DPD2794 were found to result from an interplay of induction and depletion caused by the exposure to reactive chemicals. GSH concentrations of most untreated cultures were between 10 and 15 mM, which is in the range of reported concentrations of 6 (Apontoweil and Berends 1975) and 20 mM (Loewen 1979) for other E. coli strains, grown in minimal medium to the exponential growth phase. In cultures with normal background levels of GSH between 10 and 15 mM, eight of the tested chemicals (BCl, DClP, DClB, NBCl, 3MBCl, EPI, SOX, MVIN) led to an induction of GSH (results not shown). For unknown reasons 20% of the cultures showed elevated levels of GSH that were up to four times as high as normal. In these cultures, no induction was observed, even for compounds that induced GSH in the cultures with a

26 lower GSH-level. This supports the theory that the GSH level is regulated by feedback inhibition (Penninckx and Elskens 1993).

In E. coli nearly all chemicals except DClB, PPOX, and SOX, caused GSH depletion (Table 2.2) in the tested concentration range, restricted by the water solubility of the test compounds. The depletion of GSH due to cell lysis can be excluded, as during 8 times longer incubation times (experiments with MJF276) a maximum of only 10% cell lysis was observed. Thus, some chemicals induce GSH and thereby indicate a reaction with GSH, but cause no observable depletion (DClB, SOX), and some chemicals deplete GSH without prior induction (EPOX, NOX,

EOX, ACA, ACN, EA, HEA) (data not shown). When comparing EC50 values of GSH depletion with EC50 values based on growth inhibition, GSH depletion occurs generally at concentrations higher than cell growth inhibition. Hence GSH depletion must be judged as a rather insensitive toxicity endpoint.

For most of the tested chemicals the GSH deficient strain MJF335 was more sensitive than the GSH sufficient strain MJF276 (Table 2.2). Based on the growth differences, three groups of chemicals can be deduced: chemicals with nearly identical growth curves for both strains

(0.7

(TRGSH≥1 with 95% confidence), and chemicals with growth difference of lower statistical significance (1.2

GSH had no detoxifying effect. The TRGSH for all compounds with an activated double bond was significant, indicating a detoxifying effect of GSH. Except for IBA the TRGSH for compounds with an activated double bond even exceeded the factor 4 (see Figure 2.3 b for HEA). Significant differences where also found for the epoxide NOX, and all reactive organochlorines (see Figure 2.3 c for DClP), except 3MBCl. Differences with a confidence lower than 95% were found for the epoxides EPI and MVIN. Comparing compounds with the same reactive moiety, it becomes obvious that structural similarity does not strictly result in similar toxic effects. Whereas toxicity of SOX was not influenced by GSH, NOX was detoxified by GSH. NBCl was not only more toxic than BCl but exhibited also a higher TRGSH and 3MBCl was hardly detoxified by GSH. One reason for the differences in the series of benzyl chlorides might be the difference of reaction mechanism (see Chapter 3). While BCl and 3MBCl react according to first-order nucleophilic substitution, NBCl follows second-order nucleophilic substitution.

27

100 a) Figure 2.3: Effect of a) SOX (squares), b) HEA (triangles), and c) DClP (circles) on growth of 80 MJF276 (open symbols) and MJF335 (closed symbols). Datapoints are given with the 60 standard error of the mean. 40

20

0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 log conc (mM) 100 b)

80

60

40

20

0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 log conc (mM)

100 c) 80

60

40

20

0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 log conc (mM)

Qualitatively, the results of GSH depletion correlated with growth differences between GSH sufficient and deficient strains. Two contradictions however were noticed: DClB showed a high

TRGSH but no GSH depletion, and EOX and EPOX depleted GSH but showed similar effects on both MJF276 and MJF335, implying that GSH has no detoxifying effect. The discrepancy for DClB might be explained by the concurrence of de-novo synthesis of GSH and depletion leading to a more or less balanced GSH level in the tested concentration range, which is limited by the poor water solubility of DClB . The GSH depletion of EOX and EPOX could be the result of a reduced de-novo synthesis due to decreased cell fitness caused by other toxic effects (see below discussion).

In summary, GSH depletion is not a suitable biosensor. The concurrence of GSH depletion and induction and thus the relative insensitivity of GSH depletion as well as the unspecific decrease of GSH parallel to reduced cell fitness could lead to misleading conclusions about the

28 role of GSH for cytotoxicity. A clear evidence of the detoxifying effect of GSH can however be deduced from the comparison of the growth of a GSH producing strain and a strain fully lacking GSH.

2.4.3 Evaluation of Suitable Biosensors for Assessing the Contribution of the Reaction with DNA to the Toxicity in E. coli Alkylation of DNA bases would cause a severe toxicity if remaining unrepaired. Thus most adducts cause the induction of the alkylation-specific adaptive response or the induction of the SOS response, a non-specific, error-prone, DNA repair pathway. Interaction with DNA should therefore be recognized by the induction of these repair systems and thus result in detectable growth differences between E. coli strains possessing and lacking these repair systems.

In order to screen for possible DNA interaction the measurement of the induction of the SOS response was chosen. As the induction of the adaptive response concurrently induces the SOS response (Vasil`eva et al. 1999), measuring SOS induction indirectly includes the induction of the adaptive response. Only for those compounds, which induced the SOS response a further determination of TRDNA was performed. For interpretation of induction experiments a number of recommendations can be found in the literature for unambiguous identification of inducing chemicals. These include the application of limits for a maximum fold induction of 1.3, 1.5 or 2.0, or the necessity of a concentration-response curve (Quilliardet and Hofnung 1993). Choosing a minimal limit of 1.3 fold induction, nearly all tested chemicals but the acrylates EA, IBA, and ACR, induced the SOS response (Table 2.3). Results for the induction of the SOS response for SOX, EPI and DCLP are comparable to results found in literature (Quilliardet and Hofnung 1993). EOX however was found to be negative (von der Hude et al. 1990); compared to test conditions applied here, it was tested for similar concentrations but with addition of dimethylsulfoxide. For ACN both positive (Eder et al. 1990) and negative (von der Hude et al. 1988) results were reported. Testing organochlorines and epoxides for the induction of the adaptive response, EOX and EPI gave positive results. This finding supports the theory that only small DNA adducts are effectively repaired by the adaptive response in E. coli (Volkert 1988). MVIN, although small, did not induce the adaptive response. One explanation might be the branching within the molecule that could obstruct an adduct recognition. 29

Table 2.3: Growth difference of DNA repair sufficient and deficient strain, induction of DNA repair, DNA fragmentation, and mutation rates. electrophile TR induction SOS induction adap- DNA frag- mutation ratec DNA response a tive response a mentation b

MVIN 289 11.2 (30 mM) n.o. d 50 mM n.d. e EOX 126 4.4 (20 mM) 4.4 (50 mM) 100 mM n.d. EPI 94 4.5 (5.0 mM) 34.1 (5.0 mM) 12.5 mM n.d. SOX 50 6.6 (1.5 mM) n.o. 1.25 mM n.d. EPOX 29 3.2 (1.5 mM) n.o. n.o. n.d. DClB 25 6.1 (0.25 mM) n.o. n.o. n.d. NOX 13 1.8 (0.30 mM) n.o. n.o. n.d. 3MBCl 13 2.0 (0.18 mM) n.o. n.o. n.d. BCl 8 2.2 (0.4 mM) n.o. n.o. n.d. NBCl 7 4.9 (0.06 mM) n.o. n.o. n.d. DClP 6 1.8 (0.20 mM) n.o. n.o. n.d. PPOX 3 1.5 (1.5 mM) n.o. n.o. n.d. HEA < 1.4 2.4 (1.5 mM) n.d. n.d. 1.5 (2.1 mM) ACN < 0.7 2.9 (14 mM) n.d. n.d. 1.2 (8.5 mM) ACA 0.3 2.7 (107 mM) n.d. n.d. 3.9 (83 mM) ACR n.d. 1.3 (0.06 mM) n.d. n.d. 1.0 EA n.d. n.o. n.d. n.d. 1.0 IBA n.d. n.o. n.d. n.d. 1.0 a Given is the maximal fold induction compared to the control. Dedicated in brackets is the concentration at which the maximal induction was observed. As a positive control for the adaptive response methyl methane-sulfonate was used. The induction of the positive control ranged between 70 and 100 for 3.7 mM. b Given is the lowest concentration at which fragmentation was observed, as far as no simultaneous cell lysis occurred. C Mutation rate of CC102 normalized to highest mutation rate of control (6.3/109 cells, 95% confidence interval (4.0±1.9)/109 cells). Dedicated in brackets is the concentration at which the highest mutation rate was observed. As a positive control ethyl methane-sulfonate was used. The mutation rates of the positive control ranged between 40 and 45 for 9.7 mM. d n.o.: not observed. e n.d.: not determined.

Based on the growth differences observed for the strains MV1161 and MV4108 (Table 2.2 and 2.3) all epoxides, but PPOX, significantly react with DNA resulting in a large TRDNA. Further significant differences were only observed for the organochlorine DClB; all other organochlorines caused differences with a confidence lower than 95%. For tested compounds with an activated double bond DNA repair even seemed disadvantageous. As neither the SOS induction tests nor the mutagenicity test (see below) indicates any DNA interaction for IBA, EA, and ACR, no growth difference between DNA repair sufficient and deficient strain is expected. In contrast to differences of TRGSH observed for structurally similar compounds, some kind of a pattern for

TRDNA appears: Toxicity of epoxides, except PPOX, is clearly influenced by reaction with DNA, toxicity of organochlorines might be influenced by the reaction with DNA, and toxicity of compounds with an activated double bond is likely to be independent of reaction with DNA. This pattern is exemplarily illustrated for the epoxide SOX, the organochlorine DClP, and the acrylate HEA in Figure 2.4.

30

Figure 2.4: Effect of SOX (squares), 100 HEA (triangles), and DClP (circles) on colony forming units of MV1161 (open symbols) and MV4108 (closed 10 symbols).

1

0.1

0.01

0.001 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 log conc (mM)

Comparing observed TRDNA with the SOS induction level, growth differences roughly increase with increasing induction factor. However, some compounds did not follow this correlation, indicating the problem using induction of DNA repair for mode of toxic action classification. Each chemical has its own alkylation pattern, ranging from chemicals nearly exclusively alkylating N-7 of guanine, e.g., methyl methane-sulfonate, to chemicals favoring alkylation of phosphodiester linkages, e.g., N-ethyl-N-nitrosourea (Friedberg et al. 1995). Whereas for example alkylation of O6-guanine causes misreplication because DNA polymerases bypass this lesion, N-3-adenine adducts block the normal polymerase-types and are lethal unless repaired (Friedberg et al. 1995). Although resulting in different consequences for a cell, both lesions might be recognized by the SOS response (Friedberg et al. 1995), thus leading both to an induction but only the latter lesion to potential cell death.

Only chemicals with a large TRDNA caused visible DNA fragmentation (Table 2.3). The test for DNA fragmentation is comparable to the Comet Assay, which is more and more employed for determination of DNA damage in single eukaryotic cells (Anderson et al. 1998). Both techniques enable the detection of alkali-labile sites and single strand breaks caused, e.g., by DNA repair enzymes or direct scission of chemicals of the DNA backbone (Horvathova et al. 1998). While the DNA fragmentation test was comparably insensitive, the Comet Assay applied for eukaryotic cells was judged to be a very sensitive detection method for DNA damage (Anderson et al. 1998).

HEA, ACN, and ACA showed a weakly mutagenic activity in the same concentration range as they induced the SOS response. However, results of mutagenicity studies in bacterial test systems for acrylates and other compounds with an activated double bond gave both positive and negative results (European Chemicals Bureau 2001a, b 2002b; Tsuda et al. 1993). Different results for the same test system may reflect either the presence of only a small window between concentrations causing mutagenicity and concentrations causing high cytotoxicity due to other 31 toxic mechanisms (compare further discussion) or problems of statistical significance of results for weakly mutagenic compounds. Differences between different bacterial test systems (Tsuda et al. 1993) might be the result of differences in DNA repair or metabolic (in)activation. Nevertheless, the validity of mutation rates for mode of toxic action classification seems questionable. Comparison to TRDNA suggests that the population performs better with few mutations than with the deprivation of energy to hold up DNA repair systems.

2.4.4 Mode of Toxic Action Classification The qualification of several biosensors for the assessment if and how the reaction with the biological nucleophiles GSH and DNA contributes to the toxicity observed in E. coli was evaluated. Strength and weakness of each biosensor were discussed above and contradicting results were explained. Thus, it is recommended to use combinations of GSH sufficient and deficient strains and DNA repair sufficient and deficient strains as the most reliable biosensors from the set of biosensors tested.

Using these tests systems and comparing both TRGSH and TRDNA, patterns are recognized that lead to a mode of toxic action classification of reactive chemicals (Table 2.2):

• Glutathione depletion related toxicity: Chemicals of this class are detoxified by reaction with GSH and are characterized by a significant growth difference between GSH sufficient and deficient strain and the absent growth advantage for cells capable of DNA repair. Chemicals following this pattern are exemplified by HEA (Figures 2.3b and 2.4) and include all compounds with an activated double bond.

• DNA damage: This class of chemicals, exemplified by SOX (Figures 2.3a and 2.4), for which DNA interaction is decisive for toxicity, is characterized by lacking GSH detoxification and a huge growth difference between DNA repair sufficient and deficient strains. Chemicals of this class include two further epoxides: EPOX and EOX.

• Unspecific reactivity: The toxicity of most of the chemicals is based on their reaction with both biological nucleophiles, but, as toxic ratios imply, to a very different degree. The quantification of either the effect of GSH or of DNA interaction of these chemicals with multiple modes of toxic action remains difficult, because the knowledge about toxic interactions here can only be deduced from the needlessness of a defense mechanism and no evaluation of either interaction is possible, if both defense mechanisms are needed. Chemicals with multiple reactive modes of toxic action are exemplified by DClP (Figures 2.3c and 2.4).

The reactive chemical PPOX appears to be an exception to this classification, being neither related to reaction to DNA nor to GSH. PPOX is as toxic as SOX and EPOX and one would assume that the structurally similar compound follows the same reaction pattern. In contrast to

32 the other epoxides PPOX has no terminal epoxide moiety, leading to a shift to first order substitution reactions (see Chapter 3). The mode of toxic action of PPOX remains unsolved.

From these results it can be concluded that some classes of reactive chemicals react preferentially with specific biological nucleophiles. Interestingly neither chemicals, which are detoxified by GSH, nor chemicals, for which DNA interaction causes toxicity are restricted to specific concentration ranges: DNA interacting chemicals can be as toxic or nontoxic as GSH interacting chemicals. Qualitatively the results support the HSAB concept: rather soft acrylates react with the soft GSH and rather hard epoxides with hard DNA bases.

2.4.5 Transferability of Modes of Toxic Action Classification to Higher Organisms Difficulties transferring results from mode of toxic action classification based on bacterial biosensors may arise due to differences of DNA repairing enzymes and specificities of glutathione S-transferases (GSTs). A few examples illustrate these problems.

Human health studies examining the genotoxicity of SOX in subpopulations lacking a special GST class, revealed the detoxifying effect of GSH for SOX (Shield and Sanderson 2001) contrasting results found here in E. coli. Insertion of genes encoding a specific rat GST in the bacterium Salmonella typhimurium led to an increase mutagenicity of epibromohydrin, which was suggested to be caused by the catalyzed formation of a cationic GSH-adduct (Thier et al. 1995). This cation, an episulfonium ion, is specifically reactive with DNA. The formation of episulfonium ions was also observed for dichloropropanes and -ethanes (Beaten et al. 1999). Thus a formation of an activated GSH adduct by GSTs could also be possible for the bifunctional compounds DClB and EPI. If bacterial GSTs could lead to formation of sulfur containing cations is unknown, but it might be an explanation for the higher DNA interaction of DClB compared to other reactive organochlorines here investigated.

Epoxides are typical intermediates of oxidative metabolism. While toxicity of DClP, as was shown, is hardly determined by DNA alkylation, addition of rat liver extract (S9) lead to a drastic increase of mutagenicity (Eder et al. 1982). One explanation for the higher mutagenicity was epoxidation (Boerth et al. 1991), which would yield EPI. The toxicity of allylic compounds in higher organisms thus might be determined by the stability of its epoxide intermediates.

How metabolic transformation of electrophiles in higher organisms influences the toxicity of the parent compound is unknown. This may however be easily assessed using enzyme extracts, e.g., from liver of fish or rats. The addition of vertebrate enzymes would lead to mixture of parent compound, bacterial metabolites and metabolites, specific for a chosen higher organism. Thus the same proposed biosensors could be used to examine molecular effects of toxicants that would arise in higher organisms. 33

2.5 Literature Cited

Anderson, D., Yu, T. W., and McGregor, D. B. (1998). Comet assay responses as indicators of carcinogen exposure. Mutagenesis 13, 539-555. Apontoweil, P., and Berends, W. (1975). Glutathione biosynthesis in Escherichia coli K12 - properties of enzymes and regulation. Biochim. Biophys. Acta 399, 1-9. Beaten, A., Tafazoli, M., Kirsch-Volders, M., and Geerlings, P. (1999). Use of the HSAB principle in quantitative structure-activity relationships in toxicological research application to the genotoxicity of chlorinated hydrocarbons. Int. J. Quantum Chem. 74, 315-355. Boerth, D. W., Eder, E., Rasul, G., and Morais, J. (1991). Theoretical structure-activity study of mutagenic allyl chlorides. Chem. Res. Toxicol. 4, 368-372. Carmel-Harel, O., and Storz, G. (2000). Roles of the glutathione- and thioredoxin-dependent reduction systems in the Escherichia coli and Saccharomyces cervisiae responses to oxidative stress. Annu. Rev. Microbiol. 54, 439-461. Comporti, M., Maellaro, E., Del Bello, B., and Casini, A. F. (1991). Glutathione depletion: its effects on other antioxidant systems and hepatocellular damage. Xenobiotica 21, 1067-1076. Cupples, C. G., and Miller, J. H. (1989). A set of lacZ mutations in Escherichia coli that allow rapid detection of each of the six base substitutions. Proc. Natl. Acad. Sci. USA 86, 5345-5349. D`Souza, R. W., Francis, W. R., and Anderson, M. E. (1988). Physiological model for tissue glutathione depletion and increased resynthesis after ethylene dichloride exposure. J. Pharmacol. Exp. Ther. 245, 563-568. ECOCYC (2002). http://www.ecocyc.com, accessed on 09/04/2002. Eder, E., et al. (1990). Molecular mechanisms of DNA damage initiated by a,ß-unsaturated carbonyl compounds as criteria for genotoxicity and mutagenicity. Environ. Health Persp. 88, 99- 106. Eder, E., Neudecker, T., Lutz, D., and Henschler, D. (1982). Correlation of alkylating and mutagenic activities of allyl and allylic compounds. Chem. Biol. Interact. 38, 303-315. Edwards, J. O., and Pearson, R. G. (1962). The factors determining nucleophilic reactivities. J. Am. Chem. Soc. 84, 16-24. Escher, B. I., and Hermens, J. L. M. (2002). Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201-4217. Escher, B. I., Snozzi, M., Häberli, K., and Schwarzenbach, R. P. (1997). A new method for simultaneous quantification of the uncoupling and inhibitory activity of organic pollutants in energy transducing membranes. Environ. Sci. Technol. 16, 405-414. European Chemicals Bureau (2001a). Draft European Union Risk Assessment Report Acrylonitrile. European Commission, Joint Research Centre, Ispra, Italy. European Chemicals Bureau (2001b). European Union Risk Assessment Report Acrylaldehyde. European Commission, Joint Research Centre, Ispra, Italy. European Chemicals Bureau (2002a). http://ecb.jrc.it/existing_chemicals, accessed on 10/30/2002. European Chemicals Bureau (2002b). European Union Risk Assessment Report Acrylamide. European Commission, Joint Research Centre, Ispra, Italy.

34

Evans, G. J., Ferguson, G. P., Booth, I. R., and Vuilleumier, S. (2000). Growth inhibition of Escherichia coli by dichloromethane in cells expressing dichloromethane dehalogenase/glutathione S-transferase. Microbiology 146, 2967-2975. Ferguson, G. P., McLaggan, D., and Booth, I. R. (1995). Potassium channel activation by glutathione S-conjugates in Escherichia coli: protection against methylglyoxal is mediated by cytoplasmic acidification. Mol. Microb. 17, 1025-1033. Freidig, A., Verhaar, H. J. M., and Hermens, J. L. M. (1999). Comparing the potency of chemicals with multiple modes of toxic action in aquatic toxicity: acute toxicity due to narcosis versus reactive toxicity of acrylic compounds. Environ. Sci. Technol. 33, 3038-3043. Friedberg, E. C., Walker, G. C., and Siede, W. (1995). DNA repair and mutagenesis. ASM Press, Washington DC, USA. Hansch, C., Leo, A., and Hoekman, D. (1995). Exploring QSAR - Hydrophobic, electronic, and steric constants. American Chemical Society, Washington, D.C. Hermens, J. L. M. (1990). Electrophiles and acute toxicity to fish. Environ. Health Persp. 87, 219- 225. Horvathova, E., et al. (1998). The nature and origin of DNA single strandbreaks determined with the comet assay. Mutat. Res. 409, 163-171. Iizuka, M., Inoue, Y., Murata, K., and Kimura, A. (1989). Purification and some properties of glutathione S-transferase from Escherichia coli B. J. Bacteriol. 171, 6039-6042. Koch, W. H., and Woodgate, R. (1998). The SOS response. In DNA repair in prokaryotes and lower eukaryotes, Vol.1 of DNA damage and repair, J. A. Nickoloff and M. F. Hoekstra, eds., pp. 107-134. Humana Press, Totowa, NJ. Loewen, P. C. (1979). Levels of glutathione in Escherichia coli. Can. J. Biochem. 57, 107-111. McCarthy, T. J., Hayes, E. P., Schwartz, C., and Witz, G. (1994). The reactivity of selected acrylate toward glutathione and deoxyribonucleosides in vitro: structure-activity relationships. Fundam. Appl. Toxicol. 22, 543-548. McKim, J. M., Bradbury, S. P., and Niemi, G. J. (1987). Fish acute toxicity syndromes and their use in the QSAR approach to hazard assessment. Environ. Health Persp. 71, 171-186. Miller, J. H. (1972). Experiments in molecular genetics. Cold Spring Harbor Laboratory, Cold Spring, USA. Motulski, H. J. (1995). Intuitive biostatistics. Oxford University Press, New York, USA. Penninckx, M. J., and Elskens, M. T. (1993). Metabolism and function of glutathione in micro- organisms. Adv. Microb. Physiol. 34, 239-301. Quilliardet, P., and Hofnung, M. (1985). The SOS chromotest, a colorimetric bacterial assay for genotoxines: procedures. Mutat. Res. 147, 65-78. Quilliardet, P., and Hofnung, M. (1993). The SOS chromotest: a review. Mutat. Res. 297, 235- 297. Reed, D. J. (1990). Glutathione–toxicological implications. Annu. Rev. Pharmacol. Toxicol. 30, 603-631. Rupani, S. P., et al. (1996). Characterization of the stress response of a bioluminescent biological sensor in batch and continuous cultures. Biotechn. Progr. 12, 387-392. Russom, C. L., et al. (1997). Predicting modes of toxic action from chemical structure: Acute toxicity in the fathead minnow (Pimephales promelas). Environ. Toxicol. Chem. 16, 948-967. Shield, A. J., and Sanderson, B. (2001). Role of glutathione S-transferase Mu (GSTM1) in styrene-7,8-oxide toxicity and mutagenicity. Environ. Mol. Mutagen. 37, 285-289. 35

Thier, R., et al. (1995). Enhancement of bacterial mutagenicity of bifunctional alkylating agents by expression of mammalian glutathione S-transferase. Chem. Res. Toxicol. 8, 465-472. Tsuda, H., et al. (1993). Acrylamide; induction of DNA damage, chromosomal aberrations and cell transformation without gene mutations. Mutagenesis 8, 23-29. van Welie, R. T., van Dijck, R. G., Vermeulen, N. P., and van Sittert, N. J. (1992). Mercapturic acids, protein adducts, and DNA adducts as biomarkers of electrophilic chemicals. Crit. Rev. Tox. 22, 271-306. Vasil`eva, S. V., Makhova, E. V., and Moshkovskaya, E. Y. (1999). Expression and functions of adaptive response genes in Escherichia coli treated with mono- and bifunctional alkylating agents: interference with SOS response. Russ. J. Gen. 35, 364-369. Verhaar, H. J. M., van Leeuwen, C. J., and Hermens, J. L. M. (1992). Classifying environmental pollutants. 1: Structure-activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471-491. Volkert, M. R. (1988). Adaptive response of Escherichia coli to alkylation damage. Environ. Mol. Mutagen. 11, 241-255. Volkert, M. R., Nguyen, D. C., and Beard, K. C. (1986). Escherichia coli gene induction by alkylation treatment. Genetics 112, 11-26. Vollmer, A. C., et al. (1997). Detection of DNA damage by use of Escherichia coli carrying recA`::lux, uvrA`::lux, or alkA`::lux reporter plasmids. Appl. Environ. Microbiol. 63, 2566-2571. von der Hude, W., Behm, C., Gürtler, R., and Basler, A. (1988). Evaluation of the SOS chromotests. Mutat. Res. 203, 81-94. von der Hude, W., Seelbach, A., and Basler, A. (1990). Epoxides: comparisons of the induction of SOS repair in Escherichia coli PQ37 and the bacterial mutagenicity in the Ames test. Mutat. Res. 231, 205-218. Vuilleumier, S. (1997). Bacterial glutathione S-transferases: What are they good for? J. Bacteriol. 179, 1431-1441. Wenzel, A., Nendza, M., Hartmann, P., and Kanne, R. (1997). Testbattery for the assessment of aquatic toxicity. Chemosphere 35, 307-322.

36

3 Nucleophilic Substitution Reactions of Organochlorines and Epoxides with Biological Nucleophiles: Classification According to the Reaction Order

Abstract

Nucleophilic substitution reactions of electrophiles comprising aromatic epoxides, aliphatic epoxides, benzyl chlorides and allyl chlorides with three different nucleophiles were characterized. Nucleophiles included water and the biological nucleophiles glutathione and 2´- deoxyguanosine, which are susceptible targets for chemicals with reactive mode of toxic action.

Characterization encompassed the classification according to SN1 and SN2 and the determination of rate constants. Classification was based on determination of the electrophiles and the nucleophiles concentration as a function of time, on using principles deduced from the Swain- Scott equation, and on analyzing product distributions and mechanism attributions given in literature. A further tool was the application of structure-activity relationships, not to set up quantitative relationships but to assist the classification. Results underline the importance of SN1 and SN2 classification both for abiotic transformation products formed in the environment and for the toxicity of the compounds.

38

3.1 Introduction

Small organochlorines and epoxides are widely used as intermediates in organic synthesis or as monomers in polymer synthesis. The direct use of these reactive compounds, however, is limited to few and rare applications as insecticides or rodenticides. Thus, despite their abundance and industrial importance, the application patterns restrict emissions to industrial sites, accidental releases, e.g., during transport, or diffusive losses from polymers with fractions of not polymerized monomers. Once released, the environmental fate is determined by volatilization, and by abiotic and biotic transformation reactions, particularly reactions with nucleophiles including hydrolysis (Larson and Weber 1994; Schwarzenbach et al. 2003).

The type of reaction of organochlorines and epoxides with nucleophiles is a nucleophilic substitution reaction. This reaction may either be described as a bimolecular reaction mechanism

(SN2), in which the formation of a penta-valent transition state of the central atom determines the reaction rate, or described as a two step reaction, in which the rate is determined by the formation of a carbocation (SN1), and thus is independent of the nucleophile´s presence. SN1 and SN2 are, however, only models of two extremes of nucleophilic substitution reactions and “pure” SN1 or

SN2 reactions seldom occur. Most often substitution reactions have characteristics of both, SN1 and SN2, with varying degree. Nevertheless it is useful to classify chemicals according to their dominant reaction mechanism with a given nucleophile, since the SN1 or SN2 tendency influences the types of products formed in the environment and the toxicological behavior of the electrophile.

In seawater or groundwater the concentration of other nucleophiles than water, e.g., reduced sulphur compounds or chloride, may be so high that they compete with water in SN2 reactions, and additional transformation products besides hydrolysis products are formed. This was demonstrated for the transformation of methyl iodide to methyl chloride, the formation of mercaptans from several alkyl bromides (Schwarzenbach et al. 2003), and the formation of 1,3- dichloro-2-propanol from epichlorohydrin in seawater (Krijgsheld and van der Gen 1986).

The kind and extent of toxicity of an electrophile is both determined by the reaction rate and the reaction mechanism with cellular biological nucleophiles. Important biological nucleophiles include peptides, proteins, and DNA. A crucial reaction is the alkylation of the cysteine containing tripeptide glutathione. The following decrease of cellular glutathione concentrations leads to diminished defense against oxidative stress, and destruction of ion, and redox homeostasis (Comporti et al. 1991; Reed 1990). After glutathione is depleted, cysteines and other nucleophilic sites of proteins may be alkylated, which may impair enzymatic functions. The alkylation of DNA bases or the phosphodiester bridge may result in instability and following strand breaks (Chovanec et al. 1998; Kolman et al. 1997), or cause mutations and subsequent carcinogenesis (Uziel et al. 1992). 39

Freidig et al. (1999) showed a direct correlation between fish toxicity of acrylates and their reaction rate constants with glutathione. Reaction rate constants of organochlorines (Hermens et al. 1985) with 4-(p-nitrobenzyl)-pyridine (NBP, used as a surrogate for biological nucleophiles), and reaction rate constants of epoxides with NBP (Deneer et al. 1988) together with an hydrophobicity term, could be used to model toxicity in fish. For DNA alkylating compounds the degree of SN1 reaction is postulated to influence the level of mutagenicity. DNA provides several different nucleophilic reaction sites. Most DNA adducts are formed with N-7 guanine [e.g., 88% of adducts formed by epichlorohydrin (Koskinen and Plna 2000) and 76% of styrene oxide adducts (Savela et al. 1986)]. At this site adduct formation causes depurination most often followed by correct replacement with unaltered guanine. Mutations however are caused, e.g., by alkylation of 6 6 O guanine, and the proportion of O alkylation is assumed to rise with increasing SN1 character of the electrophile (Dipple et al. 1982). Reactions of typical electrophiles with either glutathione or the DNA nucleoside guanosine are depicted in Figure 3.1.

Figure 3.1: Nucleophilic OH O attack of glutathione to the H O O N organochlorine benzyl

O : chloride (left) and of 2´- N

: NH deoxyguanosine to the SH HN epoxide styrene oxide N (right). ClCl N NH2 HO O O NH2 H H

H H OH H

OOH

While no general models exist which can assist the classification of reactions according to

SN1 or SN2 character, experimental reaction rate constants of both SN1 and SN2 were modeled successfully with different structure-activity relationships, taking steric, polar, and resonance effects of substituents in aliphatic or aromatic molecules into account. For nucleophilic substitutions these effects are of different relevance for SN1 and SN2. Whereas the formation of a carbocation for SN1 is primarily influenced by polar effects and the possibility of resonance contribution, the reaction rate of SN2 is in general significantly influenced by steric effects. SN1 reactions of aromatic electrophiles could therefore be well described by Hammett equations with σ+ constants (Hansch and Leo 1995), which express the influence of the substituents on the quality of stabilizing a positive charge. Bimolecular SN2 reactions are either described by Hammett equations using σ values of aromatic substituents or by applying the Taft equation with * the steric substituent constant Es and the polar substituent constant σ for reactions that include aliphatic substances (Hansch and Leo 1995). Beside the application of Hammett or Taft equations for prediction of reaction rate constants, the relationship of even a small series of

40 reaction rate constants to steric, polar or resonance describing constants is of valuable diagnostic assistance to discriminate SN1 and SN2 character of studied reactions (Okamoto and Brown 1956; Parker and Isaacs 1959).

The aim of this chapter is to characterize the reaction of reactive organochlorines and epoxides with water and the important biological nucleophiles glutathione and guanosine both by reaction order and rate constant, and thereby provide information required in assessing the toxicity of substances, which is discussed in Chapter 4. The results of this characterization are also used to evaluate limitations of structure-activity relationships for the description of reactivity of structurally related compounds that may react by somewhat different reaction mechanism.

3.2 Experimental Section

3.2.1 Chemicals The set of electrophiles comprised benzyl chloride (BCl, CAS 100-44-7), 3-methylbenzyl chloride (3MBCl, CAS 620-19-9), 4-nitrobenzyl chloride (NBCl, CAS 100-14-1), 2,3-dichloro-1- propene (DClP, CAS 78-88-6), trans-1,4-dichloro-2-butene (DClB, CAS 110-57-6), styrene oxide (SOX, CAS 96-09-3), 2-(4-nitro-phenyl)-oxirane (NOX, CAS 6388-74-5), (2,3-epoxypropyl) benzene (EPOX, CAS 4436-24-2), (1S, 2S)-(–)-1-phenylpropylene oxide (PPOX, CAS 4518-66- 5), 1,2-epoxybutane (EOX, CAS 106-88-7), epichlorohydrin (EPI, CAS 106-89-8), 2-methyl-2- vinyloxirane (MVIN, CAS 1838-94-4). The nucleophiles were water (de-ionized and ultra-filtrated) - 2- including 15 mM H2PO4 /HPO4 (pH 7.65), reduced L-glutathione (GSH, CAS 70-18-8) and 2´- deoxyguanosine (GUA, CAS 961-07-9). BCl, NBCl, EPI, EOX, and GSH were purchased from Fluka Chemie AG, Buchs, Switzerland. 3MBCl, DClP, SOX, and GUA were obtained from Sigma- Aldrich Chemie AG, Steinheim, Germany. DClB, NOX, EPOX, PPOX, and MVIN were bought from Aldrich Chem. Co. Inc, Milwaukee, USA. All chemicals were of highest purity available (≥i95%) and were used as received.

3.2.2 Measurement of Hydrolysis Rate Constants Hydrolysis rate constants of electrophiles were determined at 30°C in autoclaved 15 mM phosphate buffer at pH 7.65 using sterilized serum vials with crimped sterile viton rubber stoppers. Liquid electrophiles were directly dissolved in buffered water in concentrations suitable for analysis. The solid electrophiles NBCl and NOX were dissolved in cyclohexane and different quantities of stock solution were pipetted into serum vials. After evaporation of cyclohexane, crystallized NBCl and NOX could be redissolved quickly in the buffer. All experiments were performed in triplicate. 4 to 7 samples were taken with fine sterile canulas in hour to week intervals, depending on the reactivity of the compound. All samples were immediately analyzed. During the experiments the headspace in the serum vials rose. Taking the air-water partitioning coefficients of the compounds with the highest volatility, i.e. DClP and DClB (Albanese et al. 41

1987), into account the calculated loss to the headspace did not exceed 2%; therefore nominal concentrations needed not to be corrected. The time course of the disappearance of the parent compound was used to calculate hydrolysis rate constants. DClP, DClB, EOX, EPI, and MVIN were analyzed by direct GC-FID detection (GC 8000, Fision Instruments, Milano, Italy) using a Stabilwax column (30 m x 0.32 mm, 1 µm; BGB Analytik, Anwil, Switzerland) with on-column injection. BCl, 3MBCl, NBCl, SOX, NOX, EPOX, and PPOX were analyzed by HPLC-UV-VIS detection (pump M480, Gina 160 autosampler, Gynkotek, Germering, Germany; 875-UV detector, Jasco, Gross-Umstadt, Germany) using C-8 (LiChrosphere, 125 x 4 mm, 5 µm spheres; Merck, Darmstadt, Germany) and C-18 (Nucleosil, 250 x 4 mm, 5 µm spheres; Macherey-Nagel, Dueren, Germany) reversed phase columns with differing methanol-water mixtures as mobile phase.

3.2.3 Measurement of Reaction Rate Constants with Glutathione The reaction rate constants of electrophiles with GSH were determined under the same conditions as the hydrolysis rate constants. 1 mM GSH stock solutions in buffer, containing 50 µM complexating sodium-EDTA to prevent oxidation, were prepared daily. The reaction was initiated by adding an aliquot of the GSH stock solution to predissolved electrophiles. At the start of the reaction, the concentration of GSH was 100 µM and the concentration of the electrophiles ranged from 400 to 2000 µM. Immediately after adding GSH, a 0.1 ml sample was withdrawn. The reaction was stopped by acidification, adding 0.9 ml phosphate buffer (1 mM, pH 3.0). Experiments were performed at least in triplicate, and, depending on the rate of the reaction, 5 to 7 samples were taken in intervals of 45 minutes to 10 hours. The reaction was stopped as outlined before. The disappearance of GSH was used to calculate reaction rate constants. The oxidation of GSH during 74 hours was determined in control experiments without electrophiles. In the case of SOX, NBCl, and BCl, the compounds were analyzed simultaneously with HPLC-UV- VIS as described before.

GSH was determined as a fluorescent adduct with o-phthaldialdehyde based on a fluorometric assay first described by Cohn and Lyle (1966) and adapted to HPLC analysis by Fujita et al. (1993). GSH was separated from electrophiles on a reversed phase C-18 column (Discovery, 150 x 4.6 mm, 5 µm spheres; Supelco, Bellefonte, USA) using isocratic elution at 0.8 ml/min with 10% methanol / 90% aqueous phosphate buffer (1 mM, pH 3.0). The 6.7 µM solution of o-phthaldialdehyde (9% methanol, pH 12.0) was added at 0.5 ml/min post-column and the formation of the fluorescent isoindole took place at room temperature in a reaction coil placed between column and detector. The fluorescence was monitored at an emission wavelength of 420 nm after excitation at 350 nm with a fluorescence detector (RF 1002, Gynkotek, Germering, Germany).

42

3.2.4 Measurement of Reaction Rate Constants with 2´-Deoxyguanosine The reaction rate constants of electrophiles with GUA were determined under the same conditions as the hydrolysis rate constants and the reaction rate constants with GSH. 15 mM GUA stock solutions in buffer were prepared weekly and stored in the refrigerator. Adding GUA from stock solution to predissolved electrophiles started the reaction. Concentration of GUA at the reaction start was 3 to 5 µM; the concentration of the electrophiles ranged from 200 µM to 1 mM. Immediately after adding GUA, samples were withdrawn and the concentration of GUA was determined by HPLC-VIS at 252 nm after separation from electrophiles on a C-8 reversed phase column (LiChrosphere, 125 x 4 mm, 5 µm spheres; Merck, Darmstadt, Germany) using isocratic elution at 0.8 ml/min with 8% methanol / 92% phosphate buffer (50 mM, pH 5.5). Experiments were performed in triplicate and depending on speed of the reaction and the hydrolysis rate 4 to 5 samples were taken in intervals of 2 to 36 hours. Stability of GUA during 7 days was checked in buffer only. The disappearance of GUA was used to calculate reaction rate constants. As the reaction time the time between reaction start and the injection of the GUA sample in the HPLC was defined.

3.2.5 Derivation of (Pseudo) First-Order Hydrolysis Rate Constants Hydrolysis rate constants were calculated consistent with (pseudo) first-order kinetics. The reaction rate constant k’hyd was derived from the slope of the linear regression of the natural logarithm of the measured concentration of electrophiles (“el” for electrophile) depending on the reaction time (t) according to Equation 3.1.

' ln c el = ln c 0,el − khyd ⋅t (3.1)

3.2.6 Derivation of Second-Order Reaction Rate Constants with Glutathione and 2´-Deoxyguanosine The time course of the depletion of GSH and GUA (“nuc” for the nucleophile GUA or GSH) could be described by a pseudo first-order rate law (Equation 3.2).

' ln cnuc =ln c 0,nuc −knuc,tot ⋅ t (3.2)

For GSH the pseudo first-order reaction rate constant k’nuc,tot = k’GSH,tot had to be corrected . -5 -1 with the apparent oxidation rate constant of GSH k’GSH,ox (Equation 3.3, k’GSH,ox = 6.06 10 min ± 12.3%), which was measured without addition of electrophiles under the same reaction conditions. The pseudo first-order reaction rate constant k’GSH, of the reaction with electrophiles, then could be calculated according to Equation 3.4. For GUA this correction was unnecessary as the nucleophile did not react at significant rates with other electrophiles including oxygen in control experiments (k’nuc,tot = k’GUA,tot = k’GUA).

' lnc GSH,control =lnc 0,GSH,control −k GSH,ox ⋅ t (3.3) 43

' ' ' k GSH =k GSH,tot −k GSH,ox (3.4)

Since the hydrolysis rates of some electrophiles were rather high (see Results and

Discussion, Table 3.1), this process could not be neglected in the derivation of kGSH or kGUA. Therefore for the correct calculation of second order substitution rate constants of the electrophile with either GSH or GUA, kGSH and kGUA, hydrolysis of the electrophile has to be taken into account (Equation 3.5).

dc − el = k' ⋅c + k ⋅c ⋅c (3.5) dt hyd el nuc el nuc

Since the time dependent concentration of the electrophile was only determined for few compounds no definite mathematical solution for the depletion of electrophile simultaneously caused by hydrolysis and substitution with GSH or GUA could be used. Therefore the substitution rate constant knuc was calculated numerically as an average value by division of k’nuc with decreasing electrophile concentration (Equation 3.6 and 3.7). The decrease of the electrophile´s concentration due to substitution was considered equivalent to the decrease of the concentration of the nucleophile. For the reaction with GUA this loss however could be neglected was as it was insignificant compared to the loss of the electrophile due to hydrolysis. For the reaction with GSH the observed decrease of the concentration of GSH had to be corrected with the loss due to oxidation of GSH. Dividing the observed reaction time into 79 sections and calculating the decreasing electrophile concentration for n=80 points of time was sufficient to reach a constant knuc from Equations 3.6 and 3.7.

 n  '     k GUA    ∑  '    i=1  c 0,el ⋅exp(−khyd ⋅ ti )   k = (3.6) GUA lim n  n→∞       

 n  '     k GSH    ∑ ' '    i=1  c 0,el ⋅ exp(−khyd ⋅ ti ) + c GSH − c 0,GSH ⋅exp(−k GSH,ox ⋅ ti )   k = (3.7) GSH lim n  n→∞       

44

Table 3.1: Suggested dominant reaction mechanism, (pseudo) first-order hydrolysis rate, and reaction rate constant with 2´-deoxyguanosine and reduced glutathione of electrophiles investigated in this study. a electrophile hydrolysis 2′-deoxyguanosine glutathione b – n =0.0 (H2O) n=3.5 (N-7) n=5.1 (GS )

BCl Cl SN1 SN1 SN1 1.94·10-3 min-1 ±3.6% 1.94·10-3 min-1 c 1.94·10-3 min-1 d

3MBCl SN1 SN1 SN1 Cl 3.09·10-3 min-1 ±3.5% 3.09·10-3 min-1 c 3.09·10-3 min-1 d

NBCl Cl SN1 SN1 → SN2 SN2

O -5 -1 e -1 -1 + 5.74·10 min ±3.0% 2.70 M min ±4.3% N no decrease

O–

DClP SN1 SN1 SN2 Cl -6 -1 -6 -1 d -1 -1 Cl 3.24·10 min ±3.0% 3.24·10 min 0.20 M min ±5.7%

DClB Cl SN1 → SN2 SN2 SN2 Cl -4 -1 -2 -1 –1 -1 -1 3.40·10 min ±3.1% 4.16·10 M min ±7.9% 2.27 M min ±6.0%

O EOX SN2 > SN1 SN2 SN2 4.69·10-5 min-1 ±0.9% 9.62·10-4 M-1 min–1 ±5.8% 0.15 M-1 min-1 ±3.1%

O EPI SN2 SN2 SN2 -4 -1 -3 -1 –1 -1 -1 Cl 2.15·10 min ±2.2% 5.03·10 M min ±2.4% 1.41 M min ±1.3%

EPOX SN2 SN2 SN2 O 4.76·10-5 min-1 ±3.0% 8.09·10-3 M-1 min–1 ±3.3% 0.38 M-1 min-1 ±2.6%

O MVIN SN1 SN2 SN2 1.92·10-3 min-1 ±1.8% no decrease e 0.28 M-1 min-1 ±2.2%

O SOX SN1 SN2 > SN1 SN1 > SN2 3.82·10-4 min-1 ±2.8% 7.06·10-3 M-1 min–1 ±3.7% 1.28 M-1 min-1 ±1.3%

O NOX O SN2 SN2 SN2 N+ -5 -1 -2 -1 –1 -1 -1 O– 8.51·10 min ±1.1% 1.30·10 M min ±8.0% 0.76 M min ±5.9%

PPOX O SN1 SN1 SN1 1.71·10-4 min-1 ±2.8% 1.71·10-4 min-1 d 1.71·10-4 min-1 d

a All reaction rate constants, given with the relative standard deviation of the mean, were determined at 30°C and pH 7.65. b Nucleophilicity constants defined according to Swain and Scott (1953). Value for N-7 of 2′-deoxyguanosine and esterified cysteine from Vogel and Nivard (1994). c No decrease of nucleophile measured and the reaction rate constant is assumed to be the same as the hydrolysis rate constant, because the mechanism of the reactions with H2O, GUA, and GSH is likely to be SN1. d Reaction is likely to follow SN1 and rate constant is assumed to be equivalent to the hydrolysis rate constant. From the observed decrease of nucleophile “pseudo-second” order reaction rate constants were calculated: BCl: 2.51 M-1 min-1 ±7.8%, 3MBCl: 3.34 M-1 min-1 ±4.8%, DClP: 3.38·10-3 M-1 min-1 ± 12%, PPOX: 6.14·10-3 M-1 min-1 ±7.8% with GUA, and PPOX: 0.78 M-1 min-1 ±2.6% with GSH. e Reaction rate constants could not be determined, because no decrease of nucleophile was measured. 45

3.3 Results and Discussion

3.3.1 Classification of Reactions According to SN1 and SN2

The classification according to SN1 and SN2 of reactions investigated in this study is based on measuring both the depletion of the electrophile and the nucleophile, on using principles deduced from the Swain-Scott equation, application of structure-activity relationships as an evaluative tool, and on analyzing product distributions and mechanistic information given in literature. Results of our proposed classification according to SN1 and SN2 and respective reaction rate constants of the compounds investigated here are summarized in Table 3.1.

Second order reaction rate constants of SN2 reactions, knuc, might be predicted applying the Swain Scott equation (Swain and Scott 1953) (Equation 3.7), if the substrate (electrophile) constant s and the nucleophilic strength n of the nucleophile are known:

k log nuc = s ⋅ n (3.7) ' khyd /55.5M The higher the n-value of a given nucleophile, the higher the tendency of an electrophile- nucleophile combination to react according to SN2, i.e., an electrophile, which reacts according to

SN2 with a weak nucleophile will most likely react according to SN2 with nucleophiles of higher nucleophilicity. Hence, classification of reaction order of reactions with the nucleophiles water, 2´- deoxyguanosine, and glutathione, can be carried out by the principles illustrated in Figure 3.2.

increasing nucleophilicity Figure 3.2: Deduction of substitution mechanisms GSH from known reaction mechanism. H2O GUA

SN1 ??

S 1 S 1 ? N N

S 1S1 S 1 N N N

SN2 SN2SN2

? SN2 SN2 The known reaction mechanisms are given in the boxes. Therefrom derived mechanisms are depicted in italics. ??S 2 Principles are only applicable given the same temperature and N solvent, since SN1 and SN2 are differently influenced by both parameters.

46

3.3.1.1 Organochlorines

BCl is assumed to hydrolyze according to SN1 (Hyne et al. 1962; Ohnishi and Tanabe 1971). In the reaction with GSH, following both, the decrease of BCl and GSH, the first order reaction rate constant of BCl (2.12·10-3 min-1 ±1.02%) was only 1.1 times higher than in the hydrolysis experiments, i.e., even with the strong nucleophile GSH, BCl seems to react predominantly according to SN1. Consequently BCl should also react according to SN1 with the weaker nucleophile GUA. Since the positive inductive effect of the in 3MBCl enhances the tendency to form a carbocation 3MBCl reacts as well according to SN1. For NBCl the situation is different. The negative inductive effect of the nitro group reduces the stability of the corresponding carbocation, which could lead to a reaction mechanism with less SN1 character. In the following the classification of hydrolysis reactions and reactions with GSH of 3MBCl, BCl, and NBCl is evaluated by correlation of the reaction rate constants with Hammett σ and σ+ constants (Table 3.2), respectively.

Table 3.2: σ and σ+ Hammett substituent constants used for classification of reaction rate with structure-activity correlations. electrophile substituent σ and σ+ a BCl –para-H 0.00

3MBCl –meta-CH3 -0.07

NBCl –para-NO2 0.78 a Recommended σ constants derived from acidity constants of a series of substituted benzoic acids, phenols, and anilines (Hansch et al. 1995). σ+ constants as average values from reactions of aromatic electrophiles (Okamoto and Brown 1956).

The linear relationship of hydrolysis rate constants towards σ+ (Figure 3.3) clearly points to

SN1 hydrolysis of all investigated benzyl chlorides. Our results contradict conclusions made from changes of activation energies of hydrolysis of 4-methylbenzyl chloride, BCl, and NBCl, in ethanol-water mixtures (Hyne et al. 1962). Comparing the first-order hydrolysis rate constant with the first-order depletion rate constant of NBCl in reaction with GSH, the hydrolysis rate constant is 3.4 times lower than the depletion rate constant (1.93·10-4 min-1 ±3.4%). This can be explained by a shift towards SN2 character with the strong nucleophile GSH. The change of reaction mechanisms from SN1 for 3MBCl and BCl to SN2 for NBCl can be illustrated by the change of the slope in the correlation of (pseudo) second-order reaction rate constants towards σ (Figure 3.3). A comparable change of reaction order from methyl- to nitro-substituted compounds was found for the reaction of benzyl bromides with different thiophenols (Hudson and Klopman 1962), which are comparable to GSH. The reaction of NBCl with GUA can therefore be assumed to have both

SN1 and SN2 characteristics. 47

-0.5 0.55 Figure 3.3: Dependence of hydrolysis rate cons- -1.0 3MBCl + 0.5 tants ( ) upon σ and

-1.5 ) reaction rate constants -1 )

-1 with GSH (z) upon σ. NBCl 0.45 -2.0 min

3MBCl -1

’(min BCl

-2.5 0.4 (M hyd

-3.0 BCl GSH log k log 0.35

-3.5 log k 0.3 -4.0 NBCl

-4.5 0.25 -0.2 0.1 0.3 0.5 0.7 σ+, σ

The hydrolysis of DClP has not been examined in detail. For 1,3-dichloropropene, however,

SN1 hydrolysis was observed in ethanol and formic acid (Eliel 1956). Comparing the stability of the carbocations of DClP and 1,3-dichloropropene, the former carbocation is presumably better stabilized, therefore it is likely that DClP hydrolysis follows SN1 reaction. Boerth et al. (1991) examined the reaction of DClP and 1,3-dichloropropene with 4-(p-nitrobenzyl)pyridine (NBP). As the NBP test (described in detail in Hermens et al. (1985)) allows no discrimination between SN1 and SN2, isodesmic reaction enthalpies were used to determine the mechanism. DClP was found to react according to SN1 and 1,3-dichloropropene according to SN2. The nucleophilicity of NBP should be almost the same as the nucleophilicity of pyridine (n=3.6 (Eliel 1956)) and N-7 GUA (n=3.5 (Vogel and Nivard 1994)). Because of equal nucleophilicity and even higher steric requirements for the reaction with GUA, DClP is likely to react according to SN1 with GUA. In the reaction with GSH the depletion of GSH was much higher than the formation of carbocations from DClP determined in the hydrolysis reaction; GSH depletion can consequently only result from a

SN2 reaction with DClP. The hydrolysis of DClB was found to follow SN2 and SN1 kinetics, part of the product resulted from rearrangement of the carbocation (Eliel 1956). Thus, the reactions with

GSH and GUA are likely to follow SN2.

3.3.1.2 Epoxides

The rate determining step of SN1 reactions of the examined epoxides is supposed to be the formation of a carbocation at the α-C, which is stabilized by one or both substituents, R2 and R3.

The SN2 reactions of epoxides is assumed to be determined by the formation of a penta-valent transition state at the sterically less hindered β-C (Figure 3.4) (Parker and Isaacs 1959). Thus, from the adduct proportion of α- and β-substituted epoxide the dominant reaction mechanism can be deduced.

48

Figure 3.4: Rate determining step of S 1 reaction (top) and S 2 N N HO R2 HOHR2 reaction (bottom) of epoxides. +H+ CC CC β α -H+ R1 R3 R1 R3

δ− HO R2 HO R2

CC CC α Note that R =H for all examined epoxide, β 1 R1 R3 R1 − R3 except PPOX and R3=H for all, except Nuc δ MVIN. Nuc

At pH 7, reaction of EOX and EPI with chloride has been shown to follow dominantly second order kinetics (Addy and Parker 1965). The positive inductive effect of the in EOX led to a minor proportion of 16% SN1 product, whereas the negative inductive effect of the chloromethyl group in EPI resulted in 100% SN2 product formation. As chloride (n=3.0 (Schwarzenbach et al. 2003)) is of comparable nucleophilicity as GUA, reactions with GUA and

GSH are very likely to follow SN2. In product studies of GUA-adducts of EPI only SN2 adducts were found (Koskinen and Plna 2000; Singh et al. 1996). Hydrolysis of EOX very likely follows both SN1 and SN2. Transferring results found for the hydrolysis (Long and Pritchard 1956) and reaction with chloride (Addy and Parker 1965) of propylene oxide, around 40% of EOX might hydrolyze according to SN1. The methylene group of EPOX between the phenyl and the epoxy group inhibits SN1 supporting resonance stabilization. It can therefore be presumed that neutral hydrolysis and reactions with GUA and GSH follow primarily SN2 mechanism. Hydrolysis of MVIN was determined to follow SN1 mechanism (Bleasdale et al. 1996). Although a carbocation of MVIN is presumably well stabilized by the inductive and resonance contribution of its vinyl substituent on the α-C, it was found to react according to SN2 with amino and thiolate nucleophiles in a one to one solution of deuterium oxide and acetonitrile and in methanol (Bleasdale et al. 1996). As both reaction media are less polar than the aqueous solution used in this study, part of the reaction of MVIN with GUA and GSH may have followed SN1.

Reactions with SOX can either follow SN1 or SN2 kinetics. Depending on steric requirements of the nucleophile, SN1 reaction on the α-C may be favored due to good resonance stabilization or SN2 reaction on the less hindered ß-C may be preferred. In analogy to BCl, electron-donating groups in the phenyl ring should enhance SN1 reactions and electron-withdrawing groups should increase the proportion of SN2. Blumenstein et al. (1993) followed the hydrolysis reaction of SOX 18 18 with H2 O and found that 95% of O is incorporated at the α-C. The classification towards SN1 based on product analyses was supported by a linear correlation of different para-substituted styrene oxides towards Hammett σ+ constants (Blumenstein et al. 1993). In this correlation NOX was found to be an outlier, indicating a change in hydrolysis mechanism from SN1 to SN2. 49

Assuming SN2 hydrolysis for NOX, reactions with GSH and GUA should follow second order kinetics as well. PPOX was shown to follow predominantly SN1 reacting with thiophenol- triethylamine and toluene-α-thiol-triethylamine (Marples et al. 1986). The reactive moieties are comparable with GSH, thus PPOX should react according to SN1 in reactions with GSH, GUA, and water. It is noteworthy that the cis-isomer gave a great extent of SN2 adduct (Marples et al. 1986).

To evaluate the classification according to SN1 and SN2 of some epoxides, comparisons based on structure-activity relationships including steric constants are helpful. Because for certain substituents investigated in this study, the often-used Taft constants Es describe not only steric influences but also electronic resonance contribution (Schwarzenbach et al. 1993), steric effects are described by a molecular volume term (Table 3.3):

* Table 3.3: Molecular volume (MV), and σ Taft substituent constants used for classification of reaction rate with structure-activity correlations. a b electrophile substituent MV (l/mol) σ *

EOX –C2H5 0.076 -0.10

EPI –CH2Cl 0.055 1.01

EPOX –CH2C6H5 0.176 0.22

SOX –C6H5 0.142 0.60

NOX –C6H4-para-NO2 0.172 1.14 a Molecular volume of the substituent was derived by addition of atomic volumes (Abraham and McGowan 1987). b Recommended σ* constants given in Hansch and Leo, where no recommendation was made the constant which was derived in most studies was chosen (Hansch et al. 1995).

The correlation for the hydrolysis of terminal epoxides with one substituent at the α-C (Figure 3.5), showed the clear distinction between EPI, EOX, NOX, EPOX and SOX. Thus, EOX together with EPI, EPOX and NOX could be described with a model for second order mechanism, indicating that the presumed proportion of SN1 hydrolysis is not dominant. The linear correlation is ∗ 2 given by log khyd =0.42·σ -2.71·MV-4.02, r =0.94. Note that the intention of the correlation was not to set up a quantitative structure property relationship, but to provide assistance for the classification to the reaction order. Steric and polar substituent constant influence the reaction rate constant, which is enhanced by electron-withdrawing substituents and decreased with increasing size of the substituent.

50

Figure 3.5: Correlation of 5 3.5 ) -1 hydrolysis rate constants ) ( ) and reaction rate -1 3 min constants with GUA (▲) 4 -1 )(min 2.5 to the molecular volume EPOX )(M EOX * EOX σ (Mv) and Taft σ* constant. 3 2 * *- σ σ *- σ )/( SOX 1.5 )/( 2 , EOX ’ 1 hyd EPI SOX EOX GUA, 1 NOX 0.5 ’-log k ’-log

-log k hyd In order to show the depen- 0 EPI GUA dence of reaction rate constants 0 NOX on two parameters in a one- (log k -0.5

EPOX (log k dimensional plot, Taft relations were normalized to reaction rate -1 -1 constants and Mv and σ* for -0.05 0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 EOX. *- * -1 (MV-MV,EOX)/(σ σ EOX) (l mol )

Experimental results of reactions of SOX with GSH and GUA reported in literature illustrate the problems of reaction order classification for compounds that may react according to both SN1 and SN2. Reaction of salmon testis DNA with SOX resulted in 41 to 60% α-N-7-GUA-adduct and 40 to 59% ß-adduct (Koskinen et al. 2000; Phillips and Farmer 1994; Savela et al. 1986). Following the principles shown in Figure 3.2, one would conclude that reaction with GSH leads to an even higher proportion of ß-adduct. However, analysis of reaction products revealed that 60 to

70% of the GSH resulted from SN1 reaction and only 30 to 40% from SN2 (Hernandez et al. 1981; Pacheka et al. 1979). The results of this chapter support these findings. First order depletion rate constants of SOX in experiments with GSH (4.89·10-4 min-1 ±0.6%) were only 1.3 times higher than in hydrolysis experiments, i.e., the reaction is dominated by SN1. For borderline reacting compounds, like SOX, other factors than the relative nucleophilicity of either GUA or GSH may become important. Overlay of π-systems of GUA and SOX may both lead to a facilitated approximation of reaction centers and a decreased activation energy of the transition state, and thus to a greater probability of SN2. Because two effects may influence the reaction rate constant with GUA, namely size of the substituent and π-system, it is not possible to set up standard structure-activity relationships, and derived correlations strongly depend on the selection of compounds. As “large” molecules of our test set possess an extended π-system, the positive influence of the steric parameter in the linear correlation of reaction rate constants given by * 2 kGUA =0.54·σ +4.88·MV-3.21, r =0.91 (Figure 3.5) might reflect the influence of overlay of π- orbitals. Second order reaction rate constants with GSH could not be modeled with a sufficient quality. 51

3.3.2 Comparison of Experimental Reaction Rate Constants with Literature Data Comparison of hydrolysis rate constants presented in this Chapter with rate constants reported in literature is most often limited due to different compositions of aqueous solution and reaction temperature. Whereas differences in temperature can be handled if the frequency factor and activation energy of the Arrhenius equation are known, differences due to the use of organic solvent water mixtures can not be easily accounted for. For comparison of hydrolysis rates, a further prerequisite is either identical pH or a pH-independency of the hydrolysis rate in the respective pH range of comparison; that is relevant hydronium ion or hydroxide ion catalysis must be excluded. Table 3.4 gives an overview of reaction rate constants found for reactions in aqueous solutions. Given are data of experiments that were nearest to our experimental conditions of pH 7.65 and 30°C, differences in composition of the aqueous solution are also indicated. All reported rate constants lie well within the range of here reported data. A few comments however should demonstrate how composition and pH of the aqueous solution might influence the hydrolysis rates.

Table 3.4: Reaction rate constants reported in literature for some electrophiles in aqueous solution. elec. nuc. rate constant pH T (°C) composition reference -3 -1 BCl H2O 1.16·10 min 7 30 no additions (Ohnishi and Tanabe 1971) 1.46·10-3 min-1 7 30 1 mM KCl (Robertson and Scott 1961) -7 -1 DClP H2O 9.20·10 min 7 25 O2 free (Milano et al. 1988) -4 -1 DClB H2O 2.41·10 min 7 25 O2 free (Milano et al. 1988) -4 -1 a SOX H2O 2.82·10 min 7.65 25 0.2 M NaClO4 (Ross et al. 1982) 5.42·10-4 min-1 7.4 37 10 mM Tris-HCl, (Koskinen et al. 2001) 0.15 M NaCl SOX GUA 1.72·10-3 M-1min-1 b 7.4 37 10 mM Tris-HCl, (Koskinen et al. 2001) 0.15 M NaCl -5 -1 NOX H2O 3.70·10 min 7.0 25 0.2 M NaClO4 (Blumenstein et al. 1993) -5 -1 EOX H2O 7.41·10 min 7.4 37 1 mM Tris-HCl (Gervasi et al. 1985) -4 -1 c EPI H2O 1.67·10 min 7.65 30 10% ethanol (Piringer 1980) EPI GUA 2.95·10-3 M-1min-1 b 7.4 37 50 mM Tris-HCl (Landin et al. 1999) -3 -1 MVIN H2O 9.24·10 min 7.4 37 1 mM Tris-HCl (Gervasi et al. 1985) a Calculated from neutral hydrolysis constant of 4.1 s-1 and acid hydrolysis constant of 26.7 M-1 s-1, base catalyzed reaction is insignificant for SOX at pH 7.65 (Ross et al. 1982). b Calculated from data given for GUA in l/(g DNA h) assuming 41% GC-content in salmon and herring testes DNA (Sigma- Aldrich product information). Rate constants for quantitatively important sites in GUA (93% N-7, 2% N2) are summed up. c Calculated with Arrhenius frequency factor of 7.1·106 s-1 and activation energy of 72 kJ/mol for neutral hydrolysis, base and acid catalyzed hydrolysis is insignificant for EPI at pH 7.65 (Piringer 1980).

52

Hydrolysis of allylic halides (DClP, DClB) and benzylic halides (3MBCl, BCl, NBCl) follow SN1 and are therefore independent of pH. Hydrolysis of epoxides however is pH dependent, whereby acid-catalyzed and neutral hydrolysis are most important for pH-values around 7; base-catalyzed hydrolysis can usually be neglected (Larson and Weber 1994). Hydrolysis rates of EPI are relatively constant between pH 6.5 and 8.5 (Piringer 1980). The contribution of acid catalyzed reactions is important for SOX, though the factor of hydrolysis rates between pH 7.4 and 7.65 is only 1.1 (Ross et al. 1982). Hydrolysis of EOX is likely to be pH independent over a large pH scale, transferring results found for hydrolysis studies of propylene oxide (Mabey and Mill 1978).

This additionally should hold true for the SN2 hydrolysis of NOX. The SN1 hydrolysis of MVIN however is supposed to be susceptible towards the hydronium ion concentration. The extent of this influence could not be estimated, because both temperature and pH varied.

Besides the deviation obviously resulting from temperature difference, some of the deviation might result from the used buffers. In this study relatively high concentrations of hydrogen phosphate (11 mM) were used, which acts as an additional nucleophile (n=3.8, Schwarzenbach et al, 2003). Applying the Swain-Scott relationship (Equation 3.7), hydrogen phosphate could lead, e.g., for EPI (s=0.93, (Swain and Scott 1953)), to a 1.7 times higher observed hydrolysis rate constant. Measured hydrolysis rates were however only 1.3 times higher than in 10% - ethanolic solution of the same pH and temperature. Whereas the nucleophilicity of ClO4 is negligible, Cl- is a concurring nucleophile (n=3.0, Schwarzenbach et al. 2003). Assuming an average substrate constant s = 1 for epoxides, 150 mM Cl- (compare Table 3.4) should increase pure hydrolysis by a factor of 3.7.

Literature data on reaction rate constants of electrophiles with biological nucleophiles are rare. Most uncatalyzed reaction studies were conducted with 4-(p-nitrobenzyl)-pyridine, which was used as a surrogate for all biological nucleophiles (Deneer et al. 1988; Hermens et al. 1985) or for DNA bases (Eder et al. 1982; Hemminki et al. 1983; Hemminki and Vainio 1984). Reactions have usually been carried out in organic solvents and at elevated temperatures. Thus, despite the similarity of the reactive moiety of 4-(p-nitrobenzyl)-pyridine and GUA, no comparison can be made to reactions of GUA in aqueous solution. Furthermore, the comparison of reaction rate constants reported in this chapter is restricted to reaction rate constants reported for bases in double stranded DNA. Reaction rate constants found for EPI (Landin et al. 1999) and SOX (Koskinen et al. 2001) in DNA for guanidine are 1.7 and 4.0 times lower than that found with the molecule GUA. These differences cannot be accounted to differences of temperature or composition of aqueous solution, but clearly point to a huge steric influence of DNA, which is more important for the large SOX than for the small EPI. Thus, although it was attempted to determine reactions rate constants relevant for toxicity in biological systems, one only got a glance at the relative but not at the absolute reactivity of electrophiles. We still are confronted with the complexity of steric influences and influences of differing cytosolic composition in living cells 53 that influence, besides the numerous enzymatic facilitated reactions, the reactivity of electrophiles in organisms.

3.4 Literature Cited

Abraham, M. J., and McGowan, J. C. (1987). The use of characteristic volumes to measure cavity terms in reversed phase liquid chromatography. Chromatographia 25, 243-246. Addy, J. K., and Parker, R. E. (1965). The mechanism of epoxide reactions. Part VII. The reactions of 1,2-epoxybutane, 3,4-epoxybut-1-ene, 1,2-epoxy-3-chloropropane, and 1,2-epoxy-3- methoxypropane with chloride ion in water under neutral and acidic conditions. J. Chem. Soc. Jan., 644-649. Albanese, V., Milano, J. C., and Vernet, J. L. (1987). Etude de l´évaporation de quelques hydrocarbures halogènes de faible masse moléculaire dissous à l´état de traces dans l´eau. Environ. Tech. Lett. 8, 657-668. Bleasdale, C., Small, R. D., Watson, W. P., Wilson, J., and Golding, B. T. (1996). Studies on the molecular toxicology of buta-1,3-diene and isoprene epoxides. Toxicology 113, 290-293. Blumenstein, J. J., Ukaschukwu, V. C., Mohan, R. S., and Whalen, D. L. (1993). Effects of para- substituents on the mechanism of solvolysis of styrene oxides. J. Org. Chem. 58, 924-932. Boerth, D. W., Eder, E., Rasul, G., and Morais, J. (1991). Theoretical structure-activity study of mutagenic allyl chlorides. Chem. Res. Toxicol. 4, 368-372. Chovanec, M., Naslund, M., Spivak, I., Dusinska, M., Cedervall, B., and Kolman, A. (1998). Rejoining of DNA strand breaks induced by propylene oxide and epichlorohydrin in human diploid fibroblasts. Environ. Mol. Mutagen. 32, 223-228. Cohn, V. H., and Lyle, J. (1966). A fluorometric assay for glutathione. Analytical 14, 434-440. Comporti, M., Maellaro, E., Del Bello, B., and Casini, A. F. (1991). Glutathione depletion: its effects on other antioxidant systems and hepatocellular damage. Xenobiotica 21, 1067-1076. Deneer, J. W., Sinnige, T. L., Seinen, W., and Hermens, J. L. M. (1988). A quantitative structure- activity relationship of some epoxy compounds to the guppy. Aquat. Tox. 13, 195-204. Dipple, A., Moschel, R. C., and Hudgins, W. R. (1982). Selectivity of alkylation and aralkylation of nucleic acid components. Drug Metabolism Reviews 13, 249-268. Eder, E., Neudecker, T., Lutz, D., and Henschler, D. (1982). Correlation of alkylating and mutagenic activities of allyl and allylic compounds. Chem. Biol. Interact. 38, 303-315. Eliel, E. L. (1956). Substitution at saturated carbon atoms. In Steric effects in organic chemistry M. S. Newman, ed., pp. 61-163. John Wiley & Sons, Inc.; Chapman & Hall, New York, London. Freidig, A., Verhaar, H. J. M., and Hermens, J. L. M. (1999). Comparing the potency of chemicals with multiple modes of toxic action in aquatic toxicity: acute toxicity due to narcosis versus reactive toxicity of acrylic compounds. Environ. Sci. Technol. 33, 3038-3043. Fujita, M., Sano, M., Takeda, K., and Tomita, I. (1993). Fluorescence detection of glutathione S conjugate with aldehyde by high-performance liquid chromatography with post-column derivatization. Analyst 118, 1289-1292. Gervasi, P. G., Citti, L., Del Monte, M., Longo, V., and Benetti, D. (1985). Mutagenicity and chemical reactivity of epoxidic intermediates of the isoprene metabolism and other structurally related compounds. Mutat. Res. 156, 77-82. Hansch, C., and Leo, A. (1995). Exploring QSAR - Fundamentals and applications in chemistry and biology. American Chemical Society, Washington, DC.

54

Hansch, C., Leo, A., and Hoekman, D. (1995). Exploring QSAR - Hydrophobic, electronic, and steric constants. American Chemical Society, Washington, D.C. Hemminki, K., Falck, K., and Linnainmaa, K. (1983). Reactivity, SCE induction and mutagenicity of benzyl chloride derivatives. J. Appl. Toxicol. 3, 203-207. Hemminki, K., and Vainio, H. (1984). Genotoxicity of epoxides and epoxy compounds. In Industrial Hazards of plastics and synthetic elastomers, pp. 373-384. Alan R. Liss, Inc, New York. Hermens, J. L. M., Busser, F., Leeuwach, P., and Musch, A. (1985). Quantitative correlations studies between the acute lethal toxicity of 15 organic halides to the guppy (Poecilla reticulata) and chemical reactivity towards 4-nitrobenzylpyridine. Toxicol. Environ. Chem. 9, 219-236. Hernandez, O., Yagen, B., Cox, R. H., Smith, B. R., Fourman, G. L., Bend, J. R., and McKinney, J. D. (1981). Stereospecificy and regioselectivity in the reaction of epoxides with glutathione. In Environmental Health Chemistry, J. D. McKinney, ed., pp. 425-444. Ann Arbor Science Publishers, Collingwood. Hudson, R. F., and Klopman, G. (1962). Nucleophilic reactivity. Part II. The reaction between substituted thiophenols and benzyl bromides. J. Chem. Soc. March, 1062-1067. Hyne, J. B., Wills, R., and Wonka, R. E. (1962). Specific solvation in binary mixtures. Part II. Dependence of activation energy of solvolysis in benzyl chloride in ethanol-water mixtures on temperature and ring substitution. J. Am. Chem. Soc. 84, 2914-2919. Kolman, A., Spivak, I., Naslund, M., Dusinska, M., and Cedervall, B. (1997). Propylene oxide and epichlorohydrin induce DNA strand breaks in human diploid fibroblasts. Environ. Mol. Mutagen. 30, 40-46. Koskinen, M., and Plna, K. (2000). Specific DNA adducts induced by some mono-substituted epoxides in vitro and in vivo. Chem.-Biolog. Interact. 129, 209-229. Koskinen, M., Vodicka, P., and Hemminki, K. (2000). Adenine N3 is a main alkylation site of styrene oxide in double stranded DNA. Chem.-Biol. Interact. 124, 13-27. Koskinen, M., Vodickova, L., Vodicka, P., Warner, S. C., and Hemminki, K. (2001). Kinetics of formation of specific styrene oxide adducts in double-stranded DNA. Chem.-Biol. Interact. 138, 111-124. Krijgsheld, K. R., and van der Gen, A. (1986). Assessment of the impact of the emission of certain organochlorine compounds on the aquatic environment. Part III: Epichlorohydrine. Chemosphere 15, 881-893. Landin, H. H., Segerback, D., Damberg, C., and Osterman Golkar, S. (1999). Adducts with haemoglobin and with DNA in epichlorohydrin-exposed rats. Chem.-Biol. Interact. 117, 49-64. Larson, R. A., and Weber, E. J. (1994). Reaction mechanisms in environmental organic chemistry. Lewis Publishers, Boca Raton. 18 Long, F. A., and Pritchard, J. G. (1956). Hydrolysis of substituted ethylene oxides in H2 O solutions. J. Am. Chem. Soc. 78, 2663-2667. Mabey, W., and Mill, T. (1978). Critical review of hydrolysis of organic compounds in water under environmental conditions. J. Phys. Chem. Ref. Data 7, 383-415. Marples, B. A., Saint, C. G., and Traynor, J. R. (1986). Regiochemistry of nucleophilic opening of ß-substituted styrene oxides with thiolate anions: Model experiments in the synthesis of leukotriene analogues. J. Chem. Soc. Perkin Trans. I 4, 567-574. Milano, J. C., Guibourg, A., and Vernet, J. L. (1988). Evolution non biologique dans l`eau de composés organohalogènes a trois et quatre atomes de carbone: hydrolyse et photolyse. Wat. Res. 22, 1553-1562. 55

Ohnishi, R., and Tanabe, K. (1971). A new method of solubility determination of hydrolyzing solute-solubility of benzyl chloride in water. Bull. Chem. Soc. Jpn. 44, 2647-2649. Okamoto, Y., and Brown, H. C. (1956). A quantitative treatment for electrophilic reactions of aromatic derivatives. J. Org. Chem 22, 485-494. Pacheka, J., Gariboldi, P., Cantoni, L., Belvedere, G., Mussini, E., and Salmona, M. (1979). Isolation and structure determination of enzymatically formed styrene oxide glutathione conjugates. Chem.-Biol. Interact. 27, 313-321. Parker, R. E., and Isaacs, N. S. (1959). Mechanisms of epoxide reactions. Chem. Rev. 59, 737- 799. Phillips, D. H., and Farmer, P. B. (1994). Evidence for DNA and protein binding by styrene and styrene oxide. Crit. Rev. Tox. 24, S35-S46. Piringer, O. (1980). Die Kinetik der Hydrolyse von Epichlorhydrin in verdünnten wässrigen Lösungen. Deutsche Lebensmittel Rundschau 76, 11-13. Reed, D. J. (1990). Glutathione–toxicological implications. Annu. Rev. Pharmacol. Toxicol. 30, 603-631. Robertson, R. E., and Scott, J. M. W. (1961). The neutral hydrolysis of some alkyl and benzyl halides. J. Chem. Soc. Apr., 1596-1604. Ross, A. M., Pohl, T. M., Piazza, K., Thomas, M., Fox, B., and Whalen, D. L. (1982). Vinyl epoxide hydrolysis reactions. J. Am. Chem. Soc. 104, 1658-1665. Savela, K., Hesso, A., and Hemminki, K. (1986). Characterization of reaction products between styrene oxide and deoxynucleosides and DNA. Chem.-Biol. Interact. 60, 235-246. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (1993). Environmental Organic Chemistry. John Wiley and Sons, Inc., New York. Schwarzenbach, R. P., Gschwend, P. M., and Imboden, D. M. (2003). Environmental Organic Chemistry. John Wiley & Sons, Inc., New York. Singh, U. S., Decker-Samuelian, K., and Solomon, J. J. (1996). Reaction of epichlorohydrin with 2'-deoxynucleosides: Characterization of adducts. Chem.-Biol. Interact. 99, 109-128. Swain, C. G., and Scott, C. B. (1953). Quantitative correlations of relative rates. Comparison of hydroxide ion with other nucleophilic reagents toward alkyl halides, esters, epoxides and acyl halides. J. Am. Chem. Soc. 75, 141-147. Uziel, M., Munro, N. B., Katz, D. S., Vo Dinh, T., Zeighami, E. A., Waters, M. D., and Griffith, J. D. (1992). DNA adduct formation by 12 chemicals with populations potentially suitable for molecular epidemiological studies. Mutat. Res. 277, 35-90. Vogel, E. W., and Nivard, M. J. M. (1994). The subtlety of alkylating agents in reactions with biological macromolecules. Mutat. Res. 305, 13-32.

56

4 Methods and Tools for Prediction of the Toxicity of Electrophilic Chemicals

Abstract

The appropriate selection and application of quantitative structure-activity relationships (QSARs) is based on the prior assignment of a chemical to its mode of toxic action. This classification is often derived from structural characteristics, with the underlying assumption that chemically similar compounds have similar mechanisms of action, which is often but not necessarily the case. Instead of using structural characteristics for classification towards a mode of toxic action we used Escherichia coli based biosensors to identify reactive chemicals. Analyzing a series of reactive organochlorines, epoxides, and compounds with an activated double bond, three subclasses of reactive toxicity were distinguished: “glutathione depletion related toxicity”, “DNA damage”, and “unspecific reactivity”. For both subsets of specifically reacting compounds a direct correlation between effects and chemical reactivity was found. Reaction rate constants with either glutathione or 2´-deoxyguanosine, which was used as a model for complex DNA, served well to set up QSARs for either glutathione depletion related toxicity or toxicity based on DNA damage in the model organism E. coli. Correlation of bacterial toxicity values of all examined compounds to toxicity in algae, daphnids, and fish yielded a linear regression with unit slope, independent of the assigned mode of toxic action. Whereas the application of QSARs for reactive chemicals which are based on reaction rate constants is limited to small subsets of compounds with strictly identical mechanism of toxic action and similar metabolic rates, the proposed biosensors not only allow the experimental identification of molecular mechanisms underlying the observable toxicity, but their toxicity values can also be used to quantitatively predict toxic effects in higher organisms by linear correlation models. 58

4.1 Introduction

Quantitative structure activity relationships (QSARs) are widely used to predict toxicity from chemical structure and corresponding physico-chemical properties. The development and application of QSARs started with the prediction of toxicity caused by baseline toxicants (for a historical review see Lipnick (1995)). Species-specific toxicity of baseline toxicants was sufficiently described by a hydrophobicity term, e.g., the octanol-water partition coefficient. Recent work showed that species differences of baseline toxicity can be explained by differences of membrane lipid content, and lethal membrane concentrations for algae, daphnia, and fish are nearly identical (Escher and Schwarzenbach 2002). QSARs using membrane-water partition coefficients are thus mechanistically meaningful and are based on a descriptor describing the crucial event leading to observed toxicity, the interaction of the toxicant with its target site, the biological membrane.

A prerequisite for a correct predictive assessment of the toxicity of a chemical by using QSARs is the accurate assignment of the mode of toxic of action. Structural characteristics (Verhaar et al. 1992) and physico-chemical properties of a chemical are helpful information to identify baseline toxicants and to discriminate them from potentially reactive chemicals (EC 1996). In contrast to baseline toxicants, the toxicity of reactive chemicals is determined by the intrinsic reactivity of the toxicant and the target occupation (Legierse et al. 1999; Verhaar et al.

1999). Important targets for electrophilic chemicals are nucleophilic sites (e.g., -SH, -NH2, -OH) in peptides and proteins as well as in DNA, which may either have approximately constant cellular concentration (DNA) or are subject to a regular turnover (peptides and proteins). Toxicity of reactive chemicals is therefore time-dependent and effects are irreversible unless defeated by an active cellular defence system, whereas baseline toxicity is reversible and occurs whenever a critical membrane concentration is exceeded, independent of the time of exposure (Escher and Hermens 2002).

Electrophiles may react with biological nucleophiles through various mechanisms including nucleophilic substitution, Schiff´s base formation, or Michael addition (Hermens 1990). Thus a shared mechanistic basis, which is a prerequisite for a common QSAR, can not be fulfilled a priori, which is reflected by numerous QSARs that are restricted to narrow groups of structurally related or congeneric chemicals that are likely to react according to the same reaction mechanism (Lipnick 1995). However, as was shown in Chapters 2 and 3 structurally related nucleophiles do not necessarily have the same reactive mode of toxic action nor do they necessarily react according to the same reaction mechanism with different nucleophiles. These are additionally complicating factors that may hamper setting up reliable QSARs with a sound mechanistic basis. 59

QSARs for prediction of toxicity of reactive chemicals typically account for the intrinsic reactivity by measured or calculated reaction rate constants. However, no parameter explicitly accounts for the target site occupation that is influenced by toxicokinetic processes like uptake and metabolism. Thus, the improvement of the quality of QSAR-equations for certain reactive chemicals, e.g., for epoxides (Deneer et al. 1988), achieved by adding a hydrophobicity term was in fact explained by the influence of uptake. However, it remains unclear why reactive chemicals with other reactive moieties, e.g., organochlorines, but the same range of hydrophobicity are satisfactorily modelled without this additional parameter (Freidig et al. 1999; Hermens et al. 1985). Of course, it is always possible that within a given test set certain properties related to uptake and metabolisms are constant, e.g., uptake is not rate-limiting or catalytic enhancement of reactivity due to the glutathione S-transferases is proportional for all compounds.

Bacteria were initially chosen as model systems to study the direct effects of electrophilic chemicals on nucleophilic biological targets, because reactions with target sites are presumably less influenced by toxicokinetic processes. This assumption was made because catalysis by glutathione S-transferase is less efficient in bacteria than in higher organisms (Kerklaan et al. 1985; Penninckx and Elskens 1993) and DNA is not compartmentalized and less shielded by proteins than in higher organisms. The bacterial biosensor set presented in Chapter 2 allowed a clear classification of a test set of epoxides, reactive organochlorines, and compounds with an activated double bond into three subclasses of reactive toxicity: “GSH depletion related toxicity”, “DNA damage”, and “unspecific reactivity”, which is governed by both of the aforementioned mechanisms.

It was the goal of this study to set up QSARs for each class of reactive toxicity to describe the toxicity by appropriate reactivity parameters, i.e., to test if GSH depletion related toxicity can be predicted by the reaction rate constant towards GSH and, analogously, DNA damage by the reaction rate constant towards 2´-deoxyguanosine. In order to check if possible influences of uptake and other toxicokinetic processes are relevant for bacterial QSAR models, the octanol- water partition coefficients were additionally measured. Further it was the goal to evaluate if bacterial toxicity data correlate with acute toxicity data for algae, daphnia, and fish, despite the toxicokinetic differences. Finally we will conclude with recommendations on how to approach prediction of reactive toxicity in general.

4.2 Methods and Data Sets

4.2.1 Determination of Octanol-Water Partition Coefficients

Octanol-water partition coefficients (KOW) were measured for the following reactive organochlorine compounds and epoxides: benzyl chloride (BCl, CAS 100-44-7), 4-nitrobenzyl chloride (NBCl, CAS 100-14-1), 2,3-dichloro-1-propene (DClP, CAS 78-88-6), trans-1,4-dichloro- 60

2-butene (DClB, CAS 110-57-6), styrene oxide (SOX, CAS 96-09-3), 2-(4-nitro-phenyl)-oxirane (NOX, CAS 6388-74-5), (2,3-epoxypropyl) benzene (EPOX, CAS 4436-24-2), 1,2-epoxybutane (EOX, CAS 106-88-7), epichlorohydrin (EPI, CAS 106-89-8), 2-methyl-2-vinyloxirane (MVIN, CAS 1838-94-4). BCl, NBCl, EPI, and EOX were purchased from Fluka Chemie AG, Buchs, Switzerland. DClP and SOX were obtained from Sigma-Aldrich Chemie AG, Steinheim, Germany. DClB, NOX, EPOX, MVIN, and n-octanol (CAS 111-87-5) were obtained from Aldrich Chem. Co. Inc, Milwaukee, USA. All chemicals were of highest purity available (≥ 95%) and were used as received.

n-Octanol of ACS reagent grade and deionized water filtered by a Millipore filter system were mutually saturated for 12 h on a shaking incubator. Octanol-saturated water and water-saturated octanol phases separated within four days. KOW-values were determined in duplicate in 1:5.3, 1:2.6, and 1:1.3 water:octanol mixtures, using serum vials with crimped viton rubber stoppers. Liquid electrophiles were directly added to the mixture. Different masses of solid electrophiles, i.e. NBCl and NOX, were weighed prior to addition of octanol and water. As some of the examined electrophiles hydrolyze quite quickly (compare Chapter 3) the whole procedure of determination of partition coefficients had to be very fast, but sufficiently long for the compounds to attain equilibrium. The following procedure satisfied both requirements: Mixtures with liquid electrophiles were vigorously mixed for six times 20 s on a vortex. 30 min of ultrasonic treatment lead to dissolution of the slowly hydrolyzing solid electrophiles, thereafter, similarly to the liquid electrophiles they were additionally mixed on a vortex. Subsequently, the mixtures were centrifuged for 2 min at 2500 rcf, leading to a fast separation of phases. Samples of the octanol and water phases were taken using syringes with very fine canulas. To avoid traces of octanol in the water phase, air was gently expelled while passing the octanol-layer. Samples were analyzed immediately afterwards. Aliphatic electrophiles, i.e., DClP, DClB, EOX, EPI, and MVIN, were analyzed by GC-FID detection (GC 8000, Fision Instruments, Milano, Italy) using a Stabilwax column (30 m x 0.32 mm, 1 µm; BGB Analytik, Anwil, Switzerland) with direct on-column injection. The aromatic electrophiles BCl, NBCl, SOX, NOX, and EPOX were analyzed by HPLC with UV- VIS detection (pump M480, Gina 160 autosampler, Gynkotek, Germering, Germany; 875-UV detector, Jasco, Gross-Umstadt, Germany) using C-8 (LiChrosphere, 125 x 4 mm, 5 µm spheres; Merck, Darmstadt, Germany) and C-18 (Nucleosil, 250 x 4 mm, 5 µm spheres; Macherey-Nagel, Dueren, Germany) reversed phase columns with differing methanol-water mixtures as mobile phase. Concentrations in the water phase could be measured directly. For HPLC analysis, the octanol-phase had to be diluted with methanol. GC measurements of octanol-phases could be done directly using a very slow on-column injection. KOW-values were calculated as the ratio of the concentration in the octanol phase to the concentration in the water phase. Reported values (see Data Sets) are average values from six octanol-water mixtures. 61

4.2.2 Data Sets To examine the relationship of reaction rate constants, octanol-water partition coefficients, and toxicity of electrophiles with respect to their reactive mode of toxic action, toxicity values for four aquatic species groups were selected and complemented with reaction rate constants and KOW- values as far as available (Table 4.1). Examined electrophilic classes comprise compounds with an activated double bond, epoxides and reactive organochlorine compounds. Toxicity values include concentrations resulting in 50% growth inhibition (EC50) of Escherichia coli (strain CC102,

Chapter 2), EC50-values of the inhibition of the photosystem II quantum yield measured by chlorophyll fluorescence of the unicellular algae Scenedesmus vacuolatus after 2 h of exposure, concentrations causing 50% lethality (LC50) after 48 h for the crustacean Daphnia magna, 96 h

LC50 for the fish fathead minnow (Pimephales promelas), and 14 days LC50 values of the guppy (Poecilia reticulata). Physico-chemical descriptors include the octanol-water partition coefficient, and second-order reaction rate constants with the nucleophiles glutathione (GSH) and 2´- deoxyguanosine (GUA). Additionally, the classification of the examined chemicals towards the reactive mode of toxic action is given. The classification is based on growth differences observed for a set of bacterial biosensors either lacking the nucleophile GSH or lacking DNA repair systems. Comparison of toxicity observed in those strains to the toxicity of their unaltered parent strains, allowed the distinction of three modes of action for electrophilic chemicals: DNA damage, glutathione depletion related toxicity, and unspecific reactivity (for details see Chapter 2).

4.3 Results and Discussion

4.3.1 Prediction of Bacterial Toxicity Using Reaction Rate Constants with Biological Nucleophiles Based on the assignment of a chemical towards a reactive mode of toxic action (Table 4.1) the relationship of bacterial toxicity towards the corresponding reaction rate constants was examined. Toxicity values of compounds with glutathione depletion related toxicity log linearly correlated with the chemical reaction rate constant with glutathione kGSH (Equation 4.1 and Figure

4.1). Toxicity values showed no dependence on the hydrophobicity term log KOW as a single 2 descriptor (r =0.11, F=0.5); and the weight of log KOW is negligible in correlation with log kGSH and 2 log KOW (r =1.00, F=322). The log linearly correlation of toxicity to kGSH suggests that the chemical reaction of examined compounds with glutathione is the limiting step the limiting step in a series of events leading to observed toxicity.

 EC E.coli   k  log  50  =−0.87(±0.04)⋅log  GSH  +1.60(±0.08)  mM   M−1 min−1  (4.1) n = 6, r2 = 0.99, F = 440

Table 4.1: Toxicity values for E. coli, algae, daphnia, and fish, mode of toxic action classification, octanol-water partition coefficients, and reaction rate constants with glutathione (GSH) and 2´-deoxyguanosine (GUA). h electro- log EC50 log EC50 log EC50 log 96 h LC50 log 14 d LC50 mode of log KOW log kGSH log kGUA phile a E. coli (mM) b algae (mM) c daphnia (mM) d fish (mM) e fish (mM) f action g (M-1 min-1) i (M-1 min-1) j ACR -1.78 -2.99 -3.60 GSH -0.01 3.92 IBA 0.16 -1.79 GSH 2.22 1.47 HEA 0.24 -1.38 GSH -0.21 1.71 EA 0.25 -1.60 -2.16 GSH 1.32 1.60 ACN 0.73 -0.73 -0.80 GSH 0.25 0.87 ACA 1.89 0.35 0.19 -0.31 GSH -0.067 -0.33, -0.23 k EPOX 0.45 0.99 DNA 2.00 (1.97-2.02) -0.42 -2.09 SOX 0.56 0.88 -1.01 -1.42 -1.23 DNA 1.65 (1.64-1.67) 0.11 -2.15 EOX 1.67 2.13 -0.34 DNA 0.64 (0.62-0.68) -0.82 -3.02 NBCl -0.93 u. r. 2.39 (2.38-2.40) 0.43 not detectable DClB -0.42 0.03 -3.16 u. r.l 2.18 (2.14-2.21) 0.36 -1.38 m m BCl -0.33 0.05 -0.94 -2.51 u. r. 2.57 (2.51-2.64) SN1 SN1 m DClP -0.29 -1.99 u. r. 2.15 (2.14-2.15) -0.70 SN1 NOX -0.26 u. r.l 1.68 (1.67-1.69) -0.12 -1.89 EPI 0.53 1.14 -0.64 -1.43 -2.15 u. r.l 0.42 (0.42-0.43) 0.15 -2.30 MVIN 1.51 u. r. 1.03 (1.01-1.05) -0.55 not detectable a Abbreviation for compounds with an activated double bond: ACR: acrolein, IBA: isobutyl acrylate, HEA: 2-hydroxyethyl acrylate, EA: ethyl acrylate, ACN: acrylonitrile, ACA: acryl amide. Abbreviations for other electrophiles are given in section 4.2.1. b EC50 values for growth inhibition of the microorganism Escherichia coli strain CC102 are from Chapter 2. c 2 h EC50 values for photosystem II inhibition of the algae Scenedesmus vacuolatus (Niederer 2002). d 48 h LC50 values for the crustacean Daphnia magna from the EPA database ECOTOX (EPA 2002). Data were derived in different systems: ACA: flow-through experiments, ACN and SOX: semi-static, ACR and EPI: static. e 96 h LC50 values for the fish fathead minnow (Pimephales promelas) from the EPA database ECOTOX (EPA 2002). Selected were lowest values reported with preference for data derived in flow-through systems (ACA, IBA, HEA, EA, ACN, ACA, SOX). Data for BCl and EPI are from static experiments. f 14 d LC50 toxicity values for the fish guppy (Poecelia reticulata) from semi-static experiments. Data for EA and HEA from (Hermens and Leeuwangh 1982), data for SOX, EOX, EPI from (Deneer et al. 1988), data for DClB, BCl, and DClP from (Hermens et al. 1985). g Mode of toxic action of electrophilic chemicals. Classification: GSH – GSH depletion related toxicity, DNA – DNA damage, u. r. – unspecific reactivity (see Chapter 2). h log KOW for ACR, EA, ACN, ACA recommended values from (Chemfate 2002), values for HEA and IBA from (Tanii and Hashimoto 1982). Log KOW for the other compounds are experimental results (Section 4.2.1) with 95% confidence intervals in parentheses. i Values for reaction with glutathione (GSH) for ACR to ACA measured at 20°C, pH 8.8 (Freidig et al. 1999). Values for all other compounds were determined at 30°C, pH 7.65 (Chapter 3). j Values for reaction with 2´-deoxyguanosine (GUA) measured at 30°C at pH 7.65 (Chapter 3). k Measured at 30°C, pH 7.65 (Tobler 2002). l Correlation to reaction rate constants indicate the higher importance of DNA damage compared to GSH depletion related toxicity (Equation 4.2, Figure 4.2). m Reaction of electrophile follows first-order nucleophilic substitution (SN1) and is therefore independent of the concentration of nucleophile the GSH or GUA, respectively. 63

Correspondingly, the toxicity of DNA damaging compounds, i.e., EPOX, SOX, and EOX, log linearly correlated with the reaction rate constants with 2´-deoxyguanosine. However, as the data include two compounds of comparable reactivity and toxicity the correlation is governed by two points (Figure 4.1).

Figure 4.1: Plot of log EC50 values of E. coli of 2 EOX ACA specifically reacting compounds versus logarithmic reaction rate constants with 1 ACN glutathione (kGSH) (z) and with 2´- SOX deoxyguanosine (k ) („). EA GUA EPOX HEA 0 IBA

-1

ACR -2 -3 -2 -1 0 1 2 3 4

-1 -1 log kGUA, log kGSH (M min )

As demonstrated for ACA the order of magnitude of reaction rate constants with GSH measured either at 20°C and pH 8.8, or at 30°C and pH 7.65 is comparable, indicating that the higher temperature compensates the higher proportion of the glutathione-anion at higher pH. This comparability allowed a combined comparison of the toxicity of unspecifically reactive compounds and compounds with glutathione depletion related toxicity to the reaction rate constant with glutathione. The toxicity of unspecifically reacting compounds could not be described by the reaction rate constants with GSH (Figure 4.2), as was anticipated for compounds whose toxicity in E. coli is influenced by the reaction with both important nucleophilic targets, GSH and DNA.

However, EC50 values of DClB, EPI, and NOX (Figure 4.2) reacting with GUA lay well on the extension of the line combining reactivity and toxicity of DNA damaging compounds, i.e., EOX, SOX, and EPOX (Figure 4.1). These three compounds – the organochlorine DClB, and the epoxides EPI and NOX – happen to be the only three compounds with a statistically significant difference of growth between the DNA repair sufficient and deficient strain (see Chapter 2) and a detectable reaction rate with GUA (compare Chapter 3). The good agreement of toxicity and reaction rate constant with GUA of these compounds suggests that toxicity is more influenced by the reaction with DNA bases than with GSH. The evaluation of either the proportion of GSH depletion related toxicity and the proportion of DNA damage for compounds of unspecific reactivity thus seems feasible including information about reaction rate constants. Equation 4.2 shows the QSAR of E. coli EC50 values to the reaction rate constant with GUA for the combined data set of EOX, SOX, and EPOX with EPI, NOX and DClB. The reaction rate constant with GUA was sufficient to describe the toxicity and no inclusion of a hydrophobicity term was needed 2 (r =0.94, F=24 for correlation with kGUA and log KOW.) 64

 EC E.coli   k  log 50  =−1.34(±0.18)⋅log GUA  − 2.43(±0.40)  mM   M−1min−1  (4.2) n = 6, r2 = 0.93, F = 52

The unspecifically reacting compound MVIN could not be included in the comparison, because due to the fast hydrolysis the reaction rate with GUA could not be determined.

Figure 4.2: Plot of log EC50 values of E. coli of 2 unspecifically reacting compounds versus MVIN logarithmic reaction rate constants with 1 glutathione (kGSH) (epoxides F, organochlorines Ì) and with 2´- EPI EPI deoxyguanosine (k ) (epoxides „, GUA 0 DClP NOX organochlorines ▲). NOX DClB

DClB -1 NBCl

-2 -3 -2 -1 0 1 2 3 4 -1 -1 log kGUA, log kGSH (M min )

Thus, it can be concluded that the toxicity of all examined epoxides (EPOX, SOX, EOX, NOX, EPI) is primarily based on the reaction with DNA and that the toxicity of all examined compounds with an activated double bond (ACR, IBA, HEA, EA, ACN, ACA) is triggered by reaction with glutathione. These results are in line with the hard and soft acid and base concept (Edwards and Pearson 1962), confirming the hypothesis that hard acids (electrophiles), e.g., epoxides, tend to react with hard bases (nucleophiles), e.g., DNA, and soft acids (electrophiles), e.g., compounds with an activated double bond, tend to react with soft bases (nucleophiles), e.g., glutathione (van

Welie et al. 1992). For DClB, the direct correlation between the reaction rate constant kGUA and toxicity implies that DClB is a directly alkylating chemical, which does not need prior activation, e.g., by epoxidation of the allyl-bond (Their et al. 1995) or by formation of a reactive glutathione conjugate (Beaten et al. 1999), as was observed for other bifunctional organochlorines.

Note that the difference in reactivity between the nucleophiles GSH and GUA is not constant for different electrophiles. This seems trivial, as this can be readily deduced from the Swain-Scott relationship (Swain and Scott 1953), which describes the dependence of second-order reaction rate constants on a nucleophilicity constant n and a substrate (electrophile) constant s. Since the difference of nucleophilicity between GSH and GUA for the examined compounds can be assumed to be constant, the difference of reaction rate constants is higher for electrophiles with a large substrate constant than for compounds with a small substrate constant. Thus, not only the intrinsic reactivity of electrophiles but also the selected nucleophiles influence QSARs set up to describe toxic effects of electrophiles. Reaction rate constants with different nucleophiles are consequently not interchangeable. Therefore, it is mandatory to select the “right” nucleophile for the description of toxic effects. In order to get a closer link to observed toxicity of electrophiles, 65 reactions rates with target nucleophiles like GSH or GUA thus seem to be most suitable descriptors. This was demonstrated by a good correlation of reaction rate constants with glutathione and 2´-deoxyguanosine with toxicity values of substances assigned to GSH depletion related toxicity and DNA damaging chemicals, respectively (Equation 4.1 and 4.2). In the context of toxicity studies we advise to characterize electrophilic reactivity by reaction rate constants with target nucleophiles or nucleophiles closely related to targets instead of the often used model nucleophile 4-(p-nitrobenzyl)-pyridine (NBP) (Deneer et al. 1988; Eder et al. 1982; Hemminki et al. 1983; Hemminki et al. 1980; Hermens et al. 1985; von der Hude et al. 1992)

4.3.2 Comparison of Bacterial Toxicity with Toxicity in Algae, Daphnia, and Fish With the aim to evaluate if toxicokinetic processes in higher organisms influence the target occupation of electrophilic chemicals, toxicity values of E. coli were correlated with consistent data sets for algae, daphnia, and fish. For six compounds, for which data were available, a comparison between bacterial toxicity and algae toxicity was made. As can be seen from the intercept of the linear correlation between algal and bacterial toxicity (Equation 4.3 and Figure 4.3), the bacterial toxicity endpoint was slightly more sensitive. Indicated by the unit slope of the regression, the relative toxicity of the examined compounds however is identical in both organisms. This observed sensitivity difference does not allow the conclusion to be drawn that bacteria are more sensitive than algae, as the two endpoints are arbitrarily chosen. Noteworthy is that bacterial tests are performed within a time frame of up to 3 generation times of the E. coli strain used (3 h of exposure, specific growth rate µ=0.57 h-1, experiments not shown), whereas the duration of the algal tests span only 10% of the generation time (2 h of exposure, specific growth rate µ=0.039 h-1, (Niederer 2002)). Thus, the sensitivity difference might reflect the relative differences of exposure times.

2.5 Figure 4.3: Linear correlation with 95% confidence interval between EC50 growth EOX 2.0 inhibition of E. coli and EC50 of photosystem II inhibition of Scenedesmus vacuolatus. The 1.5 broken line marks the 1:1 correlation. EPI 1.0 EPOX SOX 0.5

BCl 0.0 DClB

-0.5

-1.0 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5

log EC50 E. coli (mM)

66

 EC S.vacuolatus   EC E.coli  log 50  =1.02 (±0.07) ⋅log 50  + 0.45 (±0.06)  mM   mM  (4.3) n = 6,r2 = 0.98,F = 225

Comparison of toxicity in Daphnia magna with toxicity in E. coli is based on 48h LC50 values for the compounds ACR, ACN, ACA, SOX, and EPI. Despite the mixed quality of data for daphnia, derived in static, semi-static and flow-through tests a linear 1:1 correlation could be set up (Equation 4.4 and Figure 4.4), indicating the constant relative toxicity of the examined compounds.

 LC Daphniamagna  EC E.coli  log  50  =0.91(±0.06) ⋅log 50  −1.36(±0.09)  mM   mM  (4.4) n = 5,r2 = 0.99,F = 215

Figure 4.4: Linear correlation with 95% 1.0 confidence interval between EC growth ACN 50 0.5 inhibition of E. coli and 48 h LC50 of Daphnia magna. The broken line marks 0.0 the 1:1 correlation. -0.5 EPI ACA -1.0 SOX -1.5

-2.0

-2.5 ACR -3.0

-3.5

-4.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

log EC50 E. coli (mM)

Comparison of toxicity in fish with toxicity in E. coli was made with 96 h LC50 values of

Pimephales promelas and 14 d LC50 values of Poecilia reticulata. Because EPI and especially BCl hydrolyze quite rapidly (Chapter 3), and because toxicity values were determined in static experiments, the data points were omitted from the correlation to 96 h toxicity of the fish Pimephales promelas, but are still shown in Figure 4.5a. Whereas for data from flow through tests a sound correlation could be set up (Equation 4.5), the correlation with toxicity data from 14id semi-static tests was poor (Equation 4.6 and Figure 4.5b), probably reflecting depletion of fast hydrolyzing substances, like BCl. Toxicity of SOX in the 14 d semi-static tests was surprisingly lower than in the 96 h experiments. Fish toxicity endpoints were more sensitive as could be deduced from the intercept of the linear correlation between fish and bacterial toxicity. Indicated by the slope of the regression, the relative toxicity of examined compounds however is identical for both fish species and bacteria. 67

 LC P. promelas   EC E.col i  log 50  =1.04(±0.07)⋅log 50  −1.79(±0.07)  mM   mM  (4.5) n = 7,r2 = 0.96,F = 242

 LC P.reticulata  EC E.coli  log 50  =1.06(±0.17)⋅ log 50  − 2.25(±0.17)  mM   mM  (4.6) n = 8,r2 = 0.86, F = 38

1.0 1.0 (a) (b) 0.5 ACA 0.5

0.0 0.0 EOX

-0.5 -0.5 ACN ACA EPI -1.0 -1.0 SOX HEA -1.5 BCl SOX -1.5 EA DClP -2.0 IBA -2.0 EPI -2.5 -2.5 BCl EA

-3.0 -3.0 DClB -3.5 ACR -3.5

-4.0 -4.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 -2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0

log EC50 E. coli (mM) log EC50 E. coli (mM)

Figure 4.5: Linear correlation with 95% confidence interval between EC50 growth inhibition of E. coli and a) 96 h LC50 of the fish Pimephales promelas (● flow through tests, { static tests) and b) 14 d LC50 of the fish Poecilia reticulata (■ semi-static tests). The broken line marks the 1:1 correlation.

The compounds in the QSAR for algae (Equation 4.3) include DNA damaging compounds (EPOX, SOX, EOX, DClB, EPI) and an unspecifically reacting compound (BCl). Daphnia and fish toxicity data also comprise compounds from different reactive modes of toxic action. The log linear regressions of bacterial toxicity to toxicity in higher organisms show unit slope and thus indicate the same relative sensitivity in all here investigated organisms. This constant relative sensitivity is independent of the electrophile´s mode of toxic action which suggests that the mode of toxic action is the same in all organisms. There are as well arguments that support this hypothesis as well as considerations that oppose it.

A supporting study was conducted by Freidig et al (1999). He observed that compounds with an activated double bond, which are assigned to GSH depletion related toxicity, cause a critical depletion rate constant of glutathione resulting in 50% lethality of fish, which suggests that the mode of toxic action in bacteria and fish remains the same. In a toxicity study of EA in rat no DNA adducts but protein adducts were found (Ghanayem et al. 1987). Van Welie at al. (1992) found that exposure to ethylene oxide and 1,2-dibromoethane in rat resulted in ratios of 1:10 and 1:107 of DNA to GSH adducts, respectively. This underlines the retained DNA interaction of epoxides in 68 higher organisms and indicates that haloalkanes might, as haloalkenes, react unspecifically with both important biomolecules.

However, a comparison of cytotoxicity and glutathione depletion in algae and E. coli indicates a potential influence of metabolism on the occurrence of reactive modes of toxic action. Comparing concentrations of cytotoxicity (Figure 4.3 and 4.6) for algae and bacteria, algae were found to be less sensitive than bacteria. Glutathione depletion (Niederer 2002) in algae, however, occurred at much lower toxicant concentrations than in bacteria (Chapter 2) (Figure 4.6). Whereas in bacteria no GSH depletion was observed for DClB and SOX, both compounds depleted GSH in algae. Testing GSH depletion in E. coli for DClB was however limited due to the poor water solubility. A GSH deficient strain was much more sensitive towards DClB than a GSH competent E. coli strain (Chapter 2)

Figure 4.6: Comparison of sensitivity of 2.5 toxicity endpoints for electrophiles in algae 2 and bacteria. 1.5

1 (mM) 50 0.5

EC log 0

-0.5 ● log EC50 values of photosystem II inhibition of S. vacuolatus, { log EC values of GSH depletion of S. 50 -1 vacuolatus, ▲ log EC50 values of growth inhibition of EPOX EOX BCl EPI E. coil, Ì log EC50 values of GSH depletion of E. coli.

One explanation for the discrepancy of the GSH depleting potency in algae and E.coli might be the influence of glutathione S-transferases (GSTs). Bacteria are known to possess only small concentrations of glutathione S-transferases (Penninckx and Elskens 1993), catalyzing the reactions of GSH with electrophiles. But not only the concentrations are low, additionally bacterial GSTs were found to have lower activities than for example mammalian GST (Kerklaan et al. 1985), which enhance chemical reaction rate constants up to 104 – 105 times (Douglas 1988). Thus, it can be assumed that the higher susceptibility of the GSH homeostasis in algae compared to bacteria is determined by the activity of GSTs. An additional point is the difference of substrate specificity. For example, whereas mammalian GSTs effectively catalyze the substitution of SOX, GSTs of E. coli were found to have no activity with SOX (Kerklaan et al. 1985). Reaction of SOX with glutathione in higher organisms was found to be an important detoxification mechanism (Pacifici et al. 1987) reducing the extent of toxic effects on DNA (Shield and Sanderson 2001).

How big can the influence of GSTs on the target site occupation be? Naturally, GSTs could have a high impact comparing organisms with GSTs of different substrate specificities (see discussion above). However, the relative influence on GST catalyzed reactions is small, as can be seen in a comparison of the enhancement factor of chemical and enzymatically catalyzed 69 reactions (Table 4.2). The difference of enhancement of very dissimilar reaction mechanisms was shown to be a factor of only 7. Thus, model systems describing the biological reaction with GSH with chemical reaction rate constants are applicable, as long as the impact of differences in GST catalysis efficiency for different substrates is negligible, because the chemical reaction rate constants itself span more than one order of magnitude. This is the case for the series of the reaction of compounds with an activated double bond with glutathione. Their reaction rate constants span a range of 104.

Table 4.2: Enhancement of aliphatic and aromatic nucleophilic substitution and Michael addition by glutathione S-transferases from rat liver (adapted from figure and text, Ketterer 1982). substance reaction mechanism with GSH enzymaticalrate

chemical rate 4-nitrobenzyl chloride nucleophilic substitution of chlorine 1.3 · 104 1-chloro-2,4- nucleophilic aromatic substitution of chlorine 8.6 · 103 dinitrobenzene 1,2-dichloro-4-nitrobenzen nucleophilic aromatic substitution of chlorine 3.6 · 103 1-(p-nitrophenoxy)propane- nucleophilic substitution of epoxide 1.8 · 103 2,3-oxide t-4-phenyl-3-butene-2-one Michael addition on activated double bond 2.0 · 103

However, when the range of chemical reaction rate constants is small, catalysis by GSTs may change the relative ranking of reactivity. For the epoxides EOX, EPI, and MVIN, that react according to second-order nucleophilic substitution (see Chapter 3), chemical reaction rate constants increase in the order EOX < MVIN < EPI and differ by a factor of 12. In contrast, enzymatic catalysis of a GST from the microorganism Rhodococcus sp. for EOX and EPI was three times lower than that for MVIN (van Hylckama Vieg et al. 1999).

Thus, QSARs using chemical reaction rate constants may be obstructed by differences in metabolism, e.g., by differences of GST catalysis. The influence of enzyme catalysis on the relation of chemical reactivity to toxicity is high if the chemical reactivities of the examined compounds differ by only one order of magnitude, which is approximately equivalent to the variation of the relative catalytic enhancement (Table 4.2). In this case, reaction rate constants can only then successfully be applied when the examined compounds are proportionally catalyzed to their chemical reactivity. From the relative enhancement of chemical reaction by enzyme catalysis it can also be concluded that the influences of catalytic enhancement by GSTs on intended QSARs diminishes if the chemical reactivities of the examined compounds differ by more than one order of magnitude.

The reaction with DNA is however not catalyzed and alkylation adducts are mainly repaired by the adduct unspecific SOS response (see Chapter 2). Thus, if steric factors in DNA result in only a small change of the reaction rate constant in DNA compared to the reaction rate constant with the model nucleophile 2´-deoxyguanosine (see Chapter 3), the toxicity caused by DNA 70

damage might be well described by kGUA (Equation 4.2). Note that “DNA damage” is not synonymous with mutagenicity. Besides alkylation of specific nucleophilic sites of DNA bases, which results in the exchange of bases when replicated, DNA damage also comprises alkylation of all nucleophilic sites in DNA that may cause merely physiological stress due to induction of DNA repair processes. Because mutations are induced by alkylation of specific sites and reaction rate constants with nucleophiles, e.g., with 4-(p-nitrobenzyl)-pyridine (Hemminki et al. 1983; Hemminki et al. 1980) or guanine (Hemminki et al. 1980) reflect reactions of a totally different type, failures to predict mutagenicity from reaction rate constants of electrophiles with mechanistically irrelevant nucleophilic sites can be explained.

Here, the influence on the outcome of a QSAR of only one detoxification process has been discussed, and GSTs were found to have limited modulating influence when considering chemicals with a wide range of reactivity. However, other detoxification processes may have a still unevaluated impact on the target site occupation of electrophiles. An important example is the detoxification of epoxides by epoxide hydrolases (Fretland and Omiecinski 2000; Swaving and de Bont 1998).

QSARs for reactive chemicals that are intended to describe toxicity by chemical reaction rate constants can therefore be applied for only a limited range of compounds, for which the mode of toxic action and a rough perception of metabolic influences are known. Because the correct application of such QSARs already requires important toxicological knowledge, the claim that QSARs provide a predictive tool must, at least for reactive chemicals, be challenged.

4.3.3 Hazard Assessment Toxicity of electrophilic chemicals is determined by their reaction with susceptible biological nucleophiles, i.e., peptides and proteins and DNA. As was shown with a set of E.coli biosensors (Chapter 2 and this chapter), epoxides tend to react with DNA while compounds with an activated double bond prefer the tripeptide GSH as a target. For other “unspecifically reacting” compounds, reactions with both targets determine their toxicity. This clear distinction of reactive modes of toxic action diminishes in higher organisms, because they may have a broader metabolic competence, as was shown for glutathione S-transferases catalyzing the reaction of GSH with SOX. However, as is shown by multiple studies (Ehrenberg and Hussain 1981; Giri 1997; Griem 2001; Uziel et al. 1992), the hazard of chemicals recognized as “DNA damaging” chemicals to impair DNA remains. As was shown by correlations of toxicity of electrophilic chemicals between different organisms, the relative toxicity of chemicals of different mode of toxic action remains the same. This is an indication that the pattern of glutathione depletion related toxicity and toxicity based on DNA damage is constant for microorganisms, algae, daphnids, and fish, despite certain differences in metabolic transformation reactions. However, clear evidence is missing, and research on toxicity of electrophiles in higher organisms should concentrate on finding toxicity 71 indicators that clearly link observable toxicity with mechanisms of toxic action. Despite lacking a mechanistic proof, the observed linearity (Equations 4.3 - 4.6) of toxicity in algae, daphnia, and fish provides a practical tool to predict toxicity with a simple bacterial test. Independent from the final proof, the mode of toxic action classification provides a clear indication of genotoxic hazards.

QSARs based on reaction rate constants as descriptors are suitable as exploratory tool but lack suitability as predictive tool. Such QSARs are only valid for very small sets of chemicals that act not only according to the same mechanism in a given biological organism but also have identical chemical reaction mechanism. Consequently, correct assignment of a compound to the appropriate QSAR requires such detailed information on the mode of action and toxicokinetic processes that experimental data on a more integrative level, e.g., the E. coli biosensors, are better suited as descriptors in predictive models.

4.4 Literature Cited

Beaten, A., Tafazoli, M., Kirsch-Volders, M., and Geerlings, P. (1999). Use of the HSAB principle in quantitative structure-activity relationships in toxicological research application to the genotoxicity of chlorinated hydrocarbons. Int. J. Quantum Chem. 74, 315-355. Chemfate (2002). Chemfate data base: http://esc.syrres/efdb/Chemfate.htm, accessed on 11/21/2002. Deneer, J. W., Sinnige, T. L., Seinen, W., and Hermens, J. L. M. (1988). A quantitative structure- activity relationship of some epoxy compounds to the guppy. Aquat. Tox. 13, 195-204. Douglas, K. (1988). Reactivity of glutathione in model systems. In Glutathione Conjugation: Mechanisms and Biological Significance, H. Sies and B. Ketterer, eds., pp. 1-41. Academic Press, London, Great Britain. EC, ed. (1996). Technical Guidance Document in Support of Commission Directive 93/67/EEC on Risk Assessment for New Notified Substances and Commission Regulation (EC) No 1488/94 on Risk Assessment for Existing Substances. Office for Official Publications of the European Communities, Luxembourg. Eder, E., Neudecker, T., Lutz, D., and Henschler, D. (1982). Correlation of alkylating and mutagenic activities of allyl and allylic compounds. Chem. Biol. Interact. 38, 303-315. Edwards, J. O., and Pearson, R. G. (1962). The factors determining nucleophilic reactivities. J. Am. Chem. Soc. 84, 16-24. Ehrenberg, L., and Hussain, S. (1981). Genetic toxicity of some important epoxides. Mutation Res. 86, 1-113. EPA (2002). http://www.epa.gov/ecotox, accessed on 11/17/2002. Escher, B. I., and Hermens, J. L. M. (2002). Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201-4217. Escher, B. I., and Schwarzenbach, R. P. (2002). Mechanistic studies on baseline toxicity and uncoupling of organic compounds as a basis for modeling effective membrane concentrations in aquatic organisms. Aquat. Sci. 64, 20-35. Freidig, A., Verhaar, H. J. M., and Hermens, J. L. M. (1999). Comparing the potency of chemicals with multiple modes of toxic action in aquatic toxicity: acute toxicity due to narcosis versus reactive toxicity of acrylic compounds. Environ. Sci. Technol. 33, 3038-3043. 72

Fretland, A. J., and Omiecinski, C. J. (2000). Epoxide hydrolases: biochemistry and molecular biology. Chem. Biol. Interact. 129, 41-59. Ghanayem, B. I., Burka, L. T., and Metthews, H. B. (1987). Ethyl acrylate distribution, macromolecular binding, excretion, and metabolisms in male Fischer 344 rats. Fundam. Appl. Toxicol. 9, 389-397. Giri, A. K. (1997). Genetic toxicology of epichlorohydrin: A review. Mutation Res. Rev. 386, 25-38. Griem, H. (2001). Use of covalent binding in risk assessment. In Biological Reactive Intermediates, P. M. Dansette, R. Snyder, M. Delaforge, G. G. Gibson, H. Griem, D. J. Jollow, T. J. Monks and I. G. Sipes, eds., Vol. VI, pp. 715-722. Kluwer Academic/Plenum Publishers, New York, NY, USA. Hemminki, K., Falck, K., and Linnainmaa, K. (1983). Reactivity, SCE induction and mutagenicity of benzyl chloride derivatives. J. Appl. Toxicol. 3, 203-207. Hemminki, K., Falck, K., and Vainio, H. (1980). Comparison of alkylation rates and mutagenicity of directly acting industrial and laboratory chemicals: epoxides, glycidyl ethers, methylating and ethylating agents, halogenated hydrocarbons, hydrazine derivatives, aldehydes, thiuram and dithiocarbamate derivatives. Arch. Toxicol. 46, 277-85. Hermens, J. L. M. (1990). Electrophiles and acute toxicity to fish. Environ. Health Persp. 87, 219- 225. Hermens, J. L. M., Busser, F., Leeuwach, P., and Musch, A. (1985). Quantitative correlations studies between the acute lethal toxicity of 15 organic halides to the guppy (Poecilla reticulata) and chemical reactivity towards 4-nitrobenzylpyridine. Toxicol. Environ. Chem. 9, 219-236. Hermens, J. M. L., and Leeuwangh, P. (1982). Joint toxicity of mixtures of 8 and 24 chemicals to the guppy (Poecilia reticulata). Ecotoxicol. Environ. Safety 6, 302-310. Kerklaan, P. M., Zoetemelk, C. E. M., and Mohn, G. R. (1985). Mutagenic activity of various chemicals in Salmonella strain TA 100 and glutathione-deficient derivatives - on the role of glutathione in the detoxification or activation of mutagens inside bacterial cells. Biochem. Pharm. 34, 2151-2156. Ketterer, B. (1982). The role of nonenzymatic reactions of glutathione in xenobiotic metabolism. Drug Metab. Rev. 13, 161-187. Legierse, K. C. H. M., et al. (1999). Analysis of the time-dependent acute aquatic toxicity of organophosphorus pesticides: The critical target occupation model. Environ. Sci. Technol. 33, 917-925. Lipnick, R. L. (1995). Structure-activity relationships. In Aquatic Toxicology, G. M. Rand, ed., pp. 609-655. Taylor & Francis, London, Great Britain. Niederer, C. (2002). Toxische Effekte und Wirkmechanismen von Reaktivchemikalien mit Chlor- und Epoxidfunktionen in Grünalgen. Diploma thesis, Eidg. Techn. Hochschule (ETH), Zürich, Switzerland. Pacifici, G. M., et al. (1987). Detoxification of styrene oxide by human liver glutathione transferase. Human. Toxicol. 6, 483-489. Penninckx, M. J., and Elskens, M. T. (1993). Metabolism and function of glutathione in micro- organisms. Adv. Microb. Physiol. 34, 239-301. Shield, A. J., and Sanderson, B. (2001). Role of glutathione S-transferase Mu (GSTM1) in styrene-7,8-oxide toxicity and mutagenicity. Environ. Mol. Mutagen. 37, 285-289. Swain, C. G., and Scott, C. B. (1953). Quantitative correlations of relative rates. Comparison of hydroxide ion with other nucleophilic reagents toward alkyl halides, esters, epoxides and acyl halides. J. Am. Chem. Soc. 75, 141-147. 73

Swaving, J., and de Bont, J. A. M. (1998). Microbial transformation of epoxides. Enzym. Micr. Techn. 22, 19-26. Tanii, H., and Hashimoto, K. (1982). Structure-toxicity relationships of acrylates and methacrylates. Toxicol. Lett. 11, 125-129. Thier, R., et al. (1995). Enhancement of bacterial mutagenicity of bifunctional alkylating agents by expression of mammalian glutathione S-transferase. Chem. Res. Toxicol. 8, 465-472. Tobler, N. B. (2002). Toxizität und Reaktivität von Acrylaten in E. coli als Modell für die ökotoxikologische Risikobewertung. Diploma thesis, Eidg. Techn. Hochschule (ETH), Zürich, Switzerland. Uziel, M., et al. (1992). DNA adduct formation by 12 chemicals with populations potentially suitable for molecular epidemiological studies. Mutat. Res. 277, 35-90. van Hylckama Vieg, J. E. T., Kingma, J., Kruizinga, W., and Janssen, D. B. (1999). Purification of a glutathione S-transferase and a glutathione conjugate-specific dehydrogenase involved in isoprene metabolism in Rhodococcus sp. strain AD45. J. Bacteriol. 181, 2094-2101. van Welie, R. T., van Dijck, R. G., Vermeulen, N. P., and van Sittert, N. J. (1992). Mercapturic acids, protein adducts, and DNA adducts as biomarkers of electrophilic chemicals. Crit. Rev. Tox. 22, 271-306.

Verhaar, H. J. M., et al. (1999). An LC50 versus time model for the aquatic toxicity of reactive and receptor-mediated compounds. Consequences for bioconcentration kinetics and risk assessment. Environ. Sci. Technol. 33, 758-763. Verhaar, H. J. M., van Leeuwen, C. J., and Hermens, J. L. M. (1992). Classifying environmental pollutants. 1: Structure-activity relationships for prediction of aquatic toxicity. Chemosphere 25, 471-491. von der Hude, W., Carstensen, S., Gürtler, R., and Obe, G. (1992). Structure-activity relationships of epoxides: induction of sister-chromatid exchange in V79 cells by enantiomeric epoxides. Mut. Res. 278, 289-297.

74

5 Outlook

This thesis has shown that electrophilic chemicals react to a different extent with different nucleophilic targets in biological organisms. Examined electrophiles included reactive organochlorines, epoxides, and compounds with an activated double bond. Using a set of bacterial biosensors derived from E. coli that were either proficient or deficient in glutathione, an important detoxifying peptide, or capable and incapable of DNA repair, a close link between these mechanisms of action and observed cytotoxicity could be made. From the pattern of response of these biosensors three different classes of reactive modes of toxic action were discriminated: glutathione depletion related toxicity, DNA damage, and unspecific reactivity. The specific reactive toxicity could be correlated to either the chemical reaction rate constant with the target molecule glutathione or to 2´-deoxyguanosine, which stands as a model for DNA. Correlation of bacterial toxicity to toxicity in algae, daphnia, and fish yielded a linear relationship with slope one, thus the relative toxicity of different electrophiles was identical in all organisms. This finding suggests but does not give a final proof that the mode of toxic action in higher organisms might be the same as in bacteria, despite modulating effects of uptake, distribution, and metabolism. Whereas the importance of either reaction to glutathione and DNA in bacteria could be evaluated using genetically modified strains, such an approach is not possible in higher organisms. Thus, alternative methods need to be developed to validate the findings of this study and to relate them to higher biological level.

Promising methods that could be used to trace the mechanistic links between initial reactions of electrophiles and nucleophiles and observable toxic effects in higher organisms include expression profiling, e.g., gene arrays, and translational profiling (Figure 5.1). Gene arrays determine the transcriptional activity of genes, whose function is experimentally resolved. Translational profiling includes the identification and quantification of proteins as a complete complement to the genome. From the comparison of transcriptional or translational patterns and specific modes of toxic action, identified by bacterial biosensors, complying results would not only confirm the assumption of the same mode of toxic action in bacteria and higher organism but would also allow one to deduce, which genes and proteins participate in defense or repair mechanisms and would give hints if important modes of toxic action were overlooked. The complete genome of E. coli has already been sequenced and gene arrays are available (Wisconsin 2002), which would make a good start for a direct comparison between transcriptional patterns and the biosensors developed in this thesis. Since complete genomic information of the algal species Chlamydomonas reinhardtii (Duke 2002) and the zebra fish (Danio rerio) (Sanger 2002) is or will soon be available, correlations between different levels of biological organization 76 will be possible. However, the interpretation of gene arrays and translational profiles is difficult, e.g., one third of genes of yeast Saccharomyces cervisiae, was found to be influenced by the treatment with DNA alkylating agents (Jelinsky et al. 2000). Though, the study also demonstrated that, once the mechanisms of action are precisely known, e.g., the reaction of the carcinogen methyl methane-sulfonate with DNA, function of new genes might be identified.

Figure 5.1: Comparison of electrophile damage sensor expression and translational mode of pattern of ? toxic profiles with mode of toxic gene induction action classification by action bacterial biosensors for nucleophile adduct pattern of defense identification of the mode of and repair proteins toxic action in higher damage sensor organisms electrophile mode of pattern of ? toxic gene induction action nucleophile adduct pattern of defense and repair proteins

toxicokinetic initiating reaction expression profiles translational profiles

The proposed biosensors not only offer a more detailed insight in toxic mechanisms of selected reactive chemicals, but might also be used as a part of a mechanism based screening set for the risk assessment of single compounds and complex mixtures.

This screening set is suggested to include a number of biosensors and other subcellular and cellular test systems, which identify the mechanism of action of tested chemicals and quantify their effects. The majority of the chemicals introduced into the environment act as baseline toxicants (Russom et al. 1997). They may pose a problem in the environment if they are additionally persistent and bioaccumulative but their toxicity per se is no problem. It is important to identify the remaining fraction that pose particular hazards, and which includes reactive and specifically acting compounds. The set of biosensors and test systems needs to be carefully designed to identify most relevant interactions with biomolecules that can trigger toxic effects (for systematic compilation of such interactions see Escher and Hermens (2002)). Such interactions include noncovalent interaction with certain enzymes leading to receptor mediated toxicity like the inhibition of acetylcholinesterase, and interference with biological membrane`s structure and functioning resulting, e.g., in photosynthesis inhibition, uncoupling, or inhibition of ATPase. Suitable test systems for toxic effects related to mechanisms of action in energy transducing membranes have already been evaluated (Escher et al. 1999; Escher et al. 1997; Hunziker et al. 2002). Further examples include tests for identification of hormone disrupting chemicals (Andersen et al. 1999). Alike the here presented biosensors for identification and assessment of reactive toxicity, a crucial feature for being included in such a test battery is to be able to link observed effects to the primary interaction of toxicants with the target sites. 77

In the environment, we are mostly confronted with assessing effects of complex mixtures, e.g., effluents from municipal or industrial wastewater treatment plants, which we can not simply be ascribed to a single chemical. Normally a mixture of chemicals with the same mechanism of action reacts according to concentration addition, i.e. one component of a mixture can be replaced by an equipotent concentration of another chemical. A mixture of chemicals with dissimilar action most likely will react according to independent action, i.e. the chemicals can be treated independently and we observe effect addition. Synergistic or antagonistic effects with respect to concentration addition were only seldom reported for mixtures (Escher and Hermens 2002). The effects of mixtures of reactive chemicals have been rarely studied. Such studies are complicated by the fact that reactive chemicals may or may not have different targets, and that they additionally could react with each other. For mixtures of acrylates in rat hepatocytes dose- addition was found, supporting the finding that acrylates mainly react with glutathione (Freidig et al. 2001). However, binary mixtures of reactive chemicals were shown to have both antagonistic and synergistic effects in a test with the luminescent bacteria Vibrio fischeri (Chen and Yeh 1996). Thus, results are inconclusive and no common pattern of response for reactive chemicals was found. The proposed biosensors provide a perfect tool to investigate the possibly interactive affects of reactive chemicals in mixtures because effects at specific target sites can be independently identified.

5.1 Literature Cited

Andersen HR, et al. (1999). Comparison of short-term estrogenicity tests for identification of hormone-disrupting chemicals. Environ. Health Persp. 107, 89-108, Suppl. 1. Chen, C. Y., and Yeh, J. T. (1996). Toxicity of binary mixtures of reactive toxicants. Environ. Toxicol. Wat. Qual. 11, 83-90. Duke (2002). http://www.biology.duke.edu/chlamy_genome/microarrays.html, accessed on 11/02/2002. Escher, B. I., and Hermens, J. L. M. (2002). Modes of action in ecotoxicology: their role in body burdens, species sensitivity, QSARs, and mixture effects. Environ. Sci. Technol. 36, 4201-4217. Escher, B. I., Hunziker, R., and Schwarzenbach, R. P. (1999). Kinetic model to describe the intrinsic uncoupling activity of substituted phenols in energy transducing membranes. Environ. Sci. Technol. 33, 560-570. Escher, B. I., Snozzi, M., Häberli, K., and Schwarzenbach, R. P. (1997). A new method for simultaneous quantification of the uncoupling and inhibitory activity of organic pollutants in energy transducing membranes. Environ. Sci. Technol. 16, 405-414. Freidig, A., Hofhuis, M., Van Holstijn, I., and Hermens, J. (2001). Glutathione depletion in rat hepatocytes: a mixture toxicity study with α, β-unsaturated esters. Xenobiotica 31, 295-307. Hunziker, R. W., Escher, B. I., and Schwarzenbach, R. P. (2002). Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH. Environ. Toxicol. Chem. 21. 78

Jelinsky, S. A., Estep, P., Church, G. M., and Samson, L. D. (2000). Regulatory networks revealed by transcriptional profiling of damaged Saccharomyces cerevisiae cells: Rpn4 links base excision repair with proteasomes. Mol. Cell. Biol. 20, 8157-8167. Russom, C. L., et al. (1997). Predicting modes of toxic action from chemical structure: Acute toxicity in the fathead minnow (Pimephales promelas). Environ. Toxicol. Chem. 16, 948-967. Sanger (2002). http://www.sanger.ac.uk/Projects/D_rerio/blast_server.shtml, accessed on 11/27/2002. Wisconsin (2002). http://www.genome.wisc.edu/functional/microarray.htm, accessed on 11/27/2002. Acknowledgments

I want to express my gratitude to René Schwarzenbach for his guidance, advice, and for the support writing this thesis. I owe much to Beate Escher who supervised my work, and continuously gave me encouragement, confidence, and great freedom throughout my dissertation.

The financial support of my dissertation by the Center of Excellence on Risk and Safety Science (KOVERS) at the ETH Zürich is gratefully acknowledged.

Sincere thanks to Joop Hermens and Paolo Landini for valuable advices and critical discussions during the dissertation and for co-examining this work.

Many thanks to Werner Angst for various discussions about reaction mechanisms, Michael Volkert for sharing his expert knowledge in DNA repair, Stéphane Vuilleumier for his discussions about glutathione S-transferases, and Ivonne Rietjens for calculating some quantum chemical parameters and her advice to refrain from modelling with them.

Nicole Tobler and Christian Niederer greatly contributed with their diploma thesis to this work. I am thankful for their careful work and took pleasure in their critical approaches. Thanks to Andreas Freidig for showing me how much fun and how efficient experimental work could be. Thanks also to Jakov Bolutin for assisting me in the lab even after he bathed in octanol, to Andreas Brunner for plating and counting numerous colonies of E. coli, and to Karin Rüfenacht for introducing me in some genetic techniques.

René Hunziker, Bianca Wisner, Nina Schweigert, and all unnamed former and current members of the Schwaba- and the MIX-group a big thank for making my time at the EAWAG memorable and enjoyable.

Thanks to all my friends for keeping patience and understanding especially at the end of my PhD time. I am much obliged to my mother who packed her things and travelled to Zurich every time I needed her helping hand. A special thank to my beloved ones Saska and Michael who kept me from working and busy otherwise.

Curriculum Vitae

July 20, 1968 Born in Hamburg, Germany

1975-1988 Primary and Secondary School in Hamburg, Germany

1988-1994 Degree in Biotechnology at Technical University of Braunschweig, Germany

1993-1994 Diploma Thesis in Technical Chemistry, Dow Germany Inc., Stade, Germany

1994-1995 Research Assistant in Biotransformation and Biosensoric, Technical University Hamburg-Harburg, Germany

1995-1998 Assistant Lecturer in Environmental Management and Technology, Technical College of Wedel, Germany

1999-2000 Graduate Studies in Risk and Safety, ETH Zürich, Switzerland

1999-2001 Teaching Assistant in Environmental Science Department, ETH Zürich, Switzerland

1998-2002 Doctoral Studies in Environmental Chemistry, EAWAG/ETH Zürich, Switzerland