Quantification of the Membrane Toxicity of Hydrophobic Ionogenic Organic Compounds (Hiocs): Role of Uptake and Speciation for Single Compounds and Binary Mixtures

Diss. ETH No. 14066

Quantification of the membrane toxicity of hydrophobic ionogenic organic compounds (HIOCs): Role of uptake and speciation for single compounds and binary mixtures

René W. Hunziker

Zürich, 2001

Diss. ETH No. 14066

Quantification of the membrane toxicity of hydrophobic ionogenic organic compounds (HIOCs): Role of uptake and speciation for single compounds and binary mixtures

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

Presented by René W. Hunziker

Dipl. Natw. ETH born October 10th, 1968 citizen of Moosleerau (AG)

Accepted on the recommendation of Prof. Dr. R.P. Schwarzenbach, examiner Prof. Dr. U.P. Fringeli, co-examiner Dr. R. Altenburger, co-examiner Dr. B.I. Escher, co-examiner

Zürich, 2001

Table of contents

Summary...... VII Zusammenfassung...... IX

1. General introduction...... 1 1.1 Biomembranes as site of toxic action...... 2 1.2 Membrane toxicity of hydrophobic ionogenic organic compounds (HIOC) ...... 4 1.3 Objectives...... 6 1.4 Outline...... 7

2. The pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water...... 9 2.1 Introduction...... 10 2.2 Materials and methods...... 12 2.2.1 Chemicals ...... 12 2.2.2 Aqueous solutions ...... 12 2.2.3 Preparation of liposomes...... 12 2.2.4 Determination of liposome-water and chromatophore-water distribution ratios of tributyltin and triphenyltin...... 13 2.2.5 Analytical procedures ...... 14 2.2.6 Calculation of triorganotin concentrations in the liposomes ...... 15 2.3 Results and discussion...... 16 2.3.1 Liposome-water partitioning ...... 16 2.3.2 Chromatophore-water partitioning ...... 21 2.4 Conclusion...... 22

3. Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH...... 23 3.1 Introduction...... 24 3.1.1 Effects of tinorganic compounds on the energy metabolism of cells...... 24 3.2 Materials and methods...... 28 3.2.1 Chemicals ...... 28 3.2.2 Stock solutions...... 28 3.2.3 Preparation of the chromatophores ...... 29

3.2.4 Protonophoric shuttle and bc1 complex inhibition...... 29 3.2.5 ATP synthesis inhibition...... 30

3.2.6 Definition of the nominal chromatophore concentration, ccph,nom ...... 31 3.3 Results and discussion...... 32

3.3.1 Inhibition of the bc1 complex ...... 32

– III – 3.3.2 Comparison of OH–/Cl– antiport and OH– uniport ...... 33 3.3.3 OH– uniport ...... 33 3.3.4 Inhibition of ATP synthesis ...... 38 3.3.5 Comparison of the different mechanisms of action ...... 39

4. Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects ...... 41 4.1 Introduction...... 42 4.2 Material and methods ...... 44 4.2.1 Chemicals ...... 44 4.2.2 Determination of the uncoupling activity ...... 44 4.2.3 Experiments with single compounds...... 45 4.2.4 Experiments with binary mixtures ...... 45 4.2.5 Physicochemical parameters ...... 46 4.3 Quantitative evaluation and representation of the experimental data...... 46 4.4 Results and discussion ...... 48 4.4.1 Evaluation of the mixed dimer model ...... 48 4.4.2 Effect of pH...... 54 4.4.3 Effect of substitution pattern ...... 55 4.4.4 Comparison with concentration addition...... 56 4.4.5 Comparison with data from literature ...... 58 4.4.6 Significance for environmental risk assessment of mixtures...... 61

5. General discussion and conclusions ...... 63

References...... 67

Appendix A: The energy-transducing membran of R. sphaeroides and the derivation of the uncoupling activity by time-resolved spectroscopy ...... 77 Chromatophores of R. sphaeroides as model for energy-transducing membranes...... 78 Determination of build-up and decay kinetics of the membrane potential with time-resolved spectroscopy ...... 80 Deduction of the uncoupling activity ...... 82 References ...... 85

Appendix B: Kinetic model to describe the intrinsic uncoupling activity of substituted phenols in energy transducing membranes...... 87

Acknowledgments ...... 100 Curriculum Vitae...... 101

– IV – Tables

Table 2.1: pKa, liposome-water, chromatophore-water, and octanol-water distribution ratios of tributyltin and triphenyltin...... 19

Table 4.1 Physicochemical parameters of phenolic uncouplers and anisols ...... 50

Table 4.2 Uncoupling activities of the single phenolic uncouplers and anisols at various pH values...... 51

Table 4.3 Parameter describing the synergistic effect of binary mixtures of phenolic uncouplers ...... 52

Figures

Figure 1.1 Sequence of processes from the introduction of a chemical in the environment to the toxic effects exerted by this chemical depicted for a hydrophobic membrane toxicant ...... 2

Figure 1.2 Simplified picture of the energy conversion in energy-transducing membranes and possible mechanisms of toxic action...... 3

Figure 2.1 pH dependence of the speciation of tributyltin and triphenyltin in aqueous solution ...... 11

Figure 2.2: Liposome-water distribution of tributyltin and triphenyltin...... 16

Figure 2.3 Comparison of the pH dependence of the apparent liposome-water and the octanol-water distribution ratio of tributyltin and triphenyltin ...... 18

Figure 2.4: Chromatophore-water distribution of tributyltin ...... 21

Figure 3.1 Chemiosmotic coupling of energy-transducing membranes and possible modes of toxic action of triorganotin compounds ...... 25

Figure 3.2: Ubiquinol reduction site, Qn, and ubiquinol oxidation site, Qp, in the

bc1 complex...... 27

Figure 3.3: Effect of tributyltin on the kinetics of the reduction and oxidation

of cyt b561 ...... 32

Figure 3.4: Effect of nigericin and tributyltin on the membrane potential Dy...... 34

– V – Figure 3.5 Dependence of the OH– uniport and the overall ATP synthesis inhibition on the concentration of tributyltin and of triphenyltin at pH = 7.5 and pH = 6.1 ...... 36

Figure 3.6: Effect of tributyltin and triphenyltin on the ATP synthesis ...... 38

Figure 4.1 Model of the protonophoric shuttle in an energy-transducing membrane...... 42

Figure 4.2 pH dependence of the uncoupling activities of 3,4-dinitrophenol and 3,4,5-trichlorophenol...... 49

Figure 4.3 Dependence of the uncoupling activity on the membrane concentration of 3,4-dinitrophenol and 3,4,5-trimethylphenol at pH = 7.4 ...... 53

Figure 4.4: Effect of pH and substitution pattern on the observed interaction in the uncoupling activity...... 55

– VI – Summary

An increasing number of the organic pollutants in the environment belong to the group of hydrophobic ionogenic organic compounds (HIOCs). Among them are pesticides, drugs, and personal care products. Often, an acid or base group is introduced in the development process to make the compound less hydrophobic and less persistent. However, due to this ionogenic group, HIOCs can also exert specific toxic effects in organisms. Due to their hydrophobicity, HIOCs accumulate in the biomembranes where they may interfere with fundamental membrane functions. In this thesis compounds from two classes of HIOCs have been investigated: Two triorganotin compounds (tributyltin and triphenyltin) and binary mixtures of a series of substituted phenols. Although some of the compounds are used as pesticides or biocidals, they were mainly chosen for structural reasons as model compounds. All of them are known to interfere with the energy conversion chain in membranes. Three primary modes of action can be differentiated by which energy conversion is disturbed: (i) The unspecific accumulation of any hydrophobic compound in the phospholipid bilayer perturbs the bilayer structure and increases its permeability (baseline toxicity). (ii) Various enzyme complexes in the membrane can be inhibited by a specific interaction of a HIOC with binding sites of the enzyme. (iii) Some HIOCs can exert a transport of charged species, mostly protons, thereby short-circuiting the chemiosmotic proton cycle which is essential for the synthesis of ATP and transport purposes. This mechanism is termed uncoupling. The effects of the model compounds on energy-transducing membranes have been investigated on membrane vesicles (chromatophores) of the photosynthetic purple bacterium Rhodobacter sphaeroides with time-resolved spectroscopy. This method enables one to distinguish the above-mentioned primary modes of action. The effects were related to the concentration of the compound present at the target site, i.e. the chromatophore membrane. Concentrations in the chromatophore membrane were deduced from liposome-water and chromatophore-water distribution ratios. While these values are available for most of the phenols investigated, they needed to be determined for tributyltin and triphenyltin. The liposome-water distribution of both, tributyltin and triphenyltin, showed a weak pH dependence. This dependence was modeled assuming pH-independent distribution ratios for the triorganotin cation species and the hydroxo complex. The distribution ratios of both, the cation and the hydroxo complex of triphenyltin

– VII – exceeded those of tributyltin by a factor of ten. The distribution ratio of the triorganotin cation exceeded that of the neutral hydroxo complex by a factor of two. It is postulated that the sorption of the cation is governed by complex formation with ligands in the phospholipids, presumably the phosphate group. In addition, the chromatophore- water distribution ratio was investigated for tributyltin. At high pH it was found to be lower than the liposome-water distribution ratio. Therefore, the sorption of the hydroxo complex is assumed to be restricted to the lipid fraction of the chromatophore membrane. Yet, at low pH, the chromatophore-water distribution ratio exceeded the liposome-water distribution ratio indicating complex formation of the tributyltin cation with ligands of the protein fraction. The effect study has confirmed that both, tributyltin and triphenyltin, can inhibit

the bc1 complex in the energy-transducing membranes. However, inhibition was observed for concentrations above those which inhibited ATP synthesis. Inhibition of

the bc1 complex is therefore not of toxicological relevance. Furthermore, tributyltin was found to affect the membrane potential which was attributed to an OH– uniport over the membrane similar to a protonophoric shuttle. The OH– uniport was observed in the same concentration range as inhibition of ATP synthesis. At pH = 6.1 both, – inhibition of ATP synthesis and the OH uniport, were more effective than at pH = 7.5, pointing to the relevance of the cationic species. In the mixture experiments with substituted phenols the uncoupling activities of 14 binary mixtures were investigated at different pH values. All of the single compounds induced a protonophoric shuttle. In 13 of the 14 mixtures, interactions of the substituted phenols led to increased protonophoric activity compared to the sum of the effects of the single compounds. These findings are significant regarding the mechanism of protonophoric action and regarding the risk assessment process of chemical mixtures in the environment. The observed increase in the uncoupling activity gives further evidence for the formation of a dimer of neutral phenol and phenoxide as part of the protonophoric shuttle as it is postulated in the literature. The observed effects of the mixture could be described in a model which includes a term for the contribution of a mixed dimer. It was found that opposite speciation favors interaction and that ortho substituents abate it. The effects of six combinations with ‘optimal’ speciation and ‘optimal’ substitution pattern were found to be higher than calculated by the concept of concentration addition, which is a reference concept generally applied for similarly acting compounds. This thesis shows that mechanism based investigations of toxic effect are a prerequisite for reliable modeling of effects of single compounds and of mixtures.

– VIII – Zusammenfassung

Eine wachsende Anzahl von organischen Schadstoffen gehört zu der Gruppe der hydrophoben, ionogenen, organischen (HIO) Verbindungen. Dazu gehören Pestizide, Medikamente und Körperpflegeprodukte. Oft wird im Laufe der Neuentwicklung eines chemischen Hilfsstoffs eine saure oder basische Gruppe in das Molekül einge- baut, mit der Absicht, die Verbindung dadurch weniger hydrophob und weniger persistent zu machen. Doch die zusätzliche ionogene Gruppe im Molekül kann auch dazu führen, dass die Verbindung zusätzliche spezifische toxische Effekte im Organismus verursacht. Aufgrund ihrer Hydrophobie reichern sich HIO-Verbindungen in den Biomembranen an, wo sie die wichtigen Aufgaben der Membran stören können. In dieser Arbeit wurden Verbindungen aus zwei Klassen von HIO-Verbindungen untersucht: Zwei Triorganozinn-Verbindungen (Tributylzinn und Triphenylzinn) und binäre Mischungen einer Auswahl von substituierten Phenolen. Obwohl einige Ver- bindungen als Pestizide oder Biozide eingesetzt werden, wurden sie vor allem aufgrund ihrer strukturellen Eigenschaften als Modellverbindungen ausgewählt. Von al- len Verbindungen war bekannt, dass sie den membranständigen Energiestoffwechsel stören. Drei unterschiedliche Mechanismen können zu dieser Störung führen: (i) Die unspezifische Anreicherung einer hydrophoben Verbindung in der Phopholipid- Doppelschicht stört deren Struktur und damit ihre Barrierenwirkung (Basistoxizität). (ii) Verschiedene Enzymkomplexe in der Membran können durch eine spezifische Wechselwirkung der HIO-Verbindung mit Rezeptoren der Enzyme inhibiert werden. (iii) Einige HIO-Verbindungen ermöglichen einen Transport von Ionen, in der Regel Protonen, was zu einem Kurzschluss des chemiosmotischen Protonenzyklus führt. Dieser stellt die treibende Kraft für die ATP-Synthese und für Transportvorgänge durch die Membran. Letzterer Mechanismus wird als Entkopplung bezeichnet. Die unterschiedlichen Wirkungen der untersuchten Verbindungen auf den Energiemetabolismus in der Membran wurden an Membranvesikeln (Chromato- phoren) des Photosynthese betreibenden Purpurbakteriums Rhodobacter sphaeroides untersucht. Mit zeitaufgelöster Spektroskopie können die obigen drei primären Mechanismen unterschieden werden. Die beobachteten Wirkungen wurden vergli- chen mit der Konzentration der Verbindung an ihrem Wirkort, d.h. in der Chromato- phorenmembran. Die Konzentration in der Chromatophorenmembran wurde über Liposom/Wasser-Verteilungsverhältnisse und Chromatophoren/Wasser-Vertei- lungsverhältnisse berechnet. Für die Gruppe der substituierten Phenole standen

– IX – bereits viele dieser Werte zur Verfügung. Für die Triorganozinnverbindungen muss- ten sie eigens gemessen werden. Die Verteilung zwischen Liposomen und Wasser zeigte bei beiden Triorganozinn- Verbindungen eine leichte Abhängigkeit vom pH-Wert. Der Verlauf der Verteilungs- verhältnisse mit dem pH-Wert konnte durch ein Modell beschrieben werden, in wel- chem pH-unabhängige Verteilungsverhältnisse für das Triorganozinn-Kation und den Hydroxokomplex angenommen wurden. Die Verteilungsverhältnisse von Kation und Hydroxokomplex von Triphenylzinn waren ungefähr zehnmal grösser als die ent- sprechenden Verteilungsverhältnisse von Tributylzinn. Bei beiden Verbindungen war das Verteilungsverhältnis des Kations ungefähr zweimal grösser als das des Hydroxokomplexes. Es wird postuliert, dass eine Komplexbildung mit dem Phospho- lipidliganden (wahrscheinlich mit der Phosphatgruppe) die Verteilung in die Liposo- men bestimmt. Für Tributylzinn wurde zusätzlich die Chromatophoren/Wasser- Verteilung untersucht. Das Chromatophoren/Wasser-Verteilungsverhältnis war bei hohem pH kleiner als das Liposom/Wasser-Verteilungsverhältnis. Es wurde deshalb gefolgert, dass die Verteilung des Hydroxokomplexes sich auf die Lipidfraktion der Chromatophoren beschränkt. Bei tiefem pH ist die Verteilung in die Chromatophoren jedoch grösser als die Verteilung in die Liposomen. Dies weist auf zusätzliche Ligan- den in der Proteinfraktion hin, welche durch das Tributylzinn-Kation komplexiert werden. Die Untersuchung der Effekte von Tributylzinn und Triphenylzinn auf die Energiekonversionskette in der Membran hat gezeigt, dass beide Verbindungen den

bc1-Komplex inhibieren. Zur Inhibition des bc1-Komplexes wurden jedoch Konzentra- tionen benötigt, die deutlich grösser waren als die Konzentrationen, bei denen Inhibi-

tion der ATP-Synthese beobachtet wurde. Die Inhibition des bc1-Komplexes ist deshalb toxikologisch nicht von Bedeutung. Tributylzinn beeinträchtigt auch das Membran- potenzial. Diese Störung kann mit einem OH–-Uniport durch die Membran erklärt werden, welcher analog zu einem Protonophoren-Shuttle funktioniert. Dieser OH–- Uniport und die Inhibition der ATP-Synthese traten im selben Konzentrationsbereich auf. Sowohl OH –-Uniport als auch die Inhibition der ATP-Synthese waren stärker bei pH = 6.1 als bei pH = 7.5, was auf die Wichtigkeit des Kations für die spezifische toxische Wirkung hinweist. In den Mischungsexperimenten mit substituierten Phenolen wurde die Ent- kopplungsaktivität von binären Mischungen bei verschiedenen pH-Werten untersucht. Alle Verbindungen wirken als protonophore Entkoppler. In 13 der 14 untersuchten Mischungen führten Wechselwirkungen zwischen den Verbindungen zu

– X – einer höheren Entkopplungswirkung als die Summe der Effekte der Einzel- verbindungen. Diese Beobachtung ist in zweierlei Hinsicht bedeutend: Einerseits hinsichtlich des Mechanismus, welcher der Entkopplung zugrundeliegt, und anderer- seits hinsichtlich der Bewertungskonzepte, welche in der Risikoabschätzung von Mischungen in der Umwelt zur Anwendung kommen. Hinsichtlich des Mechanismus bedeutet die Zunahme der Entkopplungsaktivität in den binären Mischungen ein wichtiges Indiz für die in der Literatur postulierte Bildung eines Dimers zwischen dem neutralen Phenol und dem Phenoxid-Anion während des Protonophoren- Shuttles. Die beobachteten Effekte der Mischungen konnten in einem Modell beschrieben werden, welches einen Term beinhaltet, der einen zusätzlichen Ladungstransport durch ein gemischtes Dimer beschreibt. Es zeigte sich, dass gegensätzliche Speziierung die Interaktion verstärkt, und dass Substituenten in ortho Position für eine Interaktion hinderlich sind. Sechs Kombinationen, welche diese Bedingungen ‘ideal’ erfüllten, zeigten eine Wirkung, die grösser war, als mit dem Konzept der Konzentrationsaddition berechnet wurde. Konzentrationsaddition ist ein Referenzkonzept, welches für Chemikalien mit gleichem Wirkmechanismus angewendet wird. Hinsichtlich der Bewertungskonzepte von Mischungen in der Umwelt zeigte diese Arbeit deshalb deutlich, dass das Konzept der Konzentrationsaddition nicht als ein allgemeines worst-case-Szenario betrachtet werden kann. Allgemein zeigte diese Arbeit, dass die Untersuchung der grundlegenden Wirk- mechanismen eine Voraussetzung ist für die Modellierung von Effekten von Einzel- stoffen und von Mischungen.

– XI –

General introduction General introduction

1.1 Biomembranes as site of toxic action

The amount of a contaminant in an aquatic environment is related to the observed effects on the biocenosis through a complex relationship. Various distribution processes in the environment determine the free aqueous concentration to which an aquatic organism is exposed (bioavailable concentration). After a compound has been taken up by an organism, distribution processes in the organism, transformation reactions and excretion determine its final concentration at the target site, i.e. the site where the compound provokes its primary toxic effect. In Figure 1.1 the link between the environmental fate and the toxic effects is depicted for a membrane toxicant. A compound is considered to be of particular concern when it belongs to the group of PBT compounds, which means that it is 12ersistent, that it Q.ioaccumulates, and that it is !oxic. Any hydrophobic compound tends to accumulate in the fatty tissue and in the membranes where it can severely impair the membrane function. Membranes are therefore important target sites of toxic action (Ahlers et al. 1991; Sikkema et al. 1995; Van Wezel et al. 1995).

target site primary observable concentration modes of effects environment toxic action dissolved organic ~ or anism matter ~ membrane =9• baseline =9 • narcosis concentration toxicity • decreased p~~~~~~e • ~ f;\ 4upt :k~ (x'\ .. ... r;\m • inhibition - repro- matter \Y excretion ~ ~ • protono- duction phoric - growth minerals / trani ~age shuttle • increased and formation lipids lethality sediments --~~~~~~~~~~....:...... :....~~~~..:,._~~~~---'

Figure 1.1 Sequence of processes from the introduction of a chemical in the environment to the toxic effects exerted by this chemical depicted for a hydrophobic membrane toxicant. The concentration of a contaminant in an aquatic system which is freely dissolved in water (Xw) is defined by distribution processes to various sorption sites in the environment (left side). The concentration of the toxicant in the aqueous phase of an organism (Xap). e.g. the cytosol, is defined by the ratio of uptake and excretion and by distribution to hydrophobic sites in the organism. The resulting concentration at the target site (the membrane concentration, Xm) provokes the primary modes of toxic action which in the end lead to the directly observable effects.

- 2- General introduction

Biomembranes are composed of a phospholipid bilayer with embedded enzyme complexes. Membranes fulfill central functions of a cell: (i) They enable a subdivision in small compartments with different defined chemical environments and (ii) energy- transducing membranes, like membranes of bacteria, mitochondria, or chloroplasts, play a major role in the energy metabolism of the cell. The process of energy conversion in such a membrane is depicted in Figure 1.2. The energy of light or chemical energy is used to transport protons through the membrane, which is performed by primary proton pumps (photosystem or electron transfer chain). The transport of the proton results in an electric potential over the membrane, ~\jl, and in a proton concentration gradient generally expressed as ~pH . Thus, the energy is converted into a chemiosmotic gradient over the membrane. This gradient serves the ATP synthase (secondary proton pump) as the driving force for the synthesis of ATP. In addition, the chemiosmotic gradient is used to maintain chemostasis by

primary proton pumps secondary proton pump

p-side H+ H+ H+ H+ H+ H+ H+

inhibition inhibition baseline protonophoric toxicity shuttle Figure 1.2 Simplified picture of the energy conversion in energy-transducing membranes and possible mechanisms of toxic action. Depicted is the lipid bilayer consisting of phospholipids with polar head groups and fatty acids chain which form the hydrophobic core of the membrane. lmbedded are the enzyme complexes of the primary and secondary proton pumps. Perturbation of the energy conversion by a membrane toxicant can result from the inhibition of the primary or secondary proton pumps or from a protonophoric shuttle or, if no specific effects occur, from the perturbation of the membrane structure (baseline toxicity).

- 3 - General introduction

symport mechanisms (Mitchell 1961; Mitchell 1966). Energy conversion and chemostasis are essential for the cell. Back diffusion of protons or other charged compounds through the membrane does not occur spontaneously, as the hydrophobic core of the lipid bilayer of the membrane is almost impermeable to charged compounds. How- ever, the accumulation of toxicants in the membrane can lead to a perturbation of the membrane structure which increases its permeability. The unspecific narcotic symp- toms exerted by neutral hydrophobic compounds have been related to this perturbation of the membrane (Van Wezel et al. 1995).

1.2 Membrane toxicity of hydrophobic ionogenic organic compounds (HIOC)

An increasing number of the chemicals introduced in the environment belong to the group of hydrophobic ionogenic organic compounds (HIOCs). Often, an acidic or basic functional group is introduced into a newly designed chemical to make it less hydrophobic and less persistent. Therefore many pesticides are moderate or weak acids (Worthing et al. 1991). Recently, attention has also turned to drugs and personal care products, many of which are also acids and bases. Increasing amounts of such compounds are detected in effluents of waste water treatment plants, in rivers, lakes and the sea (Halling-Sørensen et al. 1998; Daughton et al. 1999). Due to their hydrophobicity, HIOCs accumulate in membranes of organisms. Besides the nonspecific perturbation of the membrane structure, which is observed also with neutral compounds, HIOCs can exert specific mechanisms, which can occur at much lower concentrations compared to the concentrations causing nonspecific perturbation. Two additional types of effects can be differentiated: (i) By a specific interaction with a receptor of the primary or the secondary proton pumps HIOCs can inhibit the respective enzyme (Figure 1.2). (ii) Hydrophobic ions with aromatic systems, where the charge is delocalized, or ions with bulky substituents which shield the charge, are able to migrate through the hydrophobic core of the lipid bilayer driven by the membrane potential. Certain HIOCs can establish a shuttle mechanism of charged and neutral species thereby short-circuiting the chemiosmotic gradient (Mitchell 1961). This protonophoric shuttle is also termed uncoupling, although uncoupling, strictly speaking, denotes any disconnection of the primary and secondary proton pumps regardless of the underlying mechanism of action. Representatives of two compound classes of HIOCs have been investigated in the present work: triorganotin compounds and substituted phenols. Some of the investi-

– 4 – General introduction

gated compounds are introduced directly into the environment: Tributyltin and triphenyltin are widely applied biocides, e.g. as constituents in industrial cooling water systems or in antifouling paints (Blunden et al. 1990). Dinoterb (2-tert-butyl-4,6- dinitrophenol), Dinoseb (2-sec-butyl-4,6-dinitrophenol) and PCP (pentachlorophenol) are, or have been used as herbicides (Worthing et al. 1991). The use of tributyltin, pentachlorophenol and Dinoseb has been discontinued, restricted, or at least efforts are made to phase them out1. Yet, the compounds chosen for this thesis were not only selected for their direct environmental relevance, but also because they feature general chemical properties of HIOCs whose environmental relevance is increasing for the reasons given above. Both, triorganotin compounds and substituted phenols were shown to impair the energy metabolism of energy-transducing membranes (Aldrige 1976; Selwyn 1976; Miyoshi et al. 1987b; Escher et al. 1996b). Triorganotin compounds are cations which form strong complexes with hydroxides when dissolved in an aqueous phase. Fur- thermore, complexes are formed with various other ligands present in the environment or in biological systems (Hynes et al. 1987; Arnold et al. 1998b). For tributyltin and triphenyltin it is known that they inhibit the ATP synthase and that they can exert an OH–/anion shuttle (Gould 1976; Selwyn 1976; Papa et al. 1982). Whether they can also inhibit the primary proton pumps and whether they can act as protonophores is under debate (Klughammer et al. 1998; Bragadin et al. 2000). For substituted phenols it is known that they can inhibit a component of the primary proton pumps, the bc1 complex (Tokutake et al. 1991), and that they can act as protonophores (Miyoshi et al. 1987b). In previous work Escher et al. (1997) have shown that time-resolved spectroscopy with membrane vesicles (chromatophores) of the photosynthetic purple bacteria Rhodobacter sphaeroides is an appropriate in vitro method to investigate adverse effects of HIOCs on the energy metabolism. With this method inhibitory effects and the protonophoric mechanism can be differentiated and independently quantified. Accordingly, the most relevant mechanism of action of compounds which act according to several mechanisms concomitantly, can be deduced. Furthermore, the energy conversion chain can be induced by light instead of substrate which is advan- tageous, as reduced accessibility of the substrate can blear the effect of the toxicant (Escher 1995). Due to the simplicity of the biological system, concentration and

1 PCP is classified as Persistent Organic Pollutant (POP) in the list of the United Nations Environment Programme (UNEP) http://irptc.unep.ch/pops/default.html. Dinoseb is on the list of PIC (Prior Informed Consent) Rotterdam Convention http://www.pic.int. Both sites accessed on 10.01.01

– 5 – General introduction

speciation of HIOCs at the target site, i.e. the chromatophore membrane, can be deduced from partitioning experiments (Escher et al. 1996a). With this approach one can differentiate whether the effect observed for a given aqueous concentration of a toxicant is mainly due to its high partitioning or whether it is due to a high intrinsic toxicity, i.e. its ability to act according to a certain mechanism. Escher et al. (1996b) have applied time-resolved spectroscopy to investigate the membrane toxicity of substituted phenols. They found that for the majority of the compounds the protonophoric mechanism occurs at much lower concentration than

the inhibition of the bc1 complex. A complex dependence of the protonophoric activity on pH was found (Escher et al. 1996b) which was similar to the pH dependence of the conductance of lipid bilayers induced by substituted phenols in voltage clamp experiments (McLaughlin 1972; Benz et al. 1983). It was postulated that the protonophoric shuttle comprises the formation of a dimer of phenoxide and neutral phenol. The resulting charged dimer is thought to permeate better through the membrane than the single phenoxide (Lea et al. 1969; Finkelstein 1970).

1.3 Objectives

The major goal of this work was to deepen the knowledge on the impairment of the energy metabolism by HIOCs deduced in the previous work using time-resolved spectroscopy: On the one side by extending the investigation to a more complex class of environmentally relevant HIOCs, and on the other side by clarifying the model of interaction of substituted phenols in the protonophoric mechanism. The specific objectives were: • to determine liposome-water distribution ratios of tributyltin and triphenyltin and to draw a comparison with chromatophore-water distribution, with the goal to assess the concentrations of the compounds at their site of toxic action; • to develop a method to measure ATP synthesis as integrative parameter of impairment of energy metabolism in the chromatophores under comparable experimental conditions as time-resolved spectroscopy; • to investigate the effects of tributyltin and triphenyltin on the energy metabolism especially with regard to the mechanisms of action under debate, which are inhi-

bition of the bc1 complex and the protonophoric shuttle. In addition, a method had to be introduced, which allows one to assess the effects of the test compounds on the overall ATP synthesis by quantifying the amount of ATP synthesized.

– 6 – General introduction

• to make a contribution to the development of a model which describes the pH dependence of the protonophoric mechanism; • to combine substituted phenols with varied substitution patterns and with

various pKa values in binary mixtures to investigate whether the postulated interaction for single compounds can also be observed in mixtures.

1.4 Outline

The results are presented in the following way: Chapter 2 gives the pH dependence of the liposome-water and the chromatophore- water distribution experiments with tributyltin and triphenyltin. The results are discussed with respect to the octanol-water distribution behavior of these two compounds and with respect to distribution ratios found for other HIOCs in the same system. Chapter 3 describes the experiments performed to investigate inhibitory effects and protonophoric mechanism of tributyltin and triphenyltin. The results are discussed with respect to findings in mitochondria and in chloroplasts from the literature. A new method to measure the integrative parameter of ATP synthesis under comparable experimental conditions as applied for time-resolved spectroscopy is also presented. In Chapter 4 the uncoupling effects of binary mixtures of substituted phenols are presented. It is discussed in which way the substitution pattern and the acidity determine the ability to form dimers. Furthermore, the relevance of the interaction is discussed with respect to its significance for the toxicology of mixtures. Additional information is given in the appendices: In Appendix A a more detailed description is given on the components of the energy metabolism of R. sphaeroides, and on the principle of the determination of the protonophoric mechanism in chromatophores. The appendix contains more detailed information on this experimental method, which is only briefly referenced in the main chapters. In Appendix B a kinetic model which describes the protonophoric shuttle of substituted phenols in chromatophores is given. This model gave further indication for the formation of a dimer in the protonophoric shuttle of single compounds and was therefore fundamental for the work with binary mixtures. Part of the experimental data presented in the appendix were measured by the author of this thesis.

– 7 –

The pH dependence of the partitioning of tributyltin and triphenyltin between phosphatidylcholine liposomes and water

Abstract Triorganotin compounds are important contaminants. The site of action of their basic mechanism of action of acute toxicity is the biomembrane. Liposome-water distribution ratios of tributyltin and triphenyltin were determined between pH 3–8 with the equilibrium dialysis method in the micromolar concentration range. In addition biomembrane-water distribution ratios of tributyltin were determined with chromatophores of Rhodobacter sphaeroides. The liposome-water distribution of both compounds showed only a weak pH dependence. The distribution ratio of the triorganotin cation exceeded that of the neutral hydroxo complex by a factor of two. The distribution ratio of both, the cation and the hydroxo complex of triphenyltin, exceeded those of tributyltin by a factor of ten. It is postulated that the sorption of the cation is governed by complex formation with ligands in the phospholipids, presumably the phosphate group. This is assumed to partially similar for the hydroxo complex of triphenyltin. The biomembrane-water distribution ratio of tributyltin was found to be lower than the liposome-water distribution ratio at high pH. The hydroxo complex appears to partition only to the lipid fraction of the biomembrane. Yet, at low pH the biomembrane- water distribution ratio exceeded the liposome-water distribution ratio which is attributed to complex formation of the cationic species with ligands of the protein fraction. pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

2.1 Introduction

Biological membranes are important sites for toxic effects exerted by organic chemicals in aquatic organisms (Ahlers et al. 1991; Sikkema et al. 1995; Van Wezel et al. 1995; Escher et al. 1996b). For the evaluation and quantification of a given compound's toxicity, it is therefore crucial to assess to what extent the compound accumulates in membranes. For hydrophobic neutral organic compounds this task is not too difficult because, within a series of structurally related compounds, membrane-water partitioning can generally be related to the partitioning of the compounds between a bulk organic solvent, (e.g. n-octanol) and water (Van Wezel et al. 1995; Escher et al. 1996a; Vaes et al. 1998). However, when dealing with hydrophobic ionogenic organic compounds (HIOCs), e.g. weak organic acids and bases, organic solvents including n-octanol are rather poor surrogates, because the charged species tend to partition very differently in membranes as compared to bulk solvents (Miyoshi et al. 1987a; Betageri et al. 1988; Smejtek et al. 1993; Escher et al. 1996a). For example, chloro- and nitrophenoxide species were found to partition by orders of magnitude more favorably from water into phosphatidylcholine liposomes as compared to n-octanol (Escher et al. 1996a). Similar results were also obtained for cationic species including protonated amines (Ottiger et al. 1997; Escher et al. 2000a). One explanation for these findings are possible electrostatic interactions of the charged organic species with charged moieties, e.g. the phosphate group (for organic cations) or the ammonium group (for organic anions) of the phospholipids, making up the lipid bilayers. Hence, for evaluation of membrane-water partitioning of HIOCs, more appropriate model systems than bulk solvents have to be used. The simplest and most widely investigated model systems for the lipid fraction of biomembranes are liposomes, which are artificial lipid bilayer vesicles of known composition and controlled size (Escher et al. 2000b). In earlier work, we have successfully applied liposome-water distribution ratios for assessing the concentration and speciation of phenols in the lipid bilayer fraction of photosynthetic membranes of the purple bacterium Rhodobacter sphaeroides (Escher et al. 1996a). This information has been pivotal for the qualitative and quantitative interpretation of the uncoupling activities determined in this test system by time-resolved spectroscopy for a large number of phenols (Escher et al. 1996b) as well as for binary mixtures of phenols (cf. chapter 4). Recently, we have extended this study to the investigation of the effects of triorganotin (TOT) compounds, in particular, of tributyltin (TBT) and triphenyltin (TPT), on energy transduction in photosynthetic membranes (cf. chapter 3). TBT and TPT have been

– 10 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

chosen as model compounds because they are highly toxic to aquatic organisms (Fent 1996) and, despite significant restrictions in their use as antifouling agents in boat paints, they are still found in the aquatic environment, particularly in sediments of areas of high boating activities (Tolosa et al. 1996; Arnold et al. 1998a; Harino et al. 1999). Furthermore, application of TPT as fungicide in agriculture remains a continuous source of TOT contamination in freshwaters (Blunden et al. 1990). When TOT compounds are dissolved in water they form positively charged diaqua complexes, TOT+, which may dissociate to the neutral hydroxo complex TOT-OH:

+ + TOT + H2O TOT-OH + H (2.1)

At low pH the diaqua cation dominates, whereas at pH values above pKa the neutral hydroxo complex dominates. In Figure 2.1 the pH dependence of the speciation of TBT and TPT is depicted. By ligand exchange of water or hydroxide, TOT may form strong complexes with various other ligands present in the environment or in biological systems. Such ligands include carboxylate and phenoxide groups present in natural organic matter (Arnold et al. 1998b; Berg et al. 2001), or biologically relevant ligands such as sulfide groups in glutathion or cystein (Hynes et al. 1987; Shoukry 1993), or phosphate containing groups present in nucleosides (Barbieri et al. 1995), nucleic acids (Barbieri et al. 1992), ATP (Hynes et al. 1985), and phospholipids (Grigoriev et al. 2000). Hence, in contrast to the situation encountered with many other HIOCs, for compounds such as TOTs, complex formation may be an additional factor

f(TBT+) f(TPT-OH) 1.0

f(TPT+) f(TBT-OH) 0.8

0.6 fraction of the species

0.4

0.2 Figure 2.1 pH dependence of the speciation of 0.0 tributyltin and triphenyltin in aqueous 23456789 solution. pKa (TPT) pKa (TBT) pKa values (Table 2.1) are from Arnold et al. (1997). pH

– 11 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

to be considered for membrane-water partitioning as well as membrane toxicity. In the work presented in this paper, we have investigated the liposome-water partitioning behavior of TBT and TPT as a function of pH with the equilibrium dialysis method (Escher et al. 2000b). The liposomes used contained only phospholipids with choline head groups, so that the speciation of the lipids was invariant to changes of pH. For comparison, the sorption of TBT to membrane vesicles of R. sphaeroides (chromatophores) was also determined. Chromatophores contain approximately 30 % lipids and 70 % proteins. The major goals of this study were (i) to evaluate to what extent complex formation governs the uptake of TOT compounds into phospholipids, and (ii) to check whether liposomes are suitable model systems to describe the partitioning of the type of HIOCs represented by TOT compounds between biological membranes and water.

2.2 Materials and methods

2.2.1 Chemicals The following compounds were obtained from Fluka (Buchs, Switzerland): Tributyltin chloride (96 %), triphenyltin chloride (97 %), 2-morpholino-ethanesulfonic acid (MES, BioChemika), Tris(hydroxymethyl)aminomethane acetate (TRIS acetate, BioChemika), 4-(2-Hydroxyethyl)piperazine-1-propanesulfonic acid (HEPPS, BioChemika), potas- sium hydroxide (puriss p.A.), Triton X-100, citric acid, oxalic acid and morin. From Merck (Dietikon, Switzerland): Perchloric acid (60 % p.A.) and matrix modifiers

Mg(NO3)2 and Pd(NO3)2 in 0.1 M HNO3. Egg yolk lecithin (≥ 98 % phosphatidylcholine) was purchased from Lipoid (Ludwigshafen, Germany). Methanol (HPLC grade) was obtained from Scharlau Chemicals (Barcelona, Spain).

2.2.2 Aqueous solutions Aqueous solutions were prepared from a perchloric acid stock solution (1 M) which was titrated twice to determine its exact acid concentration. For pH 2.0–4.5 no buffer was used. From pH 5.0–8.5, 10 mM of the following zwitterionic buffer were added: for pH = 5.0 TRIS acetate; for pH = 5.5, 6.0 and 6.5 MES; for pH = 7.5 TRIS acetate; for pH = 8.5 HEPPS.

2.2.3 Preparation of liposomes

Phosphatidylcholine was dissolved in CHCl3 and dried to a film on a glass vessel in a

rotary evaporator. Residual traces of CHCl3 were removed under high vacuum. The film was slowly hydrated (25 mg.ml–1) by hand-shaking with an aqueous solution of

– 12 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

10 mM KClO4. Small aliquots were submitted to a freeze-and-thaw procedure. After being frozen for the third time the aliquots were transferred to –20 °C immediately and stored for a maximum of two months. Unilamellar vesicles were prepared daily from the frozen stock by membrane extrusion in a thermobarrel extruder from Lipex biomembranes (Vancouver, CA). The suspension was extruded ten times through two polycarbonate membranes, one upon the other, with 0.1 µm pore diameter from Nuleopore (Pleasanton, CA). The vesicle stock suspension was diluted in pH buffer to the desired concentration. Characterization of vesicles is given in (Escher et al. 2000a).

2.2.4 Determination of liposome-water and chromatophore-water distribution ratios of tributyltin and triphenyltin Liposome-water distribution ratios were determined with specially designed dialysis cells made of two glass chambers. The chambers were separated by a regenerated cellulose membrane with a cut-off of 10'000–20'000 Dalton from Reichelt Chemietechnik (Heidelberg, Germany). At a given pH, distribution ratios were determined at four to five different concentrations and two different lipid concentrations. TOT concentration in the stock solution varied due to decreasing solubility with increasing pH. For TBT at all pH values, 5–65 µM with 24, 48 or 96 mg.L–1 phospholipids were used. For TPT –1 at pH = 2.0 concentration varied from 2–70 µM with 45 or 446 mg.L phospholipids –1 and at pH = 7.5 from 1–8 µM with 50 or 100 mg.L phospholipids. Six cells were prepared as measurement cells and three cells as reference cells. 1.3 ml of TOT solution was added to one chamber of each measurement cell. The other chamber received 1.2 ml TOT solution and 0.1 ml of diluted vesicle solution. Reference cells were prepared in the same manner as the measurement cells, excepting the addition of buffer instead of the vesicle solution. The reference cells served to determine losses of TOT due to sorption to the glass surface and to the dialysis membrane. Refer- ence cells and measurement cells were shaken for 24±1 hours at 25 °C. Experiments designed to evaluate the diffusive kinetics through the dialysis membrane have shown that after 24 hours, concentrations were equilibrated when one chamber was filled with a TPT solution and the other chamber was filled with buffer. After equilibration the chambers were separated and the contents were transferred to glass vials for TOT analysis. For the determination of the chromatophore-water distribution ratio of TBT the same procedure and the same TBT concentrations were used as in the liposome- water distribution experiments. However, the chromatophore concentration was kept constant at 96 mg.L–1. Preparation of chromatophores is described in detail in Escher et al. (1997).

– 13 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

2.2.5 Analytical procedures High pressure liquid chromatography (HPLC) was used to determine TOT concentrations in the reference cells and in the chamber of the measurement cells containing no liposomes. HPLC was performed isocratically using a cation exchange column (Cation 1-2, Metrohm, Herisau, Switzerland) and TOTs were detected by fluorescence after post column derivatisation with a morin solution. Details of the method and the equipment are given in Arnold et al. (1998b). Briefly, the eluent consisted of 25 mM citric acid and 20 mM oxalic acid in 50:50 methanol:water (V:V). The derivatisation solution consisted of 2 % (w:V) Triton X-100, 160 µM morin, 360 mM citric acid, and 810 mM lithium hydroxide in water. Eluent and derivatisation solutions were pumped at a flow rate of 1 ml.min–1. Excitation and detection wavelengths were 403 nm and 533 nm respectively. Bandwidth of excitation and detection slits were 18 nm. Sample injection volume was 200 µL.

For both compounds stability was confirmed at pH = 7.5 (lowest recovery rates) with a gradient program which separates mono-, di- and triorganotin. Eluents and gradient are described in Arnold et al. (1998b). No increase in the amount of mono- or diorganotin compound was observed.

For TPT samples with pH > pKa (i.e. pH = 6.0 and pH = 7.5) and for a mass balance experiment, the concentrations were determined by graphite-furnace atomic absorption (GAAS) spectroscopy. The following equipment was used: 5100-PC atomic absorption spectrometer equipped with a 5100-ZF zeeman furnace module with a THGA graphite tube, a AS-71 autosampler and a electrodeless tin discharge lamp. (Perking-Elmer, Norwalk, USA). The following instrument parameters were used: Wavelength 286.3 nm, lamp current 20 A, slit width 0.7 nm, signal type zeeman AA, read time 5 s. Samples were diluted in the same buffer as used in the partitioning experiments and pipetted into 1.8 ml glass vials (Omnilab; Mettmenstetten, Switzerland). The glass vials were placed directly into the Perking Elmer polystyrene sample cups to prevent loss of TOT compounds due to sorption. 15 µL of sample, 5 µL of 0.1 M HCl and 5 µL of matrix modifier were transferred into the graphite tube by the autosampler. The . –1 . –1 matrix modifier consisted of Mg(NO3)2 0.1 g L and Pd(NO3)2 0.1 g L in 0.1 M HNO3. The following temperature program was applied: A) two-step drying for 30 s at 110 °C and for 30 s at 130 °C, B) charring with a 10 s ramp from 130 °C to 1000 °C and holding for 20 s, and C) atomisation for 4 s at 2300 °C and 1 s at 2400 °C (during atomisation the argon gas flow was interrupted).

– 14 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

2.2.6 Calculation of triorganotin concentrations in the liposomes With the HPLC method used, the concentration of TOT compounds in the lipid chamber could not be measured directly. Only the free aqueous concentration in the aqueous chamber of the measurement cell i, cw,i, and the free aqueous concentration in the reference cells j, cw,ref,j, were directly available. The amount of TOT sorbed to liposomes was calculated by equation 2.2. Subtraction of the concentration in the aqueous chamber from the average concentration of the corresponding reference cells yielded the amount of sorbed TOT expressed as mol per volume aqueous phase. Dividing the difference by the concentration of liposomes in the aqueous phase, c(PL) in kg.L–1, yields . –1 the concentration in the liposomes of the cell i, clip,i, in mol kg .

Ê 1 n ˆ Á ◊Âccw,ref,j˜ - w,i Á n j=1 ˜ Ê mol ˆ c = Ë ¯ (2.2) lip,i c()PL Á kg ˜ Ë lip ¯

Recoveries in the reference cell were pH-dependent and decreased from 80 %

(pH < pKa) to 50 % (pH > pKa). No increase of the concentration of di- and monoorganotin compounds was detected after the 24 hours equilibration. Furthermore, a mass balance performed for TPT at pH = 3.0 with GAAS showed that the loss of TPT was the same in both the measurement cells and the reference cells. Therefore, loss of TOT can be attributed primarily to sorption to the glass surface of the dialysis cell, and to the dialysis membrane. However, for distribution experiments of TPT with pH > pKa and low solubility, accuracy in determining clip,i was increased by measuring the total concentration in the lipid chamber, ctot, directly with GAAS. The concentration of TPT in the lipids was calculated without reference cells using equation 2.3.

cctot- w,i Ê mol ˆ c = (2.3) lip,i c()PL Á kg ˜ Ë lip ¯

– 15 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidykholine liposomes and water

2.3 Results and discussion

2.3.1 Liposome-water partitioning Liposome-water partitioning was determined in the pH range 3.0- 8.5 for TBT and 2.0- 7.5 for TPT in 10 mM perchlorate solution. The concentration of TOT present as perchlorate complex is less than 2 % of the total concentration of TOT in the aqueous phase, and can be neglected. Figure 2.2 shows some representative experimental liposome-water partitioning data for TBT (Figure 2.2- A) and for TPT (Figure 2.2- B) at low and high pH values. Similar results were obtained for the pH values in-between. As can be seen, particularly for TBT, due to experimental difficulties there is a considerable scatter in the data. Nevertheless, a quantitative analysis of the whole data set provides some important information about the uptake of the two TOTs into the lipid bilayers.

0.10 0.30 I I O> A)TBT O> • B)TPT ~ ~ 0 pH=3.0 0 0.25 pH=2.0 §_ 0.08 §_ .Q. 0.. 6 pH=7.5 6 0.20 0.06 • 0.1 5 0.04 0.1 0 pH=7.5 0.02 0.05 • 0.00 • 0.00 0 10 20 30 0 2.5 5.0 7.5 Cw (µM) Cw (µM) Figure 2.2: Liposome-water distribution of tributyltin and triphenyltin. Experiments were performed in presence of 10 mM perchlorate at the indicated pH. Data points represent means of three cells with identical stock solution concentration. At pH=?.5 10 mM TRIS acetate was used as pH buffer. Sorption isotherms correspond to linear regression of the experi- 1 mental data. Values with c1ip. tot> 0.1 mol·kg- were excluded from the linear regression due to saturation. Apparent D1ipw values in Table 2.1 correspond to the slopes of the sorption isotherms.

- 16 - pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

The apparent liposome-water distribution ratios at the respective pH

clip,tot Ê L ˆ D = (2.4) lipw c Á kg ˜ w,tot Ë lip ¯ were obtained from sorption isotherms established for the concentration range considered which were determined by a linear regression. Note that data points at high . –1 TBT concentrations (i.e. clip > 0.1 mol kg ), above which saturation was observed, have been excluded for the derivation of the isotherm. In addition, the isotherms have not been forced through the origin, because it cannot be excluded that at very low concentrations, higher Dlipw values would apply, which is in fact indicated by the significant positive intercept of several isotherms. However, this should not be too critical because the derived Dlipw values can be assumed to describe correctly the general features of TOT sorption to the liposomes. Furthermore the concentration range of the present experiments corresponds to the concentrations which caused effects on energy transduction in chromatophores (cf. Chapter 3) and are also typically used in many other toxicity assays. The simplest approach for a quantitative description of the pH dependence of the liposome-water partitioning is to define a pH-independent distribution ratio for each of the species of the two TOTs. This approach presumes that the speciation of the phosphatidylcholines does not change in the experimental pH range. With a pKa of the phosphatidylcholine below 2 (Escher et al. 2000a) this condition is fulfilled for most experiments and accordingly the overall distribution ratio (equation 2.4) can then be expressed as:

Df=()TOT+++ ◊ D () TOT1 f () TOT ◊ D (TOT OH ) (2.5) lipw w lipw +-()w lipw - + With the species distribution, Dlipw(TOT-OH) and Dlipw(TOT ), ratios defined as:

c()TOT- OH lip Ê L ˆ Dlipw()TOT-= OH (2.6) c()TOT OH Á kg ˜ - w Ë lip ¯ and

+ + c()TOT lip Ê L ˆ Dlipw()TOT = (2.7) c()TOT+ Á kg ˜ w Ë lip ¯ + pH–pKa –1 f(TOT )w= (1 + 10 ) is the fraction of the TOT compounds in the acidic (cationic) + form. Rearrangement of equation 2.5 shows a linear dependence of Dlipw on f(TOT )w: DD()TOT++ D (TOT OH )()◊ fTOT D ( TOT OH ) (2.8) lipw=--() lipw lipw w +-lipw

– 17 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

Thus the distribution ratios of the TOT+ and TOT-OH were derived from a linear regression. As can be seen from Figure 2.3 (solid lines), the simple model according to equa-

tion 2.5 describes the experimental Dlipw data in both cases quite well. A more complex model which accounts for the build-up of a potential at the membrane surface due to the sorption of charged species (Escher et al. 2000a) has also been applied on the present data set. The data can successfully be described with this complex model. How- ever, due to the small concentration range available no definite information about the electrostatic effects and the partitioning in the low concentration range could be deduced. The following discussion is therefore restricted to the simple model. Inspection of Figure 2.3 reveals several important quantitative findings. First, in

contrast to the octanol-water distribution ratios, Dow, the Dlipw values of both TBT and

) 5 lip

-1 A) TBT B) TPT kg

• D (pH)

lipw lipw D

(mol 4 log Dlipw(pH)

ow 3 D (pH)

log D cphw

2 Dow(pH)

Dow(pH)

1 23456789 23456789 pKa pKa pH pH Figure 2.3 Comparison of the pH dependence of the apparent liposome-water and the octanol-water distribution ratio of tributyltin and triphenyltin. Data points represent the slope of the sorption isotherm (cf. Figure 2.2) at a given pH for TBT ({) and TPT(). The solid lines corresponds to the modeled pH dependence of the Dlipw values with + equation 2.8 and the values for Dlipw(TOT ) and Dlipw(TOT-OH) given in Table 2.1. Dcphw values of

TBT (X) represent the slope of the isotherms given in Figure 2.4. The pH profiles of Dow were calculated for an aqueous solution with 10 mM perchlorate using the values from Arnold et al. (1997). Note that the octanol-water distribution at low pH is solely due to the distribution of the perchlorate . –1 . –1 complex. Values of Dlipw are in units of L kg , values of Dow are L L . However, because the . –1 density of liposomes is almost 1 L kg , a direct comparison of Dlipw with Dow is possible (Escher et al. 2000a).

– 18 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

TPT exhibit only a very weak pH dependence. This means that the cationic species, TOT+, and the neutral hydroxo complex, TOT-OH, partition very similarly into the liposomes (for speciation, see Figure 2.2). Second, the TOT+ species sorb even somewhat more strongly as compared to the TOT-OH’s. This is in strong contrast to the octanol-water partitioning of the two compounds where the neutral species partitions much more favorably into the organic phase. A higher liposome-water distribution ratio of the charged species as compared to that of the neutral is also in contrast to findings with substituted phenols, anilines and aromatic and aliphatic amines (Ottiger et al. 1997; Escher et al. 2000a). Finally, over the whole pH-range, TPT exhibits significantly higher Dlipw values than TBT, although the TPT species are less hydrophobic

(compare Dow-curve in Figure 2.3). Table 2.1 summarizes the distribution ratios of the TOT+ and TOT-OH species + together with the corresponding Dow values. As can be seen, D lipw(TOT ) and

Dlipw(TOT-OH) are about an order of magnitude higher for TPT as compared to TBT. + + In the case of Dlipw(TOT ) this findings can be rationalized by the fact that TPT forms a much stronger complex with oxygen ligands than TBT +. For example TPT+ exhibits about one order of magnitude higher complexation constant for OH– as compared to 8.8 7.75 that of TBT (KOH(TPT) = 10 , KOH(TBT) = 10 ; Arnold et al. 1997). The same trend is found for carboxylate and phenoxide ligands present in particulate and dissolved natural matter (Arnold et al. 1998b; Berg et al. 2001). Furthermore the uptake of the TPT cation by larval midge was found to be almost one order of magnitude higher than the uptake of the TBT cation (Looser et al. 1998; Looser et al. 2001). Hence it seems reasonable to assume that the sorption of the cationic species by the liposomes is

Table 2.1: pKa, liposome-water, chromatophore-water, and octanol-water distribution ratios of tributyltin and triphenyltin. + Dlipw(TOT ) and Dlipw(TOT-OH) were determined by linear regression of the apparent Dlipw values according to equation 2.8. Dlipw values were determined by linear regression of the sorption isotherm measured in presence of 10 mM perchlorate. Dcphw of TBT were determined at pH=5.5 for + TOT and at pH=8.5 for TOT-OH. Dow was calculated for an aqueous solution with 10 mM perchlorate using the values from Arnold et al. (1997). Note that the apparent octanol-water distribution ratio given for TOT+ is almost solely due to the distribution of the perchlorate complex.

TOT+ TOT-OH

pKa, Dlipw Dcphw Dow Dlipw Dcphw Dow L.kg–1 L.kg–1 L.L–1 L.kg–1 L.kg–1 L.L–1 TBT 6.25 4'100 6’200 200 2'000 500 12’300 TPT 5.20 70'000 n. d. 50 22'000 n. d. 3’400

– 19 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

governed by complex formation with the phosphate groups and not just by electrostatic interactions. The results obtained for the neutral species are somewhat more difficult to inter- pret. The striking feature is the much higher sorption of the hydroxo complex of TPT as compared to the one of TBT, which is in contrast to the octanol-water partitioning behavior of the two species (Table 2.1). It is instructive to compare these data with the partitioning behavior of phenols. In a preceding study with the same type of liposomes, Escher et al. (2000a) have derived the following linear free energy relationship (LFER) for the neutral species of substituted phenols:

logKKrn=078 . ◊ +== 112 . (2 092 . , 20 ) (2.9) lipw ow The LFER of equation 2.9 includes compounds with similar or even higher octanol-

water partitioning constants than the hydroxo complex of TBT and TPT. Using the Dow

values given in Table 2.1, equation 2.9 would predict Dlipw(TOT-OH) values of about . –1 . –1 20’000 L kg for TBT and only about 7’000 L kg for TPT. Hence, Dlipw(TBT-OH) is

smaller by a factor of five as compared to a phenol with identical Kow. Whereas,

Dlipw(TPT-OH) is higher by a factor of three compared to the prediction for phenols.

The much lower value of Dlipw(TBT-OH) cannot be considered to be compatible with equation 2.9. A quite reasonable argument for this finding is that, due to steric reasons, TBT-OH cannot be optimally accumulated into the hydrophobic core of the lipid bilayer. However, this argument should also hold for TPT-OH. Consequently, it seems reasonable to assume that sorption of TPT-OH is enhanced by an additional mechanism, i.e. by assuming that, in addition to hydrophobic partitioning, complex formation with the phospholipids group may also occur for TPT-OH. In fact, using fluorescence probes covalently linked either to the hydrophobic core or to the polar head groups regions of liposomes, Ambrosini et al. (1996) found that TPT-OH altered the fluorescence in both regions. A conceivable reason why such complex formation could be favorable in this case is the possibility of the stabilization of the OH– by electrostatic interaction with the trimethyl ammonium group of the choline function.

– 20 – pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidykholine liposomes and water

2.3.2 Chromatophore-water partitioning In contrast to the phosphatidylcholine liposomes the surface of biomembranes are generally negatively charged at neutral pH. For the chromatophores used in this study, the surface potential was found to decrease almost linearly from zero at pH= 5.5 to - 97 m V at pH= 7.8 (Matsuura et al. 1979). Figure 2.4 shows the chromatophore sorption isotherms determined for TBT at three pH values. The intercept of the linear regression clearly deviates from zero. This can be attributed to the presence of sorption sites with high affinity which are saturated already at lower concentration. At higher concentrations, the remaining ligands appear to be more similar in their affinity, resulting in the observed linear sorption isotherm. Apparent distribution ratios in Table 2.1 represent the slope of the isotherm either at pH= 5.5 for TBT'" or at pH= 8.5 for TBT-OH. Consequently, the contribution of the high affinity sites (axis intercept) are not explicitly represented in Dcphw· Chromatophores contain only about 30 % lipids. If sorption occurred primarily into the lipid bilayer, and if any charge did not have a significant effect on the process, chromatophore-water distribution ratios of only about one third of the corresponding liposome-water distribution ratios would be expected. This has been found for the sorption of neutral as well as of anionic phenolic species (Escher et al. 1996a), and seems to be the case also for the neutral TBT-OH species (Table 2.1). However, for TBT at lower pH values, where the overall partitioning to chromatophores is governed by the sorption of TBT+, Dcphw is even larger than D1ipw(Figure 2.3-A and Table 2.1). This suggests that in the chromatophores a significant number of ligands, which are

0.30 I ..r::: a. 0 TBT ~ 0.25 pH=5.5 0 .._..E Figure 2.4: 0.20 Chromatophore-water distribution of tributyltin. 0.15 Experiments were performed in presence pH=6.5 of 1O mM perchlorate at pH 5.5 (• ), 6.5 (e ), 0.10 8.5 (+ ). Ccph denotes the amount of TBT bound to the chromatophores per kg of dry 0.05 pH=8.5 mass of chromatophores. Data points 0 represent means of three cells with identi- .... • cal stock solution. (0) denotes an outlier at 0.00 • pH=6.5. Sorption isotherms correspond to 0 10 20 30 40 50 linear regression of the experimental data. Cw (µM)

- 21 - pH dependence of the partitioning of triphenyltin and tributyltin between phosphatidylcholine liposomes and water

probably present in the protein fraction, are available for the sorption of TBT+. From the results of the few measurements made in this study, it is, however, not possible to identify and quantify these sites. Farrow et al. (1978) have investigated triethyltin binding to mitochondria at pH = 7.6, pH = 6.1 and pH = 5.7 by Mössbauer spectroscopy which enabled investigation at low concentrations. Consistent with the implication of the present work they found sorption to high affinity sites at low concentrations and almost linear isotherms . –1 for concentrations above 0.002 mol kgprotein . Further consistent with the present findings, distribution ratios decreased with increasing pH (Farrow et al. 1978).

2.4 Conclusion

The results of this study have shown that sorption of TOT compounds to phospholipid bilayers cannot be directly related to the partitioning behavior of other HIOCs, e.g. phenols, anilines and aliphatic and aromatic amines. The major reason is that the TOT cation, and additionally in the case of TPT presumably the hydroxo complex, may form complexes with ligands present in membrane constituents. The result also shows that for TOT compounds liposomes are reasonable surrogates for assessing the

biomembrane-water partitioning in chromatophores only for pH values above the pKa, at which the hydroxide species dominates. For a proper assessment of TOT sorption to a given biomembrane at pH values at which the cationic species dominates, biomembrane-water distribution ratios need to be determined experimentally.

– 22 – 3

Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

Abstract Triorganotin compounds exhibit several modes of toxic action on the energy metabolism in energy- transducing membranes. The inhibition of the ATP synthase and the OH–/Cl– antiport have been extensively investigated but there is still debate on whether further mechanisms are relevant. In this work two possible – further effects have been investigated: Inhibition of the bc1 complex and the OH uniport, and in addition, the overall inhibition of the ATP synthesis were investigated in chromatophores of the photosynthetic purple bacterium Rhodobacter sphaeroides at pH=7.5 and at pH=6.1. Experimental conditions were chosen in order to – exclude the OH /anion antiport. Inhibition of the bc1 complex was detected; however, at such high concentration that it is without significance for acute toxicity. Tributyltin was found to induce a decrease of the membrane potential, which can be attributed to an OH– uniport, whereas for triphenyltin no such activity was observed. For both compounds, inhibition of the ATP synthesis was higher at pH=6.1 than at pH=7.5. Also the OH– uniport activity of tributyltin was higher at lower pH. The contribution of the OH– uniport of tributyltin to the overall inhibition of the ATP synthesis cannot be quantified, however, OH– uniport occurred in the same concentration range as inhibition of the ATP synthesis. For triphenyltin, inhibition of the ATP synthesis can be attributed to the inhibition of the ATP synthase. Chromatophores of R. sphaeroides were found to be a useful system to discriminate various effects of toxicants on the energy metabolism of a cell. Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

3.1 Introduction

Triorganotin (TOT) compounds, in particular tributyltin (TBT) and triphenyltin (TPT), are widely applied biocidals. TPT is used as a fungicide in potato and rice fields. Both TBT and TPT are used as biocides constituents in industrial cooling water systems and antifouling paints. These applications have led to continuous release of TOT compounds in the environment. The use of TOT compounds in boat paintings has been stringently regulated (EU 1998; review: Champ 2000) which has resulted in a continuous decrease of TOT concentrations in coastal waters (Tolosa et al. 1996; Stewart 1997; Harino et al. 1999) and in lakes (Berg et al. 2001). Nevertheless, the practice of the application of TOT compounds in agriculture can lead to significant concentrations in adjacent surface waters (Stäb et al. 1993). Despite major concern due to their chronic effects at low concentrations, the acute toxicity of TOT is still relevant for contaminated sites (Fent 1996). The basic effect of TOTs is a disturbance of the energy metabolism of the cell. Several secondary toxic effects may result from this basic effect (Fent 1996). While the acute toxicity of TOT compounds towards bacteria, invertebrates, and fish correlates well with effects on submitochondrial particles, which directly reflect the effects on the energy metabolism (Argese et al. 1998), no meaningful correlation

was found between any of these effects and the Kow or other physicochemical or structural descriptors used for Quantitative Structure Activity Relationships, QSARs (Vighi et al. 1985; Nagase et al. 1991; Argese et al. 1998). Argese et al. (1998) therefore concluded that the acute toxicity indeed is due to the basic effect of the impairment of the energy metabolism, yet, this impairment is caused by different primary modes of toxic action. It is therefore necessary to differentiate the primary modes of toxic action which concomitantly lead to the observed inhibition of ATP synthesis.

3.1.1 Effects of tinorganic compounds on the energy metabolism of cells In undisturbed energy-transducing membranes (membranes of bacteria, mitochondria, or chloroplasts) either the electron transfer chain or the photosystem transports protons through the membrane (primary proton pumps). This results in an electric field over the membrane, Dy, and in a proton concentration gradient generally expressed as DpH. Thus the energy is converted into a chemiosmotic gradient over the membrane. This gradient provides the ATP synthase (secondary proton pump) a driving force for the synthesis of ATP (Mitchell 1961; Mitchell 1966). Further the

– 24 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

gradient is used to maintain chemostasis of the cell, e.g. the uptake of substrate by substrate I H+sympor t mechanisms. Tinorganic compounds inhibit oxidative phosphorylation and photophosphorylation in the low micromolar concentration range. Inhibition of oxidative phosphorylation was first described for di- and triethyltin by Aldrige et al. (1955) in rat mitochondria. Inhibition of photophosphorylation was first described by Kahn (1968; 1970) on chloroplasts. TOT compounds were found to be most effective, whereas no effect was observed for tetraorganotin in the same concentration range. In addition, the effect of di- and monoorganotin was smaller and less specific (Aldrige 1976). The following overview will focus on the effect of TOT compounds, nevertheless some considerations are also valid for di- and monoorganotin compounds. In principal, TOT compounds can disturb the complex interplay of the primary and secondary proton pumps by four primary modes of action which are depicted in Fig- ure 3.1. The overall inhibition of oxidative phosphorylation and photophosphorylation is mainly attributed to the inhibition of the F1F0 and C1C0 ATP synthase respectively (Figure 3.1-D), and to an o H-/anion shuttle (Figure 3.1-B) across the energy- transducing membrane (Aldrige 1976; Selwyn 1976).

Figure 3.1 Chemiosmotic coupling of energy-transducing membranes and possible modes of toxic action of triorganotin compounds. TOT compounds can : A) inhibit the primary proton pumps, B) short-circuit the chemiosmotic grad ient by inducing an oH-/anion antiport, C) act as an OH- uniporter, and D) inhibit the F1F0 ATP synthase.

- 25 - Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

The F-type ATP synthase is of bipartite structure. The F0 moiety consists of several membrane spanning subunits which enable proton transfer through the membrane.

The F1 headpiece is completely exposed to the aqueous phase and includes the

ADP/Pi and the ATP binding site (Fillingame 1990). TOT compounds can block the

proton transfer through the F0 moiety (Gould 1976; Selwyn 1976; Papa et al. 1982). The inhibition by TOT compounds can be reversed by the addition of thiol compounds, e.g. dithiothreitol (Yagi et al. 1984). The second mechanism, the OH–/anion antiport (Figure 3.1-B) impairs the chemiosmotic gradient. The TOT cation can form a hydrophobic complex with OH– and diffuse along the OH– concentration gradient to the p-side where exchange of anion can take place. Back diffusion of the complex with the counterion closes the circle. The overall process leads to a conversion of DpOH (i. e. DpH) to an anion concentration gradient, however, Dy remains unchanged. Resulting swelling of the vesicle has been described for mitochondria (Stockdale et al. 1970), erythrocytes (Selwyn et al. 1970), and artificial phospholipids (Selwyn et al. 1970). In addition, the concomitant loss of DpH disables substrate uptake by symport mechanisms (Skilleter 1975). For the OH– uniport (Figure 3.1-C) permeation of the TOT cation is obligatory to close the shuttle. In contrast to the OH–/anion antiport, the OH– uniport also depletes Dy. A similar mechanism occurs for several aromatic acids, which induce a protonophoric shuttle. Distribution of the negative charge in the aromatic system enables permeation of the phenoxide anion (Miyoshi et al. 1987b; Terada 1990; Escher et al. 1996b). In the following the term „protonophoric shuttle“ is also used for the OH– uniport, although the postulated underlying mechanism is an OH– shuttle. Due to the dipole potential of the membrane, migration of cations across bilayers is orders of magnitude smaller than migration of structurally similar anions (Flewelling et al. 1986). Nevertheless, a protonophoric shuttle was postulated for acridones (Horváth et al. 1996). Selwyn et al. (1970) excluded migration of the tripropyltin cation in liposomes. Yet, for tributyltin, migration of the cation as part of a OH– uniport shuttle was postulated by Bragadin et al. (1997; 2000) in mitochondria. Effects of TOTs on the primary proton pump (Figure 3.1-A) are under debate. Gould (1976) described complete inhibition of ATPase, while no effect on the electron transfer was observed in chloroplasts for TPT concentrations ranging from 0.1 to 10 µM. Complementing the observation of Gould, Klughammer et al. (1998) described a decline of the electron transport rate induced by TPT or TBT with concentrations ranging from 10 to 150 µM in uncoupled chloroplasts. They explain their findings with

the binding of TOTs to a postulated proton channel (analogous to the F0 moiety).

– 26 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

According to their model this channel connects the quinone binding sites located in the hydrophobic core of the membrane with the adjacent water phase. The ubiquinol- cytochrome c oxidoreductase (bc1 complex) is similar in many respects to the bf complex (Furbacher et al. 1996). The bc 1 complex (Figure 3.2) features two ubiquinol binding sites. Analogous to the bf complex these two binding sites must be interconnected with the adjacent water phase. TOT sensitive, interconnecting proton chan- nels may therefore be suggested for the bc1 complex as well. Bragadin et al. (2000) described a decrease in the respiratory rate of completely uncoupled mitochondria in presence of TBT, which they ascribed to an inhibition of the electron transfer chain. This work aims to give insight into the relevance of the two mechanisms under debate: OH- uniport and inhibitory effects on the bc1 complex, for the overall effect of TOT on ATP synthesis. Tinorganic compounds are known to form complexes with various inorganic or organic lignads, e.g. phosphate or dithiothreitol (Hynes et al. 1987). Such ligands are often present in biological assays and can have significant effects on the observed activities of TOT compounds (Yagi et al. 1984). Consequently, efforts were made to obviate unnecessary ligands in the assay and to maintain consistency among all components present in order to compare the effects found in different assays. Strongest complexes are formed with OH- (Hynes et al. 1987). Thus, the activities are expected to be dependent on the pH. Therefore the influence of pH was investigated on the overall inhibition of ATP synthesis and on the OH- uniport, which was found to be the more relevant of the two investigated mechanisms. All experiments were performed with chromatophores of the purple bacterium Rhodobacter sphaeroides by means of time-resolved spectroscopy. In previous work, we have used chromatophores to differentiate protonophoric effects and inhibition of the electron transfer chain induced by substituted phenols (Escher et al. 1997). To enable a

Figure 3.2: 2H+

Ubiquinol reduction site, Q 0 , and ubiquinol oxidation site, Q P, in the bc1 complex. Outlined arrows: subsequent flow of two electrons from Op to 0 0 via cytochrome b566 and cytochrome b561 , and from Op to cyto- ch rome c1 via the FeS (Rieske) cluster. Filled arrows: mass flow of quinone, 0 , and quinoles, OH 2, respectively; Bold lines: Inhibition site of antimycin (ant) and 2c (red) myxothiazol (myx). 2

- 27 - Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

qualitative comparison of the effects on overall ATP synthesized based on the mechanisms 1–4 in Figure 3.1, determination of the amount of ATP synthesized was integrated in the test system by means of a luciferin/luciferase system.

3.2 Materials and methods

3.2.1 Chemicals Tributyltin chloride (96 %), triphenyltin chloride (97 %), 2-morpholino-ethanesulfonic acid (MES, BioChemika), Tris(hydroxymethyl)aminomethane acetate (TRIS acetate, BioChemika) were obtained from Fluka (Buchs, Switzerland). Luciferase (lyophilized) and D(–)-luciferin, both from Photinus pyralis, adenosine-5'-diphosphate monopotassium salt (ADP), adenosine-5'-triphosphate (ATP, standard lyophilized), (P-1,P-5-di(adenosine-5')-pentaphosphate (dAPP) were obtained from Boehringer (Mannheim, Germany). Sources of redox mediators, inhibitors and ionophores used

for the determination of protonophoric activity and for the cyt bc1 inhibition experiments are described in (Escher et al. 1997).

3.2.2 Stock solutions TBT and TPT were dissolved in methanol containing 0.01 M HCl and stored at 4 °C. Maximum concentration was 2.15 mM (TPT) and 23.7 mM (TBT). TRIS acetate or MES

were dissolved in 5 mM Mg(ClO4)2 to a final concentration of 10 mM and pH was adjusted by addition of KOH to 7.5 and 6.1. Ethanolic solutions were prepared of the inhibitors, (antimycin 10 mM (5 mg/ml), myxothiazol 4 mM (2 mg/ml)), of the ionophores (valinomycin 2.5 mM (2 mg/ml), nigericin 2.5 mM (2 mg/ml)) and of the redox mediators (2,3,5,6-tetramethyl-p-phenylene diamine (DAD) 20 mM, N-methyl phenazonium methosulfate (PMS) 20 mM, and the following quinones in one single stock solution each of them 10 mM: duroquinone, 1,2-naphtoquinone, 1,4-naphtoquinone, 1,4-benzoquinone). For storage, lyophilized luciferase was dissolved in 0.5 M TRIS acetate pH = 7.5 in a concentration of 1 mg/ml. Prior to the experiment the luciferase stock was diluted 1:10 in water. Luciferin stock solution was prepared daily in pre- cooled 10 mM TRIS acetate pH = 7.5 in a concentration of approx. 1 mg/ml. The exact . 4 . –1. –1 concentration was determined photometrically at 327 nm (e327 = 1.82 10 L mol cm ; Bowie 1978). The redox buffer succinic and fumaric acid were dissolved together in water to a final concentration of 0.5 M of each. By successive addition of KOH the pH

was set to 7.5. Stock solutions of KH2PO4 (0.5 M) and dAPP (2 mM) were prepared in water. All aqueous solutions, except luciferin and pH buffer, were aliquoted and stored at –20 °C.

– 28 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

3.2.3 Preparation of the chromatophores R. sphaeroides, strain GA, were cultivated in a batch reactor in a medium according to Sistrom (1960), and the conditions described in Escher et al. (1997). Cell free extract was prepared by French pressing and the chromatophores were prepared by several wash and centrifugation steps, frozen in liquid nitrogen, and stored at –80 °C. Details are given in Escher et al. (1997). In all steps the following buffer was used:

50 mM MOPS, 75 mM KCl, 10 mM MgCl2, and approx. 25 mM KOH resulting in a final pH of 7.0. The overall chloride concentration resulting in the storage buffer is 75 mM. Final dry weight of chromatophores was 86 mg/ml. Phosphor content was determined according to Escher et al. (1996a). Corresponding phospholipid content was approximately 30 % of the dry weight, the rest can be attributed to membrane proteins.

3.2.4 Protonophoric shuttle and bc1 complex inhibition.

Protonophoric shuttle and bc1 complex inhibition were determined in a specially designed, single-beam spectrophotometer, equipped with a flash excitation unit (Escher et al. 1997). Experiments were performed under anaerobic conditions in a glass cuvette at a redox potential adjusted to 120–130mV by addition of sodium dithionit. All experiments were performed in 6.6 ml buffer containing 25 mg/ml chromatophores and the redox mediators: 20 µM DAD, 20 µM PMS, 10 µM of each of the quinones. Chloride concentration calculated from the dilution of the chromatophore storage buffer was 2.2 mM. TRIS acetate and MES were used at pH = 7.5 and pH = 6.1 respectively. The pH was remeasured at the end of the experiment. Inhibition experiments were conducted under completely uncoupled conditions by the addition of 2.5 µM valinomycin and 2.5 µM nigericin. Protonophoric shuttle experiments were conducted with the bc1 complex inhibited by the addition of 10 µM antimycin and 4 µM myxothiazol. Before starting the experiment, the chromatophore suspension was equilibrated for one hour in the dark. Activity of the quinone binding sites QP and Qn of the bc1 complex were investigated by measuring the kinetics of the oxidation- reduction cycle of the cytochrome b561. Concentration of the reduced cytochrome b561 is proportional to the difference in absorbance at l = 561 nm and l = 569 nm. Absorption traces at the respective wavelength were recorded one after another and subtracted subsequently according to ctAtAt() () () (3.1) cyt b561, reduced µD561 -D 569

The oxidation-reduction cycle of the bc1 complex was induced by a series of 8 single turnovers of the reaction center which were induced by 8 short xenon flashes with a frequency of 40 ms.

– 29 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

The build-up and decay of the membrane potential was measured as electrochromic

shift of the carotenoids at l = 503 nm. Dy is proportional to DA503. Acceleration of the decay of the membrane potential was quantified as a pseudo first-order rate constant which was determined by normalizing each trace to the control trace, followed by fitting a first-order decay law to the normalized trace (Escher et al. 1997).

3.2.5 ATP synthesis inhibition ATP synthesis inhibition was determined as the decrease of the luciferase luminescence in-situ (Stanley et al. 1989). The luminescence meter from Walz (Effeltrich, Ger- many) consists of a photomultiplier and signal amplifier unit, PM-101/D, a photomultiplier control unit, PM-101/N, and a light source (high power LED HPL-L875, orthogonal to the photomultiplier). Experiments were performed in a quartz cuvette (diameter 1 cm) with two mirroring sidings. The luciferin/luciferase assay was mixed from scratch in order to prevent loss of free TOT due to sorption to albumin or complex formation with dithiothreitol (Yagi et al. 1984) present in many commercial assays. TRIS acetate and MES buffer were prepared as described above. For the experiments at pH = 7.5, 1.2 ml of buffer were pipetted into the cuvette. Subsequently the following

components were added to the respective final concentration: KH2PO4, 1 mM; succinic and fumaric acid, 1 mM of each; and 15 µM dAPP, to inhibit adenylate kinase (Lienhard et al. 1973). Then, the TOT stock solution, or pure solvent, and 0.66 mg/ml chromatophores were added. The mixture was then incubated for 10 minutes. Thereafter, luciferin (final concentration: 30 µM), luciferase (0.95 mg/ml), and ADP (0.11 mM) were added and the recording of the signal was started. After one minute, ATP synthesis was induced by turning the LED light on for 6 s. After the emission signal attained its maximum, a defined amount of ATP was added for signal calibration. Control experiments, in which ATP was added first and the LED was turned on subsequently, showed the same results. To compensate loss of signal inten-

sity at pH = 6.1 the amount of chromatophores and the concentration of KH2PO4 had to be increased to 2 mg/ml and 5 mM respectively. Chloride concentration calculated from the dilution of the chromatophore storage buffer was 0.5 mM at pH = 7.5 and 1.75 mM at pH = 6.1.

3.2.6 Definition of the nominal chromatophore concentration, ccph,nom Concentration of chromatophores varied between the different assays and the different pH values. In order to compare the results of the different assays, concentrations were normalized to nominal concentration of TOT in the chromatophores:

– 30 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

n c = tot cph,nom (3.2) mcph where ntot denotes the total amount of TOT added to the assay and mcph denotes the amount of chromatophores as mass of the dry weight. Neglecting the amount of freely dissolved TOT is justified considering the high concentration of chromatophores in the assay and the high chromatophore-water distribution ratio as was determined in a parallel work for TBT (cf. chapter 2). According to this study, a chromatophore-water distribution ratio of approximately 1000 can be assumed for TBT at pH = 7.5 (cf. chapter 2). With this value, the maximum fraction of TBT present freely dissolved in the aqueous phase is below 10 %. For lower pH the fraction in the aqueous phase decreases. The affinity of TPT to ligands in the chromatophores can be assumed to be higher and accordingly the fraction in the aqueous phase to be lower.

– 31 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

3.3 Results and discussion

3.3.1 Inhibition of the bc1 complex

In the present system, we investigated inhibition of the bc1 complex under uncou-

pled conditions in terms of oxidation and reduction of the cytochrome b561 (cyt b561). Uncoupled conditions are necessary to prevent inhibitory effects of the membrane

potential on the electron transfer in the bc1 complex. Complete uncoupling was

achieved by the addition of nigericin and valinomycin. Reduction of cyt b561 was detected by kinetic difference spectroscopy (Escher et al. 1997) at pH = 7.5. Single turnover of the reaction center was induced by a series of 8 flashes in the dark. Subsequent

reduction of the ubiquinol pool drives the redox processes in the bc1 complex. Cyt b561 is reduced and oxidized after each flash (cf. Figure 3.2). As the kinetics of these reactions exceeds the time resolution of our set-up, the control stays nearly unchanged

(negative control in Figure 3.3-A). If the reoxidation of cyt b561 is inhibited due to direct

inhibition or due to the inhibition of Qn, the kinetic trace reflects the reduction of

cyt b561 (positive control by the addition of antimycin in Figure 3.3-A). TBT and TPT

A: inhibition of Qn B: inhibition of Qp

positive control

569 negative control (10 µM antimycin) (10 µM antimycin) D A

–

0.001 abs. units 561 21 mmol/kg TBT

D A 21 mmol/kg TBT

positive control negative control (4 µM myxothiazol) cyt b 561 c

single-turnover single-turnover flashes flashes 100 ms time Figure 3.3:

Effect of tributyltin (TBT) on the kinetics of the reduction (A) and oxidation (B) of cyt b561. Electron transfer is induced by a series of 8 single-turnover flashes. Assay at pH=7.5 contained 25 mg/ml chromatophores, 10 mM TRIS acetate, 10 mM perchlorate, redox mediators (see methods), 2.5 µM nigericin, 2.5 µM valinomycin, NaS2O4 < 2mM, and, if applied, 10 µM antimycin and 4 µM myxothiazol.

– 32 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

disturb the oxidation of cyt b561 (effect of 21 mmol/kg TBT is shown in Figure 3.3-A).

Inhibition of QP prevents the reduction of cyt b561. QP inhibition can be detected if antimycin is added previously (negative control in Figure 3.3-B). Subsequent addition of a QP inhibitor (e.g. myxothiazol) prevents reduction of cyt b561 and the signal flattens (positive control by the addition of myxothiazol in Figure 3.3-B). TBT and TPT also disturb the reduction of cyt b561, yet the effect is only evident after the first flash (effect of 21 mmol/kg TBT in Figure 3.3-B). Even at the highest concentration after the second flash, the electron cycle is able to completely reduce cyt b561. Due to the low solubility of TOT compounds in the solvent used for the stock solutions (methanol), the volumes of the added solvent were relatively high (2 % V/V). Although the applied volumes themselves do not disturb the bc1 complex, as was determined in the control experiments, perpetuation of the redox potential is delicate. Additionally, nonspecific effects seemed to overlay when higher TOT concentrations were present. Therefore, the experiment was not continued to 100 % inhibition (given the effect of antimycin or myxothiazol as 100 %). Nevertheless, it was obvious that up to a concentration when

ATP synthesis was inhibited by 50 %, no significant inhibition of the bc1 complex occurred. The present observations correspond with the findings of Klughammer et al. (1998). They described inhibition of the plastoquinol-plastocyanin oxidoreductase (bf complex) in chloroplasts for concentrations above those that inhibit the ATP synthase.

Further consistently with the present findings, Klughammer et al. (1998) found that Qn was more sensitive than Qp.

3.3.2 Comparison of OH–/Cl– antiport and OH– uniport Build-up and decay of the membrane potential in chromatophores was determined by time-resolved spectroscopy. The potential Dy should not be sensitive to a charge- neutral antiport. This assumption was confirmed in an analogous experiment with nigericin. Nigericin induces a charge-neutral H+/K+ antiport (Gomez-Puyou et al. 1977). Figure 3.4-A shows the effect of the addition of 10 µM nigericin. Besides a slight delay in the build-up kinetic the two curves are identical. No net charge is transported in both the H+/K+ antiport induced by nigericin and the OH-/Cl– antiport of TOT. Therefore a possible OH–/Cl– antiport, with remaining chloride (ca. 2.2 mM), should not interfere with the signal of the membrane potential at 503 nm.

3.3.3 OH– uniport – OH uniport (Figure 3.1-C) was determined at pH = 7.5 (±0.1) and pH = 6.1 (±0.1) as acceleration of the decay of the membrane potential. The acceleration of the decay of the membrane potential induced by the addition of 9.6 mmol.kg–1 TBT is shown in

– 33 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

Figure 3.4-B. Build-up of the membrane potential after a single turnover flash can be split into two phases: (i) a fast phase (1 µs) resulting from the charge transfer by the reaction center and (ii) a slower phase (1 to 5 ms) resulting from the subsequent con-

tribution of the bc1 complex (control curve in Figure 3.4-A). As TBT and TPT interfere

with the bc1 complex, antimycin and myxothiazol were added to the assay to inhibit the second phase in the control trace of Figure 3.4-B. Despite this precaution, after addition of TBT the initial membrane potential decreased to approximately 85 % of the initial potential of the control trace (see Figure 3.4-B). This could be due to a specific inhibition of the reaction center complex. However, similar effects also occurred with linear alcohol ethoxylates (Müller et al. 1999) and with a series of nonspecifically acting hydrophobic compounds (unpublished data from our lab). The observed decrease of the initial potential could therefore also be attributed to a nonspecific effect. Yet, the concentrations in the membrane are low compared to the concentrations where unspecific effects are observed with other compounds. The effect of the OH– uniport, i.e. the acceleration of the decay rate is quantified as – pseudo first-order constant, kobs. The concentration dependence of the OH uniport of – TBT and TPT expressed as kobs is depicted in Figure 3.5. For TBT, the OH uniport was

A: charge-neutral antiport B: OH- uniport control (no inhibitors) control 10 µM antimycin, 4 µM myxothiazol 2.5 µM nigericin 0.002 abs. units 503 D A 9.6 mmol/kg TBT D

single turnover single turnover flash flash

100 ms time Figure 3.4: Effect of nigericin and tributyltin on the membrane potential Dy. The membrane potential was built-up by one single-turnover flash. The assay at pH=7.5 contained 25 mg/ml chromatophores, 10 mM TRIS acetate, 10 mM perchlorate, redox mediators (see methods), NaS2O4 < 2mM, and, if applied, 2.5 µM nigericin, 10 µM antimycin and 4 µM myxothiazol.

– 34 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

– more effective at pH 6.1 than at pH = 7.5. For TPT, no OH uniport was observed at pH = 7.5 and exhibited only minimal activity at pH = 6.1, The observed increase of the – OH uniport at lower pH can be attributed to the speciation of TBT. At pH = 7.5, 95 % of TBT in the aqueous phase is present as a hydroxide complex and only 5 % is present as cation. Yet, at pH = 6.1 the fraction of the cation is 55 % (fractions are calculated with – pKa = 6.25 given in Arnold et al. 1997). The OH uniport, according to the mechanism depicted in Figure 3.1-C has the highest activity when the concentrations of the cation and the hydroxide complex in the lipid phase in the membrane are equal (Escher et al. 1999). The increase in activity with lower pH can therefore be attributed to a more favorable speciation. Unfavorable speciation might in part also be the reason for the inactivity of TPT. However, at pH = 6.1, the fraction of TPT present as the cation is

10 % (pKa = 5.2 from Arnold et al. 1997), thus higher than the fraction of TBT present as the cation at pH = 7.5. Therefore from the point of view of speciation a comparable activity could be expected. The inactivity of TPT may result from its different distribution behavior as compared to TBT. The liposome-water distribution ratio of the TPT cation exceeded that of TBT by one order of magnitude, which was attributed to a higher affinity of TPT to the phosphate ligand in phosphatidylcholine (cf. chapter 2). From the present findings it can be hypothesized that the TPT cation, which is com- plexed with biological ligands, is not available for the OH– uniport shuttle. Thus, the higher affinity to the ligands rather decreases the OH– uniport activity. Complex formation with inorganic ligands in the aqueous phase might also influence the efficiency of the OH– uniport. For this reason the effect of phosphate and chloride were evaluated for TBT. The effect of phosphate was tested mainly to confirm that the results of the OH– uniport experiments are comparable with the ATP synthesis experiments, where a certain phosphate concentration was a prerequisite. Addition – of 5 mM KH2PO4 did not influence the observed OH uniport. Chloride was added to test for possible competition with a OH–/Cl– antiport. Addition of 0.5 M Cl– reduces the observed effect at both pH values (only 0.5 M was tested). The concentration of 0.5 M may seem high with respect to osmotic pressure, however the addition had no effect on the initial membrane potential. Bursting of chromatophores would lead to a decrease of the initial signal. The decrease of the OH– uniport activity after addition of chloride can thus be attributed to the reduction of the fraction present as hydroxide complex or as cation after formation of a chloride complex. Selwyn (1992) has described a detergent-like, lytic action on mitochondrial membrane by TPT. Heywood et al. (1989) have shown that TBT affects the integrity of the structure of phosphatidylcholine liposomes. High levels of TOT were found to induce the release of soluble proteins from the mitochondrial membrane (Tzagoloff et al. 1968;

– 35 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

Byington 1971; Wulf et al. 1975). The question therefore arises as to whether the observed decrease of Dy is due to a specific shuttle, or whether it can be attributed to a nonspecific loss in membrane structure. There are three arguments which favor a specific effect: (1) no acceleration in the decrease of Dy was observed with TPT in the same range of concentration in the membrane, (2) the concentration of TBT in the membrane was two orders of magnitude below the concentrations where acceleration in the decrease of Dy, due to nonspecific effects, was observed with other hydrophobic compounds (unpublished data from our lab), (3) the reversing effect of chloride is a strong indication that a specific mechanism involving the TBT cation is responsible for the observed acceleration in the decrease of Dy. A shuttle, as depicted in Figure 3.1-C, is the most convincing explanation for the observed effect of chloride. Selwyn et al. (1970; 1976) questioned permeation of the cation for tripropyltin for a phospholipid liposome bilayer. Alternatively, Bragadin et al. (1997; 2000) observed a decrease of Dy in mitochondria induced by three different alkyltins by measuring the quenching of the absorbance of the dye safarin. The results of the present study clearly show that TBT induces transfer of a charged species through the membrane and decreases Dy. The permeating charged species is presumably the TBT cation.

Figure 3.5 Dependence of the OH– uniport and the overall ATP synthesis inhibition on the concentration of tributyltin and of triphenyltin at pH = 7.5 and pH = 6.1. – {/ OH uniport expressed as kobs (left axis). Data points are measurements at the respective concentration in two individual assays. Trend line for TBT is 2nd order least square fit of the data. Q amount of ATP synthesized in percent of control (right axis). Data points are single measurements. Trend line is drawn manually without underlying model.

– 36 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

-I OJ ~"""' "O :::I II ...... u,

...... 0 I\)0 0""'" 0O> 0CX> 0 0 I\)0 0""'" 0O> 0CX> 0 ATP synthezised (% of control) ATP synthezised (% of control)

0 ¢ ...... 0 .,,-t 0 ~-t ¢ "C "O :::c 0 :::I II II !'l :"' .... ¢ U1

OC) "'O ::r w :::, 0 0 3 3 3 0 ::::::: 7' !0--"" + ...... -o ...... I\) ""'" O> CX> 0 O> CX> 0 0 0 0 0 0 0 0 0 0 0 ATP synthezised (% of control) ATP synthezised (% of control)

- 37- Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

3.3.4 Inhibition of ATP synthesis ATP synthesis was determined in situ by the light emission of the added luciferin- luciferase system. ATP synthesis was induced by a light pulse of 6 s. The inhibitory effect of TBT and TPT on the overall ATP synthesized at pH = 7.5 is shown in Fig- ure 3.6. As TBT and TPT might also inhibit the luciferase, e.g. by complexing ATP (Hynes et al. 1987), each assay was calibrated by addition of a defined amount of ATP. TBT did not influence the luciferase in the applied concentration whereas TPT did, which is reflected in a smaller increase in the signal after ATP addition (cf. Figure 3.6). At the maximum TPT concentration the inhibition of luciferase was approximately 50 %. As the amount of ATP synthesized was calculated from the ATP signal, the inhibitory effect on the luciferase was corrected for every single batch. Figure 3.5 shows the concentration dependence of the inhibition of the ATP synthesis at pH = 7.5 and at pH = 6.1. Both compounds were found to be more effective at pH = 6.1 than at pH = 7.5 which agrees with the findings of Dawson et al. (1974) for trimethyltin.

A) TBT B) TPT control TBT control 0.8 mmol.kg–1 TBT TPT 1.7 14.5 mmol.kg–1 mmol.kg–1 TPT 28.8 ATP mmol.kg–1 ATP emission (V) ATP ATP ATP ATP light light light light light light

time 25 s Figure 3.6: Effect of tributyltin and triphenyltin on the ATP synthesis. Signal shows light emission by luciferin-luciferase. ATP synthesis was induced by a light pulse of 6 s. After 20–40 s ATP was added for calibration of signal (note the difference in amount of ATP added: TBT experiments 0.3 nmol, TPT experiments 0.4 nmol). Media at pH=7.5 contained 0.66 mg/ml chromatophores, 10 mM TRIS acetate, 1 mM fumarate and succinate, 1 mM phosphate, 0.11 mM ADP, 0.95 mg/ml luciferase, ca. 0.03 mM luciferin and 10 mM perchlorate.

– 38 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

3.3.5 Comparison of the different mechanisms of action The inhibition of the ATP synthesis is a sum parameter. It integrates inhibitory effects on every link of the energy conversion chain including the ATP synthase (Figure 3.1). Inhibition of the primary proton pumps, i.e. inhibition of the reaction center and of the bc1 complex, were shown to be irrelevant: At concentrations which inhibited synthesis of ATP to 20 % of the control, the initial membrane potential which is built-up by the reaction center complex was more than 90 % of the control. Significant inhibitory effects on the bc1 complex were only detected for concentrations above the ones which inhibited ATP synthesis to 50 %. As in all assays, only 10 mM of perchlorate and negligible amounts of chloride were present, an OH–/anion antiport can be precluded. For TPT no OH– uniport was observed and thus, the observed inhibition of the ATP synthesis can be attributed to the inhibition of the ATP synthase. For TBT, the OH– uniport was observed in the same concentration range as inhibition of ATP synthesis (Figure 3.5). Thus it is expected to be relevant for the overall inhibition of ATP synthesis. However, the exact contribution of the OH – uniport to the decrease in ATP synthesis cannot be quantified.

The following effect concentration for half maximum inhibition, EC50, of the ATP synthesis were deduced from the graph in Figure 3.5: for TBT EC50(pH = 7.5) = 3.2,

EC50(pH = 6.1) = 2.4; for TPT EC50(pH = 7.5) = 22, EC50(pH = 6.1) = 8, all values are in units of mmol.kg–1. The effectiveness of the two compounds as inhibitors differs less than one order of magnitude. This agrees with findings in the literature. In a comparison of Selwyn (1976) on the inhibitory activity of various TOT on mitochondrial and chloroplast phosphorylation, the activity of TBT exceeded the activity of TPT by a maximum factor of three. In a comparison of the inhibitory effects of TBT and TPT on reverse electron transfer in submitochondrial particles (SMP) and on the luminescence in the Microtox assay, TPT was found to be more effective by a factor of three than TBT in SMP whereas in the Microtox assay, TBT was found to be more effective by a factor of two (Argese et al. 1998). The findings of this work agree with the general knowledge pertaining to these two compounds. The hydroxide complex of TBT has a higher hydrophobicity, in terms of octanol-water distribution ratio Kow, than that of TPT (Arnold et al. 1997). A higher hydrophobicity favors a mechanism which necessitates permeation of the hydrophobic core of the membrane. This also agrees with the finding that TBT has a higher activity in the OH–/Cl– shuttle (Selwyn et al. 1970; Selwyn 1976). Alternatively, the TPT cation has a higher affinity to organic ligands, e.g. to phosphatidylcholine (cf. chapter 2), which favors binding to sensitive sites in enzymes. The different mecha-

– 39 – Acute toxicity of triorganotin compounds: Different specific effects on the energy metabolism and role of pH

nisms of action all result in the inhibition of the ATP synthesis, however they necessitates different chemical properties. This may explain why correlations of acute toxicity

data of TOT compounds, with Kow as a descriptor for hydrophobicity, fail to describe the toxic effects satisfactorily.

– 40 – 4

Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

Abstract The uncoupling activities of 14 binary mixtures of substituted phenols and of 4 binary mixtures of phenols and anisols were investigated at different pH values. Experiments were performed with time-resolved spectroscopy on membrane vesicles of the photosynthetic bacteria Rhodobacter sphaeroides (chromatophores). Phenols are known to destroy the electrochemical proton gradient in energy-transducing membranes by a protonophoric mechanism. In 13 of the 14 mixtures of substituted phenols, interactions led to increased uncoupling activity compared to the sum of the effects of the single compounds. These findings are significant regarding both, (i) the mechanism of protonophoric action and (ii) the risk assessment process of chemical mixtures in the environment. (i) It was postulated in the literature that in the mechanism of the protonophoric action of a single compound, the formation of a dimer between the phenoxide and the neutral phenol enhances its protonophoric activity. The observed increase in the uncoupling activity of the binary mixtures gives further evidence for this postulate because the overall uncoupling activity of the 13 mixtures with interaction can be described by a model which includes a term for the contribution of a mixed dimer. Furthermore, it was found that opposite speciation favors interaction and ortho substituents abate interaction, which gives evidence for dimer formation via a hydrogen bond between the phenol-OH and the phenoxide. (ii) When assessing the effect of mixtures, concentration addition is regarded as a concept to estimate effects of similar acting compounds. The substituted phenols in this work act according to the same mechanism of action of uncoupling. Nevertheless, the overall effect of 6 of the investigated mixtures exceeded the effect calculated according to concentration addition considerably. This synergistic effect in vitro will need to be validated in in vivo to deduce its implications for the risk assessment process. Interaction of phenolic uncouplers in bina1y mixtures: Concentration-additive and synergistic effects

4.1 Introduction

Hydrophobic ionogenic organic compounds, HIOCs, like organic acids and bases, are of increasing environmental concern since many pesticides (Worthing et al. 1991) and many pharmaceuticals and personal care products (Halling-S0rensen et al. 1998; Daughton et al. 1999) contain ionogenic groups. We have chosen substituted phenols as representatives of HIOCs to investigate mechanisms of interaction with respect to their toxic effect on energy metabolism. The chosen compounds act predominantly as uncouplers of the chemiosmotic phosphorylation, i.e. they destroy the electrochemical proton gradient of energy-transducing membranes (Miyoshi et al. 1990; Terada 1990; Escher et al. 1996b). The electrochemical proton gradient is a system of energy conservation and allotment common to all cells. Enzyme complexes in the membrane transport protons through the membrane under energy consumption (primary proton pumps). This results in a proton concentration gradient over the membrane, expressed in terms of i:1pH, and further in an electric potential over the membrane, L1 'If. The electrochemical proton gradient is used to drive energy-dependent processes, e.g. ATP synthesis by the ATP synthase. Weak aromatic acids like phenols short-circuit the electrochemical proton gradient by a specific proton shuttle.

H+ H+ n-side n-side A(1)- ~ ~ HA(1) A(2)- HA(2) ~ I A(2)HA(1 )- 1 A(2)HA(2)-

A(1l)- ~ HA(1) jAb A!2- HA(g} p-side p-side H+ H+ rH+ H+ IH+ H+ I H+ H+ H+ H+ I I A c B Figure 4.1 Model of the protonophoric shuttle in an energy-transducing membrane. A: monomeric shuttle mechanism of a first-order uncoupler with the phenoxide species, A(1r. and the phenolic species, HA(1). B: dimeric shuttle mechanism of a second-order uncoupler with the dimer, A(2)HA(2f, as additional charge transferring species. C: postulated formation of a mixed dimer A(2)HA(1 t as additional charge transferring species in a binary mixture of two substituted phenols.

- 42 - Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

The present model of this protonophoric shuttle mechanism is sketched in Fig- ure 4.1. Because of their hydrophobicity, phenols and phenoxides partition into the outer region of the membrane. Driven by the membrane potential, Dy, the phenoxides migrate through the membrane from the negatively charged n-side to the positively charged p-side (see Figure 4.1-A), thereby decreasing Dy. On the p-side, the phenoxides bind a proton from the adjacent aqueous boundary layer. Then the protonated neutral phenols passively diffuse back to the n-side, where they release a proton, and the circuit is closed. Except for the fast protonation step, all steps are of first-order, and overall first- order kinetics are expected for this monomeric shuttle mechanism. However, several uncouplers were shown to have a second-order dependence of uncoupling activity on their concentration. Because of these findings a dimeric shuttle mechanism was proposed (Lea et al. 1969; Finkelstein 1970). According to this mechanism, a formation of a dimer of the phenoxide and the phenolic species takes place in the membrane (Fig- ure 4.1-B). The resulting dimer is thought to migrate through the membrane more easily than the phenoxide because of its reduced charge density (Finkelstein 1970). Monomeric and dimeric shuttle mechanisms can work concomitantly (Escher et al. 1996b). As both species are required for the described shuttle, the uncoupling activity exhibits a strong dependence on the pH in the surrounding aqueous phase. The activity follows more or less a Gaussian curve with a maximum at the pH at which both species are present in the membrane at approximately equal concentrations (Escher et al. 1996b). The pH dependence of the uncoupling activity can be described in a kinetic model, in which the permeation steps of the different species are treated as first-order reactions. Such models were first developed to describe the conductance induced by weak organic acids in black lipid bilayers (Cohen et al. 1977; McLaughlin et al. 1980). In our previous work we have extended the kinetic model and adapted it to the uncoupling data measured with time-resolved spectroscopy on energy-transducing membranes of Rhodobacter sphaeroides (Escher et al. 1999).

Given a mixture of two compounds, of which one has a high acidity constant, Ka,

(i.e. a low pKa) and is present mainly as an anion and the other has a high pKa and is present mainly as a neutral species, the formation of a mixed dimer can be postulated (see Figure 4.1-C). This mixed dimer can act as an additional carrier of charge through the membrane. In analogy to the single compound dimer, a mixed dimer would have diminished charge density and could enhance the activity of the mixture significantly as compared to the single compounds.

– 43 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

This hypothesis was tested with 18 binary mixtures of substituted phenols and anisols using compounds that cover a wide range of acidity constants and substitution patterns (e.g. ortho versus para substituents). The results are discussed on the one hand with regard to a simpliefied kinetic model of uncoupling of mixtures and on the other hand with regard to the mixture toxicity concept of concentration addition which is a reference concept for chemicals with similar mode of toxic action. Finally potential implications for the risk assessment process are inferred.

4.2 Material and methods

4.2.1 Chemicals The compounds were purchased from the following companies: from Riedel-de Haën (Seelze, Germany): 2-tert-butyl-4,6-dinitrophenol (Dinoterb), 2,6-dichlorophenol (26DCP), 2,6-dibromo-4-hydroxybenzonitrile (Bromox); from Fluka (Buchs, Switzer- land): 3,4,5-trichlorophenol (345TCP), 3,4-dinitrophenol (34DNP), 2,4-dinitrophenol (24DNP), 3,4,5-trimethylphenol (345TMP); from Sigma-Aldrich (Buchs, Switzerland): 3,5-dimethylanisole (35DMA), pentachloroanisole (PCA); from EGA-Chemie (Steinheim, Germany): pentachlorophenol (PCP); from Lancaster (Mühlheim, Ger- many): 3,5-dibromo-4-methylphenol (35DBC); the following zwitterionic pH buffers

from Fluka: 2-morpholino-ethanesulfonic acid (MES, pKa = 6.15); 3-(N-morpho-

lino)propanesulfonic acid (MOPS, pKa = 7.2); N-2-hydroxyethylpiperazine-N'-3-pro-

panesulfonic acid (HEPPS, pKa = 7.8); 2-(cyclohexylamino)-ethanesulfonic acid (CHES,

pKa = 9.55); the following redox buffers from Fluka: succinic acid and fumaric acid. Chemicals used for time-resolved spectroscopy are described in Escher et al. (1997). Egg yolk phosphatidylcholine was purchased from Lipoid (Ludwigshafen, Germany).

4.2.2 Determination of the uncoupling activity The effect of the compounds on the membrane potential was detected by time- resolved spectroscopy using membrane vesicles (chromatophores) of the photosynthetic purple bacterium R. sphaeroides. Cultivation of bacteria, preparation and characterization of the chromatophores and the specially designed single-beam spectrophotometer equipped with a flash excitation unit and kinetic data acquisition capabilities are described in detail in Escher et al. (1997). Measurements were performed at different pH values either in an anaerobic cuvette at a redox potential adjusted to 120– 130 mV, or with the redox potential buffered by the addition of 1 mM of each succinate and fumarate (pH = 7.4 only). The pH was buffered by using a mixture of MES, MOPS, HEPPS, and CHES, adjusted to a total buffer- and K+ concentration

– 44 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

(KCl/KOH) of 50 mM and 100 mM, respectively. At the end of each measurement series, the pH was measured.

4.2.3 Experiments with single compounds Concentration response curves were determined in one single assay by stepwise addition of the respective compound. During one measurement cycle, the membrane potential was built up by a single turnover of the bacterial photosystem after its excitation by a short flash (µs) in the dark. The following decay of the potential was recorded over a period of 150 ms. Flash and record sequence was repeated four times in an interval with 1 min. pause. The resulting four kinetic traces were averaged (1 cycle). At the beginning of every measurement series, the intrinsic membrane potential decay was determined in three subsequent cycles which were used as control. Then the respective compound was added. The assay was equilibrated after each addition with stirring for 15 min. and the kinetic traces were recorded in two cycles. The acceleration of the decay kinetics by the phenolic compound was calculated by normalizing each trace to the control trace, followed by fitting a first-order decay curve to the normalized trace (Escher et al. 1997). The corresponding pseudo-first order rate constant, kobs, represents the uncoupling activity induced by the uncoupler. Two different batches of chromatophores were used. Difference of activity for a given compound in two different batches varied in the range of ± 20 %. For 345TCP, which was used with both batches at pH = 7.4, individual control curves were determined for each batch.

4.2.4 Experiments with binary mixtures

Mixture experiments were performed in a n•n-design according to which in several series the concentration of one compound was kept constant and the concentration of the other was altered. Thus, the assay was prepared and the controls were recorded as described above for the single compound experiments. Then compound-1, in general the weaker uncoupler, was added and the resulting uncoupling activity was determined. Thereafter, compound-2 was added stepwise, and the measurements were performed analogously to the measurements of the single compounds. Each combination with fixed concentration of compound-1 was measured in duplicate and combinations with at least three different concentrations of compound-1 were analyzed. For each combination, the activity of the single components were determined at the respective pH in duplicate.

– 45 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

4.2.5 Physicochemical parameters

For Bromox and 35DBC the acidity constants pKa and the octanol-water partition

coefficients, Kow,HA, were measured with the pH-metric method (Avdeef 1992; Avdeef 1993) on a PCA 101 automatic titrator (Sirius Analytical Instruments, East Sussex, Great Britain).

The liposome-water partition coefficient of the non-dissociated species, Klipw,HA, of

Bromox and 35DBC were estimated from a linear regression of log Klipw,HA versus

log Kow,HA (equation 4.1) from data for a series of substituted phenols given in Escher et al. (2000a).

log KKrn=078. ◊ +== 112. (2 092 . , 20 ) (4.1) lipw,HA ow,HA

The liposome-water partition coefficient of the non-dissociated species Klipw,HA of 345TMP was determined by solid phase micro extraction (SPME; Vaes et al. 1997). Liposomes were prepared from egg yolk phosphatidylcholine (≥ 98 %) as described in Escher et al. (2000a). The distribution experiments were performed at pH = 4.8 with 10 mM acetate buffer and 10 mM NaCl. The liposome concentration used was 5.10–4 kg.L–1 and the concentration of 345TMP varied from 10–5 M to 10–4 M. The aqueous concentration of 345TMP was determined with GC-FID (Fision, Carlo Erba Instruments, Schlieren, Switzerland) after 40 min. extraction by a 1 mm polyacrylate fiber and desorption for 5 min. at 270 °C. Extraction and desorption was performed with a Combi PAL autosampler (CTC, Zwingen, Switzerland). Under the chosen conditions the depletion of the nominal 345TMP concentration in the vial was negligible.

The liposome-water partition coefficients of the phenoxide species, Klipw,A, of Bro-

mox, 35DBC and 345 TMP were estimated from a linear regression of log Klipw,A versus

log Klipw,HA (equation 4.2) from data for a series of substituted phenols given in Escher et al. (2000a).

logKK=090 . ◊ 061. ( rn2 073 . , = 20 ) (4.2) lipw,A lipw,HA -= All other physicochemical parameters were taken from Escher et al. (2000a).

4.3 Quantitative evaluation and representation of the experimental data

The overall uncoupling activity of a single compound is determined by its concentration and speciation inside the membrane and by its intrinsic activity, i.e. the translocation of protons induced by the uncoupler already present in the membrane. The

– 46 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

intrinsic activity of a compound can be understood as the permeability of its charged and uncharged species in the membrane and the ability to form a dimer (Escher et al. 1999). At constant pH and accordingly constant species fractions, the complex kinetic model can be simplified, and the uncoupling activity of a single compound can be described by a simple second-order polynomial equation according to equation 4.3 (Escher et al. 1999). The uncoupling activity resulting from a given concentration of a –1 compound i is expressed as kobs,i(clip,i) in units of s . This is the experimentally determined normalized pseudo first-order decay rate constant of the membrane potential. To enable a direct comparison of the intrinsic activities of the compounds, the observed uncoupling activity is related to the effective concentration of the com- . –1 pound i in the lipids of the membrane, clip,i, in units of mol kg : kc()= acbc ◊ + ◊ 2 (4.3) obs,ii lip,i lip,i i lip,i ai and bi are compound-specific fit parameters for the contribution of the first- and the second-order steps of the protonophoric shuttle mechanism (Escher et al. 1996b). For compounds that do not form dimers, the parameter bi is zero. As the simplified second-order model can only be applied at constant pH, a separate set of parameter ai and bi is derived for each pH.

The uncoupling activity of a binary mixture, kobs,mix, can be described in a simplified kinetic model according to the mixed dimer mechanism dipicted in Figure 4.1 as the sum of the effects of the single compounds and a third term, the product of the membrane concentration of the two compounds and a dimer parameter d:

kccacbcacbcdcc(, )= ◊ + ◊2 + ◊ + ◊2 + ◊◊ (4.4) obs, mix 12 1 lip,1 12 lip,1 lip,2 2 lip,2 lip,1 lip,2 where ai and bi are the parameters derived from experiments with the single compounds. The parameter d was determined by a least-square fit of the entire data set of a given binary mixture at a given pH, i.e. equation 4.4 describes several concentration effect curves. In terms of the mixed dimer mechanism (see discussion), d represents the contribution of the mixed dimer to the uncoupling activity, i.e. its ease of formation and its permeability in the membrane.

The concentration of the compound i in the membrane lipids, clip,i, and the free aqueous concentration, cw,i, were calculated from the known total concentration, ctot,i, and the liposome-water distribution ratio at the respective pH. It was shown previously that sorption of phenolic compounds to membrane proteins can be neglected (Escher et al. 1996a).

– 47 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

clip,i was calculated by applying the liposome-water distribution ratio, Dlipw,i, in the mass balance.

clip,i Ê L ˆ Dlipw,i()pH = Á ˜ (4.5) cw,i Ë kg¯

c=+ c lip◊◊ c mol L-1 (4.6) tot,ii w, lip,i () yielding

ctot,i Ê molˆ c = lip,i 1 Á ˜ + lip Ë kg ¯ (4.7) D ()pH lipw,i where lip is the concentration of the membrane lipids in the assay in units of kg.L–1.

Dlipw,i is calculated by

Ê L ˆ DKlipw,i()pH =aai ◊ lipw,HA,i +-(1 i )◊ Klipw,A,i Á ˜ (4.8) Ë kg¯

where Klipw,HA,i: = (clip,HA,i / cw,HA,i) and Klipw,A: = (clip,A,i / cw,A,i) are the liposome-water parti-

tion coefficients of the non-dissociated and the dissociated form respectively, and ai

and (1 – ai) are the fraction of the non-dissociated and the dissociated form in the

aqueous phase at the respective pH. ai is calculated from the pH and the pKa in the aqueous phase according to equation 4.9. 1 ai = ()pH pK (4.9) 110+ - a

4.4 Results and discussion

4.4.1 Evaluation of the mixed dimer model Of the 14 binary mixtures of substituted phenols, 13 showed an increase in the uncoupling activity compared to the sum of the activities of the single compounds. We postulate that this increase in activity is due to the formation of a mixed dimer of the phenolic species of one compound and the phenoxide of the other (Figure 4.1-C). As an example, in Figure 4.2 the contribution of the individual species to the overall uncoupling activity are depicted for each of the two: 34DNP and 345TCP. The activities, as a function of pH, were calculated according to the kinetic model of uncoupling from Escher et al. (1999) for a concentration of 3.4 mmol.kg–1 for 34DNP and of –1 –1 6.2 mmol.kg for 345TCP. At pH = 7.4 the concentration of 3.4 mmol.kg 34DNP alone

– 48 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

–1 would yield a kobs = 1.3 s , which comprises approximately equal contributions from the protonophoric shuttle mechanism by the anion and by the dimer. . –1 –1 The concentration of 6.2 mmol kg 345TCP alone would result in a kobs = 0.6 s . In this case the contribution of the free phenoxide to the total translocation of charge is negligible. The low dimer activity, which is equal to the overall activity, is due to the lack of the phenoxide species. When 34DNP and 345TCP are applied together, an overall –1 activity of kobs = 4.8 s results, which is more than double the sum of the effects of the

) 8 –1 (s

obs

k 7

6 34DNP 345TCP

0 4 56789101112 pH Figure 4.2 pH dependence of the uncoupling activities of 3,4-dinitrophenol and 3,4,5-trichlorophenol. Modeled contribution of the dimer (– –) and the phenoxide (— - -) to the overall uncoupling activity (——) of the respective compound when 3.4 mmol.kg–1 34DNP and 6.2 mmol.kg–1 345TCP are applied separately. X indicates addition of the effects of the two compounds at pH= 7.4, • indicates the experimentally determined activity of the mixture. Contribution of the species are calculated according to the full kinetic model with the parameters given in Escher et al. (1999). While for 34DNP the calculated activities agreed exactly with the activities measured in the present work, for 345TCP the calculated activities were 10 % above the measured activities at pH = 7.4. For reason of consistency the calculated activities were scaled linearly to fit the measured activities.

– 49 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

single compounds (1.9 s–1). We postulate that the excess activity is due to the additional translocation of charge resulting from the mixed dimer formed of the phenoxide species of 34DNP and the phenolic species of 345TCP both of which are present in excess. As in the illustrative example above, the entire data set of binary mixtures can be interpreted in terms of the mixed dimer model. Overall, binary mixtures of nine phenolic uncouplers and two anisols (see Table 4.1) have been investigated in 18 different combinations of components and pH values (see Table 4.3). Since not all of the single compounds were characterized by the full kinetic model, data evaluation was performed for every pH value separately with the simplified second-order model

(equation 4.3). The constants ai and bi, which describe the concentration-effect curves of the single components with respect to the concentration in the membrane lipids at a given pH (equation 4.3), are listed in Table 4.2.

Table 4.1 Physicochemical parameters of phenolic uncouplers and anisols.

Compound abbr. pKa log Kow,HA log Klipw,HA log Klipw,A 3,4-dinitrophenol 34DNP 5.38a 1.79d 3.17j 1.90j 2,4-dinitrophenol 24DNP 7.40b 1.67b 3.30j 2.47j 3,4,5-trichlorophenol 345TCP 7.73c 4.41c 4.71j 3.16j 2-tert-butyl-4,6-dinitrophenol Dinoterb 4.80h 3.54d 4.10j 3.59j pentachlorophenol PCP 4.75c 5.24c 5.09j 4.49j 3,5-dibromo-4-hydroxybenzonitrile Bromox 4.09e 2.97e 3.44k 2.48l 3,5-dibromo-4-methylphenol 35DBC 8.28e 5.44e 5.41k 4.24l 3,4,5-trimethylphenol 345TMP 10.25f n.d.m 2.66e 1.78 l 2,6-dichlorophenol 26DCP 6.97g 2.64g 2.87j 1.42j 3,5-dimethylanisol 35DMA - n.d.m n.d.m - pentachloroanisol PCA - 5.96i 4.85i -

a Data from Escher et al. (1999). b Data from Schwarzenbach et al. (1988). c Data from Schellenberg et al. (1984). d Data from Escher et al. (1996a). e This work (cf. methods). f Data from Ko et al. (1963). g Data from Lipnick et al. (1985). h Data from Miyoshi et al. (1987a). i Data from Escher (1995). j Data from k l Escher et al. (2000a). Estimated from Kow,HA by use of equation 4.1. Estimated from Klipw,HA by use of equation 4.2. m n.d. not determined.

– 50 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

Table 4.2 Uncoupling activities of the single phenolic uncouplers and anisols at various pH values.

log ECw ECw a b c d e f f Compound pH Dlipw(pH) a b (kee=1.4) (kee=6) (L.kg–1) (kg.mol–1.s–1) (kg.mol–2.s–2) (µmol.L–1) (µmol.L–1) 34DNP 6.1 2.48 270g 60'300g 9.6 26.4 7.4 1.98 300g 26'700g 35.3 110 24DNP 7.4 2.47 50h 0.00h 88 400 345TCP 6.4 4.70 0.00g 2'800g 0.43 0.93 7.4 4.55 0.00g 15'650g 0.25 0.55 7.4 4.55 0.00h 10'950h 0.30 0.65 8.0 4.28 0.00g 15'850g 0.47 1.02 8.5 3.94 0.00g 25'150g 0.82 1.76 Dinoterb 7.4 3.59 1'600g 0.00g 0.21 0.95 PCP 7.4 4.50 510g 0.00g 0.08 0.38 Bromox 7.4 2.38 120h 0.00h 40 160 35DBC 7.4 5.38 36h 240h 0.14 0.44 345TMP 6.1 2.66 0.00g 9.18g 823 1'770 6.4 2.66 0.00g 9.18g 823 1'770 7.4 2.66 0.00g 9.20g 823 1'770 8.0 2.66 0.00g 9.25g 824 1'770 8.5 2.65 0.00g 9.39g 827 1'780 26DCP 7.4 2.34 3.34g 5.75g 1'200 3'490 35DMA 7.4 n. d.i ---- 8.5 n. d.i ---- PCA 7.4 4.85 - -- - a Names for abbreviation are given in Table 4.1. b Mean values of the pH at the end of the measurement series. c Calculated with equation 4.8 and the values given in Table 4.1. d Fit parameter for the first- order step of the uncoupling mechanism cf. equation 4.3. e Fit parameter for the second-order step of the uncoupling mechanism cf. equation 4.3. f Total aqueous concentration needed to induce the respec- g tive endpoint, kee, calculated with equations 4.3 and 4.5. Determined with chromatophores from batch 1 in an anaerobic cuvette (cf. Methods). h Determined with chromatophores from batch 2 with succinate/fumarate redox buffer (cf. Methods). i n.d. not determined.

– 51 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

Table 4.3 Parameter describing the synergistic effect of binary mixtures of phenolic uncouplers.

ÂTU b c Mixture Partners pH d (kee=6) compound-1a compound-2a (kg2.mol–2.s–1) 345TMP 345TCP 6.4 350d 1.0 345TMP 345TCP 8.0 610d 1.1 345TMP 345TCP 8.5 450d 1.2 35DMA 345TCP 6.4 - -f 35DMA 345TCP 8.5 - -f 35DMA 34DNP 7.4 - -f PCA 34DNP 7.4 - -f 345TMP 34DNP 6.1 740d 1.1 345TMP 34DNP 7.4 960d 1.0 345TCP 34DNP 7.4 137'650d 0.7 26DCP Dinoterb 7.4 0d 1.1 345TMP Dinoterb 7.4 3'500d 0.9 26DCP PCP 7.4 500d 1.0 345TMP PCP 7.4 1'400d 0.8 24DNP 345TCP 7.4 9'700e 0.7 24DNP 35DBC 7.4 2'130e 0.7 Bromox 35DBC 7.4 24'500e 0.5 Bromox 345TCP 7.4 89'200e 0.6 a Names for abbreviation are given in Table 4.1. b Fit parameter quantifying the interaction of the two compounds according to equation 4.4. c Toxic units were calculated for a hypothetical binary combination given the components were present in the ratio of the effect concentrations of the single com-

pounds, EC(kee=6), which were backcalculated with Equation 4.3; equation 4.4 was used to calculate the effect concentrations of the mixture. d Determined with chromatophores from batch 1 in an anaerobic cuvette (cf. Methods). e Determined with chromatophores from batch 2 with succinate/fumarate redox buffer (cf. Methods). f The toxic unit concept cannot be applied because no activity was observed for the anisol compounds separately.

A typical concentration-effect plot of a mixture with interaction is shown in Fig- ure 4.3. Each series of data points represents a combination with constant concentration of compound-1, 345TMP, and an increasing concentration of compound-2, 34DNP. The data points on the y-axis show the activity of the first compound separately, the line ✕✌ gives the activity of the second compound separately. The solid lines were fitted with equation 4.4, using the a and b parameters for the single compounds, which are listed in Table 4.2, and d as only adjustable parameter. As is evident from Figure 4.3, the experimental data can be well described by equation 4.4. The

– 52 – Interaction of phenolic uncouplers in bina1y mixtures: Concentration-additive and synergistic effects

dashed line ® gives the activity for the combination with the highest concentration of compound-1 with the contribution of the mixed dimer set to zero, which is equal to the addition of effects.

The last term in equation 4.4, d·clip,1 ·clip.2' is analogous to the second-order term of equation 4.3. The product of d and the membrane concentrations of the two compounds can be interpreted as the contribution of the mixed dimer to the total uncoupling activity. The product of clip,i and clip,2 is proportional to the concentration of the mixed dimer in the membrane. The proportionality factor, i.e. the dimer formation constant, is included in parameter d and cannot be resolved from the intrinsic activity, i.e. the mobility of the dimer across the membrane. From Figure 4.3 it can be seen that

~ 10 I .._..(f) (/) .0 9 .::.:.0 m 8 ®

7 (}) ® 6 / .& / 5 / / 4

0 15 crip(34DNP) (mmol·kg-1)

Figure 4.3 Dependence of the uncoupling activity on the membrane concentration of 3,4-dinitrophenol and 3,4,5-trimethylphenol at pH= 7.4. Full lines are calculated according to equation 4.4. CD : 34DNP separately. Activities of mixture with([): 0.17; ®: 0.33; CD : 0.67 mol·kg-1 345TMP. Dashed line ®:calculated according to equation 4.4 with d set to zero (which is equal to the addition of effects) and 0.67 mol·kg-1 345TMP.

- 53 - Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

the second-order dependence of compound-2 (34DNP) decreased with increased concentration of the mixture partner (345TMP). Hence the interactive effect depends linearly on the concentration of one of the compounds. Transformation of equation 4.4 shows this linear dependence.

kadccbcacbc=+◊ ◊ + ◊2 + ◊ + ◊ 2 (4.10) obs,mix ()22lip,1 lip,2 lip,2 11m,1 lip,1 . The term (a2 + d clip,i) represents an apparent increase of the first-order contribution of compound-2. Whether there is competition between the formation of the single compound dimer and the mixed dimer cannot be resolved. A competition would result in a decrease of the second-order contribution on an absolute scale (smaller b value). It is plausible to assume that the contribution of the single compound dimer is constant, as the fraction of the compounds present as single compound dimer or mixed dimer is small and thus the concentrations of the monomer species remain constant.

Interactions of the components, i.e. d > 0, were found in 13 of the 18 combinations (Table 4.3). One combination showed no interaction most probably due to unfavorable steric constraints as is discussed below. The non-acidic anisols cannot uncouple according to the shuttle mechanism and did not enhance the activity of 345TCP in a binary mixture. The role of the speciation of the compounds, i.e. pH, as well as the role of the substitution pattern on the phenolic ring for the activity of the binary mixtures are discussed in detail below.

4.4.2 Effect of pH According to the mixed dimer model, formation of a mixed dimer can be expected

mainly in combinations of compounds with low and high pKa and at pH values, at which the complementary species predominate. This condition is fulfilled for the combination of 34DNP and 345TMP at pH = 7.4, whereas at pH = 6.1 a considerable fraction of the 34DNP is present in the neutral form and a single compound dimer can be formed thus, decreasing the relative contribution of the mixed dimer. Accordingly, the contribution of the d term to the overall effect should be higher at pH = 7.4 than at pH = 6.1. The higher d value also results in a higher relative increase of the total activity at pH = 7.4 as compared to pH = 6.1, which is shown in Figure 4.4-A. In the example of the combination of 345TCP with 345TMP (Figure 4.4-A), 345TCP is the H-acceptor in the mixed dimer, although at all three pH values the neutral species is present in excess over the phenoxide species. Consequently the activity of the single compound is dominated by the dimeric shuttle mechanism (cf. Figure 4.2).

345TMP has a high pKa so that the fraction of 345TMP anion in the membrane is less than 2 % even at the highest pH value (pH = 8.5). Therefore, 345TMP can be con-

– 54 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

sidered as the H-donor. The d values of this mixture are rather small and do not vary significantly between pH = 6.4 and pH = 8.5 (Table 4.3). However, as can be seen in Figure 4.4-A, the activity at pH = 6.4 is doubled as compared to the addition of effects, but is only 1.5-fold at pH = 8.5. This finding might appear counter-intuitive because, at the higher pH, there is a larger fraction of phenoxide species available for the mixed dimer formation. Obviously, the contribution of the mixed dimer to the overall activity decreases because, at the same time, the activity of 345TCP alone increases due to the larger fraction of the phenoxide.

4.4.3 Effect of substitution pattern Speciation is not the only determining factor for dimer or mixed dimer formation. Compounds with two bulky ortho substituents or with ortho substituents that form strong intramolecular H-bonds (e.g. 24DNP) do not show dimer formation in single compound experiments (Escher et al. 1996b; Escher et al. 1999). The present study tested the hypothesis, whether such compounds are able to form mixed dimers with phenols which have no ortho substituents. PCP and Dinoterb, which are both pure monomeric uncouplers, showed interaction in the combination with 345TMP (Figure 4.4-B). Likewise, Bromox in combination

A: effect of pH B: effect of substitution pattern

34DNP+345TMP 345TCP+345TMP pH=7.4 pH=7.4 PCP Dinoterb pH=6.4 + + ) 6 –1 pH=8.0

(s 5 pH=7.4 345TMP obs pH=8.5 345TMP k 4 pH=6.1 26DCP 26DCP 3

0 Figure 4.4: Effect of pH and substitution pattern on the observed interaction in the uncoupling activity . The bars represent calculated uncoupling activities according to equation 4.4 for mixtures with each of the components present at its effect concentration EC(kee=1.4) at the indicated pH. The thin line depicts the effect of the single compounds. The thick line gives the sum of the effects of the single compounds, which corresponds to a mixture with no interaction (d=0). Values for ai, bi, d and EC(kee=1.4) are given in Tables 4.2 and 4.3.

– 55 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

with 35DBC or 345TCP yielded high d values. The combination of two ortho- substituted compounds, PCP with 26DCP (Figure 4.4-B) showed minor interaction and the combination of Dinoterb and 26DCP (Figure 4.4-B) exhibited exact addition of effects (d = 0). Hence, in analogy to the dimer, ortho substituents on both partners hin- der the formation of a mixed dimer, whereas in combinations of partners with and without ortho substituents a mixed dimer can be formed. The good agreement of the entire set of experimental data with the mixed dimer model confirms indirectly the perception of a dimer of phenol and phenoxide as additional species in the protonophoric uncoupling mechanism of single compounds. The role of the ortho substitution is an additional indication that the two species in the dimer are interconnected by a hydrogen bond between the phenoxide oxygen and the phenol hydroxide. For pentachlorophenol, formation of a dimer was confirmed spec- troscopically in aprotic solvents and the observed spectra indicated an H-bonded structure (Barstad et al. 1993).

4.4.4 Comparison with concentration addition Plackett and Hewlett (1952; 1967) have introduced a set of four categories to distinguish joint action of chemicals. According to their terminology the proposed formation of a dimer of similar acting compounds is termed complex similar action. Predic- tion of the effects of such a mixture is difficult because the interaction of the compounds may result in an overall effect ranging from antagonism to synergism depending on the type and degree of interaction. Complex similar action has to our knowledge not yet been proposed for environmental chemicals but is only known for pharmaceutical combinations. For mixtures of similar acting compounds without interaction, termed “simple similar” by Plackett and Hewlett (1952; 1967), concentration addition is used as a reference concept (Loewe 1953; Sprague 1970). On a mechanistic level this concept can be understood as the description of two compounds with different binding affinities for the same receptor (Bliss 1939). The response of the biological system is dependent on the fraction of receptors with toxicant bound independently whether it results from a low concentration of a compound with high binding affinity or vice versa. Accord- ing to this concept, any effect in a mixture can be obtained by replacing one substance by the equi-effective amount of the other. The concept of concentration addition was proposed as a general reference concept irrespective of the underlying modes of toxic action (Sprague 1970; Berenbaum 1985). Concentration addition was found to describe the effects of mixtures of baseline toxicants (Könemann 1981; Hermens et al. 1985; Van Wezel et al. 1996) and was also

– 56 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

applied as a reference concept for mixtures of uncouplers (see discussion below). Whether or not concentration addition is an appropriate concept for mixtures of compounds with different mode and different site of action is still under debate (Pöch 1991; Altenburger et al. 1996). As the investigated phenols all act according to the same mechanism, concentration addition is used as reference concept in the following, although its mechanistic interpretation of receptor binding is not applicable to the shuttle mechanism. The term “synergism” is used for mixtures with higher observed activity than predicted from concentration addition. To test the consistency of the experimental data with the concept of concentration addition a certain endpoint effect needs to be chosen and the concentration needed to induce the chosen endpoint, i.e. the effect concentration, EC, needs to be determined. There is no a priori endpoint for the uncoupling activity. In the following two end- –1 point effects, kee, are used. The first, kee = 1.4 s , was shown to correlate well with several biochemical tests and with EC50 values of algae growth inhibition, the Microtox test, and furthermore, covers similar concentration ranges in sensitivity (Escher et al. 1997). However, the experimental mixture data needed to be extrapolated to lower –1 –1 concentrations for kee = 1.4 s . In contrast, the second endpoint, kee = 6 s , represents the intersection of the minimal activity of the most active mixtures and the highest activity –1 of the single compounds. Hence, for most mixtures, data for kee = 6 s were inter- polated from the experimental data. Therefore, this endpoint was exceptionally selected for this mixture study. Effect concentrations were backcalculated for the two endpoints by using equation 4.3 which yielded membrane concentrations Eclip,i . These were converted into the corresponding free aqueous concentrations, ECw,i, by use of the liposome-water distribution ratio (equation 4.5). To test consistency with concentration addition, toxic units (TU) were calculated. A toxic unit denotes the concentration of a compound present in a mixture as a fraction of its effect concentration (Brown 1968; Sprague 1970):

clip,i cw,i ctot,i TU === (4.11) i EC EC EC lip,i w,i tot,i Toxic units are not dependent on the unit of the concentration, i.e. TU are equal for clip,i, cw,i and ctot,i. According to the concept of concentration addition a mixture will give the same endpoint effect if the sum of the TU of the components present in the mixture (equation 4.11) equals one:

– 57 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

c1 c2 +=1ŸkECkkobs,i()i =ee ﬁ obs,mix(, cck1 2 )= ee (4.12) EC 1 EC2 Toxic units, TUs, were calculated for hypothetical binary combinations given the components were present in the ratio of the effect concentrations

(cw,1:cw,2 = ECw,1:ECw,2). Concentration values were backcalculated using equation 4.3 and 4.4 and the values for a, b and d given in Table 4.2 and 4.3. The ÂTU ranged from 0.5 to 1.2 (Table 4.3). ÂTU = 0.5 is a significant synergistic effect with respect to concentration addition. ÂTU = 1.2 shows slight underestimation by concentration addition. ÂTU ≥ 1 was observed for the combinations of the second-order uncouplers that have no substituents in ortho position and for which speciation was not clearly opposite. Concentration addition was also found for combinations of two diortho substituents that exhibit small or zero d values, i.e. that are not forming strong mixed dimers. It should be emphasized that not all binary combinations with d > 0 are synergistic with respect to concentration addition

Synergism (ÂTU < 1) was observed when speciation and/or steric effects were favorable for interaction between the two mixture partners. The combination of 34DNP and 345TCP showed a significant synergistic effect. In this case, the main reason for the synergism appears to be the speciation (cf. Figure 4.2 and discussion above) while the mixed dimer formation is sterically equivalent with the dimer formation of the single compounds. The argument of speciation holds also true for the combination of 24DNP with 345TCP or with 35DBC. The synergistic effect is even more pronounced for Bromox with 35DBC or 345TCP (ÂTU = 0.5), for which both, speciation and the steric effect favor the formation of a mixed dimer.

4.4.5 Comparison with data from literature Comparison of observed effects of mixtures of uncouplers with the concept of concentration addition on a higher biological level were performed by Könemann (1981)

and by Altenburger et al. (2000). In the study of Könemann (1981) the LC50 value (7 and 14 days) of mixtures of 10 moderate uncouplers (phenol and various chloro-

phenols) were investigated with guppies (Poecilia reticula) at pH = 6.1. The LC50 of the mixture was in accordance with concentration addition (ÂTU = 1.0). Not all compounds in their test set were investigated here. Yet, according to the criteria derived

above, for the pK a values and the substitution pattern of the phenols used in Könemann’s study, concentration addition would also be expected in the test system used here. In a study of Altenburger et al. (2000) 16 substituted phenols and phenylhydra- zones, most of them potent uncouplers, were investigated in the bioluminescence test

– 58 – Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

with Vibrio fischeri (30 min.). Complete dose response curves (inhibition from 0– 100 %) were examined for the individual compounds and for two mixtures with different component ratios. For both component ratios, the experimental findings were in agreement with the prediction according to the concept of concentration addition up to an effect level of approximately 50 % inhibition, but for higher effect levels, the observed effects were slightly underestimated by the concept of concentration addition (further details are given in Grimme et al. (1998)). In Figure 4.5, calculated concentration-effect curves are plotted for the binary mixtures of 345TCP with 34DNP and 34DNP with 345TMP with a mixture ratio equivalent to the ratio of their ECw,i(kee = 6) values. Predictions for concentration addition were calculated according to equation 4.12. For a better comparison with common literature data, the concentration scale was converted to aqueous concentrations. As is discussed in detail above, the effect of the mixture of 34DNP with 345TMP is in agreement with the prediction for concentration addition, and 345TCP with 34DNP shows a considerable synergistic effect. This observation is an indication that the slight synergism detected in the study of Altenburger et al. (2000) is not an artifact but has a mechanistic basis. It is interesting to note that the degree of synergism increases with increasing effect level both in the present study and in the toxicity tests with Vibrio fischeri. Final conclusions regarding this comparison can only be drawn once multiple mixtures are investigated with the in vitro test system presented here. In addition further work should be directed to binary mixtures with clear synergistic effects to resolve if the results of the in vitro test system can be extrapolated to higher organisms.

– 59 – Interaction of phenolic uncouplers in bina1y mixtures: Concentration-additive and synergistic effects

10 I {/) -..cIll 9 ~o • • 8 • 7 • • 6 CD ®®@ • 5 • 4

0 10E-9 100E-9 1E-6 10E-6 100E-6 1 E-3 10E-3

Figure 4.5 Calculated concentration effect curves of three single phenolic uncouplers and their binary mixtures at pH= 7 .4 in a comparison with calculated activities predicted by concentration addition. Activities of the single compounds (calculated using equation 4.3 and the values for ai and bi from Table 4.2): -

• ·®: 345TCP with 34DNP, with Cw,1:cw,2 = ECw,1: ECw,2 = 1:200

• ·®: 34DNP with 345TMP, with Cw,1:cw,2 = ECw,1: ECw,2 = 1:0.062 Note that each point represents the prediction according to concentration addition at the respective kobs as endpoint and that the Effects are drawn versus the sum of the free aqueous concentration of the two compounds.

- 60 - Interaction of phenolic uncouplers in binary mixtures: Concentration-additive and synergistic effects

4.4.6 Significance for environmental risk assessment of mixtures To date, environmental risk assessment is generally focused on single compounds. In the regulatory process of the EU, compounds will be excluded from the following steps of the risk assessment process (European Commission 1996) if the predicted environmental concentration (PEC) of a compound of interest is below the predicted no effect concentration (PNEC). This approach implies that the toxicants act independently. However, the present work and many others show that it is a misleading simplification if the compounds in a mixture are processed one after the other. The present work also shows that concentration addition is a very likely but not a worst- case scenario for compounds with similar mode of action. In the future, risk assessment processes for mixtures need to be established. Most efforts in this direction are based on the concept of concentration addition, e.g. for non-specifically acting compounds (Könemann 1981; Hermens et al. 1985; Van Wezel et al. 1996). Further examples include the hydrocarbon block method for oil components (European Commis- sion 1996) or the concept of toxic equivalent concentrations for dioxin-like compounds (Safe 1998). There is agreement in human health risk assessment that in the long term the overall effect of single compounds and of mixtures should be modeled in two steps (El-Masri et al. 1997; Bucher et al. 1998; Haddad et al. 1998). In a first step absorption, distribution, metabolism, and excretion of a compound should be modeled to predict the resulting concentration of the compound at its target site (toxicokinetics). The effect of the compound present at its target site (toxicodynamics) should then be modeled in a second step. This approach, known as pharmacokinetic / pharmaco- dynamic or toxicokinetic/toxicodynamic modeling, has been applied successfully in several mixture studies (Krihnan et al. 1991; McCarty et al. 1992). The present results can be a basis for a comparable approach for uncoupling compounds and for applications in environmental toxicology. Thanks to the simplicity of the chosen set-up, the interaction of the compounds can be observed without blearing effects of different accumulation kinetics and metabolism, and other possible mechanisms of toxic action of the compounds can be excluded.

– 61 –

General discussion and conclusions General discussion and conclusions

This work contributes to the mechanism-based effect assessment of environmental pollutants that interfere with energy conversion in membranes, a vital process in all cells. The development of this mechanistic approach was furthered both in breadth and in depth – it was expanded from simple organic acids, substituted phenols, to complexing acidic compounds, here triorganotin compounds, and it was extended from effect assessment of single compounds to binary mixtures. To differentiate the contribution of the partitioning and the intrinsic effects, the observed effects were related to the effective concentration of the compounds at the target site, i.e. the energy- transducing membrane. Membrane concentration were deduced from liposome-water and from biomembrane-water distribution ratios. The distribution experiments with tributyltin and triphenyltin confirmed the crucial role of the charged species in the partitioning behavior of HIOCs between biomembranes and water. The anisotropic phospholipid vesicles (liposomes) were found to be a superior surrogate for the biomembrane than the well-known bulk solvent octanol. Yet, the distribution experiments with tributyltin in biomembrane-water systems also showed that when specific interaction, e.g. complex formation with biological ligands, have a higher impact on the overall partitioning as compared to hydrophobicity, the applicability of an uniform surrogate is generally limited. The mechanism of action which was relevant for the overall adverse effect of tributyltin and triphenyltin on energy conversion was investigated by time-resolved spectroscopy. A distinct difference in the effect pattern exerted by the structurally similar compounds was found. Tributyltin is able to exert an OH– uniport similar to the protonophoric shuttle whereas triphenyltin showed no OH– uniport activity. Triphenyltin exerts its adverse effect on energy conversion mainly by inhibiting the ATP synthase, as was concluded by indirect evidence. Both mechanisms result in the inhibition of the ATP synthesis, however they necessitate different chemical properties. Considering the discrepancy in the partitioning behavior between liposomes and water as well as octanol and water of the investigated triorganotin compounds and considering further their distinct effect pattern, it seems evident that attempts in the

literature have failed to correlate effects of triorganotin with Kow or other physicochemical or structural descriptors used for Quantitative Structure Activity Relation- ships (QSARs) (Vighi et al. 1985; Nagase et al. 1991; Argese et al. 1998). It is also interesting to compare the activity of tributyltin to exert an OH– uniport with the protonophoric activity of substituted phenols. 2,3,4,5-tetrachlorophenol has a

comparable pKa and also a comparable chromatophore-water distribution ratio. The

– 64 – General discussion and conclusions

uncoupling activity in terms of the effect concentration in the chromatophores

EC(kee = 1.4), are in the same order of magnitude for the two compounds. Thus, the activity of the tributyltin cation to induce charge translocation is comparable with the activity of the phenoxide anion. This confirms a hypothesis from the literature that hydrophobic cations are also potential uncouplers. However, equally hydrophobic and acidic arylamines are much less effective uncouplers (Brandt et al. 1992). The increase in the uncoupling activity of binary mixtures of substituted phenols gave further evidence for the formation of a dimer of phenol and phenoxide in the protonophoric shuttle mechanism of uncoupling. The kinetic model of the uncoupling mechanism confirmed that monomeric and dimeric shuttles can act concomitantly. As a consequence of the formation of a mixed dimer, some binary mixtures of substituted phenols with similar mode of action (i.e. uncoupling) were found to exert a higher activity than calculated by concentration addition, which is a reference concept generally applied for similarly acting compounds. This is relevant also from a con- ceptual point of view: Generally, synergistic effects are assumed to occur mainly with integrating toxic endpoints, i.e. reduced growth, reproduction or increased lethality, because synergism is thought to result from the concomitant action of the mixture components on dependent systems of an organism. The synergistic effect found in this work is in contrast to this view. Synergism occurred on the level of primary effects and was exerted by compounds with identical mode of action, due to physicochemical interaction of the components. The findings of the mixture experiments give rise to three questions: (i) What can be expected for multiple mixtures of phenols? According to a simple view, one could expect that an increasing number of dimers would be formed in a multiple mixture and thus, that even stronger synergistic effect would occur. Yet, only a few combinations among those multiple mixtures can have opposite speciation or a complementing substitution pattern. The synergistic effect will therefore probably be most evident in binary mixtures of ‘optimal’ partners. (ii) Are the effects observed in vitro also of relevance in vivo? There are two prerequisites, that the effects observed in vitro in principle can also occur in vivo: The compounds must not have any other relevant specific mechanism of action, which would supersede the uncoupling effect. Further- more, the two compounds need to be present at the same time in relevant concentrations at the target site, which is dependent on the uptake, the transformation and the excretion kinetics of the compounds. This is most probable given in

– 65 – General discussion and conclusions

single celled organisms. The observation of slightly synergistic effects of multiple mixtures of uncouplers determined in the bacteria Vibrio fischeri might indicate a certain relevance of the in vitro findings for the mixture toxicity in vivo (Altenburger et al. 2000). Promising consistency was also found in preliminary experiments with Daphnia magna (Kühnholz 2000). (iii) Are there further examples of interaction of compounds that may lead to an increased activity when they are present in mixtures? The author is not aware of further examples of possible interactions of chemicals with similar mode of action. An interesting interaction was found for dis- similarly acting compounds. In combinations of copper with catechol, increased membrane toxicity was observed, which was attributed to the formation of copper-catechol complexes (Schweigert et al. 2001). Furthermore, when the step of translocation through the membrane, which is part of the protonophoric shuttle, is regarded as the model for the uptake of a contaminant, similar interactions of chemicals with dissimilar mode of action can most probably occur, for example the uptake of heavy metals in charged chelates with organic contaminants. These considerations make evident that the processes of uptake, transformation and excretion of a chemical need to be considered to complement the knowledge on the effects of toxicants at the target site. Attempts to model the time course of uptake, distribution, transformation and excretion are made in toxicokinetic models. By using such models, the concentration in the aqueous phase of an organism can be assessed. The combination of toxicokinetic models with the toxicodynamic data determined in this thesis should enable a realistic prediction of effects in vivo.

– 66 – References

Abrahams, J. P., Leslie, A. G. W., Lutter, R. and Walker, J. E. (1994): Structure at 2.8 Å resolution of F1-ATPase from bovine heart mitochondria. Nature, 370, 621–628. Ahlers, J., Cascorbi, I., Gies, A., Köhler, M., Pauli, W. and Rösick, E. (1991): Interaction with functional membrane proteins - a common mechanism of toxicity for lipophilic environmental chemicals? Comp. Biochem. Physiol., 100c, 111–113. Aldrige, W. N. (1976): The influence of organotin compounds on mitochondrial functions. In: «Organotin compounds: New chemistry and applications», Ed.: J. J. Zuckerman. Washington, American Chemical Society. pp. 204–226. Aldrige, W. N. and Cremer, J. E. (1955): The biochemistry of organo-tin compounds – diethyltin dichloride and triethyltin sulphate. Biochem. J., 61, 406–418. Altenburger, R., Backhaus, T., Bödeker, W., Faust, M., Scholze, M. and Grimme, L. H. (2000): Predictability of the toxicity of multiple mixture to Vibrio fischeri: Mixtures composed of similarly acting chemicals. Environ. Toxicol. Chem., 19, 2341–2347. Altenburger, R., Bödeker, W., Faust, M. and Grimme, L. H. (1996): Regulations for combined effects of pollutants: Consequences from risk assessment in aquatic toxicology. Food Chem. Toxicol., 34, 1155–1157. Ambrosini, A., Bertoli, E. and Zolese, G. (1996): Effect of organotin compounds on membrane lipids: Fluorescence spectroscopy studies. Appl. Organomet. Chem., 10, 53–59. Argese, E., Bettiol, C., Ghirardini, A. V., Fasolo, M., Giurin, G. and Ghetti, P. F. (1998): Comparison of in vitro submitochondrial particle and Microtox assays for determining the toxicity of organotin compounds. Environ. Toxicol. Chem., 17, 1005–1012. Arnold, C. G., Berg, M., Müller, S. R., Dommann, U. and Schwarzenbach, R. P. (1998a): Determination of organotin compounds in water, sediments, and sewage sludge using perdeuterated internal standards, accelerated solvent extraction, and large-volume- injection GC/MS. Anal. Chem., 70, 3094–3101. Arnold, C. G., Ciani, A., Müller, S. R., Amirbahman, A. and Schwarzenbach, R. P. (1998b): Association of triorganotin compounds with dissolved humic acids. Environ. Sci. Technol., 32, 2976–2983. Arnold, C. G., Weidenhaupt, A., David, M. M., Müller, S. R., Haderlein, S. B. and Schwarzenbach, R. P. (1997): Aqueous speciation and 1-octanol-water partitioning of tributyl- and triphenyltin: Effect of pH and ion composition. Environ. Sci. Technol., 31, 2596–2602. Avdeef, A. (1992): pH-metric log P. I: Difference plots of determining ion-pair octanol-water partition coefficients of multiprotic substances. Quant. Struct. Act. Relat., 11, 510–517. Avdeef, A. (1993): pH-metric log P. II: Refinement of partition coefficients and ionization constants of multiprotic substances. J. Pharm. Sci., 82, 183–190. References

Barbieri, R., Ruisi, G., Silvestri, A., Giuliani, A. M., Barbieri, A., Pina, G., Pieralli, F. and Del Giallo, F. (1995): Dynamics of nuclei in alkyltin (IV)-deoxyribonucleic acid condensates by variable-temperature tin-119 Mössbauer spectroscopy. J. Chem. Soc. Dalton Trans., 467–475. Barbieri, R., Silvestri, A., Giuliani, A. M., Piro, V., Di Simone, F. and Grazia, M. (1992): Organotin compounds and deoxyribonucleic acids. J. Chem. Soc. Dalton Trans., 585–590. Barstad, A. W., Peyton, D. H. and Smejtek, P. (1993): AHA- heterodimer of a class-2 uncoupler: pentachlorophenol. Biochim. Biophys. Acta, 1140, 262–270. Benz, R. and McLaughlin, S. G. A. (1983): The Molecular Mechanism of Action of the Proton Ionophor FCCP (Carbonylcyanide p-Trifluoromethoxyphenylhydrazone). Biophys. J., 381– 398. Berenbaum, M. C. (1985): The expected effect of a combination of agents: the general solution. J. Theoretical Biol., 114, 413–431. Berg, M., Arnold, C. G., Müller, S. R., Mühlemann, J. and Schwarzenbach, R. P. (2001): Sorption and fate of organotin compounds in sediment-pore water systems: Laboratory experiments and field data. Environ. Sci. Technol., submitted. Betageri, G. V. and Rogers, J. A. (1988): The liposome as a distribution model in QSAR studies. Int. J. Pharm., 46, 95–102. Bliss, C. I. (1939): The toxicity of poisons applied jointly. Ann. Appl. Biol., 26, 585–615. Blunden, S. J. and Evans, C. J. (1990): Organotin compounds. In: «Antropogenic compounds», Ed.: O. Hutzinger. Berlin, Springer. 3 part E, pp. 1–44. Bödeker, W., Drescher, K., Altenburger, R. and Faust, M. (1993): Combined effects of toxicants: the need and soundness of assessment approaches in ecotoxicology. Sci. Total Environ., suppl., 931–939. Borghese, R., Crimi, M., Fava, L. and Melandri, B. A. (1998a): The ATP synthase atpHAGDC (F-1) operon from Rhodobacter capsulatus. Journal Of Bacteriology, 180, 416–421. Borghese, R., Turina, P., Lambertini, L. and Melandri, B. A. (1998b): The atpIBEXF operon coding for the F-0 sector of the ATP synthase from the purple nonsulfur photosynthetic bacterium Rhodobacter capsulatus. Archives Of Microbiology, 170, 385–388. Bowie, L. J. (1978): Synthesis of firefly luciferin and structural analogs. In: «Bioluminescence and chemiluminescence part A», Ed.: M. A. De Luca. New York, Academic Press. 57, pp. 15–28. Bowyer, J. R., Tierney, G. V. and Crofts, A. R. (1979): Secondary electron transfer in chromatophores of Rhodopseudomonas capsulata A1a pho. FEBS Lett., 101, 201–206. Boyer, P. D. (1993): The binding change mechanism for ATP synthase – some probabilities and possibilities. Biochim. Biophys. Acta, 1140, 215–250. Bragadin, M. and Marton, D. (1997): A proposal for a new mechanism of interaction of trialkyltin (TAT) compounds with mitochondria. J. Inorg. Biochem., 68, 75–78. Bragadin, M., Marton, D., Toninello, A. and Viola, E. R. (2000): Tributyltin and mitochondria: new evidence in favour of an uncoupling mechanism. Inorg. Chem. Comm., 3, 255–258.

– 68 – References

Brandt, U., Schubert, J., Geck, P. and von Jagow, G. (1992): Uncoupling activity and physicochemical properties of derivatives of fluazinam. Biochim. Biophys. Acta, 1101, 41– 47. Brown, V. M. (1968): The calculation of the acute toxicity of mixture of poisons to rainbow trout. Water Res., 2, 723–733. Bucher, J. R. and Lucier, G. (1998): Current approaches toward chemical mixture studies at the national institute of environmental health sciences and the U.S. national toxicology program. Environ. Health Perspect., 106, 1295–1298. Byington, K. H. (1971): Effects of triphenyltin compounds on the adenosine triphosphatase activity of beef heart submitochondrial particles. Biochem. Biophys. Res. Commun., 42, 16– 22. Champ, M. A. (2000): A review of organotin regulatory strategies, pending actions, related costs and benefits. Sci. Total Environ., 258, 21–71. Cohen, F. S., Eisenberg, M. and McLaughlin, S. G. A. (1977): The kinetic mechanism of action of an uncoupler of oxidative phosphorylation. J. Membrane Biol., 37, 361–396. Crofts, A. R. (1985): The mechanism of the ubiquinol:cytochrome c oxidoreductase of mitochondria and of Rhodopseudomonas sphaeroides. In: «The enzymes of biological membranes», Ed.: Martonosi. New York, N.Y., Plenum Press. 4, pp. 347–383. Crofts, A. R. and Wraight, C. A. (1983): The electrochemical domain of photosynthesis. Biochim. Biophys. Acta, 726, 149–185. Daughton, C. and Ternes, T. (1999): Pharmaceuticals and personal care products in the environment: agents of subtle change? Environ. Health Perspect., 107, Supp. 6, 907–937. Dawson, A. P. and Selwyn, M. J. (1974): The action of trialkyltin compounds on mitochondrial respiration. Biochem. J., 138, 349–357. Dutton, P. L., Petty, K. M., Bonner, H. S. and Morse, S. D. (1975): Cytochrome c2 and reaction center of Rhodopseudomonas sphaeroides Ga. membranes. Extinction coefficient, half reduction potentials, kinetics and electric field alterations. Biochim. Biophys. Acta, 387, 536– 556. El-Masri, H. A., Reardon, K. F. and Yang, R. S. H. (1997): Integrated approaches for the analysis of toxicologic interactions of chemical mixtures. Crit. Rev. Toxicol., 27, 175–197. Escher, B. I. (1995): The relationship between membrane-water partitioning, uncoupling, and inhibitory activity of substituted phenols in chromatophores of Rhodobacter sphaeroides. PhD thesis, Zürich, Swiss Federal Institute of Technology. Escher, B. I., Hunziker, R. W., Schwarzenbach, R. P. and Westall, J. C. (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. and Schwarzenbach, R. P. (1996a): Partitioning of substituted phenols in liposome-water, biomembrane-water, and octanol-water systems. Environ. Sci. Technol., 30, 260–270.

– 69 – References

Escher, B. I., Schwarzenbach, R. P. and Westall, J. C. (2000a): Evaluation of liposome-water partitioning of organic acids and bases. 1. Development of a sorption model. Environ. Sci. Technol., 34, 3954–3961. Escher, B. I., Schwarzenbach, R. P. and Westall, J. C. (2000b): Evaluation of liposome-water partitioning of organic acids and bases. 2. Comparison of experimental determination methods. Environ. Sci. Technol., 34, 3962–3968. 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. Toxicol. Chem., 16, 405–414. Escher, B. I., Snozzi, M. and Schwarzenbach, R. P. (1996b): Uptake, speciation, and toxic effect of substituted phenols in energy transducing membranes. Environ. Sci. Technol., 30, 3071– 3079. EU (1998): Directive 98/8/EC of the european parliament and the council of 16 February 1998 concerning the placing of biocidals products on the market. Official Journal of the European Communities, 41, 123. European Commission (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. In: Ed.: Luxembourg, Office for Official Publications of the European Communities. Falk, G. and Walker, J. E. (1988): DNA sequence of a gene cluster coding for subunits of the F0 membrane sector of ATP synthase in Rhodospirillum rubrum. Biochem. J., 254, 109–122. Farrow, B. and Dawson, A. P. (1978): Investigation of the interaction of triethyltin with rat liver mitochondria using binding studies and Mössbauer spectroscopy. Eur. J. Biochem., 86, 85–95. Fent, K. (1996): Ecotoxicology of organotin compounds. Crit. Rev. Toxicol., 26, 1–117. + Fillingame, R. H. (1990): Molecular mechanics of ATP synthesis by F1F0-type H -transporting ATP synthases. In: «Bacterial energetics», Ed.: T. A. Krulwich. San Diego, Academic Press. 12, pp. 345–391. Fillingame, R. H. and Divall, S. (1999): Proton ATPases in bacteria: comparison to Escherichia coli F1F0 as the prototype. Novartis found. symp., 221, 218–234. Finkelstein, A. (1970): Weak acid uncouplers of oxidative phosphorylation. Mechanism of action on thin lipid membranes. Biochim. Biophys. Acta, 205, 1–6. Flewelling, R. F. and Hubbell, W. L. (1986): Hydrophobic ion interactions with membranes – thermodynamic analysis of tetraphenylphosphonium binding to vesicles. Biopys. J., 49, 531–540. Foster, D. L. and Fillingame, R. H. (1982): Stoichiometry of subunits in the H+-ATPase complex of Escherichia coli. J. Biol. Chem., 257, 2009–2015.

– 70 – References

Furbacher, P. N., Tae, G.-S. and Cramer, W. A. (1996): Evolution and origines of the cytochrome bc1 complex and b6f complex. In: «origin and evolution of biological energy conversion», Ed.: H. Baltscheffsky. New York, VCH. pp. 221–253. Gomez-Puyou, A. and Gomez-Lojero, C. (1977): The use of ionophores and channel formers in the study of the function of biological membranes. In: «Current topics in bioenergetics», Ed.: D. R. Sanadi. New York, Academic Press. 6, pp. 221–257. Gould, J. M. (1976): Inhibition by triphenyltin chloride of a tightly-bound membrane component involved in photophosphorylation. Eur. J. Biochem., 62, 567–75. Grigoriev, E. V., Pellerito, L., Yashina, N. S., Pellerito, C. and Petrosyan, V. S. (2000): Organotin(IV) chloride complexes with phosphocholine and dimyristoyl-L-a-phosphatidylcholine. Appl. Organomet. Chem., 14, 443–448. Grimme, L. H., Altenburg, R., Backhaus, T., Bödeker, W., Faust, M. and Scholze, M. (1998): Vorhersagbarkeit und Beurteilung der aquatischen Toxizität von Stoffgemischen. Leipzig- Halle, Umweltforschungszentrum, Report 25. Haddad, S. and Krishnan, K. (1998): Physiological modelling of toxicokinetic interactions: Implications for mixture risk assessment. Environ. Health Perspect., 106, 1377–1384. Halling-Sørensen, B., Nielsen, S. N., Lanzky, P. F., Ingerslev, F., Lützhøft, H. C. H. and Jørgensen, S. E. (1998): Occurrence, fate and effects of pharmaceutical substances in the environment – a review. Chemosphere, 36, 357–393. Harino, H., Fukushima, M. and Kawai, S. (1999): Temporal trends of organotin compounds in the aquatic environment of the Port of Osaka, Japan. Environ. Pollut., 105, 1–7. Hermann, R. G., Steppuhn, J., Hermann, G. S. and Nelson, N. (1993): The nuclear-encoded polypeptide Cfo-II is from spinach is a real, ninth subunit of chloroplast ATP synthase. FEBS Letters, 326, 192–198. Hermens, J., Könemann, H., Leeuwangh, P. and Musch, A. (1985): Quantitative structure- activity relationships in aquatic toxicity studies of chemicals and complex mixtures of chemicals. Environ. Toxicol. Chem., 4, 273–279. Heywood, B. R., Molloy, K. C. and Waterfiled, P. C. (1989): Organotin biocides XV: Modelling the interactions of triorganotins with the cell membrane. Appl. Organomet. Chem., 3, 443– 450. Holmes, N. G., Hunter, C. N., Niedermann, R. and Crofts, A. R. (1980): Identification of the pigment pool responsible for the flash induced carotenoid band shift in Rhodopseudomonas sphaeroides chromatophores. FEBS Lett., 115, 43–48. Horváth, G., Droppa, M., Fodorpataki, L., Istókovics, A., Garab, G. and Oettmeier, W. (1996): Acridones: A chemically new group of protonophores. Proc. Nat. Acad. Sci. USA, 93, 3876– 3880. Hynes, M. and O'Dowd, M. (1985): Metal toxicity: complexing of alkyltin species with low-Mr ligands in aqueous solution. Biochem. Soc. Trans., 13, 490–491.

– 71 – References

Hynes, M. J. and O'Dowd, M. (1987): Interactions of trimethyltin(IV) cation with carboxylic acids, amino acids and related ligands. J. Chem. Soc. Dalton Trans., 3, 563–566. Jackson, J. B. and Crofts, A. R. (1971): The kinetics of light induced carotenoid changes in Rhodopseudomonas sphaeroides and their relation to electrical field generation across the chromatophore membrane. Eur. J. Biochem., 18, 120–130. Junge, W. and Jackson, J. B. (1982): The development of electrochemical potential gradients across photosynthetic membranes. In: «Photosynthesis: Energy Conversion by Plants and Bacteria», Ed.: Govindjee. New York, NY, Academic Press. 1, pp. 589–646. Kahn, J. S. (1968): Chlorotributyltin – inhibitor of photophosphorylation in isolated chloroplasts. Biochim. Biophys. Acta, 153, 203–210. Kahn, J. S. (1970): Absence of a common intermediate pool among individual enzyme chains of the energy-conservation pathway in chloroplasts of Euglena gracilis. Biochem. J., 116, 55– 60.

Klughammer, C., Heimann, S. and Schreiber, U. (1998): Inhibition of cytochrome b563- oxidation by triorganotins in spinach chloroplasts. Photosynth. Res., 56, 117–130. Ko, H. C., O'Hara, W. F., Hu, T. and Hepler, L. G. (1963): Ionization of substituted phenols in aqueous solution. J. Am. Chem. Soc., 86, 1003–1004. Könemann, H. (1981): Fish toxicity tests with mixtures of more than two chemicals: a proposal for a quantitative approach and experimental results. Toxicology, 19, 229. Krihnan, K. and Andersen, M. E. (1991): Toxicological consequences of combined exposure to environmental pollutants. Arch. Complex Environ. Studies, 3, 1–106. Kühnholz, O. (2000): Binäre Mischungen von Phenolen mit und ohne ortho-Substitution zeigen als Entkoppler synergistischee Effekte in-vitro und in-vivo. Diploma thesis, Zürich, Swiss Federal Institute of Technology (ETH). Lea, E. J. and Croghan, P. C. (1969): The effect of 2,4-dinitrophenol on the properties of thin phospholipid films. J. Membrane Biol., 1, 225–237. Lienhard, G. E. and Secemski, I. I. (1973): P1,P5-Di(adenosine-5')pentaphosphate, a potent multisubstrate inhibitor of adenylate kinase. J. Biol. Chem., 248, 1121–1123. Lipnick, R. L., Bickings, C. K., Johnson, D. E. and Eastmond, D. A. (1985): Comparison of QSAR predictions with fish toxicity screening data for 110 phenols. Aquat. Tox. Hazard Assess., 8th Symp., 153–176. Loewe, S. (1953): The problem of synergism and antagonism of combined drugs. Arzneimittel- Forsch., 3, 285–290. Looser, P., Bertschi, S. and Fent, K. (1998): Bioconcentration and bioavailability of organotin compounds: influence of pH and humic substances. Appl. Organomet. Chem., 12, 601–611. Looser, P. W., Fent, K., Berg, M., Goudsmit, G.-H. and Schwarzenbach, R. P. (2001): Uptake and Elimination of triorganotin compounds by larval midge of Chironomus riparius in the absence and presence of Aldrich humic acids (AHA). Environ. Sci. Technol., submitted.

– 72 – References

Matsuura, K., Masamoto, K., S, I. and Nishimura, M. (1979): Effect of the surface potential on the intermembrane electrical field measured with carotenoid spectral shift in chromatophores from Rhodopseudomonas sphaeroides. Biochim. Biophys. Acta, 547, 91–102. McCarty, L. S., Ozbur, G. W., Smith, A. D. and Sixon, D. G. (1992): Toxicokinetic modelling of mixtures of organic chemicals. Environ. Toxicol. Chem., 11, 1037–1047. McLaughlin, S. G. A. (1972): The Mechanism of action of DNP on phospholipid bilayer membranes. J. Membrane Biol., 361–372. McLaughlin, S. G. A. and Dilger, J. P. (1980): Transport of protons across membranes by weak acids. Physiol. Rev., 60, 825–863. Mitchell, P. (1961): Coupling of phosphorylation to electron and hydrogen transfer by a chemiosmotic type of mechanism. Nature, 191, 144–148. Mitchell, P. (1966): Chemiosmotic coupling in oxidative and photosynthetic phosphorylation. Biol. Rev., 41, 445–502. Miyoshi, H., Maeda, N., Tokutake, N. and Fujita, T. (1987a): Quantitative analysis of partition behaviour of substituted phenols from aqueous phase into liposomes made of lecithin and various lipids. Bull. Chem. Soc. Jpn., 60, 4357–4362. Miyoshi, H., Nishioka, T. and Fujita, T. (1987b): Quantitative relationship between protonophoric and uncoupling activities of substituted phenols. Biochim. Biophys. Acta, 891, 194–204. Miyoshi, H., Tsujishita, H., Tokutake, N. and Fujita, T. (1990): Quantitative analysis of uncoupling activity of substituted phenols with a physicochemical substituent and molecular parameters. Biochim. Biophys. Acta, 1016, 99–106. Müller, M. T., Zehnder, A. J. B. and Escher, B. I. (1999): Membrane toxicity of linear alcohol ethoxylates. Environ. Toxicol. Chem., 18, 2767–2774. Nagase, H., Hamasaki, T., Sato, T., Kito, H., Yoshioka, Y. and Ose, Y. (1991): Structure-activity relationship for organotin compounds. Environ. Toxicol. Chem., 4, 343–351. Noji, H., Yasuda, R., Yosida, M. and Kinosita, K. (1997): Direct observation of the rotation of F1-ATPase. Nature, 386, 299–302. Ottiger, C. and Wunderli-Allenspach, H. (1997): Partitioning behaviour of acids and bases in a phosphatidylcholine liposome-buffer equilibrium dialysis system. Eur. J. Pharm. Sci., 5, 223–231. Papa, S., Guerrieri, F., de-Gomez-Puyou, M. T., Barranco, J. and Gomez-Puyou, A. (1982): Studies on the mechanism of action of triphenyltin on proton conduction by the H+- ATPase of mitochondria. Eur. J. Biochem., 128, 1–7. Plackett, R. L. and Hewlett, P. S. (1952): Quantal responses to mixtures of poisons. S. R. Stat. Soc., 14, 141–163. Plackett, R. L. and Hewlett, P. S. (1967): A comparison of two approaches to the construction of modules for quantal responses to mixtures of drugs. Biometrics, 23, 27–44.

– 73 – References

Pöch, G. (1991): Evaluation of combined effects with respect to independent action. Arch. Complex Environ. Studies, 3, 65–74. Sabbert, D. S., Engelbrecht, S. and Junge, W. (1996): Intersubunit rotation in active F-ATPase. Nature, 381, 623–626. Safe, S. H. (1998): Hazard and risk assessment of chemicals mixtures using the toxic equivilency factor approach. Environ. Health Perspect., 106, 1051–1058. Schellenberg, K., Leuenberger, C. and Schwarzenbach, R. P. (1984): Sorption of chlorinated phenols by natural sediments and aquifers. Environ. Sci. Technol., 18, 652–657. Schwarzenbach, R. P., Stierli, R., Folsom, B. R. and Zeyer, J. (1988): Compound properties relevant for assessing the environmental partitioning of nitrophenols. Environ. Sci. Technol., 22, 83–92. Schweigert, N., Hunziker, R. W., Escher, B. I. and Eggen, R. I. L. (2001): Acute toxicity of (chloro)catechols and (chloro-)catechol-copper combinations in Escherichia coli corresponds to their membrane toxicity in vitro. Environ. Toxicol. Chem., 20, 239–247. Selwyn, M. J. (1976): Triorganotin compounds as ionophores and inhibitors of ion translocating ATPase. In: «Organotin compounds: New chemistry and Applications», Ed.: J. J. Zuckerman. Washington, American Chemical Society. pp. 204–226. Selwyn, M. J. (1992): Ionophores activity and biological effects of triorganotin compounds. In: «Chemistry and technology of silicon and tin», Ed.: V. G. Kumar Das, N. Seik Weng and M. Gielen. Oxford, Oxford university press. pp. 298–311. Selwyn, M. J., Dawson, A. P., Stockdale, M. and Gains, N. (1970): Chloride-hydroxide exchange across mitochondrial, erythrocyte and artificial lipid membranes mediated by trialkyl- and triphenyltin compounds. Biochem., 14, 120–126. Shoukry, M. M. (1993): Equilibrium study of tributyltin (IV) complexes with amino acids and related compounds. Bull. Soc. Chim. Fr., 130, 117–120. Sikkema, J., de Bont, J. A. and Poolman, B. (1995): Mechanisms of membrane toxicity of hydrocarbons. Microbiol. Rev., 59, 201–222. Sistrom, W. R. (1960): A requirement for sodium in the growth of Rhodopseudomonas sphaeroides. J. Gen. Microbiol., 22, 778–785. Skilleter, D. N. (1975): The decrease of mitochondrial substrate uptake caused by trialkyltin and trialkyl-lead compounds in chloride media and its relevance to inhibition of oxidative phosphorylation. Biochem J, 146, 465–71. Smejtek, P. and Wang, S. (1993): Distribution of hydrophobic ionizable xenobiotics between water and lipid membranes: pentachlorophenol and pentachlorophenate. A comparison with octanol-water partition. Arch. Environ. Contam. Toxicol., 25, 394–404. Sprague, J. B. (1970): Measurement of pollutant toxicity to fish. II Utilizing and applying bioassay results. Water Res., 4, 3–32.

– 74 – References

Stäb, J. A., Cofino, W. P., Hattum, B. and Brinkman, U. A. T. (1993): Comparison of GC/MSD and GC/AED for the determination of organotion compounds in the environment. Fresenius. J. Anal. Chem., 347, 247–255. Stanley, P. E., McCarthy, B. J. and Smither, R., Eds. (1989): ATP luminescence – rapid methods in microbiology. The Society for Applied Bacteriology technical series. Oxford, Blackwell Scientific Publications. Stewart, C. T.-J. (1997): Vertical distribution of butyltin residues in sediments of British Columbia harbours. Environ. Tech., 18, 1195–1202. Stockdale, M., Dawson, A. P. and Selwyn, M. J. (1970): Effects of trialkyltin and triphenyltin compounds on mitochondrial respiration. Eur. J. Biochem., 15, 342–51. Terada, H. (1990): Uncouplers of oxidative phosphorylation. Environ. Health Perspect., 87, 213– 218. Tokutake, N., Miyoshi, H. and Fujita, T. (1991): Electron transport inhibition of the cytochrome bc1 complex of rat liver mitochondria by phenolic uncouplers. Biochim. Biophys. Acta, 1057, 377–383. Tolosa, I., Readman, J. W., Blaevoet, A., Ghilini, S., Bartocci, J., Horvat, M., Gomez-Ariza, J. L., Giraldez, I., Sanchez-Rodas, D., Acuna, T. and Morales, E. (1996): Contamination of Mediterranean (Cote d'Azur) coastal waters by organotins and Irgarol 1051 used in antifouling paints. The use of transplanted Venerupis decussata to evaluate the pollution of heavy metals and tributyltin in marinas. Mar. Pollut. Bull., 32, 335–341. Tzagoloff, A., Byington, K. H. and MacLennan, D. H. (1968): Studies on the mitochondrial adenosine triphosphatase system. II. The isolation and characterization of an oligomycin- sensitive adenosine triphosphatase from bovine heart mitochondria. J. Biol. Chem., 243, 2405–12. Vaes, W., Ramos, E. U., Hamwick, C., Holstein;, I. v., Blaauboer, B. J., Seinen, W., Verhaar, H. J. M. and Hermens, J. L. M. (1997): Solid phase microextraction as a tool to determine membrane/water partition coefficients and bioavailable concentrations in in-vitro systems. Chem. Res. Toxicol., 10, 1067–1072. Vaes, W. H. J., Ramos, E. U., Verhaar, H. J. M., Cramer, C. J. and Hermens, J. L. M. (1998): Understanding and estimating membrane/water partition coefficients: Approaches to derive quantitative structure property relationships. Chemical Research In Toxicology, 11, 847–854. Van Wezel, A. P., de Vries, D. A. M., Sjim, D. T. H. M. and Opperhuizen, A. (1996): Use of the lethal body burden in the evaluation of mixture toxicity. Ecotox. Environ. Safety, 35, 236– 241. Van Wezel, A. P. and Opperhuizen, A. (1995): Narcosis due to environmental pollutants in aquatic organisms: residue-based toxicity, mechanisms, and membrane burdens. Crit. Rev. Toxicol., 25, 255–279.

– 75 – References

Venturoli, G., Fernàndez-Velasco, J. G., Crofts, A. R. and Melandri, B. A. (1986): Demonstration of collisional interaction of ubiquinol with the ubiquinol-cytochrome c2 oxidoreductase complex in chromatophores from Rhodobacter sphaeroides. Biochim. Biophys. Acta, 851, 340–352. Verméglio, A. and Joliot, P. (1999): The photosynthetic apparatus of Rhodobacter sphaeroides. Trends in Microbiology, 7, 435–440. Vighi, M. and Calamari, D. (1985): QSARs for organotin compounds on Daphnia magna. Chemosphere, 14, 1925–1932. Worthing, C. R. and Hance, R. J., Eds. (1991): The pesticide manual: a world compendium. Farnham, Surrey, British Crop Protection Council. Wulf, R. G. and Byington, K. H. (1975): On the structure-activity relationship and mechanism of organotin induced, nonenergy dependent swelling of liver mitochondria. Arch. Biochem. Biophys., 197, 176–185. Yagi, T. and Hatefi, Y. (1984): Thiols in oxidative phosphorylation: Inhibition and energy- potentiated uncoupling by monothiol and dithiol modifiers. Biochem., 23, 2449–2455.

– 76 – Appendix A: The energy-transducing membran of R. sphaeroides and the derivation of the uncoupling activity by time-resolved spectroscopy Appendix A

Chromatophores of R. sphaeroides as model for energy- transducing membranes

Chromatophores of R. sphaeroides are a suitable biological system to investigate protonophoric action. Build-up of the chemiosmotic gradient can be induced by light, i.e. without the addition of substrate, and build-up and decay of Dy can be determined photometrically. The following paragraphs will give a short introduction to the energy-transducing membrane of R. sphaeroides and summarize how in principle protonophoric activity can be determined by time resolved-spectroscopy. Details of the experimental conditions are given in the respective chapter and in Escher et al. (1997). Details of the instruments are given in Escher et al. (1997). When R. sphaeroides is cultivated under photoheterotrophic conditions the reaction

center complex and the bc1 complex convert light energy into a cyclic electron transfer which is coupled to the transport of protons through the membrane (primary proton pumps). In chromatophores of R. sphaeroides the oxidation-reduction cascade of the redox-components and the build-up of the membrane potential can be detected by time-resolved spectroscopy. This enables one to differentiate inhibitory effects on the respective components and the protonophoric mechanism (Escher et al. 1997). Figure 1 depicts the three main components of the energy conversion chain of photoheterotrophically cultivated R. sphaeroides: a) Reaction center complex (RC),

b) bc1 complex, and c) the F1F0 ATP synthase. The mechanistic description of the reac-

tion center complex and the bc1 complex, in the following paragraph summarizes reviews given in Crofts et al. 1983; Crofts 1985 and Verméglio et al. 1999. The reaction

center complex and the bc1 complex are the primary proton pumps that maintain a cyclic electron transfer driven by light. Light energy is captured by light harvesting complexes (not shown in Figure 1) and the excited state is transferred to the primary electron donor, P, a bacteriochlorophyll dimer with an absorption maximum of 870 nm. In the excited state P* reduces the primary electron acceptor bacteriopheophytin,

BPheo, in less than 4 picoseconds. From BPheo the electron is transferred via QA to QB.

QA and QB are quinone binding sites. Meanwhile, the oxidized P is rereduced by the

periplasmatic cytochrome c2. In a second electron transfer step the semiquinone at QB is fully reduced, binds two protons from the cytoplasm and is released to the qui-

none/quinole pool in the membrane. Oxidation of the quinone takes place at the QP

site of the bc1 complex. This oxidation step comprises a release of protons to the periplasm, i.e. to the opposite side of the membrane. The two electron transfers of the

– 78 – Appendix A

complete reduction of the quinole at Qp are coupled: One electron is transferred via a

Riske FeS cluster to cytochrome c1 and the other electron is transferred via cytochrome b566 and cytochrome b561 to Qn. At Qn the electron is transferred to a quinone from the pool and after a subsequent quinole oxidation at Qp the intermediate semiquinone at

Qn is completely reduced and is released to the pool. The electron at c1 reduces the periplasmic cytochrome c2 which closes the electron cycle.

The resulting chemiosmotic gradient drives a F1 F0 ATP synthase. The ATP synthase is a multisubunit enzyme. It consists of two moieties. The extrinsic F1 moiety, which contains the catalytic site is connected by a thin stalk to the membrane-embedded F0 moiety, which is responsible for proton translocation. The sequence of R. sphaeroides ATP synthase has not been described yet, but is known for the closely related Rhodo- bacter capsulatus. The F 1 moiety of R. capsulatus is formed by five subunits °'31 ~3, 'If, 8, e (Borghese et al. 1998a) which corresponds to the subunit composition of Eschericha coli

2~ (ox) 2~ 2~ red) (red)

a) reaction b) bc1 complex c) F1Fo ATP center synthase complex

Figure 1 Components of the energy conversion chain in the energy-transducing membrane of R. sphaeroides. Outlined arrows in the reaction center: Subsequent flow of two electrons from the primary electron donor bacteriochlorophyll, P'870 , to the primary acceptor bacteriopheophytin, BPheo, and subsequently to the primary; Q A, and secondary, Os, quinone acceptor. Filled arrows: mass flow e.g. of quinone, Q, and quinoles, QH 2, in the membrane from Os to the quinone pool and from the quinone pool to QP and On, and of cytochrome c2 in the aqueous phase. Outlined arrows in the bc1 complex: subsequent flow of two electrons from Q P to On via cytochrome b566 and cytochrome b561, and from Q P to cytochrome c1 via the FeS (Rieske) cluster.

- 79 - Appendix A

(Foster et al. 1982) and many other bacteria (Fillingame et al. 1999). The F0 moiety is formed by four different subunits: a, b, c and b’ (Borghese et al. 1998b) which corre-

sponds to the composition of the F0 moiety in other photosynthetic membranes (Falk et al. 1988; Hermann et al. 1993). Boyer (1993) has proposed, that a rotational mechanism links proton translocation and ATP synthesis. According to this model, translocation

of protons induces a rotation of the axial, asymmetric g subunit of the F1 headpiece. This rotation induces conformational changes in the surrounding a and b subunits

which change their biding affinity to ADP, Pi and ATP. ADP and Pi bind to the complex in the loose conformation. Rotation of the g subunit induces the „tight“ confor-

mation which enables energy neutral reaction of ADP and Pi. The energy consuming step is the third conformation change to „open“, which induces release of ATP (Boyer 1993). This model has received decisive support from the atomic structure of mito-

chondrial F1 (Abrahams et al. 1994) and is in agreement with functional experiments (Sabbert et al. 1996; Noji et al. 1997).

Determination of build-up and decay kinetics of the membrane potential with time-resolved spectroscopy

The membrane of R. sphaeroides contains high amounts of carotenoids. Some of them have an inducable dipole and show an electrochromic shift, i.e. the wavelength of maximum absorption is shifted when the dipole is induced by an external electric

field. The carotenoid pool of the light harvesting complex B800/850 has a significant red shift of approximately 7 nm which is induced by the membrane potential (Holmes et al. 1980). This electrochromic shift enables one to determine build-up and decay of the membrane potential photometrically. The change in absorption at 503 nm is proportional to the change in Dy (Junge et al. 1982). In membranes of R. sphaeroides strain GA the proportion of the carotenoids responding to the membrane potential is higher, yet the total amount of carotenoids is smaller than in the wild type. Consequently the signal to background absorption ratio is improved (Jackson et al. 1971). Nevertheless the signal to background absorption ratio is very small. Therefore the experimental device was designed particularly to detect small absorption changes on a high background in the visible region of light.

– 80 – Appendix A

The device consisted of a double monochromator with a halogen light source and a bialkali, photocathode photomultiplier (cf. Figure 2). Orthogonal to the measuring light path, a xenon flash light provided the excitation light. Absorption data could be recorded for approximately 10 s with a maximum time resolution of approximately 1 µs. A specially designed cuvette with 5 in-I outlets and equipped with a stirring device was used to perform experiments under controlled redox conditions. One inlet was used for the addition of the probe, one inlet and one outlet were used for a continuous argon stream. Two inlets were used for a platinum electrode and for a salt bridge to a silver reference electrode which were used to determine the redox potential in the chromatophore suspension. Similar systems have been used to investigate the components of the energy conversion chain of purple bacteria (Dutton et al. 1975; Bowyer et al. 1979; Venturoli et al. 1986). In a typical experiment buffer was prepared in the cuvette and dissolved oxygen was removed by bubbling with argon. Then chromatophore suspension from a frozen stock was diluted to achieve a chromatophore concentration such that one flash induces a single turnover of electrons in 95 % of the reaction centers. The corresponding concentration was determined previously according to (Escher et al. 1997). The suspension was equilibrated one hour in the dark before the experiments were started. Then a sequence of measurements was started. Single turn-over of the photosystem was induced by the xenon flash. Absorption of the chromatophore suspension was recorded 0.2 milliseconds before and 150 milliseconds after the flash. To increase the signal to noise ratio four absorption traces were generally recorded and averaged

probe inlet platin electrode------, L salt bridge- L....=:: argon outlet 1...____ argon inlet------

/ /

Figure 2 Set-up of the specially designed time-resolved spectrophotometer.

- 81 - Appendix A

according to:

n Â DAi()t i=1 DAav()t = (1) n

n is the number of records taken and DAi(t) denotes the corresponding absorption trace. Built-up of the membrane potential is shown in Figure 3-A1. The first fast phase (< 1 µs) stems from the electron/proton translocation induced by reaction center com-

plex; the second slower phase (5 ms) stems from the bc1 complex. The control trace (Figure 3 top) reflects the intrinsic decay of the membrane potential. Then the uncoupler of interest was added stepwise and the decay kinetics of the membrane potential was determined as for the control trace.

Deduction of the uncoupling activity

In a simple model, the membrane can be regarded as a parallel plate capacitor with an intrinsic resistor. The decay of the potential, Dy, in a plate capacitor, C, with resistance, R, obeys the following relation: d ()t ()t yy= (2) dt RC◊ Integration yields:

Ê t 1 ˆ yy(tt )= (0 ) ◊ exp - ◊ (3) Ë CR¯ The translocation of protons induced by the uncoupler can be regarded as an addi-

tional resistor with resistance, Runc, which is connected in parallel to the intrinsic

resistor of the membrane with resistance, Rint. The inverse of the overall resistance is then equal to the sum of the reciprocals of the individual resistances: 111 =+ (4) RRR tot int unc The decay of the membrane potential after addition of an uncoupler can therefore be described by:

Ê t Ê 11ˆˆ yytot(tt )= tot (0 ) ◊ exp - ◊ + (5) Á CRÁ R ˜˜ Ë Ë int unc ¯¯ transformation yields

– 82 – Appendix A

1 1 lJf tot (t) = lJf tot

Xune(t) = exp(- .!_ . _I_) = lJ'tot(t) . lJ'int(to) (7) C Rune lJf tot Uo) lJf int (t)

l single turnover l single turnover l flash l flash

--~~~~~~~-e1 control 0.01 0.01 (bc1-complex inhibited )

0.05 0.10 0.05 0.10 t (s) t (s) Figure 3 Kinetic trace of the carotenoid absorption change and derivation of kobs· A1 /B1: Time-resolved adsorption trace at a wavelength of 503 nm of the chromatophore suspension. liA is proportional to the membrane potential li\j!Which is induced by a single turnover flash. A2/B2: Analysis of the induced proton translocation by linear regression of the logarithm of the normalized trace. The frame denotes the time window which was annalized in the linear regression.

- 83 - Appendix A

The relative change in adsorption DA(t) is proportional to Dy by a factor of c. By dividing the trace determined after the addition of the uncoupler by the control trace the proportionality factor, c, is canceled

y tot()t y int()t0 DAtot()t c Xtunc()= ◊ = ◊ (8) y tot()t0 y int()t DAcontrol()t c . –1 The expression (C Runc) is expressed as the first order decay rate kobs. It was determined by a linear regression of the logarithm of the curve derived by dividing each

point in time of the trace of DAtot(t) by the respective point DAcontrol(t). An example of the resulting trace after normalization and taking the logarithm is shown in Fig- ure 3–A2/B2. The first 20 milliseconds after the flash, and in general the last 30 milli-

seconds of the trace were left out for the derivation of kobs.

A decrease in the initial potential, e.g. due to a partial inhibition of the bc1 complex,

leads to a constant factor in the Xunc(t) trace without affecting kobs. This can be seen in a comparison of the uncoupling effect of Dinoterb determined without inhibition of the

bc1 complex (Figure 3–A1/A2) to uncoupling determined with the bc1 complex inhibited previously by the addition of antimycin and myxothiazol (Figure 3–B1/B2).

Dinoterb is a weak inhibitor of the bc1 complex which entails a decrease in the initial potential during the experiment. As can be seen the slope of the curve of the two experiments is within the experimental error. When comparing the control traces of Figure 3–A1 and Figure 3–B1 one can see a notable difference in the decay of the two traces. This difference is attributed to a con-

tinuous small contribution of the bc1 complex to the build-up of Dy while the decay

has already commenced. Complete inhibition of the bc1 complex will also influence this continuos build-up of Dy, which will interfere with the normalization. Applying

the normalization procedure on traces with inhibited bc1 complex (control in Fig-

ure 3-B1) using traces with uninhibited bc1 complex as a control (control in

Figure 3-A1) suggests a significant uncoupling effect (kobs = –0.72, s = 0.21, n = 20)

whereas a comparison of control traces which both are inhibited yielded (kobs = +0.1,

s = 0.15, n = 20). For this reason. The experiments were performed with the bc1 complex inhibited.

– 84 – Appendix A

References

Abrahams, J. P., Leslie, A. G. W., Lutter, R. and Walker, J. E. (1994): Structure at 2.8 Å resolution of F1-ATPase from bovine heart mitochondria. Nature, 370, 621–628. Borghese, R., Crimi, M., Fava, L. and Melandri, B. A. (1998a): The ATP synthase atpHAGDC (F-1) operon from Rhodobacter capsulatus. Journal Of Bacteriology, 180, 416–421. Borghese, R., Turina, P., Lambertini, L. and Melandri, B. A. (1998b): The atpIBEXF operon coding for the F-0 sector of the ATP synthase from the purple nonsulfur photosynthetic bacterium Rhodobacter capsulatus. Archives Of Microbiology, 170, 385–388. Bowyer, J. R., Tierney, G. V. and Crofts, A. R. (1979): Secondary electron transfer in chromatophores of Rhodopseudomonas capsulata A1a pho. FEBS Lett., 101, 201–206. Boyer, P. D. (1993): The binding change mechanism for ATP synthase – some probabilities and possibilities. Biochim. Biophys. Acta, 1140, 215–250. Crofts, A. R. (1985): The mechanism of the ubiquinol:cytochrome c oxidoreductase of mitochondria and of Rhodopseudomonas sphaeroides. In: «The enzymes of biological membranes», Ed.: Martonosi. New York, N.Y., Plenum Press. 4, pp. 347–383. Crofts, A. R. and Wraight, C. A. (1983): The electrochemical domain of photosynthesis. Biochim. Biophys. Acta, 726, 149–185. Dutton, P. L., Petty, K. M., Bonner, H. S. and Morse, S. D. (1975): Cytochrome c2 and reaction center of Rhodopseudomonas sphaeroides Ga. membranes. Extinction coefficient, half reduction potentials, kinetics and electric field alterations. Biochim. Biophys. Acta, 387, 536– 556. 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. Toxicol. Chem., 16, 405–414. Falk, G. and Walker, J. E. (1988): DNA sequence of a gene cluster coding for subunits of the F0 membrane sector of ATP synthase in Rhodospirillum rubrum. Biochem. J., 254, 109–122. Fillingame, R. H. and Divall, S. (1999): Proton ATPases in bacteria: comparison to Escherichia coli F1F0 as the prototype. Novartis found. symp., 221, 218–234. Foster, D. L. and Fillingame, R. H. (1982): Stoichiometry of subunits in the H+-ATPase complex of Escherichia coli. J. Biol. Chem., 257, 2009–2015. Hermann, R. G., Steppuhn, J., Hermann, G. S. and Nelson, N. (1993): The nuclear-encoded polypeptide Cfo-II is from spinach is a real, ninth subunit of chloroplast ATP synthase. FEBS Letters, 326, 192–198. Holmes, N. G., Hunter, C. N., Niedermann, R. and Crofts, A. R. (1980): Identification of the pigment pool responsible for the flash induced carotenoid band shift in Rhodopseudomonas sphaeroides chromatophores. FEBS Lett., 115, 43–48.

– 85 – Appendix A

Jackson, J. B. and Crofts, A. R. (1971): The kinetics of light induced carotenoid changes in Rhodopseudomonas sphaeroides and their relation to electrical field generation across the chromatophore membrane. Eur. J. Biochem., 18, 120–130. Junge, W. and Jackson, J. B. (1982): The development of electrochemical potential gradients across photosynthetic membranes. In: «Photosynthesis: Energy Conversion by Plants and Bacteria», Ed.: Govindjee. New York, NY, Academic Press. 1, pp. 589–646. Noji, H., Yasuda, R., Yosida, M. and Kinosita, K. (1997): Direct observation of the rotation of F1-ATPase. Nature, 386, 299–302. Sabbert, D. S., Engelbrecht, S. and Junge, W. (1996): Intersubunit rotation in active F-ATPase. Nature, 381, 623–626. Venturoli, G., Fernàndez-Velasco, J. G., Crofts, A. R. and Melandri, B. A. (1986): Demonstration of collisional interaction of ubiquinol with the ubiquinol-cytochrome c2 oxidoreductase complex in chromatophores from Rhodobacter sphaeroides. Biochim. Biophys. Acta, 851, 340–352. Verméglio, A. and Joliot, P. (1999): The photosynthetic apparatus of Rhodobacter sphaeroides. Trends in Microbiology, 7, 435–440.

– 86 – Appendix B Kinetic model to describe the intrinsic uncoupling activity of substituted phenols in energy transducing membranes

Environ. Sci. Technol. 1999, 33, 560-570 AppendixB

activity of such a specifically acting compound, normalized Kinetic Model To Describe the to Its concentration at the target site, Is referred to as the "Intrinsic toxicity" of the compound. Intrinsic Uncoupling Activity of A large number of phenolic compounds and other hydrophobic lonogenlc organic compounds (HIOCs) act Substituted Phenols in Energy specifically by Interfering with one of the basic cellular Transducing Membranes functions, namely, energy transduction. In energy-trans- duclng membranes, HIOCs may Inhibit the electron flow by BEATE I. ESCHER .• binding directly to specific components of the electron- RENE HUNZIKER , AND transfer chain (4). and, even more Importantly, they can RENE P . SCHWARZ E NBACH destroy the electrochemical proton gradient by transporting Swiss Federal Institute for Environmental Science and protons across the membrane thereby short-circuiting the Technology (EA WAG), and Swiss Federal Institute of chemlosmotlc proton cycle and preventing ATP synthesis Technology (ETH), CH-8600 Diibendorf, Switzerland (5, 6') . This mechanism Is commonly referred to as uncoupling of oxidative- or photophosphorylation or simply "uncou- J OHN C . WESTALL pling". In Oregon State University, Corvallis, Oregon the case of phenolic compounds and other weak organic acids, the mechanism of uncoupling Is viewed as a shuttle mechanism, In which both the neutral phenol species and the charged phenoxlde act together to transport protons A new approach to understand the increased toxicity of across the membrane. In addition, a heterodlmer composed uncouplers as compared to baseline toxicity (narcosis) is of one phenol plus one phenoxlde species can participate as charge carrier In the shuttle mechanism (7). The model of presented here. The overall uncoupling activity is the protonophorlc shuttle mechanism Is sketched In Figure quantitatively separated into the contribution of membrane 1. The charged species migrate across the membrane driven concentration and speciation and intrinsic activity. This by the membrane potential. Protons are then taken up from approach is a further step toward the development of improved the aqueous phase, and the resulting neutral phenols diffuse Quantitative Structure- Activity Relationships (QSAR) and back across the membrane driven by the concentration of toxicokinetic models used in risk assessment. The gradient of phenols that has been built up by the migration protonophoric uncoupling activity of seven nitro- and processes. Hence, the overall uncoupling act!Vlty of a g!Ven chlorophenols has been investigated as a function of pH phenol is dependent not only on Its total concentration In and concentration using time-resolved spectroscopy the membrane (as Is the case for narcotic effects) but also on photosynthetic membranes. The experimental data are on Its speclatlon In the membrane (I.e., degree of dissociation and formation ofheterodlmers) and the abilities of the various described by a kinetic model that includes a monomeric species to cross the membrane. These transport character- and a dimeric protonophoric shuttle mechanisms. Input istics, or the ability of the compound already present In the parameters of the model are the experimental data membrane to relax the electrochemical proton gradient for relaxation of the membrane potential, the biomembrane- across the membrane, corresponds to the Intrinsic uncou- water distribution constants of the phenol and phenoxide pling activity. species, and the acidity constant of the phenol. Adjustable In earlier work, we have shown that time-resolved parameters are the translocation rate constants of all phenolic spectroscopy can be used to quantify the uncoupling activity species and the heterodimer formation constant. These of phenols In photosynthetic membranes of the purple parameters constitute the intrinsic uncoupling activity. bacterium Rhodobacter sphaerotdes (4). In this test system, Hydrophobicity and acidity govern the partitioning of phenols the membrane potential Is created by a brief, "slngle- tumover" flash of light. The build-up and the subsequent into the membrane but appear not to be the sole relaxation of the membrane potential Is deduced from the determining factors for the intrinsic uncoupling activity of change In absorbance at 503 nm (8). Uncouplers Increase phenolic compounds. Additional factors include steric the relaxation rate. The uncoupling activity can be expressed effects and charge distribution within the molecule. as the pseudo-first-order rate constant for the decay of absorbance (and hence membrane potential). In a subsequent study we related the first-order decay Introduction constants of a large number of chloro- and nltrophenols to For the development of predictive toxicity models It Is the concentration of the phenol and phenoxlde species Important to identify and understand the mechanism of sorbed Into the membrane at pH 7 (9) by combining the toxicity (1). The toxicity ofnonspeclfically acting compounds results of the earlier study with results of membrane-water (narcotics) Is a direct function of the amount of chemical partitioning experiments (1 O'J. This first crude analysis ofthe present In the organism (2) and can be predicted from their data already provided some Interesting Insight Into the factors hydrophobicity by Quantitative Structure- Activity Relation- that determine the overall uncoupling act!Vlty of a given ships (QSAR). Other compounds exhibit toxicity greater than compound and pointed to the fact that there are activity that predicted from QSARs of narcotic baseline toxicity (3). differences between the different phenols In addition to the For these compounds, predictive models have to account effect of membrane concentration and speclatlon. However, for the ability of a compound already present at the target the simple model failed to describe adequately the effect of site to act according to a specific mechanism. The toxic pH on the uncoupling activity of the compounds. In the work presented In this paper, we measured the • Corresponding author phone: 0041 -1-823 5068; Fax: 0041-1- pH-dependence of the uncoupling activity of seven chloro- 823 5471; e-mail address: [email protected]. and nltrophenols with the goal of assessing the relative

560 • ENVIRONMENTAL SCIENCE & TECHNOLOGY I VOL. 33, NO. 4, 1999 10.1021/es980545h CCC: S18.00 @ 1999 American Chemical Society Published on Web 01/06/1999

- 89 - AppendixB

and the" -side of the membrane are equal, as they are in the membrane phase at' and". During a typical experiment (. Because the · boundary concentrations of all uncoupler species, both in the mem- ! layer brane and in solution, are small compared to the concentrations of buffer ions in solution, and the buffer capacity in the FIGURE 1. Model for protonophoric shuttle mechanism of weak aqueous phase ls high, the charge created by the transport organic acids. Slow equilibrium processes before the experiment of protons across the membrane ls presumed to be accom- are the distribution of the neutral acid HAw and the charged modated by slight shifts in the buffer speclation, while the conjugated base Aw into the polar headgroups of the lipid bilayer pH remains constant. described by Kmw.HA and Knrw.A· During the experiment, the During the relaxation phase, only the processes occurring concentrations of HA11, Am, and the heterodimer AHAm in the within the membrane have to be considered. The charged membrane are rapidly equilibrated according to the acidity constant species, phenoxJde Am and heterodimer AHAm (where the ~ and the heterodimer formation constant KAHA· The transloca- superscript m represents the membrane), migrate along the tion processes across the membrane involved are the diffusion of membrane potential from the '-side to the "-side of the HAm, ~~. and the migration of Am, ..fA;9, and of AHAm, ..IA~\. membrane. At the membrane-water interfaces, the acld- base equilibrium ls maintained, and protons are exchanged contributions of various factors (I.e., membrane-water with the buffer or water in the aqueous boundary layer (14). partitioning, speclation inside the membrane, transport rates The acid-base and heterodimer formation reactions are fast of the various species) to the pH-dependent effect of a given compared to the diffusion of the molecules across the compound. The model described in this paper ls an extension membrane (11). Therefore, it can be assumed that these of biophysical models used in the past to describe the kinetics reactions are in equilibrium at any time during the experi- of the protonophorlc shuttle mechanism of weak organic ment. Then, the neutral species HAm diffuses back across acids in artlflcial planar lipid bilayers (11, lZ). It ls shown that the experimental data can be adequately described with the membrane from the "-side to the '-side along the this model, and that translocation rate constants can be concentration gradient that ls built up as a result of the deduced for all species involved. The results of this study migration processes. Ultimately the initial equilibrium state demonstrate that for developing improved QSARs for specif- ls regained. ically acting chemicals and for assessing possible synergistic Speciation in the Aqueous Phase (w). The organic acid effects of such compounds in mixtures, it ls necessary to dissociates in the aqueous phase according to the reaction separate target site (membrane) concentration from intrinsic actlVlty. (1)

Uncoupling Model where HAw represents the neutral protonated form, Aw the Description ofthe Protonophoric Uncoupling Process. The acid anion, and H the aqueous hydrogen ion, and K';, ls the observed rate constant of the exponential decay in absor- equilibrium constant. Representations of positive charge of bance at 503 nm, kobs. ls the experimental measure of the proton and negative charge of phenoxide are omitted for uncoupling actlVlty. To explain the dependence of koos on brevity. The mass law expression for this reaction ls pH and total concentration of phenol in the aqueous phase, we present a new kinetic model of a protonophoric shuttle mechanism of uncoupling. This model was adapted from (2) the kinetic transport model of acidic uncouplers in artificial planar lipid bilayer membranes used by McLaughlin and co-workers (11-14). Our model relies on a number of assumptions introduced by McLaughlin et al. (11-14), where C:. and Ci'.iA (mol·L-1) represent the concentrations of extends that work closer to a real biological system, and A and HA in the aqueous phase and aH the hydrogen actlVlty accounts for possible heterodimer formation. In this section, (pH= -log aH), and the total concentration in the aqueous the assumptions and the essential features of the model are phase c:;,t ls defined by described for weak organic acids. A general derivation ls presented in the Appendix. (3) Figure I gives an overview of the relevant processes considered in the kinetic uncoupling model. The neutral acid HA and the charged conjugate base A are in equilibrium Concentration and Speciation in the Membrane (m). between the aqueous phase and the surface layers on both Both the species HA and A partition between the aqueous sides of the chromatophore membrane. At equilibrium, the phase and membrane phase according to reactions of the concentrations in the aqueous phases adjacent to the'- side type

VOL. 33, NO. 4, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY • 561 -90- Appendix B

w m m HA h HA (4) is evidence that the CAHA term is negligible compared to the sum of Cm and Cm. Barstad et al. ( 15 ) detected AHA w m HA A A h A (5) heterodimers in non-hydrogen-bonding solvents but were unable to detect AHA in lipid bilayers in aqueous solution. where HA m and A m refer to the species in the membrane. The m They estimated KAHA for pentachlorophenol to lie between mass law equations for these reactions are 0.005 kg ‚mol -1 and 0.5 kg ‚mol -1 from a combination of membrane conductivity measurements and the surface Cm K ) HA (6) density of adsorbed PCP. Since the heterodimer can migrate mw,HA w more easily across the membrane than the anionic monomer, CHA it may contribute significantly to the overall process despite m its low concentration ( 7). CA K ) (7) Finally, the acid -base reaction in the membrane is defined mw,A w CA by the reaction

m m m m where CA and CHA are the effective concentrations in the HA h H+A (15) membrane phase in units of mol of HA or A per kg of phospholipid in the membrane. for which the mass law expression is Furthermore, HA and A are postulated to react in the membrane to form the negatively charged heterodimer, AHA, Cma K Km ) A H )Kw mw,A (16) a m a m m m C Kmw,HA HA +A h AHA (8) HA

m for which the mass law expression is Note that the acidity constant in the membrane phase Ka is operationally defined in terms of the proton activity in the m aqueous phase, because we assume that there are no free CAHA Km ) (9) protons in the membrane and that the protons of the acid - AHA m m CHA ‚CA base reaction in the membrane are directly exchanged with the adjacent aqueous solution. The total concentration of uncoupler in the membrane, The equations above refer to the bulk partitioning and m Ctot , is the sum of the concentrations of all these species speciation of the acid in the solution and in the membrane. However, both neutral and charged species are assumed to m m m m Ctot )CHA +CA +2CAHA (10) be located primarily near the surface of the membrane, at the interface between polar headgroups and hydrophobic and the total concentration of uncoupler in the system, Ctot , membrane core ( 16, 17 ). Therefore, the bulk membrane is concentrations of a species i can be converted to surface- normalized concentrations on each side of the membrane, m w -2 m Ctot )Ctot ‚[m] +Ctot (11) Γi(mol ‚m ), by dividing Ci by the specific surface area, s, of the phospholipids in the chromatophore membrane where [m] is the ratio of membrane lipid to aqueous phase expressed in units of kg phospholipid per liter. Cm Γ) i (17) It is convenient to define the fractions of the different i s phenolic species present in the aqueous phase and inside w m the membrane, R and R , respectively, at a given pH NAA i i s) (18) mPL Cw Rw ) i (12) i w where the average molecular weight of the phospholipids in Ctot -1 chromatophores mPL is 760 g ‚mol (18 ), NAis the Avogadro Cm number, and the average surface area Aof one phospholipid Rm ) i (13) in a membrane is approximately 5 to 6 ×10 -19 m2(19 ). The i m 5 2 -1 Ctot resulting specific surface area sis 4.4 ×10 m kg . General Membrane Transport. The flux of species i, Ji Substitution of eqs 3 and 10 in eq 11 yields a complete mass [mol ‚m2. s-1], from ′across the membrane to ′′ is defined by balance. All species concentrations are then reexpressed in the first-order translocation expression w terms of CHA and the relevant mass law equations. -dΓi′ dΓi′′ Ji) ) (19) w w Ka w w Ka dt dt Ctot )CHA +CHA +CHA Kmw,HA [m] +CHA Kmw,A [m] + aH aH K The driving force of the flux of species i is on one hand 2Cw K Cw aKm [m] (14) the concentration gradient that causes the diffusional flux HA mw,HA HA a AHA diff H Ji (translocation independent of electric field) and on the other hand the membrane potential that causes the migra- w mig The resulting quadratic equation is solved for CHA , from tional flux J (translocation of charged species in an w i which the concentrations of all species and the values of Ri electric field) m m m m and Ri can be calculated. The values of RA,RHA , and RAHA so calculated are exact for the initial equilibrium condition, diff Ji )- kiΓi′′ +kiΓi′ (20) and are a very good approximation of the speciation during excitation and relaxation if the Cm term is negligible in eq Γ′+Γ′′ AHA Jmig )- kz i i ∆u (21) 10 or the relative changes in total concentration of phenol i ii( 2 ) at both sides of the membrane are small following a flash. As will be shown, these conditions are generally met. There where kiis a homogeneous translocation coefficient expressed

562 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 4, 1999 – 91 – Appendix B

-1 0 in units of s ,ziis the charge of species i, ( Γi′+Γi′′ )/2 )Γi diff JHA )- 2kHA ∆Γ HA (26) corresponds to the total surface concentration of species i, and ∆uis the dimensionless membrane potential defined by diff The diffusional fluxes of the phenoxide, JA , and of the heterodimer, Jdiff , are defined analogously and are imple- F∆φ AHA ∆u) (22) mented as well in the uncoupling model. The modeling results RT show, however, that these latter fluxes are rather small under most experimental conditions. The migrational fluxes of the where ∆φis the membrane potential ( φ′′ -φ′)[ V], Fis the charged species, Jmig and Jmig , are defined according to Faraday constant, R is the gas constant, and T is the A AHA temperature. The sum of the diffusional and the migrational Jmig )- kΓ0∆u (27) flux constitutes the total flux Ji A A A mig 0 ∆u ∆u JAHA )- kAHA ΓAHA ∆u (28) J)Jdiff +Jmig )- kΓ′′ 1+ z +kΓ′1- z ) i i i i i( 2 i) i i( 2 i) If the back-diffusion of the neutral species did not influence -2k∆Γ -kzΓ0∆u (23) i i ii i the overall uncoupling mechanism, the uncoupling model could be simplified to the migration of the charged species, where ∆Γ i)(Γi′′ -Γi′)/2. and kobs would be a linear combination of the contribution It remains to relate the membrane potential to surface from phenoxide and heterodimer: concentrations. The membrane potential is related to the surface charge, σ, as in a parallel plate capacitor 0 0 kobs )kAΓA+kAHA ΓAHA (29) ∆φ)C-1σ (24) In this case, the experimental data, kobs , can be related to the total concentration of phenol added to the system, C tot , from where Cis the capacitance of the chromatophore membrane the quadratic equation of approximately 5.5 ‚10 -3F m -2(19 ). The change in surface charge σis a function of the flux of the charged species. 2 kobs )a‚Ctot +b‚Ctot (30) dσ This simple equation was used in a previous study ( 9)to )F∑ziJi (25) dt i describe the concentration dependence of kobs at constant pH and is still applied in the present study to identify the Equations 1 -25 define the uncoupling model. As shown relevant charged species of a given compound. Note, in the Appendix, rearrangement of eqs 1 -25 yields two however, that for all data presented here, the back-diffusion coupled differential equations (eqs 41 and 43), for which the of HA had a significant contribution to the overall uncoupling eigenvalues λ1and λ2(eq 47) are the characteristic times to process and could not be set to indefinite. be compared to kobs . The rate constants of decay are a function of the experimental conditions (total organic acid concen- Materia lan dMethods tration, total concentration of phospholipid, pH), three Chemicals. The phenols (full names and abbreviations are independently determined parameters describing membrane given in Table 1) were purchased from the following w concentration and speciation ( Kmw,HA ,Kmw,A , and p Ka ), and companies: Riedel-de Ha¨en (Seelze, Germany): 245TCP, the adjustable parameters describing the intrinsic activity DINOSEB, DINOTERB; Fluka (Buchs, Switzerland): 345TCP, m (kHA ,kA,kAHA , and KAHA ). 2345TeCP, PCP, 34DNP. The following biological buffers were The model described above predicts a biphasic decay in used: MES (2-morpholino-ethanesulfonic acid, p Ka)6.15); membrane potential with characteristic times corresponding MOPS (3-( N-morpholino)propanesulfonic acid, p Ka)7.2); to the values of λ1 and λ2. This biphasic decay could be HEPPS ( N-2-hydroxyethylpiperazine- N′-3-propanesulfonic observed in charge-pulse experiments on artificial lipid acid, p Ka)7.8); CHES (2-(cyclohexylamino)ethanesulfonic membranes ( 11 ). For the movement of tetraphenylborate acid, p Ka ) 9.55), all of which were from Fluka (Buchs, ions across black lipid membranes under charge-pulse Switzerland). Chemicals used for time-resolved spectroscopy conditions, the fast phase λ2was attributed to the displace- are described in ref 4. ment of lipophilic ions as a consequence of the suddenly Determination of the Uncoupling Activity. Membrane applied voltage, whereas the second slow voltage relaxation vesicles (chromatophores) of the purple bacterium Rhodo- λ1represents the slow discharge of the membrane through bacter sphaeroides were prepared and characterized as the external resistor and the redistribution of charges within described previously ( 4, 21 ). The single-beam spectropho- the membrane ( 13 ). tometer equipped with a flash excitation unit and kinetic However, the faster phase of the decay, which is described data acquisition capabilities is described elsewhere ( 4). The by the greater eigenvalue λ2, was too fast to be detected with measurements were performed in an anaerobic cuvette at the experimental system described here. Since the build-up a redox potential adjusted to 120 -130 mV with redox of the membrane potential is polyphasic with half-times of mediators (2,3,5,6-tetramethyl -phenylene diamine, N-methyl the spatial electron-transfer reactions from picoseconds to phenazonium methosulfate, duroquinone, 1,2-naphtho- milliseconds ( 20 ), decay rate constants g70 s -1cannot be quinone, 1,4-naphthoquinone) and ferricyanide/dithionite resolved. Consequently, the values of the adjustable param- in a buffer composed of a mixture of MES, MOPS, HEPPS, eters were derived from a nonlinear fit of the observed and CHES, adjusted to a total buffer- and K + concentration monophasic decay rate constants ( kobs ) to the calculated (KCl/KOH) of 50 mM and 100 mM, respectively. Since the values of λ1(eq 47), for sets of experiments in which both the buffer mixture did not buffer equally well at every point and pH value and the total concentration of organic acid were the stock suspension of chromatophores was prepared in a varied. buffer of pH 7, the pH was measured at the end of each series Transport Processes for Acidic Uncouplers. From the directly in the cuvette. Deviation from the initial value was general equations given above, the equations for acidic never larger than (0.2 pH-units. At each pH-value, several uncouplers that can additionally form heterodimers can be measurement cycles with different total concentrations of diff set up. The diffusional flux of the neutral species, JHA is phenol were performed. During one measurement cycle, four

VOL. 33, NO. 4, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 563 – 92 – AppendixB

TABLE 1. Summary of All Parameters That Determine the Uncoupling Model

K...,JlA• ran~ of compound abbreviation (l·kg-') pk; K...,Jt- (l·kg-1) kHA (s-1) kA (s-1) kAHA1 (s-1) Cu,, M)• ,;

3,4-dinitrophenol 34DNP 1.47·103 5.38b (7.51 ± 2.280·102 g 82 ± 23 0.07 ± 0.01 53 ± 10 5-100 102 2-sec-butyl-4,S-dinitrophenol DINOSEB 9.45·103 4.62• 2.16·103. 1005± 151 0.48 ± O.oJ (j 2-10 36 2-tert-butyl-4,6-dinitrophenol DI NOTER 1.26·104 4.80d 3.89·103. 6006 ± 697 1.45 ± 0.11 (j 0.1-40 78 B 2,4,5-trichlorophenol 245TCP 2.18·104 6.94• (1.05 ± 0.13)-104 g 142 ± 816 0.014 ± 0.001 10.0 ± 0.9 2-70 50 3,4,5-trichlorophenol 345TCP 5.13·104 7.73• (2.46 ± 0.36)-104 g 150'> 0.09± 0.01 21.6 ± 2.6 2-30 43 2,3,4,5-tetrachlorophenol 2345TeCP 7.80·104 6.35• (3.44 ± 0.59)-104 g 108 ± 29 0.19±0.02 157 ± 20 1-8 52 pentachlorophenol PCP 1.23·105 4.75• 3.09·104. 346 ± 24 0.17 ± 0.09 (j 1-50 101 1 •Data from ref 10. b Experimentally determined. c I= 50 mM (25). •I= 250 mM, 1% methanol mM (2n. • 1=10 and 50 mM (26). Standard error. g Fitted. h pl(,;' is at the limit of acceptable range that there is no convergence of fit unless kHA is fixed to an estimated value (approximately equal to the value of 245TCP). 1 Value refers to kAHA if !<,;'HA is set to 1 kg L- 1• i.e .. more precisely, the values in this column correspond to the product of kAHA"l<"'AHA. j Fixed to zero because of linear dependence of ko0s from C,01• k Concentration range of experimental measurements; membrane lipid concentration in the assay varied between 7 .5 and 8.1·10-4 kg·L-1• / Number of measurements. kinetic traces were averaged, each of which consisted of the relative absorption change at 503 nm over a 150 ms intervall, beginning 2 µs after the xenon flash, followed by 1 min of reequilibration. The absorbance change at 503 run is proportional to the membrane potential (22). The membrane potential accounts for the majority of the electrochemical proton gradient in chromatophores (after single-turnover flash ~pH""' 0.003, M>""' 70 mV) (23). The "uncoupling actiVlty" was quantified as the pseudo- first-order decay rate constant, kobs. of the absorbance at 503 nm and hence of the membrane potential. The value of kobs is normalized via a control for the properties of a particular chromatophore preparation. For more details see refs 4 and 9. No second faster phase of decay could be resolved despite the model predictions due to interference with the build-up of the membrane potential and large scattering of the kinetic traces. 0 2·1 Q"5 4'10"5 6·10·5 8· 10·5 1·10-4 Determination of Acidity Constants. In the uncoupling Ctot (M) model, the acid-base equilibrium in the aqueous phase is defined by the mixed acidity constant at an ionic strength FIGURE 2. Plot of k* versus the total concentration of uncoupler of 100 mM (eq 2). The mixed acidity constant of 34 DNP was 34DNP, Ct.~ at different pH-values, • pH 5.3, v pH 5.6, t:.. pH 6.05, determined with a potentlometric titrator (PCA 101, Sirius 0 pH 6.18, O pH 6.33, 8 pH 6.42, D pH 6.5, 0 pH 6.89, m pH 6.90, • Analytical Intruments, Riverside, UK) (24). The other p~ pH 6.94, ISi pH 7.31, • pH 7.35, •pH 7.82, .., pH 8.25. Solid lines are values were taken from literature (25-27). In all references, best fits to the simplified quadratic model (eq 30). the values refer to mixed acidity constants. Since the exact ionic strength was not given in all cases, and since the underestimation of the contribution of the charged species estimated error due to differences in ionic strength are in the uncoupling model. Therefore, Kmw.A was also used as estimated to be smaller than 1%, the values given in refs an adjustable parameter, and the Kmw.A determined from the 25-27 were directly used. relaxation experiment turned out to be somewhat higher Determination ofMembrane-Water Distribution Ratios. than the experimentally determined Kmw.A· It is possible that Llposomes prepared from phosphatidylcholine were used the experimentally determined Kmw.A. which were measured as model systems for the determination of the chromato- with pure phosphatidylcholine membranes, underestimate phore-water distribution ratios. Substituted phenols were the true uptake of charged species in the chromatophore found to partition nearly quantitatively into the lipid moiety membrane, 30% of which consists of various different of the chromatophoremembraneand uptake into the protein phospholipids (23% phosphatidylcholine, 35% phosphatl- moiety was negligible (10). Liposome-water distribution ratios dylethanolamine, 34% phosphatidylglycerol, 4% cardiolipin, of34DNP were measured as described in (JO) as a function and 3% phosphatidic acid (18)) and 70% ofwhich are integral of concentration, pH and ionic strength. All other experi- membrane proteins (9). mental data were taken from ref 10. The membrane-water An attempt was made to determine distribution ratios distribution coefficients of the neutral phenol, Kmw.HA. and directly with chromatophore membranes (unpublished of the phenoxide, Kmw.A. were calculated from the experi- results), but the pH range of the experiments was too limited mental values with an improved membrane-water partition- to allow the extrapolation of the distribution ratios of the ing model (28). The improved model is superior to the species HA and A. previously published version (10), which was just an extension of the octanol-water partitioning model. In the improved Results and Discussion version, the membrane-water distribution is treated as a Determination of the Adjustable Model Parameters. The surfacesorptlon process and not as a bulk partitioning process experimental data consisted of a set of observed rate and therefore does not include ion-pair formation in the constants, koos. determined as a function of total concentra- lipid bilayer. The resulting distribution ratios for an ionic tion of phenol in the chromatophore suspension, C1ot. and strength of 100 mM are listed in Table 1. of pH, as illustrated for 34DNP in Figure 2 (see Table 1 for For some compounds (34DNP, 245TCP, 345TCP, and abbreviations of compound names). The lines in Figure 2 2345TeCP), the experimentally determined Kmw.A led to an are the fits of the quadratic model (eq 30). This simplified

564 •ENVIRONMENTAL SCIENCE & TECHNOLOGY I VOL. 33, NO. 4, 1999 -93- AppendixB

8 A 20 0 7 6 15

A 5 .¢ 0 0 0 v ~ ~ • ~ 4 10 ~ D.. c-[ :.:.0 3 • 5 2 \ ti \ ...... , ~ ¢ 0 0 3 4 5 6 7 8 9 10 0 2 3 4 5 6 7 8 pH kobs (s.1) B 5 FIGURE 3. Calculated J..1 versus the experimentally determined ko11s • • for 34DNP; ;., were calculated with eq 47 using the parameters listed in Table 1, e pH 5.3, v pH 5.6, t>. pH 6.05, 0 pH 6.18, O pH • 4 \ 6.33, 8 pH 6.42, D pH 6.5,0 pH 6.89, !El pH 6.90,• pH 6.94, ISi pH 7.31, + pH 7.35, .._pH 7.82, .., pH 8.25. ... • model was used to obtain preliminary Information on the ~ 3 relevant charged species. A linear dependence of kobs from ~ .80 Clot indicates that the monomer ls the dominant charged :.:. 2 species, and a purely quadratic dependence of kobs on Ctot indicates that the heterodimer ls the dominant charged -'· species. If both linear and quadratic term are significant, as -~ ls the case for 34DNP, both monomer and heterodimer are ·. assumed to contribute significantly to the protonophorlc shuttle mechanism. 0 · · ~ Three compounds, DINOSEB, DINOTERB, and PCP, 4 5 6 7 8 9 10 showed a linear dependence of kobs on Ctot over the entire pH pH range investigated, indicating that the heterodimer was insignificant. For these compounds Is versus pH for (A) 34DNP at Ctot = 5·10-5 M, different species in the membrane were calculated from pH (B) PCP at Ctoc = 10-5 M, and (C) 2345TeCP at Ctot = 4·10-6 M. The and the acidity constants with eqs 12-14. solid line corresponds to the model described by eq 47 and the A typical plot illustrating the quality of fit ls shown In parameters from Table 1. The broken line and the dotted line Figure 3, In which the membrane potential decay rate represent the contribution of the heterodimer and the phenoxide constants, .A.1, which were calculated from the model, are species, respectively, to the overall activity. The symbols correspond plotted against the experimentally determined kobs for 34DNP. to the experimental values at the given Ct.~ each point is intrapolated Despite scatter In the data, there is no systematic deviation from the simplified quadratic model described by eq 30. The different from the line of unit slope. symbols for the experimental data of 34DNP correspond to Figure 4 illustrates the ability of the model to capture experiments with different batches of chromatophores. variations In kobs as a function of pH at a given value of Clot· Each of the experimental values in Figure 4 represents one ls the case for 34DNP (Figure 4A), the maximum of actiVlty data series at a given pH as shown in Figure 2. The pH- ls at a pH that corresponds to the pJ

VOL. 33, NO. 4, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY• 565 -94- AppendixB

4 A 4 10 e .0 3 1000 ~ , 0 - _o, ~ 2 , ~(I) 100 g )-·-·-0 ·--- e-- - . "! t 'if E ~ 0 ~ 10 • E 0 "' .Jditt(A, AHA) ~- b • 1 _ ~...... , : ·2 -·- 0.1 --.. - -. -·- ·3 - Jditt(HA) -·. .. ·4 0.01 .. 1 ll. ll. ll. ll. ll. .0 10·4 10·3 10·2 10·1 1 OO () () () Q) € 10 z Cl) Q) I- I- ~ ll. U') U') 0 0 0 t (s) I- c c ~ U') ~ "' ~ i5 i5 FIGURE 6. The relationship between"' the translocation rate constants kHA (D), kA (+), and kAHA (II) for the compounds investigated.

~ 0.8 the simple quadratic model of uncoupling that is based on .fl) the flux of the anions as rate-limiting step of the overall ~E 0.6 process cannot fully explain the protonophoric shuttle 0 mechanism, because the back-diffusion ofphenol contributes E 0.4 significantly to the overall process. Note also that at the higher b~ 0.2 pH values, typically above pH 8, there is a significant theoretical contribution of A.2 to the overall process which is 0 taken into account in Figure 58, although it was not possible to resolve the fast decay phase experimentally.

-0.2 Despite structure-dependent variability of the k 1 values of the different compounds investigated, which is discussed in more detail below, there are two distinct ranges of 1 2 1 o-s 1 0· 1 0- translocation rate constants as shown in Figure 6. The values I (s) of kttA are more than 3 orders of magnitude greater than the FIGURE 5. Migrational and diffusive fluxes as a function of time corresponding kA, but only about 1 order of magnitude higher for 34DNP at (A) pH &and (B) pH 8 calculated for a total concentration than kAHA· Since K"'AHA was set to 1, in fact the product of .K,:'HA of uncoupler Cia1 = 15 µMand a membrane lipid-to-water ratio [ml and kAHA was modeled. The values themselves vary within 1 of 8.Mo-4 kg·L-1• order ofmagnitude between the compounds. If a heterodimer is formed at all, the values of kAHA are almost 3 orders of shifted to the alkaline side of the p~. The maximum is magnitude greater than the values of kA, which is consistent broadened by high kttA values. If phenoxideand heterodimer with the expectation that the dimer has a larger translocation are equally important, the pH range below p~ is domi- rate constant than the anion because the solubility of an ion nated by the flux of the heterodimer, and the pH range above increases strongly with size in a medium of low dielectric p~ is dominated by the flux of the phenoxide ion, and the constant, i.e., in the intertor of the lipid membrane. It is maximum of activity is between these two maxima as in the difficult to discuss quantitatively this ratio, because the case of 2345TeCP (Figure 4C). heterodimer formation constant I

566 •ENVIRONMENTAL SCIENCE & TECHNOLOGY I VOL. 33, NO. 4, 1999 -95- Appendix B

kHA . The charge of DINOSEB and DINOTERB appears to be of two weaker uncouplers, 4-nitrophenol and 3,4-dichlo- most effectively distributed over the entire molecule, and rophenol (data not shown) as well as the activity of 2,4- the sec - and the tert -butyl groups shield the hydroxy function dinitrophenol could not be modeled satisfactorily because sterically. 34DNP has a very low kA and a significant kAHA , the effective concentrations in the membrane were in the which points to a good delocalization of the charge and a range of critical membrane burdens of the narcotic chemicals good ability to form heterodimers, but both A and AHA are (2) or slightly below. Even the activities of stronger uncouplers only slightly membrane permeable because they are rather were often underestimated by the model in the outer pH hydrophilic. region, indicating a significant contribution of the underlying The formation of heterodimers is highly unfavorable for baseline toxicity. One focus of research presently conducted DINOSEB, DINOTERB, and PCP due to the two bulky ortho in our group is to incorporate the contribution of baseline substituents. Only compounds with either no or just one toxicity into the kinetic uncoupling model. ortho substituent (i.e., 34DNP, 245TCP, 345TCP, and Comparison of Translocation Rate Constants with 2345TeCP) appear to to form heterodimers. The heterodimer Membrane Permeabilities from the Literature. The overall increases the overall activity of these compounds, but they membrane permeability of species i, Pi, is defined by Fick’s still are not as strong uncouplers as the diortho substituted Law phenols investigated here despite their generally higher hydrophobicity. Phenols without ortho substituents appear w′′ w′ Ji)Pi(Ci -Ci) (31) to have significantly lower kAand kAHA , even if the charge can be very well delocalized over the entire molecule as is the that is, the rate constant for transport of a species from the case for 34DNP. It can be hypothesized that steric shielding aqueous phase on one side of the membrane, through the of the charge plays a more important role than charge membrane, to the aqueous phase on the other side. Since delocalization for the translocation rates of hydrophobic ions. it can be assumed that transport through the membrane This view is consistent with the extraordinarily high trans- itself is the rate-limiting step ( 12 ), the translocation rate location rates of tetraphenylborate analogues ( 13 ). constants kican be transformed into membrane perme- Critical Evaluation of the Uncoupling Model. When one abilities Pi considers that the kinetic uncoupling model was initially developed from charge pulse and voltage-clamp measure- K k P) mw,i i (32) ments on black lipid bilayer membranes ( 11 ), it is quite i s impressive that the model can be applied and extended to a subcellular biological system. However, as expected, the where s, the specific surface area of the membrane, is 4.4 × model is less well defined for the biological system than for 5 2 -1 10 m kg . This transformation allows comparison of ki the planar lipid membrane. A single set of kobs data measured values with Pivalues in the literature. as a function of pH and Ctot has to be fit by the complete In Table 2 membrane permeabilities calculated from the uncoupling model with two to four adjustable parameters. kivalues given in Table 1 are compared to literature data It is straightforward to derive from this observation that the from various sources and from different types of measure- experimental data need to cover both flanks of the peak- ments. In most cases, the membrane permeabilities for the shaped pH-dependent curve to yield good results. Since the neutral acids are 3 -4 orders of magnitude greater than the experimental pH range is limited to pH 5.2 -9.0 due to permeabilities of the corresponding charged bases. denaturation of proteins at low pH and fusion of chromato- The calculated membrane permeabilities PHA of the phores at high pH, good fitting results, in particular good neutral species of substituted phenols are only slightly smaller estimates of both kA and kAHA can be obtained only with than the ones deduced for carbonylcyanide p-trifluo- w compounds with a p Ka between approximately 5.5 and 7.5, romethoxyphenylhydrazone (FCCP) by Benz et al. ( 11 ) from such as 2345TeCP (Figure 4C); exceptions are cases in which charge-pulse experiments. They are also in the same order there is a structural preference for either species, as is the of magnitude as the permeabilities of the entire membrane case for the good heterodimer former 34DNP (Figure 4A) to FCCP and other weak acidic uncouplers that were and the preferentially monomeric PCP (Figure 4B). For the determined with an alternative method based on pH- w purely monomeric uncouplers, compounds with p Kavalues dependent membrane potential measurements on planar as low as 4 still yield good modeling results. bilayer membranes ( 30 -33 ). The differences between the The experimental data of the more acidic 2,4-dinitro- values from various studies appear to be caused by the w phenol (p Kaof 3.94) (data not shown) could not be fit to the structure of the molecules and not by the type of the model for three reasons. First, the experimental values cover membrane, because the permeability of neutral species is only the far right flank of the pH-dependent activity curve. assumed to be independent of the dielectric constant of the Second, both monomeric and dimeric shuttle mechanism membrane ( 11 ). contribute to the overall effect. Third, the overall activity in The membrane permeabilities of the neutral species the experimentally available pH range is influenced by obtained in this study agree not only with the data from baseline toxicity as described below. The kAof the less acidic biophysical studies on planar lipid bilayers but also with compounds, 245TCP and 345TCP, exhibited large errors data from biological experiments. The reported diffusion because no data at pH-values higher than pH 9 could be coefficient of ubiquinone between two specific quinone measured and because both presumably form heterodimers binding sites located on opposite sides on the reaction center easily. complex and the cytochrome bc 1complex in chromatophores The model was not very sensitive to changes in values of (34 ) is in the same order of magnitude as the diffusion kHA , although the fitted values of kHA were significantly coefficient of DINOSEB and DINOTERB. different for the different compounds and the values did In contrast to the membrane permeability of neutral have to be greater than zero and not too high (otherwise the molecules, the permeability of an anion is strongly dependent model would have been reduced to the earlier quadratic on the dielectric constant . It is not clear whether there is model of eq 30). an interfacial barrier to A and AHA, so calculations of PAhave Finally, a good fit of the adjustable parameters can be to be treated with caution. Nevertheless, our average values obtained only for rather strong uncouplers, whose activity of kAagree well with direct measurements of the membrane is not disturbed by the narcotic effect that constitutes the permeability of hydrophobic anions in decane-containing baseline toxicity of any hydrophobic compound. The activity planar lipid bilayers from different phospholipids membranes

VOL. 33, NO. 4, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 567 – 96 – Appendix B

TABL E2.Membran ePermeabilitie sPHAofSubstitute dPhenol san dOthe rWeakl yAcidi cUncouplers -1 -1 -1 -1 PHA (cm ‚s ) PHA (cm ‚s ) PA(cm ‚s ) PA(cm ‚s ) compound between sorption sites d entire membrane i between sorption sites d entire membrane i 34DNP 3‚10 -2e 10 -5e DINOSEB 2e 2‚10 -4e DINOTER B 17 e 10 -3e 245TCP 1e 3‚10 -5e 345TCP 2e 5‚10 -4e 2345TeCP 2e 2‚10 -3e PCP 10 e 1‚10 -3e FCCP a 30 -60 f 50 f 2f,m CCCP b 17 g 11 g,j 0.2 g,m 2‚10 -3j DTFB c 4h picric acid 0.4 k 7‚10 -6k salicylic acid 0.7 l <10 -7l aCarbonycyanide p-trifluoromethoxyphenylhydrazone. bCarbonycyanide m-chlorophenylhydrazone. c5,6-Dichloro-2-trifluoromethylbenzimi- dazole. dFrom kinetic measurements of uncoupling. eThis work; in biological membranes; calculated from data in Table 1 with eq 32. fValues from Benz and McLaughlin ( 11 ). gValues from Kasianowicz et al. ( 12 ). hValues from Cohen et al. ( 31 ). iFrom membrane potential measurements using the method described by LeBlanc( 30 ). jPhosphatidylcholine/phosphatidylethanolamine/cardiolipin/decane bilayers ( 30 ). kLecithin/cholesterol/ decane bilayers ( 32 ). lLecithin/decane bilayers ( 33 ). m Chlorodecane-containing planar phosphatidycholine membrane.

(30, 32 ) (Table 2). The permeability of the FCCP anion, PA, shielding of the hydroxy group by the bulky alkyl substituents. is 2 orders of magnitude greater in a chlorodecane containing Phenols without ortho substituents have smaller transloca- planar lipid membrane ( ) 4.5) compared to a decane tion rates among phenol and phenoxide species, but their containing membrane ( )2.1) ( 6, 35 ). The dielectric constant overall intrinsic toxicity is high due to their ability to form of chromatophore membranes is about 3.8 ( 36 ). Conse- heterodimers, whose translocation rate constants are almost quently, permeabilities similar to chlorodecane-containing 3 orders of magnitude greater than those of the phenoxides. membranes should be expected for PA in our system. In Although the small data set of seven compounds presented contrast to expectation, PA of the phenoxides are much here does not allow to deduce any quantitative equations, smaller than PA of FCCP in chlorodecane-membranes. the generalization deduced above are an important basis for Benz ( 13 ) showed for tetraphenylborate and analogues future developments of QSARs. that the structure of lipophilic ions has a strong influence on their translocation rate constants but only a small influence The quantitative differentiation of uptake and speciation on their membrane-water distribution ratio. We also find from the intrinsic toxicity is in addition a prerequisite for the increasing translocation rate constants with increasing size development of toxicokinetic models. Toxicokinetic models of the ion. The effect is not as large as expected from the are a valuable tool for the prediction of the pH-dependent example of the tetraphenylborate analogues; tetraphenylbo- activity of HIOCs ( 40 ). In a more general sense, toxicokinetic rate ions are much larger and spherical, and their charge is models have been applied in aquatic toxicology to link in presumably much better shielded from the surroundings than vitro responses to effects on whole organisms ( 41 ). They may it is the case for the phenoxides. find in the future an application in the risk assessment of Practical Implications. The results presented here show HIOCs including the extrapolation from one organism to clearly the relevance of mechanistically based toxicity studies. another. The combination of uncoupling experiments and membrane- water partitioning experiments together with modeling of The mechanistic approach presented here should be well the uncoupling mechanism allows one to separate the overall suited to investigate the joint effect of mixtures of uncouplers uncoupling effect into various factors that can be related to and mixtures of baseline toxicants and uncouplers and to specific properties of the compound investigated. These test the hypothesis that synergistic effects could occur if factors are on one hand related to the concentration and heterodimers are formed by the phenoxide species of a strong speciation of the substituted phenols at the target site, the acid and the neutral species from another compound. membrane, and, on the other hand, to the intrinsic uncou- Finally, the approach presented here allows one to deduce pling parameters, i.e., translocation constants of all species clear classification criteria for the uncoupling potency of any across the membrane and heterodimer formation constant. weak organic acids. These criteria include not only hydro- These parameters are the basis for the development of phobicity and acidity of a given compound but also additional meaningful QSAR equations for uncouplers. Saarikoski et al. electronic and steric factors. Further investigations are (37, 38 ) developed an empirical model to predict the pH- presently being undertaken in our laboratory to extend the dependence of the toxicity of substituted phenols toward range of tested compounds and to realize the envisaged fish and derived QSARs with the octanol -water partitioning applications. constant Kow , the pH, and the acidity constant p Ka as descriptors. QSARs of the in vitro uncoupling activity of substituted phenols are usually based on one hydrophobicity Acknowledgments descriptor (e.g., log Kow ), one descriptor for the speciation We are indebted to Mario Snozzi for help with the develop- e.g., p Ka), and one descriptor for the steric effect of ortho substituents ( 21, 39 ). The results of the present study ment of the measuring method and to Philippe Pe´risset for rationalize the choice of these descriptors since the intrinsic help with the build-up of the instrument. We thank Jim toxicity is strongly dependent on the presence of shielding Anderson, Rik Eggen, Kai Goss, and Nina Schweigert for ortho substituents for higher translocation rate constants or reviewing the manuscript. Valuable comments were made the absence of ortho substituents for good heterodimer by Urs Fringeli. This project was supported by a grant of the formation. o-alkylnitrophenols exhibited the highest intrinsic Swiss Federal Institute of Technology (TH project 0-20-268- toxicity of all phenols investigated presumably due to steric 96).

568 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 4, 1999 – 97 – Appendix B

Appendix :Mathematica lDerivatio nofth eUncoupling Conversion of the differential equation for the change in Model surface charge (eq 25) to the dimensionless membrane potential ∆u(eqs 22 and 24) yields the following differential The following derivation of the uncoupling model is in equation principle applicable to organic acids and bases and may accoun tfo rheterodime rformatio nan dformatio nofother d∆u F2 complexes. For the organic acids described in this paper, ) ∑ziJi (42) species i refers to HA, A, and AHA. dt CRT i The total concentration of uncoupler molecules in the 0 -2 membrane, Γ(mol ‚m ), during the course of the experiment which can be rearranged, upon substitution of eq 39 for Ji, is assumed to be constant to

d∆u m 0 m 2 0 niΓi′+ niΓi′′ ∑ ∑ )- 2B( kiRizi)Γ∆Θ -B( kiRizi)Γ∆u i i ∑ ∑ 0 dt i i Γ ) (33) (43) 2 where where nirefers to the stoichiometric coefficient (i.e., nHA ) 1, nA)1, and nAHA )2). At any time during the experiment, F2 the surface concentration of a single species, Γi′, is a function B) (44) m CRT of the fraction of this species, Ri, in the membrane and the tota lconcentratio natth egive nsid eofth emembrane, The solution of the linear nonhomogeneous differential equation system that consists of eqs 41 and 43 is of the m Γi′)R iΓ′ (34) following form:

λ1t λ2t and the total concentration, Γ′, is the sum of the concentra- ∆Θ (t))b1e +b1e (45) tions of all species. λ1t λ1t ∆u(t))b1u1e +b2u2e (46) Γ′)∑niΓi′ (35) i The eigenvalues λ1and λ2correspond to the two expected rate constants of decay of membrane potential. Before the build-up of the membrane potential, the initial surface concentration, Γ°, is equal on both sides of the membrane and is the sum of the surface concentration of m m 2 0 all species. λ1,2 )0.5 -2( ∑nikiRi)-B(∑kiRizi)Γ ( i i 0 0′ 0′ ( Γ )∑niΓi )∑niΓi (36) i i m m 2 02 (-2( ∑nikiRi)+B(∑kiRizi)Γ)+ The function ∆Γ , the difference in surface concentrations, i i (47) m m 0 is defined by 8B(∑nikiRizi)( ∑kiRizi)Γ i i ) nΓ′′ - nΓ′ ∑ i i ∑ i i The initial conditions are ∆u(t)0) )u0and ∆Θ (t)0) ) i i ∆Γ ) (37) 0, i.e., no reaction before the potential build-up induced by 2 the flash. The relaxation amplitudes are then given by

m To make the concentration term ∆Γ dimensionless, the λ1+2( ∑nikiRi) function ∆Θ is introduced, which corresponds to the i fractional coverage of the surface: u1) (48) m nikiRizi ∆Γ ∑ ∆Θ ) (38) i 0 Γ m λ2+2( ∑nikiRi) With a combination of eqs 33 and 38, the flux equation i u2) (49) (eq 23) can be rewritten as m ∑nikiRizi m 0 m 0 i Ji)- 2kiRiΓ∆Θ -kiRiΓzi∆u (39) m -∑nikiRizi and the differential equation for ∆Θ is derived from eqs 23, i 37, and 38 and amounts to b1) u0 (50) λ1-λ2 d∆Θ 1 ) niJi (40) Since the membrane is isolated toward the aqueous phase, dt 0∑ Γ i it follows that b2)- b1.

or upon substitution of eq 39 for Ji Literatur eCited (1) Escher, B. I.; Behra, R.; Eggen, R. I. L.; Fent, K. Chimia 1997 ,51 , d∆Θ m m 915 -921. )- 2( ∑nikiRi)∆Θ -(∑nikiRizi)∆u (41) (2) van Wezel, A.; Opperhuizen, A. Crit. Rev. Toxicol. 1995 ,25 , 255 - dt i i 279.

VOL. 33, NO. 4, 1999 / ENVIRONMENTAL SCIENCE & TECHNOLOGY 9 569 – 98 – Appendix B

(3) Verhaar, H. J. M.; van Leeuwen, C. J.; Hermens, J. L. M. (26) Schellenberg, K.; Leuenberger, C.; Schwarzenbach, R. P. Environ. Chemosphere 1992 ,25 , 471 -491. Sci. Technol. 1984 ,18 , 652 -657. (4) Escher, B. I.; Snozzi, M.; Ha¨berli, K.; Schwarzenbach, R. P. (27) Miyoshi, H.; Nishioka, T.; Fujita, T. Biochim. Biophys. Acta 1987 , Environ. Toxicol. Chem. 1997 ,16 , 405 -414. 891 , 194 -204. (5) Terada, H. Environ. Health Perspect. 1990 ,87 , 213 -218. (28) Escher, B. I.; Westall, J. C.; Schwarzenbach, R. P. Manuscript in (6) McLaughlin, S. G. A.; Dilger, J. P. Physiol. Rev. 1980 ,60 , 825 - preparation. 863. (7) Finkelstein, A. Biochim. Biophys. Acta 1970 ,205 ,1 -6. (29) Wolfram, S. Mathematica ; Wolfram Research: Champaign, IL, (8) Crielaard, W.; van Mourik, F.; van Grondelle, R.; Konings, W. N.; 1996. Hellingwerf, K. J. Biochim. Biophys. Acta 1992 ,1100 ,9 -14. (30) LeBlanc, O. H. J. Membr. Biol. 1971 ,2, 227 -251. (9) Escher, B. I.; Snozzi, M.; Schwarzenbach, R. P. Environ. Sci. (31) Cohen, F. S.; Eisenberg, M.; McLaughlin, S. J. Membr. Biol. 1977 , Technol. 1996 ,30 , 3071 -3079. 37 , 361 -396. (10) Escher, B. I.; Schwarzenbach, R. P. Environ. Sci. Technol. 1996 , (32) McLaughlin, S.; Eisenberg, M.; Cohen, F.; Dilger, J. In Frontiers 30 , 260 -270. of Biological Energetics ; Dutton, P. L., Leigh, J. S., Scarpa, A., (11) Benz, R.; McLaughlin, S. Biophys. J. 1983 ,41 , 381 -398. Eds.; Academic: New York, 1978; Vol. 2, pp 1205 -1214. (12) Kasianowicz, J.; Benz, R.; McLaughlin, S. J. Membrane Biol. 1984 , 82 , 179 -190. (33) Gutknecht, J.; Tosteson, D. C. Science 1973 ,182 , 1258 -1261. (13) Benz, R. Biophys. J. 1988 ,54 ,25 -33. (34) Crofts, A. R.; Meinhardt, S. W.; Jones, K. R.; Snozzi, M. Biochim. (14) Kasianowicz, J.; Benz, R.; McLaughlin, S. J. Membr. Biol. 1987 , Biophys. Acta 1983 ,723 , 202 -218. 95 ,73 -89. (35) Dilger, J.; McLaughlin, S.; McIntosh, T.; Simon, S. Science 1979 , (15) Barstad, A. W.; Peyton, D. H.; Smejtek, P. Biochim. Biophys. 206 , 1196 -1198. Acta 1993 ,1140 , 262 -270. (36) Packham, N. K.; Berriman, J. A.; Jackson, J. B. FEBS Lett. 1978 , (16) Ba¨uerle, H.; Seelig, J. Biochemistry 1991 ,30 , 7203 -7211. 89 , 205 -210. (17) Seelig, J.; Ganz, P. Biochemistry 1991 ,30 , 9354 -9359. (18) Birrell, G. B.; Sistrom, W. R.; Griffith, O. H. Biochemistry 1978 , (37) Saarikoski, J.; Viluksela, M. Arch. Environ. Contam. Toxicol. 1981 , 17 , 3768 -3773. 10 , 747 -753. (19) Casadio, R.; Venturoli, G.; Melandri, B. A. Eur. J. Biophys. 1988 , (38) Saarikoski, J.; Viluksela, M. Ecotoxicol. Environ. Saf. 1982 ,6, 16 , 243 -253. 501 -512. (20) Crofts, A. R.; Wraight, C. A. Biochim. Biophys. Acta 1983 ,726 , (39) Miyoshi, H.; Fujita, T. Biochim. Biophys. Acta 1988 ,935 , 312 - 149 -185. 321. (21) Escher, B. I. Ph.D. thesis, Swiss Federal Institute of Technology, (40) Howe, G. W.; Marking, L. L.; Bills, T. D.; Rach, J. J.; Mayer Jr., Zu¨ rich, Switzerland, 1995. F. L. Environ. Toxicol. Chem. 1994 ,13 ,51 -66. (22) Junge, W.; Jackson, J. B. In: Photosynthesis: Energy Conversion (41) McKim, J. M.; Nichols, J. W. In: Aquatic Toxicology: Molecular , by Plants and Bacteria ; Govindjee, Ed.; Academic Press: New Biochemical ,and Cellular Aspects ; Malins, D. D., Ostrander, G. York, 1982; Vol. 1, pp 589 -646. K., Eds.; Lewis: Boca Raton, FL, 1994; pp 469 -519. (23) Melandri, B. A.; Mehlhorn, R. J.; Packer, L. Arch. Biochem. Biophys. 1984 ,235 ,97 -105. (24) Albert, A.; Serjeant, E. P. The determination of ionization Received for review May 27, 1998. Revised manuscript re- constants ; Chapman and Hall: London, UK, 1984. ceived November 4, 1998. Accepted November 12, 1998. (25) Schwarzenbach, R. P.; Stierli, R.; Folsom, B. R.; Zeyer, J. Environ. Sci. Technol. 1988 ,22 ,83 -92. ES980545H

570 9 ENVIRONMENTAL SCIENCE & TECHNOLOGY / VOL. 33, NO. 4, 1999 – 99 – Acknowledgments

My first acknowledgment is dedicated to Dr. Beate Escher who supervised the development of the present work and to Prof. René Schwarzenbach who accompanied me as referent of this thesis. Both shared satisfaction when I succeeded during the curse of this work and took their responsibility when it came to a stand still. I am in indebted to Prof. Urs Fringeli who enabled my introduction into the tech- nique of SBSR FTIR spectroscopy and who co-examined this work. I am also indebted to Dr. Rolf Altenburger who contributed conceptually in the treatment of the mixture toxicology results and who also co-examined this work. I appreciate very much the experimental contribution of Olof Kühnholz with his diploma thesis. I would like to thank Mario Snozzi, Nina Schweigert and Peter Looser for the good atmosphere in our cooperation. I would like to acknowledge those who assisted me in various parts of my work: Martin Schwarz, Jürg Mühlemann, Caroline Stengel, Madeleine Langmeier, Annemarie Mezzanotte and the "Lehrlingslabor", René Schönenberger, David Kistler, Monika Zemp, Philippe Périsset And also acknowledge those who kept me from working from time to time: Béatrice, Kai, Stefan, Werner, Dieter, Michael, Jörg, Markus U., Thomas B., Claudia, Cédric, Johanna, Thomas H., Markus M., Andi, Csaba, Bianca, Angela, Torsten, Martin, Marianne, Luc, Andrea, Chris, Fabio, Erika, Zach.

The last verse is dedicated to Ursula Bollens.

Glücklich allein Ist die Seele, die liebt.

J. W. Goethe

– 100 – Curriculum vitae

Education 10.10.1968 Born Zürich, Switzerland 1975–1982 Primary school in Thalwil 1982–1985 Middle school (Sekundarschule) Thalwil 1985–1989 High School, «Mathematisch naturwissenschaftliches Gymnasium» Rämibühl, Zürich 1990–1996 Studies in environmental sciences at the Swiss Federal Institute of Technology Zürich (ETHZ) Specialization in microbiology and aquatic systems Diploma Thesis on: «Energy metabolism of three eubacteria – Investigation on a Na+ dependent ATP synthase» Supervised by Dr. G. Kaim, and Prof. Dr. P. Dimroth, Institute of Microbiology, ETHZ 1997–2001 Doctoral thesis at the Swiss Federal Institute of Technology Zürich (ETHZ) and the Swiss Federal Institute for Environmental Science and Technology (EAWAG)

Teaching and teaching assistant 1993 Instructor at the exhibition on «Genetic engineering – pro & contra» Museum of natrual history, Basel Locher, Brauchbar und Partner AG, Basel 1998–1999 Teaching assistant in organic environmental chemistry Swiss Federal Institute of Technology (ETH), Zürich 1999–2000 Teaching chemistry in high school «Gymnasium Im Lee», Winterthur

Employment 1993–1995 Computer assistant on Apple Macintosh at the Swiss Federal Institute of Technology Zürich (ETHZ) 1995–1996 Developement and implementation of data base on clients and patents with integrated due-date control. Hunziker, Patent attorney, Zürich (my fathers office).

– 101 –