Search

This paper describes methods for the biosynthesis and chemical determination of polyfluorinated 4-hydroxyanilines. 19F NMR was routinely used to determine recoveries and metabolite patterns. Male Wistar rats exposed to ortho- and/or meta-polyfluorinated anilines were shown, by 19F NMR analysis of urine samples, to excrete a large percentage of the administered dose as the corresponding 4-aminophenyl sulphate. Using a Bio-Gel P2 column, this metabolite could be separated from the other metabolites and from (nearly) all urine contaminants. Treatment of the pooled fractions of the Bio-Gel P2 column with arylsulphatase resulted in hydrolysis of the 4-aminophenyl sulphate, giving rise to the corresponding polyfluoro-4-hydroxyaniline. The yields of this biosynthetic pathway varied between 10 and 65% depending on the substitution pattern. The yield is mainly determined by the extent to which the in-vivo biotransformation results in the desired 4-aminophenyl sulphate and its separation from other urine metabolites on the Bio-Gel P2 column. The amounts of polyfluoro-4-hydroxyanilines prepared in this way were sufficient to develop a chemical assay for their determination and quantification in biological samples in the μM range. It was demonstrated that polyfluoro-4-hydroxyanilines can be measured using the assay described for nonfluorinated 4-hydroxyaniline.

The present thesis focuses on two food borne compounds whose toxicity may be influenced by the gut microbiota. These compounds are daidzein and zearalenone (ZEN), both known to exert estrogenic effects. The isoflavone daidzein represents an important bioactive food constituent in soy products which exerts estrogenicity due to structural similarity to the natural hormone 17β-estradiol (E2) (Niu et al., 2018). The mycotoxin zearalenone is a well-known example of a toxin produced by fungi growing on contaminated crops, showing potential estrogenicity (Drzymala et al., 2015). Both daidzein and ZEN can be metabolized by gut microbiota to metabolites with higher estrogenic potential than their parent compounds, being S-equol (Mayo et al., 2019) and α-zearalenol (α-ZEL) (Zinedine et al., 2007), respectively. Exposure to these microbial metabolites may add to the effects resulting from exposure to the respective parent compounds. The aim of the present thesis was to characterize the role of gut microbial metabolism in the toxicity of daidzein and zearalenone (ZEN) by including gut microbiota in physiologically based kinetic (PBK) models and applying these models for quantitative in vitro to in vivo extrapolations (QIVIVE). The results obtained provide proofs-of-principle for application of this novel approach methodology (NAM) for alternatives in animal testing, characterizing the consequences of the metabolism by the gut microbiota for toxicity of foodborne chemicals in the host without a need for in vivo studies in experimental animals or human intervention studies.

Trace metals, such as mercury (Hg) and lead (Pb), have been reported to be immunotoxic to animals, however, relatively little is known about their effects and modes of action for avian species, especially waterfowl, which are frequently linked to the transmission of avian influenza infections. Given the fact that wild waterfowl are natural reservoirs for avian influenza viruses, impaired immune competence and pathogen resistance due to trace metal exposure in wild waterfowl may raise the risks of infections and might result in avian influenza outbreaks. Therefore, it is crucial to investigate the modulatory effects of trace metals, namely Hg(II) and Pb(II) in this thesis, on the immune responses of waterfowl. The objective of this thesis was to investigate whether realistic environmental levels of Hg(II) and Pb(II) exposure could modulate the immune responses in waterfowl upon a viral-like immune challenge and to explore the underlying modes of action. With the stepwise approach, we were able to compare the effects of Hg(II) and Pb(II) in the field study with the controlled in vivo and in vitro studies. The major effects of Hg(II) and Pb(II) were promoting inflammation and impairing B-lymphocyte functions upon a viral-like immune challenge. This disturbance of the avian immune system potentially hampers the immune response upon viral infections, which might cause a higher risk of infection for individuals and even prevalence of diseases in the population as a whole. Altogether, this thesis indicates that even at low environmentally relevant levels, Hg(II) and Pb(II) exposure affected the immune responses in waterfowl upon a viral-like immune challenge, mainly by promoting inflammation and impairing B-cell functions. The current study emphasises the potential need to include immunomodulatory effects on birds in the environmental risk assessment of chemicals. This is relevant not only for the protection of wildlife populations, but also towards sustaining environmental and human health.

Pesticide use is essential for control of pests in horticultural crops and adequate production of food supplies for the ever increasing world population as well as for control of vector-borne diseases. Currently, OPs are among the most widely used pesticides in agriculture, and their residues have been found in various foods. Regarding human risk assessment, the points of departure (PODs) to set health based guidance values have been derived from data from animal studies, mostly on OP-induced acetylcholinesterase (AChE) inhibition. The use of such animal data is undesirable due to economical, ethical and scientific limitations. The main aim of this thesis was to develop non-animal approaches for the hazard assessment of OPs to be applied in human risk assessment and application of these novel approach methods (NAMs) to predict the potential of the OPs to inhibit AChE in vivo upon acute exposure, also taking combined exposure into account. Furthermore, this thesis aimed to obtain data on the occurrence of OPs on commonly consumed vegetables in Kenya, as such data, and related estimations of human exposure, are lacking. In the first step, commonly consumed vegetables in Kenya were screened for OPs (and carbamates) and the accompanying exposure and related health risks assessed. A total of 90 samples were analyzed by liquid chromatography/high-resolution tandem mass spectrometry. Residues of acephate, chlorpyrifos (CPF), methamidophos, omethoate and profenofos (PFF) were found in 22% of the samples, at levels ranging from 10 to 1343 μg/kg. The EU MRLs for these OPs was exceeded in 21%, 10%, 8% and 22% of the samples of French beans, kales, spinach, and tomatoes, respectively. CPF levels on spinach were calculated to result in an acute Health Quotient (HQ) of 3.3 and 2.2 for children and adults, respectively, upon high consumption, implying that potential health risks with respect to acute dietary exposure cannot be excluded. Calculations for chronic dietary exposure showed that all chronic HQs were below 1. The Health Index (HI) for the pesticides, i.e. the sum of the chronic HQs, was 0.54 and 0.34 for children and adults, respectively, implying no direct health concern upon chronic average consumption of the vegetables assessed..In the next step, a combined in vitro-in silico approach to predict AChE inhibition by the OP profenofos (PFF) in rats and humans by applying physiologically based kinetic (PBK) modelling-facilitated quantitative in vitro to in vivo extrapolation (QIVIVE) was developed. A PBK model was developed for both species. PBK model parameter values for PFF conversion to 4-bromo-2-chlorophenol (BCP; detoxification pathway) were derived from in vitro incubations with liver microsomes, liver cytosol, and plasma from rats (catalytic efficiencies of 1.1, 2.8, and 0.19 ml/min/mg protein, respectively) and humans (catalytic efficiencies of 0.17, 0.79, and 0.063 ml/min/mg protein, respectively), whereas other chemical-related PBK model parameter values were derived using in silico calculations. The rat PBK model was evaluated against literature data on urinary excretion of conjugated BCP. Concentration-dependent inhibition of rat and human AChE by PFF was determined in vitro and these data were translated with the PBK models to predicted dose-dependent AChE inhibition in rats and humans in vivo. Comparing predicted dose-dependent AChE inhibition in rats to literature data on PFF-induced AChE inhibition revealed an accurate prediction of in vivo effect levels. Comparison of rat predictions (BMDL10 of predicted dose-response data of 0.45 mg/kg bw) and human predictions (BMDL10 of predicted dose-response data of 0.01 mg/kg bw) suggests that humans are more sensitive than rats, being mainly due to differences in kinetics. Results obtained demonstrate that in vivo AChE inhibition upon acute exposure to PFF was closely predicted in rats, indicating the potential of this NAM in chemical hazard assessment.In order to consider OP-mixture effects, this study assessed combination effects of PFF and chlorpyrifos’ (CPF) toxic metabolite chlorpyrifos oxon (CPO) on AChE inhibition in vitro, and the effects of CPF and CPO on PFF detoxification by human liver microsomes, cytosol, and plasma. PFF was 199-fold less potent than CPO in AChE inhibition, and combined effects of CPO and PFF followed principles of dose addition. CPF and CPO affected PFF detoxification to BCP in a non-competitive manner at high concentrations that are not expected to be reached in vivo. The PBK model of PFF was extended to include the description of CPF and CPO kinetics. PFF concentrations in the model were expressed as CPO-equivalents allowing prediction of total internal CPO-equivalents upon combined exposure to PFF and CPF. Different exposure scenarios were applied for which PBK model-predicted internal (unbound) CPO-equivalents were compared to CPO concentrations affecting AChE activity in vitro, to assess whether in vivo effects on AChE activity are expected. These results reveals that AChE inhibition upon combined exposure to these OPs follows principles of dose addition, and that no effects of CPF on PFF detoxification are expected.Finally, this study determined human interindividual differences in PFF detoxification based on the combined in vitro-in silico approach. To that end, PFF conversion to BCP was studied in in vitro incubations with liver S9 fractions of 25 different Caucasian donors and plasma samples of 25 different Caucasian donors. The obtained information on the variation in the in vitro kinetic constants Vmax and Km was applied in the PBK model using a Monte Carlo simulation modelling approach. Analysis of the predicted internal maximum concentrations of the simulated population indicated a 1.4- and 1.5-fold difference between the geometric mean (GM) and the 90th and 99th percentile, respectively, of the simulated population, being smaller than the default uncertainty factor accounting for human variability in toxicokinetics of 3.16. Concentration-response curves for inhibition of human AChE determined in vitro were translated with the PBK model to predicted dose-dependent AChE inhibition in humans in vivo. In line with the differences in Cmax values, the 99th percentile of the simulated population was predicted to be 1.5-fold more sensitive to PFF-induced AChE inhibition than the GM of the simulated population based on the BMDL10 values derived from predicted dose-response data for the virtual population. This predicted BMDL10 of 0.0057 mg/kg bw (99th percentile of population) is comparable to an ARfD reported by EFSA (0.005 mg/kg bw) based on a German evaluation, and the ARfD reported by EPA (0.002 mg/kg bw), but much lower than the ARfD adopted by JMPR (1 mg/kg bw). In conclusion, integrating PBK modelling with Monte Carlo simulations using human in vitro data provides a powerful strategy to quantify human interindividual variation in kinetics, which can be used to refine the hazard characterization of OPs.Altogether, the present thesis presents a NAM-based hazard assessment of OP-induced AChE inhibition upon single or combined exposure to CPF and PFF. Since a generic bottom-up PBK modelling approach was applied, the approach can be easily extended to include more OPs allowing a mechanism-based hazard and risk assessment of combined OP exposure.&nbsp

Glycation products comprise a heterogenous group of compounds which can be formed upon thermal processing of food products and include advanced glycation end products (AGEs), such as carboxymethyllysine, and their precursors such as the Amadori product fructoselysine and α-dicarbonyl compounds like methylglyoxal, glyoxal and 3-deoxyglucosone. Besides being present in exogenous dietary sources, glycation products can also be produced endogenously inside the human body. AGEs have been associated with the development of multiple adverse health effects, such as diabetes and cardiovascular diseases, but whether exposure to exogenous dietary AGEs and their precursors contributes to these effects remains debated. In this thesis, an overview was presented of the toxicokinetics and toxicodynamics of exogenous and endogenous AGEs and their precursors which are of relevance to consider when evaluating whether exposure to exogenous dietary glycation products can contribute to the adverse health effects associated with AGEs. It was concluded that better characterization of the tested AGEs and precursors, and to distinguish between their low molecular mass (LMM) and high molecular mass (HMM) forms, as well as quantification of the exposure from endogenous formation relative to that resulting from dietary intake are all of importance to provide a definite conclusion on whether dietary exposure to glycation products can contribute to the adverse health effects associated to AGEs. In addition, it was concluded that part of the AGEs and their precursors remain unabsorbed upon dietary exposure and can reach the colon. Consequently at least part of the AGEs and their precursors can interact with the gut microbiota in a bi-directional manner. This can affect the toxicokinetics of the AGEs and their precursors but on the other hand can also result in exerted toxicodynamic effects by altering the gut microbiota composition and possibly function. The aim of this project was to characterize these interactions of the gut microbiota with selected dietary glycation products and vice versa, and to characterize inter- and intraindividual differences in gut microbial reactions using the Amadori product fructoselysine and the AGE carboxymethyllysine as model compounds. This was achieved by the application of an in vitro model where collected individual human fecal samples were incubated with the substrate of interest. Michaelis-Menten kinetic parameters could be obtained, in addition to information on metabolite formation (short chain fatty acid; SCFA). In addition to interindividual differences, intra-individual differences were quantified as well of microbial degradation of the AGE carboxymethyllysine and its precursor fructoselysine. The gut microbiota is a dynamic ecosystem and thus intraindividual temporal differences are of possible relevance for its functioning. The results revealed pronounced inter- and intraindividual variation in both carboxymethyllysine and fructoselysine degradation while the two degradation rates were not correlated (R2=0.08), showing the need to evaluate toxicokinetics for individual AGEs and precursors instead of evaluating them on a group level. In addition, multiple bacterial genera were correlated with the individuals’ carboxymethyllysine and fructoselysine degradation activities, corroborating that showing that the potential to convert fructoselysine and carboxymethyllysine may originate from multiple bacterial genera. The effect of different dietary exposures on microbial degradation was evaluated by comparing functional differences in microbial fructoselysine degradation between breast-fed (BF) and formula-fed (FF) infants, in view of their different exclusive diets and consequent different dietary fructoselysine exposures. This was done by evaluating a publicly available metagenome dataset analysis with metagenome assembled genomes (MAGs) from infant fecal samples which showed that genes involved in microbial fructoselysine degradation were present in multiple taxa in both BF and FF infant fecal samples but were higher prevalent in fecal samples from FF infants compared to the BF infants. Further collection of individual fecal samples from exclusively BF and FF infants showed that both groups were able to degrade fructoselysine anaerobically but fecal samples from the FF infants resulted in a significantly higher degradation activity compared to the BF infants. This indicated that the infant gut microbiota adapts towards dietary fructoselysine exposure. This dynamic adaptive aspect of the gut microbiota was also observed in mice which were exposed to a heated diet high in AGEs which resulted in an altered gut microbiota composition compared to mice exposed to the control diet. Exposure to the heated diet high in AGEs followed by the control diet (i.e. the switch group) showed that the altered gut microbiota composition was reversible and adapted to the dietary exposure. This reversibility was also observed for the accumulation of the tested AGEs in plasma, kidney and liver (as analyzed in their protein-bound and free form), a result that also pointed to at least partial bioavailability of the dietary AGEs and/or their precursors that appeared to be enriched in the heated chow diet.Overall, it can be concluded that the bi-directional relation of the gut microbiota with exogenous AGEs and their precursors is of relevance when evaluating their toxicokinetic and toxicodynamic characteristics. To accurately evaluate the hazards and risks of dietary exposure to AGEs and precursors, better characterization and quantification of the applied test substances and biological outcomes, and also of the exposure from endogenous formation relative to that resulting from dietary intake, are essential.&nbsp

The use of herbal supplements for sexual improvement and recreational purposes is a rather common practice in Ghana. Although the high accessibility, acceptability and assumed benefits of herbal products has promoted growth in the herbal industry, the presence of adulterants and other contaminants in these products is a major health concern.In this thesis, selected herbal supplements were initially screened using the PDE-Glo bioassay to determine their inhibition potentials against PDE-5 enzyme activity, the mode of action underlying their effect on erectile function. Results based on the bioassay revealed that about 90% of the samples were able to inhibit PDE-5 enzyme activity at various degrees, with estimated intakes of 25 samples (62.5%) pointing at daily doses higher than 25mg sildenafil equivalents and 9 (22.5%) of these even at doses higher than the maximum recommended daily intake of 100mg sildenafil equivalents. Further investigations were carried out to confirm if the observed effects were due to inherent plant constituents or merely the result of added synthetic PDE-5 enzyme inhibitors.Further chemical analysis revealed significant amounts of sildenafil and other approved PDE-5i in 13 out of the 40 supplements. Based on data from the chemical analysis, the estimated daily intakes (EDIs) of six out of the 13 even resulted in EDIs above 100 mg sildenafil equivalents. Only one sample contained a natural ingredient (icariin) potentially able to inhibit PDE-5 enzyme activity, but its concentration (0.013 mg/g) was too low to explain the observed response in the bioassay. The estimated concentrations of PDE5 inhibitors in 35 supplements as derived from the bioassay results, were in line with those of the LC-MS/MS analysis, while discrepancies were observed for 5 supplements. A tiered approach for one of the latter supplements using the PDE-Glo bioassay to select the positive fraction and further examination with LC-MS/MS and 1H-NMR revealed the presence of hydroxythiohomosildenafil, a sildenafil analogue not yet included in the LC-MS reference library. These results highlight the importance of the bioassay to detect unknown PDE-5 inhibitors.To further investigate the safety of the herbal supplements, the presence of contaminants such as polycyclic aromatic hydrocarbons (PAHs), dioxins (polychlorinated dibenzo-p-dioxins and dibenzofurans (PCDD/Fs)) and dl-PCBs possibly introduced into the products as a result of contaminated raw materials or during processing were also investigated. Selected supplements were initially screened using the DR CALUX® bioassay in order to identify the presence of stable (acid-silica clean-up and 48 hrs exposure) and less stable (without acid-silica clean-up and 4 hrs exposure) aryl hydrocarbon receptor (AhR)-agonists. The results revealed that the selected supplements contained only low levels of stable AhR-agonists like dioxins and dl-PCBs, while some contained less stable AhR-agonists like PAHs.Ten supplements selected for confirmation with GC-HRMS analysis contained PCDD/Fs and dl-PCBs at levels varying from 0.01 to 0.19 pg toxic equivalents (TEQ)/g, while the level of the sum of 4 polycyclic aromatic hydrocarbons (Σ4PAHs), representing less stable AhR agonists, ranged from not detected (ND) to 25.5 ng/g. These concentrations were in line with the responses observed in the DR CALUX® bioassay. The concentration of PCDD/Fs and dl-PCBs corresponded to EDIs ranging from 0.001 to 0.12 pg TEQ/kg bw/week for a 70 kg bw consumer, which is below the established tolerable weekly intake (TWI) of 2 pg TEQ/kg bw/week, thus indicating low concern for consumers’ health. Similarly, the EDIs based on the detected Σ4PAHs in supplements ranged from 7.2 to 111 ng/day, or 0.1 to 1.6 ng/kg bw/day, which compared to carcinogenic effects in animals corresponds to margin of exposure (MOE) values above 10,000, indicating a low health concern.Finally, the presence of pyrrolizidine alkaloids (PAs) in the 40 supplements as a result of contaminated raw materials and/or the unintentional co-harvesting of raw materials with PA containing weeds, was examined using liquid chromatography mass spectrometry (LC/MS). Up to 17 different PAs were identified in 19 out of the 40 samples. In decreasing order, the most detected PAs were rinderine (in 15 samples), its N-oxide (10), intermedine (11), its N-oxide (11), lycopsamine (8) and its N-oxide (5). The total sum of PAs in samples ranged from 0.005 to 3.2 µg/g, resulting in EDIs from 0.01 to 12 µg/day, corresponding to exposures from 0.0002 to 0.2 µg/kg bw/day for a 70 kilogram bodyweight person. When compared to a BMDL10 of 237 µg/kg bw/day for liver tumours in rats treated with riddelliine, the margin of exposure (MOE) values ranged from 1200 to 1,400,000 with 8 samples showing MOE values below 10,000, thus indicating a health concern upon long-term regular use of these 8 supplements.The presence of contaminants and especially added adulterants in the selected supplements compromises their purity and safety, the latter depending on the concentrations present. These findings are indicative of what is pertaining to the numerous herbal products available on the Ghanaian market. The practical approaches employed in this thesis, using a combination of a bioassay and chemical analysis to determine the presence and concentrations of added pharmaceuticals and undesirable contaminants is expected to be applicable in most laboratories. These approaches can be applied in investigating other herbal products available on the Ghanaian market.The adulteration of herbal products with active pharmaceutical ingredients and the presence of contaminants continues to be a problem for regulatory agencies globally including Ghana. In the present thesis, the presence of intentionally added adulterants in the selected herbal products shed a light on on-going practices in Ghana and provides some answers to the numerous reported casualties and deaths of consumers of sexual enhancing herbal products sold on the Ghanaian markets.&nbsp

Mineral oil aromatic hydrocarbons (MOAH) have been presented as a modern emerging food safety concern by the European Food Safety Authority (EFSA). MOAH is defined as a chromatographic fraction in the analysis of mineral oil hydrocarbons (MOH) in food and does not represent well defined substances or mixtures. The sources of foodborne MOAH include environmental contamination and intentional uses of mineral oils in consumer products or production processes. Some non-substituted PAHs are known to be genotoxic carcinogens and have been extensively studied, however the majority of the compounds represented by MOAH are highly alkylated aromatic hydrocarbons for which the toxicological data are scarce. Potential concerns for genotoxicity, carcinogenicity and developmental toxicity upon MOAH exposure have been raised. However, the lack of toxicological data on related alkyl substituted polycyclic aromatic hydrocarbons (PAHs) seriously hampers a proper hazard assessment and therefore challenges the risk assessment of MOAH.The present thesis aims at (1) providing data on oxidative metabolism and resulting toxicities of a series of the selected non-substituted and alkyl-substituted aromatic hydrocarbons using in vitro assays and (2) developing a physiologically-based kinetic (PBK) modeling- facilitated quantitative in vitro-in vivo-extrapolation (QIVIVE) approach for prediction of in vivo toxicity of selected PAH. The latter is needed to define data for risk assessment as a proof of principle for using new approach methodologies (NAMs). It was hypothesized (1) that alkylated PAHs will be metabolized preferably on the alkyl side chain rather than on the condensed aromatic rings which can be expected to facilitate their excretion rather than their bioactivation to potentially DNA-reactive metabolites, and (2) that the presence of an additional bay region like aromatic motif introduced by methyl substitution may affect the genotoxic potential of the PAHs.

Green tea, which is mainly manufactured from buds and leaves of Camellia sinensis, is one of the most widely consumed beverages around the world. Catechins are the most abundant bioactive constituents in green teas, amounting up to 30% of the total dry weight. (-)-Epigallocatechin gallate (EGCG), (-)-epigallocatechin (EGC), (-)-epicatechin gallate (ECG) and (-)-epicatechin (EC) are the four major catechins among others. These catechins are believed to be responsible for various beneficial health effects that have been ascribed to green tea consumption. Though the modes of action underlying these health-promoting effects can be complicated and have not been fully understood, the Kelch-like ECH-associated protein 1/Nuclear factor E2-related factor 2 (Keap1/Nrf2) regulatory network has been reported to play a role. In the present thesis, in vitro anaerobic fecal incubation was used to characterize the microbial conversion of green tea catechins EC and EGCG. Results obtained show phenylpropane-2-ols, phenyl-valerolactones, gallic acid, pyrogallol, phenyl-valeric acids as the major metabolites, with substantial interindividual differences. Moreover, intraindividual difference in the formation of the major EC colonic metabolite 5-(3’,4’-dihydroxyphenyl)-γ-valerolactone (3,4-diHPV) was also observed, though the interindividual difference was larger than intraindividual difference. Furthermore, quantitative microbiota characterization in the fecal samples, achieved by 16S rRNA analysis, revealed substantial differences in microbiota compositions among different individuals. Correlations between specific microbial abundance and formation of certain metabolites were established. The Nrf2 signaling activation by EGCG, EC and their major microbial metabolites were tested using reporter gene assays, RT-qPCR, and proteomics technic. Results show EGCG and the microbial metabolites 3,4-diHPV and pyrogallol can induce concentration-dependent induction of Nrf2 pathway. Moreover, metabolites 3,4-diHPV and pyrogallol exhibited better Nrf2-activation potency than their respective parent compounds, EGCG and EC. Finally, the thesis also developed a human PBK model for EGCG, with sub-models for its major microbial metabolites gallic acid and pyrogallol. The model enabled prediction of in vivo kinetics of EGCG, gallic acid and pyrogallol which allowed the translation of the obtained in vitro concentration-response curves from the U2OS-Nrf2 CALUX reporter gene assay to in vivo dose-response curves for Nrf2 activation, and comparison of these data to estimated daily intake levels. Results obtained reveal, by comparison to literature data on EGCG kinetics, that the developed PBK model could adequately predict in vivo time-dependent blood concentrations of EGCG after either a single or repeated oral administration(s) of EGCG under both fasting and non-fasting conditions. The predicted in vivo dose response curve revealed that at daily intake levels of green tea or EGCG supplements, the resulting blood Cmax of EGCG was in the sub-micromolar range, concentrations at which Nrf2 activation was shown to be limited. Moreover, blood Cmax values of gallic acid and pyrogallol upon intake of EGCG were predicted to be less than 1.5% of the Cmax of EGCG, indicating that despite their higher potential for Nrf2 activation, their contribution to the overall systemic Nrf2 pathway induction upon EGCG exposure is expected to be limited. In contrast, concentrations of these metabolites in the intestinal tract may reach levels that are, expressed in EGCG equivalents for Nrf2 induction, higher than that of EGCG, and high enough to activate Nrf2 gene transcription. Taken together, combining in vitro data with a human PBK model allowed the prediction of a dose-response curve for EGCG induced Nrf2-mediated gene expression in humans, and provided insight into the contribution of gut microbial metabolites to this effect. It also provided a proof-of-principle for a NAM to study the in vivo effects of bioactive phytochemicals without the need for human intervention studies.

The aim of the present thesis was to investigate whether new approach methodologies (NAMs) and especially physiologically based kinetic (PBK) model-based reverse dosimetry using in vitro assays and in silico data, in combination with Monte Carlo simulations or the toxic equivalency factor (TEF) methodology, can be used to derive points of departure (POD) and chemical-specific adjustment factors (CSAFs) for characterizing inter-ethnic, inter-species, and inter-individual differences in toxicokinetics and red blood cell (RBC) acetylcholinesterase (AChE) inhibition by organophosphate (OP) pesticides upon single acute oral exposure.Chapter 1 provided an overview of relevant background information, including an introduction of chlorpyrifos (CPF), diazinon (DZN), and profenofos (PFF) as model compounds, their absorption, distribution, metabolism and excretion (ADME) and toxicity profiles, as well as their reported PODs and health-based guidance values (HBGVs). The aim of the thesis and the NAMs that were applied in the thesis (PBK model-based reverse dosimetry solely, or combined with Monte Carlo simulations or the TEF methodology) were also presented. At the end, the general outline of the thesis was provided.Chapter 2 assessed the inter-ethnic differences in toxicokinetics of CPF and its resulting RBC AChE inhibition between the Chinese and Caucasian populations using PBK model-facilitated reverse dosimetry based on in vitro and in silico data. The results obtained revealed a marked inter-ethnic difference in toxicokinetics of CPF, with a slower CPF bioactivation and faster detoxification of its active metabolite chlorpyrifos oxon (CPO) in the Chinese than in the Caucasian population. This could be explained by the inter-ethnic variation in enzyme profiles and enzyme activities involved in biotransformation of CPF and CPO. The differences resulted in 5- to 6-fold lower predicted POD values of CPF for the Caucasian than the Chinese population, reflecting a lower risk of CPF for the Chinese than the Caucasian at similar levels of exposure. Besides, because of inter-ethnic variation in toxicokinetics of CPF, 2-fold lower cumulative urinary 3,5,6-trichloro-2-pyridinol (TCPy) biomarker levels were observed in Chinese than in Caucasian at similar dose levels, implying the importance of taking possible inter-ethnic differences into account when characterizing CPF exposure for different ethnic groups based on their urinary biomarker levels.Chapter 3 focused on another OP and investigated the inter-species differences between rat and human in PFF-induced RBC AChE inhibition by PBK model-based reverse dosimetry. The obtained results show the predicted POD of human to be 45-fold lower than that of rat, indicating human to be more sensitive than rat to the inhibition from equivalent oral doses of PFF. Such differences appeared mainly due to marked interspecies differences in toxicokinetics of PFF, with rat being more efficient in hepatic and plasma detoxification of PFF than human, while the difference in toxicodynamics appeared limited as reflected by a similar potency of PFF to inhibit rat RBC AChE and human recombinant AChE in vitro. In conclusion, this chapter provided another proof-of-principle that integrating in vitro toxicity data and PBK model-based reverse dosimetry is a promising strategy for non-animal-based safety assessment of OP pesticides.Chapter 4 further studied the applicability of PBK model-based reverse dosimetry to predict the combined inhibitory effect of DZN and its active oxon metabolite diazoxon (DZO) on RBC AChE inhibition by integrating the TEF approach into the model. Use of the TEF coded PBK model enabled calculation of the external exposure dose of DZN that would produce a defined combined free DZN plus DZO concentration expressed in DZO equivalents in human/rat blood, and to translate the observed inhibition effects of DZO in the in vitro concentration-response curve expressed in free in vitro DZO equivalents to the in vivo situation. The obtained results show that DZN contributes substantially to the DZO equivalents, indicating it is of critical importance to take DZN into account in addition to the active metabolite DZO when conducting a risk assessment for DZN. Although no inter-species differences were observed in the in vitro concentration-response curves for rat or human RBC AChE inhibition by DZN or DZO, marked inter-species differences in toxicokinetics were observed, as rat displayed a faster metabolic rate for biotransformation of DZN and DZO than human, resulting in a 6-fold lower POD in human than in rat. In conclusion, this chapter provided a proof-of-principle for derivation of a point of departure (POD) in risk assessment for an OP pesticide for which both the parent OP and its oxon metabolite have the ability to contribute to RBC AChE inhibition.Chapter 5 characterized the inter-individual variation in toxicokinetics of CPF and its resulting RBC AChE inhibition. A CSAF for inter-individual differences in kinetics (HKAF) was quantified on the basis of the PBK model for CPF developed in Caucasian (Chapter 2), in combination with Monte Carlo simulations. To define the metabolic variation in the PBK model, two different approaches were developed and compared, namely a SupersomeTM cytochrome P450 (CYP)-based PBK model approach and a human liver microsome (HLM)-based PBK model approach. The results revealed that bioactivation of CPF exhibits biphasic kinetics due to distinct differences in Km values for CYPs involved in the CPF metabolic pathway, and these kinetic characteristics could be best identified by SupersomeTM CYP. In general, there is a good match between the two approaches, as no substantial differences were found in the predicted HKAF values and the BMDL10 values obtained by the two procedures. The predicted HKAF values for the 99th percentile obtained by the two approaches were slightly higher than the default uncertainty factor (UF) of 3.16, reflecting possible inadequate protection for extremely sensitive adult individuals by the default UF. Altogether, this chapter provided a proof-of-principle for assessing inter-individual variation in toxicokinetics of OP pesticides and its resulting RBC AChE inhibition, and the SupersomeTM CYP-based approach seems to be the preferred option for defining inter-individual differences in CYP-mediated kinetics.Chapter 6 provided an overview of the results and main findings of the present thesis, followed by an in-depth discussion of the results obtained, and the future perspectives. It was concluded that the present thesis generated proofs-of-principle for the use of NAM and especially PBK model-facilitated reverse dosimetry based on in vitro assays and in silico data, in combination with Monte Carlo simulations or the TEF methodology, to derive PODs and CSAF values for defining inter-ethnic, inter-species and inter-individual differences in toxicokinetics and the resulting RBC AChE inhibition following an OP pesticide acute single oral exposure. Altogether, the current work supports the application of such a NAM in regulatory safety decision making on compounds, and contributes to the implementation of the 3Rs (replacement, reduction, and refinement of experimental animal studies), thereby promoting the development of next generation risk assessment.&nbsp

DNA adduct formation upon exposure to genotoxic carcinogens is often referred to as a biomarker of exposure rather than as a biomarker of effect. In spite of this, increased levels of DNA adducts of a specific genotoxic carcinogen are generally assumed to increase the tumor incidences. However, the relation between the levels of DNA adducts formed and the levels of mutations or tumor formation is by no means well defined and may vary from one compound to another. This may in part be related to the fact that cells have quite efficient DNA repair systems, which may prevent the conversion of DNA lesions into mutations. At the present state-of-the-art DNA adduct persistence was found to be substantial for different genotoxic carcinogens including for example, DNA adducts derived from a pyrrolizidine alkaloid, aflatoxin B1 and some polycyclic aromatic hydrocarbons (Croy and Wogan 1981; Geacintov and Broyde 2017; Zhu et al. 2017), providing opportunities for adduct accumulation upon chronic exposure. The potential adduct accumulation may increase the chances of mutations and subsequent induction of tumors.The aim of the present thesis was to obtain better insight in the relative hazards and risks of DNA adducts formed by alkenylbenzenes, a group of compounds naturally occurring in many spices and herbs, by studying their DNA adduct formation, stability, and repair.Chapter 1 provided general information on estragole and safrole, their bioactivation resulting in DNA adduct formation, DNA repair, molecular dynamics (MD) simulations to study structural perturbations upon DNA adduct formation, and the current state-of-the-art on the risk assessment related to these genotoxic carcinogens. In Chapter 2 and 3, formation and repair of the major DNA adducts of estragole and safrole was quantified in different cell models. Results showed a limited DNA repair efficiency for both DNA adducts. Molecular dynamics simulations were used to investigate the potential conformation dependent (in)efficiency of repair of the major estragole and safrole DNA adducts. Results from molecular dynamics simulations revealed that conformational changes in double-stranded DNA upon formation of these adducts were small, providing a possible explanation for the restrained repair, which may require larger distortions in the DNA structure to activate recognition and subsequent repair. In Chapter 4, accumulation of the major DNA adducts derived from the alkenylbenzene estragole upon repeated exposure was investigated in HepaRG cells treated in repeated cycles of 2 h exposure and 22 hours repair. The results obtained showed accumulation of adducts at a rate of 17.53 adducts / 108 nts / cycle. This rate was converted to a rate expected at average human daily intake of estragole. Based on these data it was estimated that it would take 6-57 years intake at the estimated daily intake to reach levels of 10-100 adducts /108 nts, a level of DNA adducts reported at the BMD10 of the related alkenylbenzene methyleugenol. These findings revealed that the persistent nature of the major estragole DNA adducts may contribute to accumulation of substantial levels of DNA adducts upon prolonged dietary exposure. In Chapter 5, potential consequences of combined exposure to the proximate carcinogenic metabolites of the selected foodborne alkenylbenzenes safrole and estragole were evaluated in vitro and in silico. Results indicate that concentration addition adequately described the cytotoxic effects and no statistically significant differences were shown in the level of formation of the major DNA adducts upon combined as compared to single exposures. The absence of any interaction on DNA adduct formation was also predicted for combined exposure to estragole and safrole at normal dietary intake. The prediction revealed that the simultaneous presence of the two proximate carcinogens does not affect their DNA adduct formation. Chapter 6 presented a summary of the results obtained in this thesis and an overall discussion regarding to 1) considerations on the use of MD simulations to investigate the potential underlying mechanisms for the NER resistance of alkenylbenzene DNA adducts, 2) impact of the base sequence within which DNA adduct binding occurs, and the consequences for DNA structure distortion and the efficiency of NER mediated repair, 3) the consequences of DNA damage caused by estragole and safrole DNA adduct formation when not repaired, 4) accumulation of the persistent estragole DNA adducts, 5) potential combination effects affecting the DNA adduct formation, 6) implications for risk assessment.Altogether, the present thesis provides further insight in the relative hazards and risks of DNA adduct formation by elucidating the mode of action of estragole and safrole DNA adduct formation and repair in more details via combined use of in vitro and in silico novel testing strategies. The in vitro and in silico methods of the present thesis provide a way forward to study the accumulation of the DNA adducts at low repeated dose exposure levels to a further extent.

Mycotoxins are natural contaminants, unavoidable in food. In Mexico, the large consumption of maize, may place the population at risk, yet limited risk assessments of mycotoxins in Mexico are available (Chapter 1). This thesis assessed the risk of the main mycotoxins present in nixtamalized maize products consumed in Mexico City. To this end, a characterization of aflatoxin B1 (AFB1), fumonisin B1 (FB1), fumonisin B2 (FB2), and deoxynivalenol (DON) in samples collected from Mexico City was performed with an accompanying risk assessment. To enhance the use of mode-of-action based non-animal testing strategies in risk assessment of these mycotoxins, a physiologically based kinetic (PBK) model of AFB1 was developed. This enabled the use of a PBK model-based reverse dosimetry approach to predict the acute toxicity and genotoxicity of AFB1 in the liver using an in vitro–in silico testing strategy.The risk associated to the intake of AFB1 from nixtamalized maize was evaluated after analyzing its occurrence in nixtamalized maize samples from Mexico City (Chapter 2). A 4,5% of the samples contained AFB1 at levels above the limit of detection (1 ng/g) indicating a low frequency of AFB1 in the nixtamalized maize samples. The results obtained and additional AFB1 occurrence values from literature were combined with mean and P95 consumption values from literature. For a 70 kg body weight (bw) person, lower and upper bound exposure assessments resulted in estimated daily intakes (EDI) of 0.7–8.5 ng/kg bw/day, based on a mean maize consumption. Based on a P95 maize consumption, EDI values were 3.3–11.7 ng/kg bw/day. Corresponding Margin of Exposure (MOE) values amounted to 257-20 for the mean and 50-15 for the P95 consumers. The estimated increased cancer risks were 9-320 and 43-439 cases/106 individuals/lifetime of 75 years for the mean and P95 consumers, respectively. Altogether, continued risk management of AFB1 in Mexico City is needed.Given that mycotoxins can occur simultaneously in food, the levels of the mycotoxins most frequently encountered in maize, fumonisins (FB1+FB2) and DON, were also determined with an accompanying risk assessment. Chapter 3 presents the occurrence of FB1+FB2 and DON in the nixtamalized maize samples collected in Mexico City. It further analyses their distribution and the resulting EDI for Mexican consumers by a probabilistic approach using a two-dimensional Monte-Carlo simulation. The results revealed that for fumonisins 47% of the Mexican men and 30% of the Mexican women might exceed the provisional tolerable daily intake (PMTDI) of 2 µg/kg bw/day for fumonisins and for DON the high consumers (9% of men and 5% of women) would be exceeding the PMTDI of 1 µg/kg bw/day. The results raise a flag for risk managers in Mexico, to consider regulations and interventions to lower mycotoxin levels in nixtamalized maize for human consumption.To enhance the use of non-animal testing strategies in risk assessment, a generic PBK model of AFB1 was developed to predict in vivo toxicity by using a reverse dosimetry approach. Since PBK modelling-based reverse dosimetry enables quantitative in vitro in vivo extrapolation (QIVIVE), Chapter 4 presents an in vitro-in silico testing strategy to predict the doses causing acute liver toxicity of AFB1 in rats and humans. To this end PBK models for AFB1 in rats and humans were developed and evaluated. The models were further used to translate in vitro concentration–response curves for cytotoxicity in primary rat and human hepatocytes to in vivo dose–response curves using reverse dosimetry. From these data, the dose levels at which toxicity would be expected were obtained and compared to toxic dose levels from available rat studies and human case studies on AFB1 toxicity. The results showed that the in vitro–in silico testing strategy could predict dose levels observed to cause acute toxicity of AFB1 in rats and human in vivo. From this study, it was concluded that QIVIVE using PBK modelling-based reverse dosimetry can adequately predict AFB1 doses that cause acute liver toxicity in rats and human.The importance of genotoxicity in the mode of action of AFB1 made interesting to use this PBK model to predict its in vivo genotoxicity in liver. In Chapter 5, reported concentration-response curves for in vitro genotoxicity in hepatocytes were searched from literature. Suitable in vitro data, consisting of in vitro micronucleus (MN) data obtained in rat hepatocytes in vitro, were translated to an in vivo dose-response curve. Comparison of the predicted in vivo dose-response curve for MN induction with available in vivo data for MN induction in the liver of rats, revealed adequate dose predictions. The assessment of AFB1 target organ genotoxicity via PBK modelling-facilitated reverse dosimetry approach to perform QIVIVE, offers a promising alternative to obtain insight in the potential genotoxicity in rats taking in vivo kinetics into account.Despite the focus on one commodity, this study provides a promising starting point for future risk assessments in Mexico (Chapter 6). The risk assessment also revealed that the exposure to AFB1, fumonisins and DON is mainly driven by consumption rather than occurrence levels, and nixtamalization of maize may not be fully effective in reducing mycotoxins levels to a level of low concern. A PBK model approach for AFB1, and translation of in vitro data using PBK modelling-based reverse dosimetry to in vivo dose response curves was developed to estimate the acute liver toxicity without in vivo studies. Altogether, it was noticed that chronic exposure to AFB1 by Mexicans raise a concern. Also, it was observed that genotoxicity data available for AFB1 in vitro and in vivo in liver cells or liver tissue appear to be limited, and that improvement of liver cell models for studying chronic exposure in in vitro studies will also improve the PBK modelling-based reverse dosimetry approach.In conclusion, the results support the conclusion that further risk management actions to reduce potential health-risks of mycotoxins AFB1, fumonisins and DON in nixtamalized maize consumed by Mexican population are needed. QIVIVE results showed that PBK modelling-based reverse dosimetry is a promising tool to predict and study the toxicity of AFB1, but potentially also of other food borne mycotoxins.

The development of reliable non-animal testing strategies is the holy grail in current human safety testing of chemicals and drugs. However, adverse effective concentration obtained from in vitro models represent concentrations in the tissue and are not equivalent to the exposure doses. Hence concentration-response curves are inadequate for human risk and safety assessment because risk assessment requires in vivo dose-response curves from which points of departure. To bridge this gap, a so-called physiologically based kinetic (PBK) modeling based reverse dosimetry approach have been developed, which allows the translation of in vitro data to the in vivo situation. Cardiotoxicity has been considered as an important endpoint in pharmaceutical safety testing given that it has been a leading cause of drug attrition during the development. Cardiotoxicity is also a relevant endpoint in chemical/food safety risk assessment given that many natural alkaloids from botanicals and botanical preparation show the potential cardiotoxicity. The current project aims to provide a proof-of-principle that electrophysiological cardiotoxicity of chemicals or drugs for human can be predicted using a quantitative in vitro in vivo extrapolation (QIVIVE) approach, which could contribute the utility of alternatives of animal testing in risk assessment and safety evaluation of chemicals and drugs. Methadone and ibogaine, two anti-addiction drugs with known in vivo cardiotoxicity, were selected as model compounds.Chapter 1 as the introduction chapter starts with the background information on alternative test strategies/ NAM and the aim of the present PhD project. It also provides a definition on cardiotoxicity, followed by the summaries of toxicokinetic and toxicodynamic profiles of the two model compounds used for the studies and their metabolites. Subsequently the main approaches applied in the present project, including two in vitro cardiotoxicity assays, PBK modelling and Monte Carlo simulations are introduced. Chapter 2 evaluates a mouse and a human stem cell-based in vitro model, namely the mESC-CM beating arrest assay and the hiPSC-CM MEA assay, for cardiotoxicity screening of chemicals. Eleven cardiotoxic chemicals with different modes of action were used as reference compounds. The results obtained from the two models were compared to each other and to in vivo cardiotoxicity data, to provide insight into their applicability domains and to enable selection of a suitable toxicity assay for QIVIVE in subsequent chapters. In Chapter 3 the cardiotoxicity of methadone and its metabolites EDDP and EMDP were quantified using the hiPSC-CM MEA assay. A PBK model of racemic methadone was developed to enable the translation of the in vitro concentration-response curve obtained to an in vivo dose-response curve for methadone-induced QTc prolongation. The outcomes were compared with available human in vivo QTc prolongation data to evaluate the model performance. As follow-up of the work described in Chapter 3, Chapter 4 investigated the potential of the developed QIVIVE approach for the prediction of human inter-individual variability in in vivo cardiotoxicity of methadone. To this end in vitro cardiotoxicity and metabolic data were integrated with PBK models and Monte Carlo simulations to predict the effect of inter-individual and inter-ethnic kinetic variations on the cardiotoxicity of the two methadone enantiomers in the Caucasian and the Chinese population. CSAFs were defined and used to derive dose-response curves for the sensitive individuals. The kinetic variations obtained using individual human liver microsomes or recombinant cytochrome P450 enzymes (rCYPs) were compared. Chapter 5 investigates whether PBK modeling-based reverse dosimetry of in vitro data was able to adequately predict the human cardiotoxicity of the herbal alkaloid ibogaine and its metabolite noribogaine. The TEQ approach was incorporated in the PBK model enabling the evaluation of the role of noribogaine in ibogaine-induced in vivo cardiotoxicity. Chapter 6 summarizes the results obtained in the thesis, provides an overall discussion and presents the future perspectives that follow from the results obtained.

Zearalenone (ZEN) is a mycotoxin produced by Fusarium spp. present in food and feed. ZEN shows in vivo and in vitro endocrine disrupting potential via the activation of estrogen receptors (ERs). In mammals, liver and intestinal microbiota metabolize ZEN to α-ZEL and β-ZEL, with the estrogenic potency of α-ZEL being 60-times higher than that of ZEN and β-ZEL being 5-times less potent than ZEN. The species differences in sensitivity to ZEN exposure, with pigs being the most sensitive species, might be ascribed to interspecies differences in the formation of α-ZEL and β-ZEL in combination with possible interspecies differences in toxicodynamics. EFSA proposed in 2016 a group health-based guidance value (HBGV), i.e. a tolerable daily intake (TDI) for ZEN and its modified forms, of 0.25 μg/ kg bw based on the no observed effect level (NOEL) of 10.4 µg/kg bw for estrogenic effects of ZEN in young gilts. Considering the uncertainty around the effects of ZEN in humans, the aim of this PhD project was to gain further insight into the metabolism of ZEN, including its metabolism in liver and intestinal microbiota of not only experimental animals but also human, and to include this information in physiologically-based kinetic (PBK) models to enable evaluation of the role of metabolism of ZEN in its estrogenic activity. To this end, intestinal microbial metabolism of ZEN was studied in vitro using anaerobic incubations of fecal samples from rats, pigs and humans, and scaled to the in vivo situation to enable comparison to metabolism in the liver. The apparent in vivo catalytic efficiencies (kcat) for the formation of α-ZEL and β-ZEL in the three species showed that overall metabolism was higher in pigs, followed by rats and humans. The comparison to liver metabolism showed that the intestinal microbiota of pigs contributes up to 36% to the overall formation of α-ZEL from ZEN, and might thus contribute to the bioactivation of ZEN. For rats, this contribution amounted to 27%, while in humans this was less than 0.1%. These interspecies differences highlighted the importance for the development of human specific models to assess the kinetics of ZEN. In further studies of the present thesis the kinetic constants for the metabolism of ZEN by liver and intestinal microbiota were included in newly developed physiologically-based kinetic (PBK) models for ZEN and its bioactive metabolite α-ZEL, for rats and humans. These models enabled prediction of the maximum blood concentrations of ZEN and α-ZEL, as well as a comparison of these concentrations to concentrations known to induce estrogenicity in in vitro bioassays like the estrogen receptor α-mediated reporter gene (ERα-CALUX) assay and cell proliferation assays. The results obtained revealed that at a range of doses of 2.4-29 ng/kg bw of ZEN, representing the estimated daily intake in Europe for the average adult population, the maximum levels reaching the blood circulation in humans are 3 orders of magnitude below the concentrations of ZEN and α-ZEL known to be active in in vitro studies for estrogenicity of these compounds (EC10). Additionally, the human PBK model was used to study interindividual differences in toxicokinetics of ZEN. To this end in vitro derived kinetic constants for conversion of ZEN by individual liver and intestinal microbiota samples were combined to define 400 individual PBK models enabling the prediction of a distribution of the maximum blood concentrations (Cmax) for ZEN and α-ZEL. Subsequently the PBK model outcomes were combined with Monte Carlo simulation to obtain a distribution for a larger population (n=9,879). The distribution of the Cmax at a dose equivalent to the current TDI of 0.25 μg/ kg bw was used to estimate a chemical-specific assessment factor (CSAF) for human interindividual differences in toxicokinetics of 2.45 for the 95th percentile of the adult population. The estimated CSAF obtained for the simulated population indicated that the current default uncertainty factor for interindividual differences in kinetics of 3.16 is sufficiently protective. Age and reproductive status are reported to influence the sensitivity towards ZEN. In pigs, prepubertal gilts seem to be more sensitive to ZEN exposure. In humans, the presence of ZEN in urine of young girls with cases of precocious puberty was suggested to have a causal relation, while this could not be proven. It remains of interest for future studies to characterize if the CSAF also adequately covers the interindividual differences for sensitive groups and whether the relatively high sensitivity in younger populations is in part also related to differences in toxicokinetics or solely due to differences in toxicodynamics. Overall, the work of this thesis presents a proof-of-principle for the inclusion of in vitro kinetics for intestinal microbial and host metabolism in PBK models. Additionally, combining the PBK model with Monte Carlo simulations was shown to provide a strong in vitro-in silico tool for the study of interindividual differences in metabolism and to evaluate the default uncertainty factors used in the current risk assessment. Furthermore, the rapid hydrolysis by the intestinal microbiota in vitro of zearalenone-14-glucoside (ZEN-14-G) to release ZEN indicates the importance of the intestinal microbiota upon exposure to ZEN-14-G and supports the inclusion of ZEN-14-G into the group HBGV (EFSA, 2016) of 0.25 µg/ kg bw.In conclusion, the insights provided in this PhD thesis open a series of possibilities for further research including the identification of potential sensitive human subpopulations to ZEN exposure due to possible differences in toxicokinetics between individuals. The present thesis provides a way forward how to study this through the application of a combined approach of in vitro, and in silico studies.

The small intestine is a crucial component of the digestive system allowing the digestion of food and absorption of nutrients to supply the body with energy to function properly. The complex anatomy and physiology of the human small intestine poses a challenge when defining alternative testing strategies to characterise intestinal absorption and toxicity. In the present thesis, the potential of an alternative in vitro gut-on-chip model with a continuous liquid flow to study intestinal absorption and toxicity was revealed. A glass-based chip was used to culture epithelial cells under the optimized dynamic conditions. Upon morphologic characterization a selected range of model compounds were used for transport studies including 17 dioxin congeners with different physicochemical properties that are well known food contaminants, and selected pharmaceuticals (i.e. antipyrine, ketoprofen, digoxin, amoxicillin) with well-known uptake profiles from traditional static in vitro models and from human in vivo studies. These studies were followed by the comparative gene expression study of cells cultured in the gut-on-chip and in Transwells and human intestinal tissues. Finally, a comparative gene expression study was performed evaluating the effects on gene expression profiles of a nanomaterial (TiO2 and ZnO) exposure in the gut-on-chip versus the Transwell model.

Marine benthic ecosystems are considered to be the final sink of persistent organic pollutants (POPs). Antarctica is a global net receptor of POPs, therefore, quantification of fluxes of POPs to and in the Antarctic benthic ecosystems, although practically difficult, is essential to understand the overall dynamics of these contaminants on a global scale. In order to achieve that, Antarctic benthic ecosystems were studied as an integral part of the Antarctic environment as a whole. Furthermore, robust quality assurance and reporting practices are needed to establish comprehensive time trends of these pollutants, especially due to their expected relatively low concentrations. Current thesis was one of the first attempts to assess fluxes of POPs to benthic Antarctic ecosystems. The major outcome of the study is the conformation of the hypothesis on potentially increasing concentrations of POPs in the benthic ecosystems. Simultaneously, the concentrations of POPs in the pelagic ecosystems may have also stopped decreasing, which was previously indicated. Nevertheless, the fluxes of POPs and, possibly, their toxicological effects on the Antarctic benthic animals will undergo considerable changes in the future due to global change, which must be researched further in detail.

In chapter 1 of this thesis, an overview of the main applications of NPs was provided and the main properties of NPs were briefly introduced. In addition, some of the key methods that are currently used in the toxicological safety assessment of NPs were presented. The aim of the thesis was introduced which was to investigate the potential of different in vitro methods combined with high-end analytical techniques as a testing strategy to study the toxicokinetic and toxicodynamic properties of silver (Ag) and polystyrene (PS) NPs and set priorities in their further safety testing. The current state of the art of the in vitro methods used in the studies in this thesis and the outline of the thesis were also presented.In chapter 2, the influences of the size and surface chemistry of pristine PSNPs on the protein corona formation and subsequent uptake/association and transport of these NPs through a Caco-2 intestinal cell model were studied. Four negatively charged PSNPs of different sizes (50 and 200 nm) and with different surface chemistries (sulfone or carboxyl groups) were studied. The protein coronas of these PSNPs were analysed by LC-MS/MS which showed subtle differences in the protein composition of especially the two PSNPs with different surface chemistries. In further experiments, the impact of surface chemistry on the cellular uptake/association and transport was characterised using high-content imaging analysis. These experiments demonstrated that the PSNPs with sulfone surface groups were associated with the cells to a significantly higher extent than the PSNPs with carboxyl surface groups. No clear effect of the size of the PSNPs on the cellular uptake/association was noted. Also, the transport of the PSNPs with sulfone surface groups through the monolayer of cells was significantly higher than that of PSNPs with carboxyl surface groups.The general conclusion was that the composition of the protein corona and the PSNPs surface chemistry influence the cellular NPs uptake/association and transport, with the effect of the NP surface chemistry outweighing the impact of NPs size on the cellular uptake/associations and transport. These results might be predictive of the intestinal transport of NPs. Still, further studies are required to identify which corona proteins affect the uptake and transport of NPs.Chapter 3 described an investigation on the impact of the biochemical conditions within the human digestive tract on the intestinal fate of AgNPs with different surface chemistries. On top of that, the transport of these AgNPs across an intestinal in vitro model of Caco-2/HT29-MTX cells was evaluated. An in vitro digestion model was used to simulate the human digestion processes. Two 50 nm negatively charged AgNPs with different surface chemistries were used, lipoic acid (LA) AgNPs and citrate (Cit) AgNPs next to silver nitrate (AgNO3) as a source of ionic silver (Ag+). The co-culture model was exposed to different concentrations of pristine and in vitro digested (IVD) AgNPs or AgNO3 for 24 hr. Using ICP-MS and spICP-MS, the size distribution, dissolution, particle concentration (mass- and number-based) and total silver content of the AgNPs were characterized before and after digestion and in the apical, basolateral and cellular compartments of the Caco-2/HT29-MTX intestinal epithelial model. The surface chemistry of AgNPs had a significant influence on their dissolution and on their biological interactions with the Caco-2/HT29-MTX intestinal model. In general, a significant fraction of the AgNPs dissolved during the digestion up to 86 - 92% and 48 – 70% of the original amount of NPs for the (LA) and (Cit) AgNPs, respectively. Exposure of the monolayer of intestinal cells to increasing concentrations of pristine or IVD AgNPs resulted in a concentration dependent increase of total Ag and AgNPs content in the cellular fractions. The cellular concentrations were significantly lower following exposure to IVD AgNPs compared to the pristine AgNPs. The effect of the surface chemistry on the cellular concentration of Ag was only detected upon exposure to pristine AgNPs, while this difference disappeared upon exposure to IVD AgNPs.Transport of Ag across the intestinal model layer, as either total Ag or AgNPs, was limited (< 0.1%) following exposure to pristine and IVD AgNPs. De novo formation of AgNPs was shown in the exposure suspensions of IVD AgNPs and AgNO3 and in the cellular fractions, upon cellular exposure to pristine and IVD AgNO3. In conclusion, the surface chemistry of AgNPs and the biochemical conditions during in vitro digestion influences the dissolution and also influences the uptake/association within the Caco-2/HT29-MTX monolayer. This highlights the need to take in vitro digestion into account when studying nanoparticle toxicokinetics in an intestinal cellular in vitro model system. The observation that dissolution characteristics of NPs may change upon digestion could be of added value in the safe(r)-by-design NPs development.Chapter 4 presented the outcomes from combining the BeWo b30 placental transport model and the embryonic stem cell test (EST) to investigate the capability of pristine AgNPs of different surface chemistries and aged AgNPs (Ag2S NPs) to cross the placental barrier and induce in vitro developmental toxicity. AgNO3 was used as a source of Ag+. The pristine negatively charged AgNPs used in this study were similar to the ones used in chapter 3 while in addition also positively charged 50 nm branched polyethylenimine (BPEI) AgNPs were included in the study. The size distribution, dissolution, particle concentration (mass- and number-based) and total Ag content of the AgNPs in the apical, basolateral and cellular compartments of the BeWo b30 placental transport model at different time points was characterized using ICP-MS and spICP-MS. The ability of the AgNPs to cross the BeWo b30 cell layer was limited and dependent on the surface chemistry of these AgNPs. The particles detected in the basolateral compartment could result from transport of the original AgNPs and/or from the de novo formed AgNPs in the basolateral compartment from Ag+ that was transported.The in vitro developmental toxicity of the AgNPs was investigated by characterizing their potential to inhibit the differentiation of mouse embryonic stem cells (mESCs) into beating cardiomyocytes using the EST. The observed inhibitory effects of the AgNPs on differentiation of mESCs were most likely the result of cytotoxicity rather than specific effects related to developmental toxicity as the effects on differentiation of the mESCs were only detected at cytotoxic concentrations. Compared to the pristine AgNPs, the aged Ag2S NPs were significantly less cytotoxic, transported less across the BeWo cell layer and did not induce in vitro developmental toxicity. In conclusion, the combination of the BeWo placental transport model with the mESCs differentiation assay appeared to provide a valuable alternative in vitro methodology for prenatal developmental toxicity testing and prioritization for further safety testing of AgNPs, with aged Ag2S NPs appearing to present less of a hazard than pristine AgNPs.In chapter 5, the In-Cell Western (ICW)-γ-H2AX assay was evaluated as an alternative in vitro assay to detect the potential of aged AgNPs and pristine AgNPs to induce phosphorylation of H2AX in HepG2 liver cells. AgNO3 was used as source of Ag+ to test the effects of Ag+ themselves. The γ-H2AX induction detected was higher after 24 hr exposure compared to 4 hr and was accompanied by a significant cytotoxicity in the HepG2 cells. The increased induction of γ-H2AX measured could be due to the cytotoxicity that occurred at the same concentrations which can result in DNA damage resulting in an increased induction of γ-H2AX. This suggests potential false-positive confounders limiting the use of the ICW-γ-H2AX assay, in the form as applied in chapter 5, for evaluation of the genotoxicity of NPs. Additionally, the potential of the AgNPs to induce ROS production, as a potential underlying mechanism of induction of the cytotoxicity and/or DNA-DSBs, was assessed in HepG2 cells. No increase in ROS levels was measured upon exposure of the cells to the AgNPs for 4 or 24 hr and in the absence of cells, while an increase in ROS levels was detected upon AgNO3 exposure of the cells and in the absence of cells.In conclusion, the surface chemistry of AgNPs has a significant influence on their cytotoxic effects and the accompanying induction of γ-H2AX levels in HepG2 cells. The aged Ag2S NPs were biologically less active in inducing both cytotoxicity and γ-H2AX levels, as these effects were absent in the dose range tested. The absence of cellular ROS generation upon exposure to all AgNPs indicates that the observed effects were not ROS-mediated. Additional tests, to rule out apoptotic mediated false-positive signals, need to be combined with the ICW-γ-H2AX assay, to render this interesting assay into a robust screening method for the potential genotoxicity of NPs.Chapter 6 of the present thesis included a general discussion of the results of the previous chapters and highlights on future perspectives for research in the field of in vitro nanotoxicology.Overall, the work presented in this thesis illustrated the role of surface chemistry and the status of the NPs (pristine or aged and/or digested) on the toxicological behaviour of NPs. Besides, the combination of different in vitro models with high-end analytical techniques was shown to; 1) provide in-depth understanding of the biological behaviour of NPs, 2) assure the value of the alternative in vitro models as a testing strategy for potential hazards that could be induced by NPs and 3) assist in setting priorities for in vivo testing and contributing to reduction, refinement and replacement (3Rs) of animal testing required for the safety evaluation of NPs.

Chapter 1 introduced the objectives of the thesis and background information regarding the production of nanomaterials (NMs) and their release in the environment. The concepts of environmental and bio-mediated transformations of NMs were defined and their implication on the NM exposure assessment is explained. Silver nanoparticles (Ag-NPs) were selected as model NM because they are widely produced, undergo transformations that represent all the most relevant transformations of NMs in soil and specific analytical methods are available to quantify them at relatively low concentrations. A short review regarding previous toxicological studies of Ag-NPs in soil organisms and the importance of performing toxicokinetic studies was presented. Finally, selected model organisms (earthworms Eisenia fetida and Lumbricus rubellus) were briefly described. Chapter 2 reported a short-term (28 days) toxicokinetic study in E. fetida exposed to Ag-NPs, aged Ag-NPs (Ag2S-NPs), and AgNO3. A one-compartment model was applied to calculate separately the kinetic constants for uptake and elimination of particulate and ionic forms of Ag. The uptake and elimination rate constants for earthworms exposed to pristine Ag-NP or AgNO3 were not significantly different from each other. Uptake rate constants of (hardly dissolvable) Ag2S-NPs which resemble the environmental relevant form of Ag-NPs was significantly lower. spICP-MS analysis demonstrated that ~85% (average of both Ag-NP and AgNO3 treatments) of the Ag within the earthworms was present as ionic Ag, regardless of the actual form of Ag that the earthworms were exposed to. Indeed, the biogenic formation of particulate Ag (~10 % of total Ag accumulated overtime) in earthworms exposed to AgNO3 led to a kinetic pattern of particulate Ag body burden similar to pristine Ag-NPs. NP size analysis and imaging techniques showed evidences that the particles in the tissues were not the same as those to which worms were exposed, highlighting that biotransformation and/or biogenic formation took place also in the case of the Ag-NP exposure. Chapter 3 investigated the influence of dissolution on the uptake of metal NPs in earthworms by the use of bimetallic NPs. E. fetida specimens were exposed to Au core-Ag shell NPs (Au@Ag-NPs) and to a combination of Au-NPs, Ag-NPs, Ag and Au ions containing natural soil for 28 days. Our hypothesis was that Ag shell would dissolve partially or completely and that Au core would not interact with the exposure media and would therefore behave as a tracer of the particulate uptake. Analysis of earthworm tissues showed that concentrations of Ag in the earthworms were not statistically different in organisms exposed to the different forms of Ag. However, the concentration of Au in the earthworms exposed to HAuCl4 (ionic Au) exceeded around twenty times the Au concentrations in the exposures to particulate Au, which did not differ among each other. Mass measurements by spICP-TOFMS provided evidence that the uptake of the metals in their bimetallic particulate form represents approximately 5 % of the total metal amount. Size measurements by spICP-TOFMS showed that the Au core remained similar after the uptake, while the Ag shell increased in thickness suggesting that biotransformation processes took place at the surface of the NPs (e.g. aggregation, adsorption of Ag ions on the surface of existing particles). The study confirmed that dissolution is the main factor driving the uptake of (dissolving) metal NPs in earthworms. Additionally, different uptake patterns resulted from the co-exposure to Au and Ag-NP and Ag+, indicating that the Ag form can lead to different interactions with Au in the soil affecting the uptake in the earthworms. Chapter 4 presented a toxicokinetic study performed to assess the potential impacts of long-term exposure (nine months) on the uptake of pristine Ag-NP, aged Ag-NP (Ag2S-NP) and ionic Ag in earthworms E. fetida. The study was conducted with same species and conditions similar to the short-term experiment which was previously conducted for 4 weeks (chapter 2), in order to allow comparison between the two models. The accumulation of Ag in Ag-NP and AgNO3 exposed earthworms did not statistically differ after nine months exposure. In Ag2S-NPs exposed earthworms, the internalized concentrations were five times lower compared to the other treatments. The Ag concentrations in pore water did not reflect the uptake pattern and metallothionein concentrations were not different from the control group. The overall conclusion of this chapter was that even after a prolonged period of time the uptake kinetic rate constants of Ag-NP and AgNO3 were not statistically different, while the one of Ag2S-NP was statistically significant lower than the other treatments. Additionally, the short-term kinetic rate constants predicted the average bioaccumulation of pristine Ag-NP and AgNO3 in the earthworms exposed for nine months, while the bioaccumulation of Ag2S-NPs in earthworms was under-predicted somewhat. This was likely because the short-term did not take into account the late dissolution of Ag2-NPs. Ag bioaccumulation of Ag2S-NP could not be related to the concentrations of Ag measured at a specific time in pore water. Chapter 5 reports a study which demonstrated that earthworm bioturbation plays an important role in the vertical transport of Ag2S-NPs in soil. In the soil columns, daily rainfall from artificial rain water did not lead to displacement of Ag2S-NPs within 28 days indicating that in the case of hardly soluble metal NPs and unsaturated soil conditions, bio-mediated transport overcomes physical chemical transport. Bioturbation from L. rubellus was quantified by assessing the changes of the macro porosity in the soil columns. Results indicated that earthworms burrowing activity was not affected by the presence of Ag2S-NPs at the experimental concentrations. The study proposed a linear relationship between bioturbation rate and the abundance of earthworms that is applicable to future bioturbation studies. Chapter 6 presented an overall discussion of the results obtained in the thesis, and concluded with the implication of such findings in the risk assessment of NMs.

In Indonesia, the market demand for herbal products keeps growing, and as a result, herbal products increasingly provide economic and perceived clinical benefits. A risk and benefit assessment are crucial to be performed to support the safe use of herbal products although the consumers perceive herbal product as “safe” and “natural” and thus “healthy”. The aim of the PHD thesis was to perform an assessment of potential risks and some benefits of herbal products available in the Indonesian market. The model compounds chosen included especially naturally occurring genotoxic and carcinogenic botanical constituents including alkenylbenzenes (ABs) and pyrrolizidine alkaloids (PAs). Beneficial effects focussed on potential PPARg activation by the carotenoids bixin and crocetin. Existing but also novel testing strategies were used to evaluate the relevance of effects at estimated human intake levels. Altogether, it can be concluded that the risk assessment using the Margin of Exposure approach combined with Haber’s rule can be used to prioritize risk management actions to prevent the adverse health effects of consuming Indonesian herbal products containing genotoxic carcinogens. In addition, a novel testing strategy, combining in vitro and PBK modeling-facilitated reverse dosimetry was found to facilitate risk and benefit assessment of botanical compounds without the need for animal experiments and/or human intervention studies. It is important to note that this conclusion holds for herbal products collected by targeted sampling, and not for all herbal products on the Indonesian market. Many aspects, including variability in detected levels of the targeted compounds, variability in recommended daily use mentioned on the label, interindividual variation of exposure among Indonesian people, absence of a generally accepted method to take shorter-than-lifetime exposure into account, the knowledge gaps in modes of action, selection of the best in vitro model for QIVIVE, potential combination effects, the chemical-specific parameters needed for PBK modeling and availability of in vivo studies to validate the predictions should be considered for future research. Seven actions including (1) applying Good Agricultural and Collection Practices (GACP) for farmers, (2) applying good manufacturing practice (GMP) of herbal product and food safety training for manufacturers and producers, (3) development of a toxicity database of medicinal botanicals used in Indonesia, (4) restriction of the exposure to genotoxic carcinogenic compounds by establishing MPLs and refining the label requirements for botanicals and botanical preparations, (5) use of human biomonitoring (HBM) and PBK modeling for a more refined exposure, risk and benefit analysis of Indonesian herbal products, (6) incorporation of Chemical Specific Adjustment Factors (CSAFs) for interspecies and interindividual variation in kinetics within the human population in the risk assessment, and (7) exploring the beneficial effects of botanicals and botanical preparations, were proposed to improve safety and efficacy of botanicals and botanical preparations on the Indonesian market.

Diethylstilbestrol (DES) is a synthetic estrogen that has been used between the 1940s and 1970s by pregnant women to prevent miscarriages and premature delivery by stimulating the synthesis of estrogen and progesterone in the placenta. However, use of DES appeared to cause a wide range of adverse effects, such as clear cell vaginal adenocarcinoma in the daughters of women who took the drug, and developmental and reproductive toxicity. These adverse effects have often been attributed to the functional estrogen receptor alpha (ERα), since it has been reported that ERα is needed to induce DES-mediated adverse developmental and reproductive effects in neonates. The question has been raised why DES behaves differently from the endogenous ERα agonist 17β-estradiol (E2), even though the molecular dimensions and binding orientations of DES and E2 to the ERα are almost identical. The research described in this thesis aimed to investigate the possible differences in the estrogenicity and developmental toxicity between DES and E2, using different in vitro and in silico approaches, focussing on the potential role of possible differences in ERα-mediated effects in the underlying mode of action. Accordingly, first the effect of DES and E2 on ERα-mediated reporter gene expression, ERα-mediated T47D breast cancer cell proliferation, and ERα-coregulator interactions and gene expression in T47D cells were evaluated. In addition, the effects of DES and E2 in two alternative developmental toxicity assays (the ES-D3 cell differentiation assay of the embryonic stem cell test (EST) and the zebrafish embyotoxicity test (ZET)) and the potential role of ERα in these effects were evaluated. Finally, possible dose-dependent differences in internal dose levels of DES and E2 were evaluated with help of PBK modelling, in order to elucidate to what extent possible differences in kinetics could play a role in differential in vivo effects of DES and E2. Altogether, the data show that two estrogens E2 and DES differ in their biological effects related to development in a subtle but significant way. At the cellular level, DES and E2 show high similarities in the molecular pathways that relate to ERα-mediated effects with small significant differences that may contribute to the developmental toxicity in part via potential epigenetic effects of DES. The in vitro developmental toxicity assays EST and ZET can discriminate DES from E2 in terms of developmental toxicity, but at the same time do not capture the full mode of action underlying DES-induced developmental toxicity. Finally, it was shown that in addition to the subtle differences in toxicodynamics, substantial differences in internal concentrations (endogenous E2 concentrations compared to predicted DES concentrations in women that took DES as medication), add to the differential in vivo effects of E2 and DES.

Cadmium can be immunotoxic even at relatively low exposure concentrations. By modulating the immune system of environmentally exposed wild small mammals, cadmium may reduce the host resistance to infections (e.g. zoonotic diseases in wild reservoir species). Because the immune status play an important role in preventing and limiting infections transmission, this project assessed the effects of cadmium on the immune responses of small mammals using a sequential approach, from in vitro to in vivo and to field conditions.In the introductory Chapter 1 the mammalian immune system, the potential effect of environmental pollutants in modulating the immune system of wild small mammals, and cadmium as model metal because its immunotoxic potential are introduced. The immune system is briefly described emphasizing its function in preventing and controlling infections. A short review on cadmium modulatory effects on the immune system of small mammals is presented, including effects in wild rodent species, in laboratory mouse models and in in vitro models. The aim of the study, to gain insight in how cadmium can modulate the immune status of reservoir rodents, is introduced. The sequential study approach is also described, in which by including in vitro, in vivo and field data, allow to stablish causality between exposure to cadmium and effects at cellular level, individual, and even population level. The other five chapters in this thesis are introduced, including four experimental chapters and general discussion.In Chapter 2 the effects of cadmium were investigated in murine cells from the innate immune system in vitro. In the study, macrophage and mast cell lines were exposed to a range of cadmium concentrations to build dose- response relationships for different toxicological endpoints in resting cells and functional parameters in activated cells. Cadmium depleted cellular glutathione in macrophages and mast cells and had cell-specific immunomodulatory effects. Cadmium was pro-inflammatory in macrophages and suppressive in mast cells. In Chapter 3, the effects of cadmium were studied in mice cells from the adaptive immune system. T helper (Th2) cell and B cell models were exposed to a range of cadmium concentrations to build dose-response relationships for different endpoints, just like as in the previous chapter. Cadmium also led to cell-specific effects, suppressing B cell functions and stimulating T helper (Th2) cells. Cadmium reduced proliferation of activated B cells by reducing DNA/RNA synthesis, while increased IL-4 production in activated T helper (Th2) cells. The suppressive effects of cadmium on B cell proliferation were considered for the following two chapters in vivo. In Chapter 4 laboratory mice were exposed to cadmium in the drinking water for seven weeks and were immunized to sheep red blood cells (SRBC) to evaluate and compare humoral responses between individuals exposed to different cadmium concentrations in vivo. Resulting kidney cadmium concentrations were in the range of environmentally relevant concentrations and higher in female mice. Total blood IgM and IgG immunoglobulin levels were also higher in non-exposed female mice, compared to non-exposed mice. The increasing cadmium concentrations decreased IgM/IgG blood levels, in which the higher IgM/IgG levels found female mice were more susceptible to be decreased by cadmium, compared to males. The lower levels of IgM/IgG in cadmium-exposed mice were correlated with a decrease in the counts of SRBC-specific antibody-producing B cells in spleen, in line with in vitro results in Chapter 3 of inhibition of B cell proliferation by cadmium.In Chapter 5 environmentally exposed wood mice (Apodemus sylvaticus) were captured from three locations in the Netherlands with known levels of cadmium pollution, to verify the effects observed in vitro and in vivo under laboratory conditions. The animals were immunized to SRBC to evaluate the humoral responses. Kidney cadmium concentrations reflected the different levels of cadmium pollution for each capture location and increased with body mass, as a result of time of exposure (age). Flow cytometry analysis of blood samples showed that cadmium exposure can reduce the counts of B cells of female mice. Humoral responses were measured independently for each capture location, due to natural immune heterogeneity between wild rodent populations. Increasing cadmium concentrations decreased the number of SRBC-specific antibody-producing B cells in wood mice, similar to the observations made in Chapter 4 in laboratory mice, and in line with a decrease of B cell proliferation in vitro in Chapter 3. The immunosuppressive effects of cadmium in the humoral immunity of laboratory mice and wood mice were discussed as a potential mechanism in reducing the host resistance to infections.Chapter 6 presented a general discussion of the results obtained in this thesis. The discussion went through the results obtained in the different experimental chapters, showing the potential consequences and impacts of the effects of cadmium on the immunity of small mammals at different levels, from sub-cellular, to cellular and individual level. It was possible to establish causality between cadmium exposure and effects at cellular level, providing evidence in how cadmium can modulate and impair small mammal humoral immune responses in vivo, to finally connect those results with effects observed in actual environmentally exposed wood mice living in polluted environments. The immunosuppressive effects of cadmium were discussed in regard of their impacts on host resistance to infections, with emphasis in the risk of increase of zoonotic infections in small mammal reservoir host species.

Currently, wild edible plants are getting into the research focus of both food and pharmaceutical domains as they are considered valuable resources for improving food and nutritional security as well as for bioactive ingredients that can be used for functional foods or managing chronic diseases. Wild edible plants, in addition to their potential nutritional contribution, may provide important health benefits as the health-promoting components of plant based foods, so called nutraceuticals or functional food ingredients, are usually higher in wild plants. However, at the same time, these plants may contain toxic ingredients that may lead to adverse effects. At the current state of the art novel foods have to comply with novel food regulation while safety of the use of botanicals and botanical preparation as food is the responsibility of the provider and not subject to premarket safety evaluation. This implies that a concern requiring adequate consideration of both risks and benefits has to be addressed when introducing wild edible plants as novel components in a regular diet or as herbal medicine or food supplement. The present PhD thesis aims at investigating the possible hazards and potential health benefits of Maerua subcordata (Gilg) DeWolf, a wild plant locally called ‘ashkulebya’ by the Kunama ethnics of Northern Ethiopia, as a novel nutritional resource and/or as a health food/functional food. To meet this aim, the thesis tested the activity of different extracts of M. subcordata in a wide range of in vitro assays for both adverse and beneficial endpoints, along with, phytochemical studies to identify active constituents. Consideration of ethnobotanical data was also part of the study. The phytochemical studies identified glucosinolates and biogenic amines such as quaternary ammonium compounds and guanidine derivatives in the different M. subcordata extracts. The potential health benefits of M. subcordata extracts via their effects on the expression of electrophile-responsive element (EpRE), peroxisome proliferator-activated receptor gamma (PPARg), and nuclear factor kappa B (NF-κB) regulated genes was evaluated. The induction of EpRE-mediated genes have implications of maintaining redox homeostasis. The fruit, leaf, and seed extracts induced the expression of EpRE-mediated genes while induction by the root extract was minimal. Candidate glucosinolates (or their matching isothiocyanates) exhibited strong induction while some biogenic amines exhibited either no significant induction or slight inhibition. Inhibition of NF-κB regulated genes have implications for anti-inflammatory effects while induction of PPARg regulated genes have implications for control of diseases of the metabolic syndrome. The fruit, root, and seed extracts induced PPARγ-mediated gene expression while induction by the leaf extract was minimal. Instead, all extracts inhibited bacterial lipopolysaccharide (LPS) induced nitric oxide production, with the root showing the lowest potency. Selected candidates such as isothiocyanates and some biogenic amines also inhibited the LPS induced nitric oxide production. Equally, a battery of in vitro tests was applied to identify selected endpoints that could be used to assess the potential hazards associated with the use of botanical preparations or parts of M. subcordata. The Ames test and in silico (Derek Nexus) predictions were applied for initial genotoxicity screening whereas the aryl hydrocarbon receptor and the estrogen receptor alpha reporter gene assays as well as the embryonic stem cell test and the zebrafish embryotoxicity test were applied to investigate alerts for developmental toxicity. The overall hazard assessment results indicated that all extracts of M. subcordata do not point to a genotoxicity hazard; fruit, root, and seed extracts do not raise a concern with respect to developmental toxicity; but the data on the leaf extract may point towards a developmental toxicity hazard. In conclusion, the overall results obtained argue in favour of the use of the root, fruit, and seed parts of M. subcordata as (famine) food and/or functional food whereas the leaf part may be used as a herbal medicine. The outcomes of the study may also contribute scientific data to design further in vivo studies on safety and proper utilization of the plant while the methodology applied can be extrapolated to serve as an approach to the risk-benefit assessment of other plant based products.

The aim of the present thesis was to perform the risk and safety assessment of two groups of natural food-borne toxins, namely alkenylbenzenes and pyrrolizidine alkaloids (PAs) for the Chinese population as compared to the overall Caucasian population using methods including the Margin of Exposure (MOE) approach, physiologically based kinetic (PBK) modelling-based reverse dosimetry and the combination of PBK modelling and Monte Carlo simulation. Chapter 1 introduced the model compounds studied in the thesis, namely alkenylbenzenes and 1,2-unsaturated PAs, the aim of the thesis and the strategies applied to perform a mode of action based safety and risk assessment. The regulations and consumer behaviour towards botanical preparations in China and in several EU countries were also presented. In chapter 2, a risk assessment of 10 plant food supplements (PFS), 23 traditional Chinese medicines (TCM) and 38 herbal teas containing alkenylbenzenes obtained at the Chinese market was performed using the Margin of Exposure (MOE) approach. Use of about half of the samples would result in estimated intakes upon daily life-time exposure that give rise to MOE values lower than 10 000, suggesting a potential priority for risk management. For short-term exposure such as two weeks consumption, applying Haber’s rule, only one TCM (TCM 6) still had a MOE value below 10 000. It is concluded that consumption of Chinese botanical preparations raise a concern because of exposure to alkenylbenzenes, especially when exposure is for longer periods of time. Chapter 3 studied the differences in bioactivation and detoxification of the food-borne genotoxic carcinogen estragole between Chinese and Caucasians using a mode of action based PBK modelling approach. The outcomes of the model predictions showed that the detoxification of the proximate carcinogenic metabolite 1’-hydroxyestragole, mainly by conversion to 1’-oxoestragole, was similar in both ethnic groups. For the bioactivation pathway, at realistic daily intakes of estragole, in the Chinese population only 0.02 % of the dose was converted to the ultimate carcinogenic metabolite 1’-sulfooxyestragole, while this value amounted to 0.09 % of the dose in Caucasian subjects. This 4.5-fold difference accompanied by similar rates of detoxification may indicate a lower risk of estragole for the Chinese population at similar levels of exposure. In chapter 4, it was demonstrated that combining in vitro cytotoxicity data obtained with primary rat hepatocytes with PBK modelling-based reverse dosimetry could adequately predict in vivo acute liver toxicity of lasiocarpine and riddelliine for rats. The PBK models for rats for lasiocarpine and riddelliine were developed based on data derived from in silico approaches, incubation experiments using subcellular tissue fractions and literature to define the PBK model parameters. Concentration-response curves obtained from in vitro cytotoxicity assays in primary rat hepatocytes were converted to in vivo dose-response curves for acute liver toxicity by PBK-modelling based reverse dosimetry. From the in vivo dose-response curves thus obtained points of departure (PoDs) were derived that were compared to available literature data on in vivo liver toxicity of lasiocarpine. The predicted PoDs appeared to fall well within the range of PoDs obtained from the available oral single dose in vivo studies. In conclusion, this chapter showed that PBK modelling based-reverse dosimetry can translate in vitro concentration-response curves to in vivo dose-response curves to predict the acute liver toxicity of lasiocarpine and riddelliine in rats. Chapter 5 investigated the inter-species and inter-ethnic human differences in acute liver toxicity of lasiocarpine and riddelliine using the approach developed in chapter 4 for rats. Thus, PBK models for lasiocarpine and riddelliine for the average Chinese and Caucasian were developed, based on the PBK models defined and validated for these PAs in rats (chapter 4). Subsequently, the models were used to convert in vitro toxicity data obtained in pooled Caucasian primary hepatocytes to predict in vivo dose-response curves for acute liver toxicity of lasiocarpine and riddelliine in humans, from which PoDs were derived that were compared to PoDs derived in a similar manner for rats in chapter 4 to obtain insight in inter-species differences. Similarly, the PoDs obtained for the average Chinese and Caucasian were compared to provide insights in inter-ethnic differences. The inter-species differences amounted to 2-fold for lasiocarpine and 8.2-fold for riddelliine with humans being more sensitive than rats. The inter-ethnic human differences varied 2-fold for lasiocarpine and 5-fold for riddelliine with the average Caucasian being more sensitive than the average Chinese. Altogether, this chapter shows a proof-of-principle for a method to predict inter-species and inter-ethnic differences in in vivo liver toxicity for PAs by an alternative testing strategy integrating in vitro cytotoxicity assays with PBK modelling-based reverse dosimetry. Chapter 6 investigated the effect of inter-individual and inter-ethnic (Chinese and Caucasian) kinetic differences in bioactivation of lasiocarpine, and definition of chemical specific adjustment factors (CSAFs) by applying PBK modelling and Monte Carlo simulation. The results revealed an inter-ethnic variation of 2.1-, 3.3- and 4.3-fold when comparing the predicted 7-GS-DHP formation at the GM, 90th and 99th percentile of the Chinese and Caucasian population, respectively, at a dose of level of 8 ng/kg bw lasiocarpine, indicating that the Caucasian population was more sensitive. The CSAFs obtained based on the GM and 90th percentile individuals of the Chinese, Caucasian and the two populations combined were 3.3, 5.2 and 5.7, respectively. The CSAFs obtained based on the GM and the 99th percentile were 8.3, 17.0 and 19.5 for inter-individual variations in the Chinese, the Caucasian population and the two populations combined, respectively. The CSAF values obtained in this chapter indicate that the default safety factor of 3.16 for inter-individual human kinetic differences may not be sufficiently protective for any of the populations. Chapter 7 presents the general discussion based on the results of the present thesis placing the results in a wider perspective while also presenting future perspectives. It is concluded that the present thesis showed an alternative to animal testing approach to define in vivo dose-response curves for risk assessment, based on in silico and in vitro data. It was demonstrated that combining in vitro toxicity data with the kinetic processes integrated into a PBK model adequately predicts in vivo acute liver toxicity of PAs. The obtained results show the possibility to use the method to provide a PoD to perform risk assessment or to define CSAFs for inter-species and intra-species differences. Additionally, this thesis provided insight in the possibilities to build ethnic-specific PBK models that can be used to further refine the risk assessment of compounds that are genotoxic and carcinogenic and also have non-genotoxic endpoints. Moreover, this thesis illustrated that there is a potential priority for risk management for several botanical preparations containing alkenylbenzenes or PAs for the Chinese population when considering daily life-long exposure. Altogether, this thesis provided a proof-of-principle to perform a risk and safety assessment for food-borne toxins based on only in vitro and in silico data.

Scope: Botanicals and botanical preparations may contain natural constituents that are of concern for human health. One group of such natural toxic compounds that may raise a concern is the group of pyrrolizidine alkaloids (PAs). Especially 1,2-unsaturated PAs are hepatotoxic and may act as genotoxic carcinogens in humans. At the current state-of-the-art, risk assessment of botanicals and botanical preparations is generally not required before they can enter the market, thus may pose a potential risk to human health. Objective: The present thesis aimed to perform the risk assessment for PAs derived from botanical products following daily life-time exposure and also more realistic exposure scenarios. Another aim of the present thesis was to investigate whether animal-free testing strategies could be of use in tackling data and knowledge gaps for PAs by predicting in vivo toxicity of different PAs and whether proofs-of-principle for applying such alternative testing approaches could be provided for two selected PA model compounds, riddelliine and lasiocarpine. Material and methods: Herbal teas, herbal medicines and plant food supplements (PFS) were bought from different counties. LC-MS/MS was used to detect the PA levels in these botanical and botanical preparations. Because PAs are genotoxic carcinogens, Margin of exposure (MOE) approach was applied for risk assessment. For development of the physiologically based kinetic (PBK) modelling to predict in vivo liver toxicity for lasiocarpine and riddelliine in rat and human, the kinetic parameters were obtained from in vitro incubations assays. The microsomal incubation assays were performed with rat and human tissue fractions to determine Vmax and Km values for lasiocarpine and riddelliine clearance, using a substrate depletion approach. The MTT assay was used to detect in vitro liver toxicity for lasiocasrpine and riddelliine. In a subsequent step, the same approach was used for prediction of genotoxicity in rat, another endpoint relevant for PA toxicity. To this end, the in vitro concentration-responses curves obtained from in vitro genotoxicity studies using the ƴH2AX assay were translated into in vivo dose-response curves using PBK modelling-facilitated reverse dosimetry. Main results: When consumption of one cup of tea a day during a whole lifetime would result in MOE values lower than 10000 for several types of herbal teas, indicating a priority for risk management for these products, these products not only derived from PA-producing plants but also derived from non-PA-producing plants. Using MOE approach, combined with Haber’s rule was employed to analyse the risks of shorter-than-lifetime exposure for total of 39 herbal teas, 8 herbal medicines, and 19 PFS. This analysis revealed that shorter-than-lifetime use would result in MOE values lower than 10000 upon use for 40 up to 3450 weeks during a lifetime (depending on the preparation). Only for a limited number of herbal teas and medicines, use of two weeks a year (150 weeks during a 75 year lifetime) would still raise a concern. The PBK models were used for translation of in vitro concentration-response curves to in vivo dose-response curves in rat. From these in vivo dose-response curves, the predicted BMDL5-BMDU5 (lower/upper limit of the 90% confidence interval of the benchmark dose that gives a 5% response) of lasiocarpine and riddelline were obtained which were 23.0-34.4 and 4.9-8.4 mg/kg bw/day, respectively. The predicted BMDL5-BMDU5 of lasiocarpine falls well within the range of the point of departures (PoDs) derived from available in vivo toxicity data. The same Quantitative in vitro to in vivo extrapolation (QIVIVE) method was subsequently used to predict the inter-species and inter-ethnic human differences in liver toxicity of lasiocarpine and riddelliine using a human PBK model. The BMDL5-BMDU5 of lasiocarpine were found to amount to 14.7-41.2 mg/kg bw/day for Chinese and 7.4-23.7 mg/kg bw/day for Caucasian, indicating the Chinese to be less sensitive. The predicted BMDL5-BMDU5 of riddelliine were 1.0-5.9 mg/kg bw/day for Chinese and 0.2-1.2 mg/kg bw/day for Caucasian. These values were subsequently compared to those previously obtained in rat to evaluate inter-species differences. The inter-species differences amounted to 2.0-fold for lasiocarpine and 8.2-fold for riddelliine with humans being more sensitive than rats. When extending the developed PBK modelling to predicted in vivo genotoxicity for lasiocarpine and riddelliine in rat, the predicted BMD10 for lasiocarpine and riddelliine amounted to 8.82 and 3.41 mg/kg bw/day, respectively, and were in line with the experimental data on in vivo genotoxicity available in the literature for these two PAs. Conclusions and implications: The risk assessment of herbal products revealed that daily life-time consumption of some of these products would be a priority for risk management. When considering realistic exposure scenarios, exposure to most of these herbal products would have MOE values higher than 10000 indicating a low priority for risk management. Moreover, this thesis demonstrated that the combined in vitro PBK modelling-based reverse dosimetry approach could adequately predict in vivo liver toxicity and genotoxicity for two model PAs, lasiocarpine and riddelliine. Such QIVIVE methods may prove to be of use in defining more realistic relative potency values for the different food/feed-related PAs. The results obtained reveal the feasibility of this combined quantitative in vitro-in silico approach to determine a PoD for a chemical without the use of experimental animals and to address the issue of how to use in vitro data for risk assessment.

Marine biotoxins in fish and shellfish can cause a number of adverse health effects in consumers, such as diarrhoea, amnesia, and death by paralysis. Worldwide, there are monitoring programs for testing shellfish on a regular basis. In some countries, testing is performed by using the so-called mouse bioassay (MBA), an assay raising both ethical and practical concerns because of animal distress and shortcomings in respect to specificity. The MBA may result in both false negatives and false positives. A false negative does not protect the consumers as anticipated and the high amounts of false positives encountered when applying the MBA lead to unnecessary closures of extraction areas, damaging local economies. A full ban of the MBA or its total replacement by analytical chemical methods has failed because these detection methods are unable to detect all toxin analogues and newly emerging toxins and will thus result in false negatives by definition. To fully replace the MBA, there is a clear need for new functional animal-free in vitro assays with specific endpoints that are able to detect both the known and yet unknown marine biotoxins. In Europe a method based on LC-MS/MS has been developed as an alternative for the MBA and is now the reference method for lipophilic marine biotoxins (LMBs) and used in the routine monitoring. However, as outlined above safety is not fully guaranteed when relying only on such a method and, as a result, the MBA is still used for surveillance purposes. The aim of the work presented in this thesis was to develop a new strategy to fully replace the MBA for detecting LMBs without the risk of missing a contaminated sample that can lead to an intoxication. This was achieved by combining effect-based bioassays and a mass spectrometry analysis, including the official EU-RL method. Chapter 1 addresses the safety issues of the marine biotoxins produced by algae, corals and bacteria and summarises the current legislations and recommendations and the methods of detection. In Chapter 2, the neuro-2a bioassay, a cell-based in vitro bioassay that was previously shown to be sensitive for several hydrophilic and lipophilic marine biotoxins, was studied for its ability to screen seafood products for the presence of lipophilic marine biotoxins. All (regulated) LMBs and their analogues were tested, and the neuro-2a bioassay outcomes showed that all these LMBs could be detected at low concentrations. Next, blank and contaminated sample extracts were prepared and tested, showing that matrix effects led to false positive screening outcomes. Therefore, the standard extraction procedure for LMBs with methanol was modified by introducing a clean-up step with n-hexane before further extraction on the SPE-column. First, the possible recovery losses due to this extra n-hexane wash step were assessed, showing that the n-hexane did not lead to recovery losses of the LMBs and that the matrix effect was successfully removed. Finally, the applicability of the neuro-2a bioassay was assessed by testing a broad range of shellfish samples contaminated with various LMBs, including diarrhoeic shellfish poisoning (DSP) toxins. The samples were also analysed by LC-MS/MS. Overall, the neuro-2a bioassay showed screening outcomes that were well in line with the toxin levels as determined by the EU-RL LC-MS/MS reference method. In chapter 3, a study with DNA microarrays was performed to explore the effects of two diarrhoeic and one azaspiracid shellfish toxin, okadaic acid (OA), dinophysistoxin-1 (DTX-1) and azaspiracid-1 (AZA-1) respectively, on the whole genome mRNA expression of undifferentiated intestinal Caco-2 cells. In this chapter the whole genome mRNA expression was analysed in order to reveal the possible modes of action of these toxins and to select genes that can be used as potential markers in new additional bioassays for the detection and identification of these LMBs. It was observed that OA and DTX-1 induced almost identical effects on mRNA expression, which strongly indicates that OA and DTX-1 induce similar toxic effects. Biological interpretation of the microarray data showed that both compounds induced endoplasmic reticulum (ER) stress, hypoxia, and unfolded protein response (UPR). The gene expression profile of AZA-1 resulted in a different expression profile and showed increased mRNA expression of genes involved in cholesterol synthesis and glycolysis, suggesting a different mode of action for this toxin. In chapter 4, twelve marker genes were selected from the previous study and five were used to develop a multiplex qRT-PCR method. This multiplex qRT-PCR method is able to detect three toxin profiles, i.e. a OA/DTX, AZA/YTX and PTX profile. The multiplex capacity of this qRT-PCR is limited to five genes. The use of a multiplex magnetic bead-based assay was explored, allowing the use of all twelve selected marker genes and two reference genes. This 14-plex also resulted in clear profiles with sometimes higher induction factors as obtained by the 5-plex qRT-PCR method. As a result, contaminated samples could easily be distinguished from the blank samples, showing the expected profiles. These multiplex assays can thus detect these LMBs in shellfish samples and the obtained profile indicates the toxin-type present. However, compared with the neuro-2a bioassay, this assay has been shown adequate so far for only a limited number of LMBs (not all LMBs have been tested), and it is more laborious, time consuming and expensive. It should be used in cases were suspect screening outcomes from the neuro-2a bioassay cannot be explained by the toxin levels as measured with the EU-RL LC-MS/MS reference method. In chapter 5, the neuro-2a bioassay as an initial screening assay was combined with the EU-RL LC-MS/MS method for confirmation and it was investigated whether this combination is able to replace the MBA for the detection and quantification of LMBs. Samples that were tested previously in the MBA (in Chile) were used. It turned out that all samples that tested positive in the MBA were also suspect in the neuro-2a bioassay and most of these samples were confirmed to be positive for the presence of LMBs by LC-MS/MS analysis. The results confirm that the combination of the neuro-2a bioassay for screening and the EU-RL LC-MS/MS method for confirmation, is a promising alternative for the unethical MBA. The data even strongly indicated that the MBA alone probably led to false positives and the unnecessary closure of extraction areas or withdrawal of products from the market, a problem not encountered when using the neuro-2a assay in combination with LC-MS/MS. In chapter 6, a fully integrated testing strategy was presented for replacing the MBA, enabling the detection of the hydrophilic marine biotoxins. The steps and methods are discussed, and some points of attention and further developments required are addressed. Taking all together it is concluded that the proposed strategy contributes to a future with a complete animal free alterative testing strategy replacing the MBA.

The gut microbiome has become a burgeoning field of research and is proven to play a role in in many host physiological aspects, also contributing to the blood metabolome that is otherwise produced by tissues of the host organism. In order to increase the mechanistic understanding of how the microbiome influences its host health, there is a need to integrate knowledge of the composition of the gut microbiome with its functionality in terms of microbiota-mediated metabolic processes. The aim of this project was to obtain detailed insight in the mammalian-microbiome co-metabolism of endogenous metabolites and the extent to which the microbiome influences the plasma metabolome observed. This was achieved with the help of the MetaMap®Tox database in which the metabolome and toxicity data of more than 800 compounds are stored. To elucidate this interaction, metabolites produced by the gut microbiome in the blood of our animal model, Wistar rats, had to be determined. Therefore, antibiotics were used to modulate the microbial communities of Wistar-rats. After 28-day oral administration, metabolomics of plasma, feces, and cecum-content was done. Additionally, DNA was extracted from rat feces and the 16S subunit was sequenced to perform a core diversity analysis. Specific plasma metabolome patterns were established, and microbiome-related metabolites identified as key metabolites in MetaMap®Tox. Metabolites such as hippuric acid, indole derivates and glycerol appeared to be microbiome-derived or -associated plasma metabolites. In general, most changes were observed in metabolites belonging to the class of bile acids, complex lipids, fatty acids and related metabolites, as well as amino acids and related metabolites. Especially abrupt changes observed in the bile acid pool after antibiotic administration evidenced a strong influence of the microbiome on bile acid metabolism. In both feces and cecum-content, where almost all plasma metabolites could be determined as well, a treatment-related effect was observed, as well as only minimal, if any, differences between samples of male and female animals. The effects of the tested antibiotics, each possessing a different activity spectrum, could be separated from each other on the basis of the feces and cecum content metabolome. Largest changes were observed for the classes of lipids, bile acids and amino acids. It could also be shown that metabolome changes can be detected equally well in feces as in the cecum content, thus allowing to use a non invasive method for measurements of studies on metabolism by the gut microbiota. In conclusion, the functional microbial changes of the gut microbiome had to be assessed and the interactions between gut microbes and the host by applying metabolomics and taxonomic profiling had to be elucidated. The results of this thesis suggest that plasma and feces based metabolic profiling via a targeted analysis turned out to be a suitable tool to investigate the microbial functionality of the gut microbiome.

REACH requires prenatal developmental toxicity (PDT) testing for substances registered in the EU at a volume of ≥100 tonnes/year. One of the consequences is that many petroleum substances (PS) will need to be tested for their potential adverse effect on prenatal development according to the current OECD 414 testing guidelines. This will involve a huge number of experimental animals and a considerable amount of resources. Therefore, the application of in vitro alternative testing strategies may reduce the animal experimentation and resources needed to study PDT potencies of PS. Furthermore, since some PS with high concentrations of polycyclic aromatic hydrocarbons (PAH) may induce PDT whilst their gas-to-liquid (GTL) analogues, which are synthetic products completely devoid of aromatics, do not induce PDT, it was hypothesized that PDT observed for some PS is caused by certain types of PAH in these products. This hypothesis was tested in the present thesis using a battery of in vitro alternative assays. Chapter 1 provided background information and presented the aim of the thesis. In addition, the selected test substances and in vitro alternative assays used in the present thesis were also introduced. In total, 19 samples derived from 6 PS and 2 GTL product categories were tested. These samples were selected because i) they represent a series with a systematic variation in PAH content, being substances containing a range of 3- to 7-ring PAHs including extremes regarding their PAH content (with and without PAHs) and ii) in vivo PDT data for these product categories were available, enabling in vitro-in vivo comparisons. The selected in vitro alternative assays were presented, including the embryonic stem cell test (EST), the zebrafish embryotoxicity test (ZET), and a panel of CALUX reporter gene assays. Finally, the general outline of the thesis was also provided. Chapter 2 assessed the applicability of the EST to evaluate in vitro embryotoxic potencies of the DMSO extracts of 9 PS (varying in their PAH content, from 5 PS categories) and 2 GTL products (containing no PAHs) as compared to their in vivo potencies. All DMSO-extracts of PS induced a concentration-dependent inhibition of ES-D3 cell differentiation into beating cardiomyocytes at non-cytotoxic concentrations, and their potency was proportional to their 3- to 7-ring PAH content. In contrast, both GTL extracts, which are completely devoid of PAHs, tested negative in the EST. When the EST results were compared to in vivo PDT data of the corresponding PS, a good correlation was found between in vitro and in vivo results (R2: 0.97). Overall, the EST showed able to evaluate the in vitro embryotoxicity of PS, within and across categories, a result for the in vitro assay that was in line with the in vivo PDT data. The results also supported the hypothesis that PAHs are the primary inducers of the PDT resulting from PS exposure. In Chapter 3, the role of endocrine- and dioxin-like activity in the developmental toxicity of PS extracts was investigated using a panel of Chemical Activated LUciferase gene eXpression (CALUX) assays. The same set of samples as in Chapter 2 was tested in the panel of CALUX assays that included agonist and antagonist assays for the androgen, estrogen-α, progesterone, and thyroid-β receptor, and also for the aryl hydrocarbon receptor (AhR). All DMSO-extracts of the PS showed strong AhR agonist activity and weak antiprogesterone, antiandrogen, and estrogenic activities. Only minor effects were seen for thyroid-related and antiestrogenic activity with some products. PS that are grouped in the same class induced similar luciferase expression profiles, suggesting a class specific signature of effects. None of the GTL products showed a meaningful interaction with the selected receptors, thus testing negative in all CALUX assays applied. The AhR-mediated activity of the PS correlated best (R2: 0.80) with the in vitro PDT potency of the corresponding PS as quantified previously in the EST, suggesting an important role of the AhR in mediating this effect. In conclusion, a high potential for endocrine and dioxin-like activity of some PS extracts was elucidated, which correlated with their in vitro PDT, and was driven by the type and level of PAHs present in the PS extracts. The prominent AhR-mediated activity as induced by the PS extracts tested could be one of the underlying mechanisms of PDT by these substances. Chapter 4 investigated the usefulness of both the EST and the AhR CALUX assay to evaluate the in vitro PDT potency of an additional series of DMSO-extracts of HFOs, heavy PS containing mainly 3- to 7-ring PAHs, and one HRBO, a highly refined mineral oil that contains no aromatics and no PAHs. All DMSO-extracts of HFOs, but not of the HRBO, resulted in inhibition of ES-D3 cell differentiation in the EST and induced AhR-mediated activity in the AhR CALUX assay, and these potencies were was shown to be proportional to the amount of 3- to 7-ring PAHs they contain. Co-exposure of ES-D3 cells (EST) or H4IIE.luc cells (AhR CALUX assay) with the selected DMSO-extracts of PS and the AhR antagonist trimethoxyflavone (TMF), successfully counteracted the PS-induced inhibition of ES-D3 cell differentiation into cardiomyocytes as well as the AhR-mediated induction of gene expression by these substances. Moreover, also for this series of PS a good concordance was obtained when comparing the EST results with available in vivo PDT data. Altogether, the resulting data corroborate the hypothesis that PS-induced PDT is induced mainly by their 3- to 7-ring PAH content and that the observed PDT is partially mediated via the AhR. In Chapter 5, the applicability of the ZET to evaluate developmental toxicity potency of the same set of samples as tested in Chapter 2 and 3 (DMSO-extracts of 9 PS and 2 GTL products) was investigated. All PS extracts, varying in PAH level and content, were able to inhibit the development of zebrafish embryos in a concentration-dependent manner and this potency could be associated with the amount of 3-5 ring PAHs they contain. On the contrary, DMSO-extracts of both GTL products, with no aromatics, showed no effect at all in the ZET. The potencies obtained in the ZET moderately correlated with those previously reported for the EST (R2: 0.61) and the AhR CALUX assay (R2: 0.66), while the correlation with potencies reported in in vivo studies were higher for the EST (R2: 0.85) than the ZET (R2: 0.69). Combining the results obtained from the EST (Chapter 2), AhR CALUX assay (Chapter 3), and ZET (Chapter 5) ranked and clustered the test substances in line with their in vivo potencies and chemical characteristics. It was concluded that the ZET did not outperform the EST as a stand-alone assay for testing PDT of PS but confirms the hypothesis that PAHs are the major inducers of PDT by some PS, and that the ZET is a useful addition to a battery of in vitro tests able to predict the in vivo PDT of PS. In Chapter 6 we combined an exogenous biotransformation system, using hamster liver microsomes, with the EST to compare the in vitro PDT potency with and without bioactivation of two model 5-ring PAHs, benzo[a]pyrene (BaP) and dibenz[a,h]anthracene (DBA), and of PAH containing PS and GTL base oil (GTLb) extracts. In the absence of bioactivation, DBA, but not BaP, inhibited the differentiation of ES-D3 cells into beating cardiomyocytes. Upon bioactivation, BaP induced in vitro PDT, while its major metabolite 3-hydroxybenzo[a]pyrene was shown to be active in the EST as well. This indicates that BaP needs metabolic activation to exert its in vitro embryotoxic effect. The PS-induced PDT in the EST was not substantially changed following bioactivation, implying that metabolism may not play a crucial role for PS to exert their in vitro PDT effects. GTL extracts tested negative in the EST, with and without bioactivation. Altogether, although some PAH constituents require metabolic activation to be able to induce PDT, some do not and this latter also appeared to hold for the (majority of) the PS constituents responsible for the in vitro PDT of these complex substances. Chapter 7, first presented an overview of the results and main findings, which was combined with a general discussion of the data obtained and with future perspectives for follow-up studies to be performed in the near future. It was concluded that PAHs present in PS are the major inducers of PDT caused by these substances and that this was successfully and adequately assessed using several in vitro alternative assays, including the EST, ZET, and AhR CALUX assay. The results obtained in Chapter 2, 4, and 5 of the thesis were used in a QSAR (quantitative structure activity relationship) approach to predict the in vivo PDT of a series of PS based on their PAH content. More PS extracts, ideally from different PS categories than those tested in the present thesis, should be tested to broaden the applicability domain of the proposed assay battery and the related QSAR approach for PDT testing of PS UVCBs in the future.

Phthalates are diesters of phthalic acid that are widely used in industry and personal care products resulting in exposure via ingestion, inhalation and dermal routes. There is an interest in the safety evaluation of phthalate exposure because these compounds are ubiquitous environmental contaminants with endocrine-disrupting properties, suspected to interfere with developmental androgen action, possibly leading to adverse effects on reproductive function. Toxicological properties of phthalates, the presence of phthalates in polyethylene terephthalate (PET) bottles as impurities, the high and regular consumption of bottled water, and the uncertainty about the impact of storage conditions of PET bottled water on migration of phthalates into the water, initiated the interest in their presence in bottled water and the accompanying risk assessment. In this study, common Iranian brands of bottled water were screened for phthalates. The effect of storage temperature on selected target chemical concentrations was investigated. A toxicological risk assessment was conducted to determine the potential health risks associated with the consumption of the bottled water. Along with indirect exposure assessment, a human biomonitoring approach was applied to facilitate better human exposure assessment of individual phthalates and their mixtures providing important information for identifying exposure sources and the contribution of intake from bottled water to the total daily intake. Chapter 1 of the thesis presents an introduction to the topic, the toxicological properties of phthalates, risk assessment strategies and the regulatory status of phthalates. Chapter 2 of the thesis describes the development of a method to extract phthalates from bottled water by applying surface-functionalized magnetic particles (MPs) as the adsorbent used in Magnetic Solid-Phase Extraction (MSPE). Based on the results obtained, it was concluded that the MSPE-GC-MS method developed provides a new method for the determination of phthalates in water samples. To extend the work to real samples chapter 3 presents the occurrence and concentrations of common phthalates (dibutyl phthalate (DBP), butyl benzyl phthalate (BBP), and diethylhexyl phthalate (DEHP) ) in PET bottled water locally produced in the Iranian market and stored under various common storage conditions. According to the results obtained, an increase in temperature and/or in the duration of storage increases phthalate migration. The highest concentrations of all phthalates were observed when bottled water samples were kept at 40 °C for 45 days. DEHP in bottled water was the most abundant phthalate under all storage conditions, although the observed level of DEHP in the worst-case scenario (40 °C for 45 days) was still much lower than the DEHP maximum concentration limit (MCL) in bottled water (MCL= 6 µg/L) set by the U.S. Food and Drug Administration (U.S. FDA). When comparing the concentrations of DBP, BBP and DEHP with initial levels in the bottled water, the results demonstrate that the release of phthalates was not substantial under all storage conditions, and especially minimal at low temperatures ( children > lactating women > teenagers > adults > pregnant women. However, for all age groups, none of the individuals exceeds existing intake limit values for DEP. Due to the anti-androgenic activity of some phthalates, in chapter 5 the cumulative health risks in pregnant and lactating women posed by combined exposure to BBP, DBP, and DEHP via consumption of bottled water was estimated. To this end, hazard quotient (HQ) values, representing the margin between health based guidance values (EPA RfD values) and estimated exposures, and hazard index (HI) values, representing the sum of HQ values of individual phthalates, were determined. The results of the study showed that the HQ values for individual phthalate intake via bottled water consumption in pregnant and lactating women were much lower than 1, and cumulative risk assessment for combined phthalate exposure demonstrated that the HIs for anti-androgenic effects were also lower than 1 which implies that adverse effects are unlikely to occur. In chapter 6 of the thesis a systematic review method was used to investigate whether the phthalate exposure would be a factor contributing to the development of autism spectrum disorders (ASD). The results of this systematic review revealed that only a limited number of studies has addressed phthalates in relation to autism. A total of five studies met the inclusion criteria and were included in the review. Of the 5 studies, two studies were cohort studies both from the U.S.A. and three were case-control studies conducted in the U.S.A., Italy and Turkey. Because of the heterogeneity in the type of included studies, different methods of assessing exposure to phthalates and the use of different statistics for summarizing the results, a meta-analysis could not be performed to combine the results of included studies. The review showed equivocal evidence for a possible connection between exposure to phthalates and ASD. Further comprehensive research is needed with appropriate attention to exposure assessment and relevant pre and post-natal confounders. In the next step we set our goal to get better insight in the total phthalate exposure of Iranian children, and to assess the proportion of phthalate intake from bottled water to the total daily intake. This was done using biomonitoring based exposure assessment. Chapter 7 of the thesis shows the data on the levels of phthalate metabolites in the spot urine samples of children and adolescents. We applied a calculation model based on the creatinine-adjusted urinary metabolite concentrations to obtain the EDIs for DEHP, DBP and BBP. The EDI values thus obtained were compared to available health-based guidance values (RfD and TDI values based on anti-androgenic effects). Assuming additive effects, the cumulative risk for combined exposure were estimated for three phthalates based on anti-androgenicity as the critical effect. The results from the risk assessment suggest that Iranian children and adolescents are exposed to low levels of a mixture of these phthalates. Risk assessment indicates that not only the exposure to the single phthalates, but also the combined exposure would not raise a safety concern. However, people typically come into contact with several chemicals with anti-androgenic properties in addition to the investigated phthalates in this study, which may also contribute to combined anti-androgenic effects. This indicates that a risk assessment of combined exposure including other anti-androgenic chemicals would be required to determine whether combined exposure to anti-androgenic chemicals is below acceptable levels. Comparison of the exposure values obtained to those obtained based on indirect estimates in earlier chapters of the thesis, revealed that bottled water provides only a limited contribution to total daily phthalates exposure in Iran. Chapter 8 presents a discussion of the results obtained and also presents some perspectives for future research and risk management of exposure to phthalates in Iran.

The implementation of regulations for protecting both humans and the environment from potential chemical health hazards, as well as the increase of global pressure for reducing, refining and replacing animal experiments promote the development and application of alternatives to in vivo developmental toxicity studies. Due to the complexity of the reproductive cycle, combined in vitro approaches, focusing on morphological, molecular and toxicokinetic parameters, could better define the developmental toxicity of chemicals. In this thesis, azoles, which are a group of chemicals with antifungal activity, are under investigation. These compounds show marked differences in developmental toxicity potency and similarities with retinoic acid (RA)- related teratogenicity. Chapter 1 of this thesis introduced information regarding the background of reproductive and developmental toxicology, including scientific concerns and the impact of past teratogenic outcomes on the society. For screening developmental teratogens, in vitro approaches have been proposed and successfully applied. Their combination may better mimic the in vivo embryo and, therefore, increase the accuracy in predicting possible developmental toxicants. Additional co-implementation of molecular approaches may give an insight in the mode of action underlying the observed effects. Azoles were selected in the present thesis due to evidence for possibly increasing developmental toxicity through dysregulating the balance of the RA pathways in the mammalian system. The chapter also described the objectives and outline of the research. In chapter 2, we examined the time- dependent developmental effects in rat embryos exposed in vitro to flusilazole (FLU), and their link to RA mediated pathways. To this end, we assessed the effects of 4-hour exposure of whole embryo culture (WEC) embryos to 300μM FLU during four developmental time windows (0-4, 4-8, 24-28 and 44-48 h), evaluated morphological parameters, as well as expression and localization of five genes directly or indirectly linked with the RA pathway. A stage- specific gene expression response of cultured rat embryos exposed to FLU was detected, which preceded the development of morphologically observable malformations. During all the tested time windows, the most pronounced effect was observed in the regulation of RA-related genes. Therefore, it was concluded that such biomarkers can be employed as useful tools for early detection of possible teratogenic properties of compounds that belong to the triazole- group or of compounds with a similar teratogenic mode of action. Chapter 3 provides mechanistic insight into the embryotoxicity of six azoles tested in the rat WEC. Here, we evaluated dose-dependent embryotoxicity of azoles in the rat WEC, calculating the concentration at which the total morphological score (TMS) is 10% decreased (ID10). For the azoles tested we compared the in vitro ID10 for embryotoxicity to the in vivo effective doses, while we also performed a comparative analysis for understanding the toxicological and pharmacological mode of action of azoles in the rat WEC at the level of the transcriptome. Functional analysis of differential gene expression after 4 hours exposure at the ID10 revealed regulation of the sterol biosynthesis pathway and embryonic development genes, dominated by genes in the RA pathway, albeit in a differential way. FLU, ketoconazole and triadimefon were the most potent compounds affecting the RA pathway, while in terms of regulation of sterol function, difenoconazole and ketoconazole showed the most pronounced effects. A similar analysis at the 24-hour time point indicated an additional time-dependent difference in the aforementioned pathways regulated by FLU. Strong in vivo embryotoxic azoles showed also an increased regulation of the RA pathway when tested in vitro. On the other hand, weak or non- embryotoxic azoles showed a non-significant effect on genes that belong in the RA pathway. These observations led us to the conclusion that the toxicological mode of action of azoles was mediated through the RA pathway. In summary, the rat WEC assay in combination with transcriptomics could add mechanistic insight into the embryotoxic potency ranking and functional efficacy of the tested compounds, showing Cyp26a1 and Cyp51 as leader biomarkers of the off- and on- target effects, respectively. Similarly to the previous chapter, in chapter 4, the potency ranking of the majority of the twelve tested azoles obtained based on the TMS in the WEC assay was in agreement with the in vivo potency ranking. Additionally, our expanded transcriptomics data, including gene specific responses of twelve azoles tested at their ID10 in the rat WEC for 4 hours, confirmed the observations of chapter 3 with another set of azoles. Potent embryotoxicants in both in vivo and in vitro assays caused more pronounced effects on the dysregulation of RA- mediated genes. Furthermore, azoles with more pronounced effects on the sterol biosynthesis mediated pathway were tested at a higher concentration, but with the same level of effect (ID10). Due to the increased concentration needed for reaching the same level of morphological effects and the absence of RA-mediated pathway regulation, these azoles were considered as more favourable candidates for clinical and agricultural use. Focusing on monitoring the fungicidal activity of azoles, we also detected an increased sensitivity of the expression of Msmo1, which is an enzyme participating in converting lanosterol for synthesizing cholesterol in the mammalian sterol biosynthesis, together with Cyp51and Nsdhl. This observation led us to the conclusion that Msmo1 could be a better biomarker of effect on the sterol biosynthesis pathway compared to the classical biomarker of this pathway, Cyp51, and this may be of use for further improvement of the assessment of fungicidal activity of azoles or chemicals with similar mode of action. Chapter 5 shows the value of combining toxico-dynamic and -kinetic in vitro approaches for embryotoxicity testing of azoles. We also report on the alterations in gene expression induced by azoles. Both the WEC assay and the embryonic stem cells test (EST) predicted the in vivo potency ranking of the twelve tested azoles with moderate accuracy. Combining these results with relative placental transfer rates (Papp values) as determined in the BeWo cell culture model, increased the predictability of both WEC and EST, with R2 values increasing from 0.51 to 0.87 and from 0.35 to 0.60, respectively. The comparison of these in vitro systems correlated well (R2 = 0.67), correctly identifying the strong and weak embryotoxicants. Evaluating specific gene responses related with the toxicological and fungicidal mode of action of the tested azoles in WEC and EST, we observed that the differential regulation of Dhrs3 and Msmo1 reached higher magnitudes in both systems compared to Cyp26a1 and Cyp51. Establishing sensitive biomarkers across all the in vitro systems for studying the underlying mechanism of action of chemicals, such as azoles, is valuable for comparing alternative in vitro models and for improving insight in the mechanism of developmental toxicity of chemicals. Chapter 6 of this thesis presented the general discussion and future perspectives on different topics raised based on the results obtained in the previously described experimental chapters. The results suggested that the combination of in vitro assays for screening the developmental toxicity of azoles may lead to predictions that are more accurate and in agreement with the in vivo observations. The addition of toxico-kinetics, which the BeWo placental transfer model offered, notably improved the correlations of in vivo and in vitro data. Furthermore, the co- implementation of transcriptomics and the identification of gene biomarkers revealed that despite the tested azoles were classified in the same chemical group, they might have a different mode of toxicological action. In conclusion, future combination of in vitro and in silico alternative approaches appear to be of advantage for screening and prioritizing chemical testing, in the process of assessing the consequences of chemical exposure for human health and the environment.

Alkenylbenzenes including estragole, methyleugenol, safrole, elemicin, apiol, and myristicin are naturally occurring in many herbs such as parsley, dill, basil, tarragon, fennel and nutmeg (Kreydiyyeh and Usta, 2002, Smith et al., 2002, Semenov et al., 2007). Estragole, methyleugenol and safrole are genotoxic and carcinogenic in rodent bioassays inducing liver tumors (Boberg et al., 1986, Boberg et al., 1983, Drinkwater et al., 1976, Miller et al., 1983, Swanson et al., 1981, Wiseman et al., 1985, Wiseman et al., 1987, Wislocki et al., 1977). Because of that, the use of methyleugenol, safrole and estragole as pure substances in foodstuff has been prohibited in the EU from September 2008 onwards (European Commission (EC), 2008). For apiol and myristicin data for their risk assessment are limited and more research is needed to support the evaluation of the risk resulting from consumption of products containing these compounds (WHO, 2009). The aim of the current thesis was to perform a mode of action based risk assessment of exposure to low doses of apiol and myristicin by using physiologically based kinetic (PBK) modelling based read-across from other alkenylbenzenes and to use the results obtained for risk assessment of consumption of plant food supplements (PFS) and other botanical products containing parsley and dill. Chapter 1 provides general background information to alkenylbenzenes especially apiol and myristicin, a description of the chemical, metabolic and toxicity characteristics of apiol and myristicin and other structurally related alkenylbenzenes, a brief outline of the method used for their risk assessment and a short introduction to PBK modelling. Besides that, Chapter 1 include the aim of the current thesis. In Chapter 2 and Chapter 3, PBK models for respectively apiol and myristicin in male rat and human were defined, enabling prediction of dose-dependent effects in bioactivation and detoxification of these alkenylbenzenes. The PBK model based predictions were subsequently compared to those for safrole enabling estimation of a BMDL10 for apiol and myristicin from read-across from the BMDL10 available for safrole, thereby enabling risk assessment of current dietary exposure to apiol. In Chapter 4 and 5, the risk assessment of exposure to apiol and related alkenylbenzenes through drinking of parsley and dill based herbal teas and consumption of parsley and dill containing PFS was performed using the BMDL10 values derived in Chapter 2 and 3. The results showed that consumption of parsley and dill based herbal teas and PFS would be a priority for risk management if consumed for longer periods of time. Chapter 6 includes a general discussion of the thesis results obtained and the future perspectives that describe the needs to further research, based on alternatives for animals testing, to improve the risk assessment approaches for different botanical preparations. Altogether, the results obtained through different thesis chapters show that integration of different approaches provides the basis for a mode of action and PBK modelling based read-across from compounds for which tumor data are available to related compounds for which such data are lacking. This can contribute to the development of alternatives for animal testing and will facilitate the risk assessment of compounds for which in vivo toxicity studies on tumor formation data are unavailable.

Botanicals and botanical preparations have become widely available on the market in the form of food supplements and other food preparations. In Europe the safety of botanicals and derived food products placed on the market has to comply with the general requirements set out in regulation (EC) No 178/2002. In spite of this regulation, the risks of botanicals and botanical preparations are generally not assessed before they enter the market. The present thesis aimed to perform a risk assessment of some selected botanicals and derived preparations focusing on samples containing the so-called alkenylbenzenes, including myristicin, apiol, safrole, methyleugenol, elemicin, and estragole. Samples containing alkenylbenzenes may be of concern because these compounds may be genotoxic and carcinogenic displaying similar characteristics regarding mode of action (MOA) and tumor formation. The aim of the present study was to develop a MOA based approach for read-across in risk assessment from safrole, for which in vivo toxicity studies are available to myristicin and apiol for which tumor data are not reported, thus contributing to alternatives in animal testing. This was done using physiologically based kinetic (PBK) modelling. Botanical preparations included in the risk assessments performed in the present thesis were basil-containing pesto and nutmeg-based plant food supplements (PFS). Given that several of the preparations analyzed contained several alkenylbenzenes a risk assessment of combined exposure to these alkenylbenzenes was included as well, to give a better understanding of when risk management actions would be needed for botanicals and derived preparations containing these ingredients. The introduction chapter starts with a short background and definition of the aim of the thesis, followed by a description of the physicochemical properties, natural occurrence and estimated daily intake (EDI) of the alkenylbenzenes, as well as of their ADME (absorption, distribution, metabolism and excretion) characteristics, carcinogenicity and MOA, the relevant risk assessment approach, and finally PBK modelling. Chapter 2 and 3 of the thesis describe the development of PBK models for myristicin and apiol in male rat and human, enabling the prediction of dose-dependent effects in bioactivation and detoxification of these alkenylbenzenes. This allows comparison of the PBK model-based prediction of bioactivation of myristicin or apiol to the PBK model-based predictions for bioactivation of the structurally related compound safrole, enabling estimation of a BMDL10 for myristicin and apiol from read-across to the BMDL10 available for safrole, thereby facilitating risk assessment of current dietary exposure to myristicin or apiol. The results from PBK analysis support that risk assessment of myristicin may be based on the BMDL10 derived for safrole, and that risk assessment of apiol may be performed using a BMDL10 value of 3 times the BMDL10 for safrole. These results enabled a preliminary risk assessment of current exposure to these alkenylbenzenes showing that the current exposure to myristicin results in MOE values indicating a priority for risk management (Chapter 2), while current exposure to apiol does not raise a concern (Chapter 3). The results obtained illustrate that PBK modeling can facilitate a read-across in risk assessment from a compound for which in vivo toxicity studies are available to a related compound for which tumor data are not reported, thus contributing to alternatives in animal testing. Subsequently, in Chapter 4 the BMDL10 values obtained were used to perform a risk assessment for alkenylbenzenes in basil-based pesto sauce. To this end the levels of methyleugenol and other alkenylbenzenes in basil-containing sauce of pesto were quantified enabling an associated risk assessment based on the MOE approach, taking into consideration the possible combined exposure to different alkenylbenzenes and the BMDL10 values as defined in literature and in chapter 2 and 3 of the thesis. The MOE values obtained generally indicated a priority for risk management when assuming daily consumption of basil-based pesto sauce. It was concluded that consumption of pesto sauces would be of concern if consumed on a daily basis over longer periods of time. Chapter 5 evaluates the risk of exposure to myristicin and related alkenylbenzenes through human exposure to nutmeg-based PFS based on the MOE approach. Chemical analysis of various PFS samples was performed and the amount of alkenylbenzenes that would be consumed through consumption of these nutmeg-based PFS where quantified. MOE values where calculated for the individual alkenylbenzenes as well as taking into account the presence of more than one alkenylbenzenes within selected PFS samples. It was concluded that the results reveal that daily nutmeg-based PFS consumption following recommendations for daily intake for especially longer periods of time would raise a concern. Chapter 6 summarizes the results obtained in the thesis, compiles the overall discussion and presents the future perspectives that follow from the results obtained.

There is currently much interest in the development of functional foods aiming at the prevention of the development of some diseases, for example cancer, by the introduction of selected natural substances at elevated levels into the diet. The rationale for this approach is based especially on epidemiological data that indicate that food items containing such chemicals may reduce the risk of these diseases in humans. Epidemiological studies indicate, for example, that diets rich in fruit and vegetables protect against a variety of diseases, including heart diseases and certain forms of cancer. However, identification of the actual ingredient in a specific diet responsible for the beneficial health effects remains an important bottleneck for translating observational epidemiology to development of a functional food ingredient. The protection against cancer afforded by fruit and vegetables has been attributed to antioxidant micronutrients such as vitamin C, beta-carotene and vitamin E, which may act at many sites, including the stomach, intestine, lung and bladder. However, present scientific attention is focusing as well on the significance of other minor dietary components, notably the flavonoids as protectants against disease. Flavonoids are widespread in nature and are found in considerable quantities in fruits, vegetables, seeds, peel and tubers. The average Western diet may provide up to 1 g of flavonoids per day. Numerous in vitro studies show that flavonoids are potent antioxidants and metal chelators. Their potential as anti-inflammatory, antiallergic and antiviral compounds has also attracted attention. These studies provide the basis for the present rapidly increasing interest for the use of flavonoids as functional food ingredients. As a result increased human exposure to flavonoids can be expected in the near future. In shops and at the internet, food and food supplements based on (iso)flavonoids as functional ingredients are marketed. This, although hard scientific data supporting the health claims as well as data allowing a balanced risk-benefit evaluation are lacking. For flavonoids increased future human exposure regimens induce the question on their pro-oxidant chemistry. There is considerable evidence that some flavonoids are mutagenic in both bacterial and mammalian experimental systems. A high incidence of gastric cancer in some human populations has been linked to consumption of wine containing potentially mutagenic flavonoids (Tamura et al. , Proc. Natl. Acad. Sci. USA. 77, 4961-4965, 1980, Hoey et al. , Am. J. Epidemiol., 113, 669-974, 1981). Relatively little is understood about either the toxicity or protection afforded by flavonoids in humans.Since flavonoid quinone/quinone methides have been suggested as the major metabolites responsible for the possible pro-oxidant toxicity and mutagenicity of flavonoids, characterisation of flavonoid quinone chemistry is of importance. However, little information is available on the structure and reactivity of these flavonoid oxidation products. Therefore, the objective of this thesis was to investigate the pro-oxidant chemistry of flavonoids and to perform structure activity studies on the chemical behaviour of 3',4'-dihydroxyflavonoids with special emphasis on the nature and reactivity of the quinone/quinone methide type metabolites formed. Using the GSH trapping method, HPLC, LC/MS, MALDI-TOF, 1H NMR, 13C NMR and quantum mechanical computer calculations the quinone/quinone methide chemistry of a series of 3',4'-dihydroxyflavonoids could be characterised.The results provide insight in structure-activity-relationships for the pro-oxidant chemistry of these electrophilic quinone/quinone methide flavonoid metabolites. The results obtained also reveal an unexpected pH-dependent electrophilic behaviour of B ring catechol flavonoids. Furthermore the results of this thesis also reveal, for the first time, evidence for the pro-oxidative chemistry of quercetin in a cellular in vitro model. The formation of these glutathionyl-flavonoid adducts provides evidence for the actual pro-oxidative formation of reactive quinone type metabolites from B ring catechol flavonoids in the selected cellular in vitro model using melanoma cells. Oxidation of the catechols to quinones and their isomeric quinone methides generates potent electrophiles that could alkylate DNA. Interestingly, the structural requirements essential for good antioxidant activity match the requirements essential for pro-oxidant action and quinone methide formation. Altogether, the pro-oxidant behaviour of flavonoids and their quinone/quinone methides are far from straight forward and need to be re-evaluated especially in the framework of the risk-benefit evaluation of the use of these flavonoids as functional food ingredients and/or food supplements. SamenvattingEr is momenteel veel interesse voor de ontwikkeling van functionele voedingsmiddelen (functional foods), met als doel het voorkomen van het ontstaan van ziekten zoals bijvoorbeeld kanker, via het in verhoogde mate introduceren van geselecteerde natuurlijke bestanddelen in het dieet. De basis voor deze aanpak wordt momenteel met name gevonden in epidemiologische studies die laten zien dat diëten rijk aan specifieke voedselcomponenten of ingrediënten de kans op bepaalde ziekten bij de mens verlagen. Zo geven epidemiologische studies bijvoorbeeld aan dat diëten die rijk zijn aan fruit en groenten beschermen tegen een aantal ziekten zoals hartziekten en bepaalde vormen van kanker. Echter, het identificeren van de belangrijke ingrediënten in het betreffende dieet die het gezondheidsbevorderende effect tot stand brengen is een knelpunt voor het vertalen van de resultaten uit de epidemiologie naar de ontwikkeling van een functioneel voedingsingrediënt.De bescherming tegen kanker door groenten en fruit is toegeschreven aan antioxidanten zoals vitamine C, beta-caroteen en vitamine E, die op vele plaatsen in het lichaam, zoals de maag, darmen, long en de blaas actief zijn. Wetenschappelijk wordt momenteel veel aandacht besteed aan het mogelijke belang van andere belangrijke dieet componenten, zoals flavonoïden, als beschermende ingrediënten tegen ziekte. Flavonoïden komen in de natuur veel voor, en worden met name in hoge concentraties gevonden in fruit, groenten, knollen en zaden. Het gemiddelde Westerse dieet bevat ongeveer 1 gram aan flavonoïden per dag.Vele in vitro studies tonen aan dat flavonoïden goede antioxidanten en metaal chelatoren zijn. Daarnaast hebben ze anti-inflammatoire, anti-allergische en anti-virale eigenschappen die van belang worden geacht. Deze bevindingen verschaffen de basis voor de momenteel snel groeiende interesse om flavonoïden te gebruiken als functionele voedingsingrediënten. Als gevolg hiervan zou er in de nabije toekomst een toename in de opname van flavonoïden via het dieet verwacht kunnen worden. In winkels en via het internet worden voedingsmiddelen en voedingssupplementen gebaseerd op (iso)flavonoïden als functionele voedingsingrediënten verkocht. Dit, terwijl zowel de wetenschappelijke onderbouwing voor de gezondheidsclaims als gegevens die een gebalanceerde "risk-benefit" analyse mogelijk maken, nog ontbreken. In het geval van verhoogde toekomstige blootstelling van mensen aan flavonoïden worden voor de risk-benefit evaluatie vragen van belang rond hun mogelijk pro-oxidatieve chemisch gedrag. Er zijn aanwijzingen dat sommige flavonoïden mutageen zijn in zowel bacteriële als zoogdier in vitro test systemen. Een verhoogde mate aan maagkanker in bepaalde humane populaties is in verband gebracht met de consumptie van wijn met daarin mogelijk mutagene flavonoïden (Tamura et al. , Proc. Natl. Acad. Sci. USA. 77, 4961-4965, 1980, Hoey et al. , Am. J. Epidem., 113, 669-974, 1981). Alles samenvattend is er eigenlijk weinig bekend van de schadelijke maar ook van de gezondheidsbevorderende effecten van flavonoïden.Omdat flavonoid chinon/chinon methides genoemd zijn als de belangrijkste metabolieten die verantwoordelijk zouden zijn voor de mogelijke pro-oxidatieve toxiciteit en mutageniteit van flavonoïden, is karakterisering van deze pro-oxidant chemie van flavonoïden van belang. Echter er is weinig bekend over de structuur en de reactiviteit van deze flavonoid oxidatie producten. Daarom was het doel van deze studie de pro-oxidant chemie van flavonoïden te onderzoeken en een structuur-activiteits studie uit te voeren naar het chemische gedrag van 3',4'-dihydroxyflavonoïden. Daarbij werd speciale aandacht besteed aan de aard en reactiviteit van de gevormde chinon/chinon methide metabolieten. Met behulp van de GSH-trapping methode, HPLC, LC/MS, MALDI-TOF, 1H-NMR, 13C-NMR en kwantum-chemische computerberekeningen kon de chinon/chinon methide chemie van een serie 3',4'-dihydroxyflavonoiden gekarakteriseerd worden.De verkregen resultaten geven inzicht in de structuur-activteits relaties voor de pro-oxidatieve chemie van de electrofiele chinon /chinon methides metabolieten van de flavonoïden. De resultaten laten ook een onverwacht effect zien van de pH op het electrofiele gedrag van de B-ring catechol flavonoïden. Bovendien laten de resultaten van het proefschrift zien dat zelfs onder reducerende omstandigheden in een cellulair in vitro model (melanoma cellen) de pro-oxidatieve chemie van quercetine van belang kan zijn. Met name de vorming van glutathion-flavonoid conjugaten is een bewijs dat in het gekozen cellulaire model de pro-oxidatieve vorming van reactieve flavonoid chinon/ chinon methide metabolieten is opgetreden. Oxidatie van de catecholen naar chinonen en hun isomere chinon methides genereert electrofielen die DNA kunnen alkyleren. Van belang is dat de structurele randvoorwaarden die een flavonoid een goede antioxidant maken gelijk blijken te zijn aan de structurele kenmerken die essentieel zijn voor pro-oxidant gedrag en chinon methide vorming.Al met al is de pro-oxidant chemie van flavonoïden en van hun chinon /chinon methides verre van recht toe recht aan gebleken en zou de pro-oxidatieve chemie en de toxiciteit van de flavonoïden in het kader van hun gebruik als functional food ingredienten beter onderzocht en afgewogen moeten worden, rekening houdend met hun mogelijk gezondheidsbevorderende effecten.

In this thesis, microperoxidase-8, Fe III MP-8, and the manganese variant Mn III MP-8, were studied. Insight in i) the mechanism of oxygen exchange between the oxo group of porphyrin high-valent metal-oxo species and solvent, ii) their cytochrome P450 (P450) and peroxidase catalytic reactivity and iii) the formation of their catalytic reactive intermediates is provided.Determinants of the peroxidase and P450 chemistry for Fe III MP-8 and Mn III MP-8 (Chapter 1)In the Chapter 1 of this thesis, peroxidase and P450 enzymes were compared in order to understand the major differences between both types of enzymes explaining the differences in their chemistry. It was shown that peroxidases and P450s share similar intermediates in their catalytic cycles but that both types of enzymes were structurally different. Differences in the distal environment of the heme and in the nature of the axial ligand can be regarded as having an influence on their chemistry.1. Differences in the heme distal environmentA comparison of the distal heme cavity for peroxidase and P450 enzymes shows that peroxidases do not have a clearly defined hydrophobic distal pocket allowing the binding of organic substrates near the heme iron centre. With the exception of hydroperoxides which bind to the heme iron centre and are subsequently reduced by the heme group to generate catalytic reactive intermediates, the organic substrate is proposed to bind on theδ-meso heme edge thereby allowing electrons transfer to the heme prosthetic group. The fact that organic substrates cannot bind to the distal part of the peroxidase active site near the oxo group of Compound I, explains why these enzymes do not support oxygen transfer reactions. Moreover the fact that only for peroxidase mutants where some "extra space" has been created on the distal part of the active site, monooxygenases reactions are observed but catalysed at a lower rate than for heme-thiolate enzymes, corroborates the assumption that substrate binding near the heme iron favours oxygen transfer. Evolutionary speaking, peroxidases are well designed to reduce small oxygen containing molecules, hydroperoxides, and to convert them into oxidative power which may be beneficial to the cell because of the formation of useful compounds such as polymers. P450s, however, have a distal active site pocket which allows substrate binding near the heme iron center. From an evolutionary point, P450s are well designed to convert harmful xenobiotics or modifying membrane lipid components in the cell. The fact that the substrate binds close to the site where the reactive catalytic intermediates are generated favours in-situ conversion and direct oxygen transfer to the substrate. Moreover this oxygen transfer can be mediated by the various catalytic intermediates of the reaction cycle preceding the Compound I analogue and not only by the Compound I analogue. The physical proximity of the substrate and the reactive intermediates favours a reaction of the catalytic intermediates with the substrate over a transformation of a given catalytic intermediate into the following catalytic intermediate of the reaction cycle of the enzyme.2. Differences in the nature of the proximal axial ligandPeroxidases and P450s differ by the nature of their proximal axial ligand. The negative charge of the cysteinate ligand of P450s is proposed to lower the oxidation potential of the Compound I analogue intermediate thereby favouring oxygen transfer to substrate over electron abstraction from substrate . The negative charge of the sulphur atom is somewhat reinforced by the involvement of a network of hydrogen bonds from the residues of the heme proximal active site to the sulphur of the thiolate group. Theoretical MO-calculations also have shown that the sulphur orbitals of the cysteinate mix with the a 2u cationic state of the porphyrin high-valent iron-oxo intermediate along the reaction pathway. This is not observed for the proximal histidine ligand of peroxidases. In peroxidases, the neutral histidine ligand is proposed to favour electron abstraction from substrate over oxygen transfer to substrate .Compound I oxo exchange: implications for oxygen transfer (Chapters 2 and 3)In the field of heme-based catalysis, labelling studies are used to discriminate between peroxidase-type of chemistry and P450-type of chemistry. Tracing of the oxygen donor atoms in the product allows to differentiate between true oxygen transfer , where the oxygen atom inserted in the substrate originates from the primary oxygen donor, P450 chemistry, and apparent oxygen transfer where the oxygen atom inserted in the substrate originates from the solvent, peroxidase chemistry. However, heme-enzymes and the corresponding models were shown to exchange the oxo group of their Compound I intermediate, or Compound I analogue, with bulk solvent. Chapters 2 and 3 of this thesis are oriented on the mechanistic aspects of the exchangeability of the oxo group of Compound I with bulk solvent. In Chapter 2, HRP and the labelled oxygen donor are left incubating during an exchange step of variable time length before the substrate, aniline, is added. The conversion of aniline results in the insertion of one oxygen atom in the substrate by para -hydroxylation. The para -hydroxylated aniline formed during the conversion step subsequent to the exchange step was analysed by MS. This reveals an increasing percentage of the incorporation of oxygen atoms originating from the solvent when the duration of the exchange step is increased. When catalase was added between the two steps of the experiment, no product was formed. This shows that the heme catalyst induces oxygen exchange between bulk water and H 2 O 2 , to form H 2 O 2 containing oxygen atoms issued from the solvent. A mechanism explaining this oxygen exchange for an axially coordinated heme has been suggested including the reversibility of the formation of Compound I for HRP, but also for Fe III MP-8, hemin and hematin. Actually, water becomes a substrate for the porphyrin high-valent iron-oxo intermediate and competes with the organic substrate, when present, for being oxidised. In the study presented in Chapter 2, substrate oxidation and water oxidation were decoupled thereby emphasising the water oxidation step. Moreover this study indirectly suggests that oxidised heme-based systems can catalyse the formation of peroxide bonds at the (modest) apparent rate of≈1 s -1 . This is a relevant outcome of this study, since the design of molecules containing a peroxide bond is of fundamental importance for oxidation chemistry and is of particularly interest for the chemical industry.In Chapter 3, the same argumentation as proposed in Chapter 2 is supported by additional exchange experiments with iron and manganese water-soluble porphyrins. This chapter is an attempt to unify the views on porphyrin high-valent metal-oxo/solvent exchange processes. Direct oxygen exchange of the oxo group of Compound I, for a solvent oxygen atom is not found to account for the regeneration of H 2 O 2 during the exchange step. Also a mechanism such as oxo/hydroxo tautomerism where the transfer of two protons from the trans water axial ligand to the oxygen of the oxo group results in an exchange of the oxo oxygen cannot explain Fe III MP-8 supported oxygen exchange. The fact that Fe III MP-8 has an axial histidine ligand prevents solvent binding trans to the oxo group, which would be a prerequisite for proton transfer resulting in oxygen exchange between both axial ligands. Obviously this cannot explain oxygen exchange catalysed by axially ligated heme or metalloporphyrins like Fe III MP-8. The reversible formation of Compound I, proposed in Chapter 2, explains the results in a clear way.Catalytic reactivity of Fe III MP-8 and Mn III MP-8 (Chapter 4 and 5)The heme-peptide model, Fe III MP-8, shows peroxidase activity and can also be used as a heme-enzyme model for studying P450 chemistry based on the fact that, upon addition of ascorbate, the reactive species which are responsible for peroxidase chemistry are scavenged. As a consequence Fe III MP-8 can be active in two modes: the peroxidase mode where the catalysis is dominated by porphyrin high-valent iron-oxo intermediates and the P450 mode where the porphyrin high-valent iron-oxo intermediates are (partially) scavenged and catalysis is mainly performed by intermediates appearing prior to Compound I in the reaction cycle. The switch between the two modes is provided by the addition of ascorbate to the system in the case of P450 chemistry, which acts as scavenger for porphyrin high-valent iron-oxo intermediates.In Chapter 4 the reactivity of Fe III MP-8 for H 2 O 2 supported O- and N-dealkylation, a peroxidase/P450 type of reaction, has been investigated. In the peroxidase mode, i.e. without ascorbate addition, the rate of conversion of the substrates is correlated with their quantum mechanically calculated first ionisation potential. This indicates that their conversion proceeds via an initial electron abstraction from the substrate. This is corroborated by the observation of a large amount of polymerisation products together with the formation of small amounts of N-dealkylated products. In contrast however O-dealkylation was not observed. This observation that O-alkylated substrates are not dealkylated in the peroxidase mode can be related to the fact that their first ionisation potential is too low. In the P450 mode, i.e. in the presence of ascorbate, O- and N-alkylated substrates are both converted and a correlation with the calculated first ionisation potential of the substrates no longer exists. This suggests that O- and N-dealkylations in the P450 mode proceed via a non-radical type of mechanism and thus through other intermediates than Compound I and Compound II, since ascorbate is a scavenger of Compound I and II. As an alternative reactive species the non-radical type PorFe III -(hydro)peroxo intermediate, may be the species involved in the P450 mode of Fe III MP-8 supported O- and N-dealkylation. The PorFe III (hydro)peroxo intermediate is known as Compound 0 and appears before Compound I in the catalytic cycle. Mechanisms for Compound 0 supported O- and N-dealkylations are discussed in detail.The aim of the investigations described in Chapter 5 was to gain information on the nature of the different reactive species involved in MP-8 supported catalysis comparing again peroxidase and P450 reactions. For this purpose the manganese variant of Fe(III)MP-8, Mn(III)MP-8 was synthesised. Iron and manganese porphyrin complexes have a similar chemistry but differ in their reaction kinetics providing information concerning which type of intermediate is involved in catalysis. The conversion of guaiacol (2-methoxyphenol), ortho -dianisidine (3,3'-dimethoxybenzidine) and aniline catalysed by Fe III MP-8 and Mn III MP-8 was studied. The first two substrates are models for the peroxidase reactivity of heme-enzymes and the third one is a model for the P450 reactivity of heme-enzymes. The pH-dependence of the rate of conversion, k cat , of each substrate was studied for Fe III MP-8 and Mn III MP-8 supported conversions, using H 2 O 2 as oxygen donor. For the peroxidase mode it was shown that the optimal pH for Mn III MP-8 supported conversions is pH 11, about 2 units higher than for Fe III MP-8 which has an optimal pH of 9. This can be correlated to the lower reduction potential of the Mn III MP-8/Mn II MP-8 transition when compared to the iron complex. The iron atom is proposed to better stabilise the deprotonated coordinated hydroperoxide molecule than the manganese centre due to its higher electron withdrawing effect on the proximal oxygen of the (hydro)peroxo group. For the cytochrome P450 mode, i.e. in the presence of ascorbate, it was found that Mn III MP-8 was not able to catalyse the para -hydroxylation of aniline whereas it was possible for Fe III MP-8 under the same conditions. These results are in line with the conclusions of Chapter 4 and indicate that the MP-8 supported cytochrome P450 chemistry proceeds via a PorFe III -hydroperoxo intermediate. This also explains the absence of aniline hydroxylation activity for Mn III MP-8 based catalysis because the Mn III -(hydro)peroxo intermediate is much less reactive toward electrophilic hydroxylation than the corresponding iron intermediate. As a consequence the hydroxylation of aniline by Fe III MP-8 may be performed by Compound 0. However, as stressed also in Chapter 4, Compound I and II, typical of the peroxidase mode, may also play a role in the P450 mode since ascorbate competes with aniline for oxidation by Compound I and II.Characterisation of peroxidase and P450 intermediates (Chapter 6 and 7)One of the major drawbacks of MP-8 is the high inactivation rate observed for the catalyst under operational conditions in both the P450 and the peroxidase mode. In an effort aiming at understanding the reasons for the fast inactivation of Fe III MP-8, the fate of the catalyst was studied under turnover conditions in the absence of substrate. Chapter 6 presents the characterisation of a modified Fe III MP-8 with a hydroxylated His18. The modified catalyst was isolated under operational conditions in the presence of H 2 O 2 . The structure of the native and the modified Fe III MP-8 were compared by HPLC, UV/visible, ESI-MS 2 and 1 H-NMR. Analysis showed the formation of a product more hydrophilic than Fe III MP-8, with an intact heme ring and having an extra oxygen inserted on the peptide. ESI-MS 2 and 1 H-NMR suggest the extra oxygen atom to be inserted on the Nδ1 of the imidazole ring of the His18. The formation of the modified intermediate is inhibited by ascorbate and labelling studies have shown that the inserted oxygen originates from the solvent. This suggests the modified Fe III MP-8 to derive from a Compound I analogue of Fe III MP-8. The fact that the solvent-assisted hydroxylation occurs on the proximal histidine ligand suggests the second oxidation equivalent of the analogue of Fe III MP-8 Compound I to be majorly delocalised on the His18 axial ligand by mesomerism.The characterisation and the analysis of the kinetics of formation of the heme-intermediates competent in MP-8-supported peroxidase and P450 catalysis is the subject of Chapter 7. The kinetics for the formation of Compound 0, Compound I analogue and Compound II were compared for Fe III MP-8 and Mn III MP-8. Analysis reveals a one order of magnitude higher rate of formation of PorFe III -OOH (k = 1.3×10 6 M -1 .s -1 ) when compared to the rate of formation of PorMn III -OOH (k = 1.1×10 5 M -1 .s -1 ). The overall rate of the reaction for both complexes with H 2 O 2 , increases with higher pH-values. The corresponding pK a values which were found to explain this pH-dependency are in complete agreement with the optimal values for Fe III MP-8 (pH 9.2) and Mn III MP-8 (pH 11.0) supported catalysis (Chapter 5) and were found to correspond to the deprotonation of metal-bound water for Fe III MP-8. This explains how heme-peptide models which lack a distal histidine, acting as an acid/base catalyst for the deprotonation of H 2 O 2 and for the subsequent cleavage of the O-O peroxide bond, facilitate the deprotonation of H 2 O 2 . For MP-8, the peroxide deprotonation may proceeds through a concerted mechanism which results in the replacement of the hydroxyl ligand by a hydroperoxo ligand. It is not sure, however, how MP-8 catalyses the cleavage of the O-O peroxide bond, since no residue or group able to deliver protons to the distal oxygen in order to facilitate bond cleavage are present on the distal site. An eventual participation of the Nδ1 of the imidazole ring of the His18 of a second molecule of MP-8 as proton donor cannot be excluded. The manganese Compound 0 is proposed to be heterolytically cleaved into a Compound I type of intermediate, Mn IV MP-8=O(R ·+ ), where the second oxidation equivalent seems to be localised on the peptide part of the molecule. This is suggested by the analysis of the transient UV/visible and EPR spectra of oxidised Mn III MP-8. Based on the kinetic analysis and on the results of Chapter 6, iron Compound 0 is also proposed to be heterolytically cleaved into a similar Compound I type of intermediate, Fe IV MP-8=O(R ·+ ), with a cleavage rate analogue to the one of the manganese complex (k≈150 s -1 ).To summarise, in peroxidases, the hydroperoxo intermediate is rapidly converted, during the reaction cycle, into Compound I catalysed by the residues of the distal heme pocket. In P450s the corresponding hydroperoxo or peroxo intermediates may react with the substrate bound in the active site before they are converted into a Compound I analogue. In other words, P450 chemistry is not only based on the catalytic reactivity of Compound I but also on the reactivity of Compound 0 and the deprotonated form of Compound 0 respectively the PorFe III -hydroperoxo and the PorFe III -peroxo intermediates. Whereas the isoelectronic analogue of Compound I can be seen as an intermediate in electrophilic reactions such as Compound 0, the deprotonated Compound 0 is proposed to be active in both electrophilic and nucleophilic reactions.As a consequence, one might conclude that designing a biomimic for the P450 chemistry requires two majors conditions: i) the presence of an open active site and ii) the stabilisation of the catalytic reactive intermediates preceding the formation of the porphyrin high-valent iron-oxo intermediate in the reaction cycle. Fe III MP-8 and Mn III MP-8 resemble an open peroxidase active site having no distal environment. It has been shown that for both MP-8 models the co-ordination of a molecule of hydrogen peroxide on the metal is facilitated by bound water, providing the pH in the medium is high (Chapter 7). This questions the role played by the distal histidine ligand in peroxidases. Generally the hydroperoxide substrate is proposed to be deprotonated by the distal histidine ligand before it displaces metal bound water. But the preliminary deprotonation of the bound water by the distal histidine, followed by concerted deprotonation of the hydroperoxide and displacement of the hydroxyl ligand might be considered as a relevant alternative mechanism. In order to mimic P450 catalysis, the porphyrin iron-hydroperoxo and the porphyrin iron-peroxo intermediates should be stabilised with respect to the porphyrin high-valent metal-oxo intermediate. This is rendered possible by partially scavenging the porphyrin high-valent metal-oxo intermediates species analogue of Compound I using ascorbate, regenerating the native MP-8. The reductant competes with the substrate for the oxidation by the porphyrin high-valent metal-oxo intermediate and does almost not affect porphyrin iron-(hydro)peroxo-based intermediates. Therefore in the presence of ascorbate, i.e. in the P450 mode, MP-8 can be considered as a model for P450 chemistry and without ascorbate, i.e. in the peroxidase mode, MP-8 can be considered as a model for peroxidase chemistry.As a conclusion, the present work has contributed to the better understanding of the chemistry of metalloporphyrin hydroperoxo and metalloporphyrin peroxo intermediates both relevant active species of the P450 chemistry. Furthermore the peroxidase mimic, MP-8, can efficiently be used as a model for the P450 chemistry.

Theoretical simulations are becoming increasingly important for our understanding of how enzymes work. The aim of the research presented in this thesis is to contribute to this development by applying various computational methods to three enzymes of theβ-ketoadipate pathway, and to validate the models obtained by means of quantitative structure-activity relationships (QSAR). The models and the resulting QSARs provide valuable mechanistic information about the relevant (rate-limiting) steps in the reaction cycles of the enzymes studied.Two of the enzymes that have been studied in this thesis, are flavin dependent monooxygenases: para -hydroxybenzoate hydroxylase (PHBH) from Pseudomonas fluorescence , and phenol hydroxylase (PH) from Trichosporon cutaneum . These enzymes catalyse the ortho-hydroxylation of para -hydroxybenzoate and phenol, leading to the formation of catechol and protocatechuate respectively. These products are the key intermediates in the degradation of many aromatic compounds. Once the catechol or protocatechuate is formed, the aromatic ring is cleaved between the two hydroxyl-substituted carbon atoms. This intradiol cleavage is catalysed by another enzyme studied in this thesis, catechol-1,2-dioxygenase (1,2-CTD), and by protocatechuate-3,4-dioxygenase (3,4-PCD), respectively.Catechol dioxygenaseThe reaction mechanism of catechol-1,2-dioxygenase from Pseudomonas putida has been studied by means of a QSAR approach based on gas-phase molecular orbital calculations. Catechol-1,2-dioxygenase catalyses intradiol cleavage of the aromatic ring of catechol by incorporating both oxygen atoms of molecular oxygen. In addition to the native catechol, this enzyme converts several C4-substituted catechol derivatives. In this study, the 4-methyl-, 4-fluoro-, 4-chloro-, 4-bromo, 4,5-difluoro- and 4-chloro,5-fluoro-catechols were obtained biosynthetically from the corresponding phenols by using the enzyme phenol hydroxylase. The overall rate constant for their conversion by catechol-1,2-dioxygenase was determined through steady-state kinetic experiments at various oxygen concentrations and saturating catechol concentrations.The crucial step in the reaction mechanism of the enzyme catalysed reaction was considered to be the nucleophilic attack of the substrate on the oxygen molecule. Therefore, the experimental results were compared to calculated energies of the highest occupied molecular orbital (HOMO) of the various catechol substrates, which represent their nucleophilic reactivities. A (linear) correlation was found between the calculated HOMO energies and the logarithm of the experimental rate constants. This indicates that the rate-limiting step in the overall reaction cycle involves a nucleophilic reaction of the substrate. Thus, the reaction of the substrate with molecular oxygen may indeed be rate limiting. Additional calculations excluded two other steps in the reaction cycle as being rate limiting.The results for catechol-1,2-dioxygenase from Pseudomonas putida were also compared to the data from two different types of catechol-1,2-dioxygenase, a normal (type I) and a chloro-catechol dioxygenase (type II), from Pseudomonas sp. B13. It could be argued that the difference in substrate preference between both types of catechol dioxygenases is related to a differential effect of the substituents on the rate of oxygen affinity binding by the two enzymes, rather than on the rate-limiting step.p-Hydroxybenzoate hydroxylaseAn important step that has been made in this thesis is the use of a combined quantum mechanical/molecular mechanical (QM/MM) method. Using this method, the quantum mechanical (reaction pathway) calculation of the reacting compounds could be performed within the actual environment of the protein. The surrounding protein atoms are calculated at a molecular mechanical (MM) level and their electrostatic and steric effects on the quantum mechanical system are included. This QM/MM technique has been applied to the hydroxylation step catalysed by p -hydroxybenzoate hydroxylase (PHBH). It was first investigated whether the energy barriers obtained from QM/MM reaction pathway calculations could be used to explain the variation in the overall rate constants for the conversion of a series of fluorinated substrates by PHBH. Reaction pathways were calculated for the proposed rate-limiting step in the reaction cycle: the electrophilic attack of the C4a-hydroxyperoxyflavin cofactor intermediate on the substrate. The energy profiles calculated for this reaction step with the various substrates yielded barriers with different heights. A correlation was found between the natural logarithm of the experimental overall rate constants for conversion of the fluorinated substrates by PHBH and the QM/MM calculated energy barriers for the different substrates. This correlation with overall rate constants supports that the electrophilic attack of the C4a-hydroxyperoxyflavin on the substrate is indeed the rate-limiting step in the reaction cycle.The correlation also indicates that the QM/MM model provides a realistic description of the hydroxylation step, as it accounts correctly for the effect of substrate substituents on the rate of hydroxylation. This was the basis for a further and more detailed analysis of the QM/MM model, which provided detailed insight into the mechanism of substrate and cofactor activation to facilitate the electrophilic reaction. Deprotonation of the substrate, which has been observed experimentally, is shown to significantly lower the energy barrier for the calculated reaction pathway. Also, the QM/MM model allowed the analysis of the energetic effect of the individual amino acid residues on the hydroxylation reaction. The results suggest catalytic effects of a backbone carbonyl moiety (Pro293), by a specific stabilisation of the transition state, and of a (crystal) water molecule (Wat717), which stabilises the negative charge arising on the proximal oxygen of the flavin cofactor.Phenol hydroxylaseThe QM/MM technique has also been applied to phenol hydroxylase (PH). As for PHBH, the hydroxylation step, proposed to be rate limiting in the reaction cycle of PH, has been simulated for a series of halogenated substrate derivatives. The energy barriers obtained correlate well with the logarithm of the overall rate constants. This correlation supports that the electrophilic attack of the C4a-hydroperoxyflavin on the substrate is the rate-limiting step in the reaction cycle at pH 7.6 and 25°C. An additional mechanistic question addressed in this study is the protonation state of the substrate during hydroxylation. Substrate deprotonation as a mechanism of activation has not been established for PH as firmly as it has been for PHBH. Proton transfer from phenol to a potential active site base, Asp54, has been investigated by calculating a 2-dimensional potential energy surface for the two reaction coordinates, i.e. hydroxylation and proton transfer. This potential energy surface suggests that proton transfer prior to hydroxylation is the most favourable mechanism, which indicates that in the PH reaction substrate deprotonation is important as well. The QM/MM model was further analysed to provide insight into the effect of the protein environment on the simulated reaction steps. Some catalytic effects on the hydroxylation step, i.e. of a proline carbonyl moiety and of a crystal water in the active site of PH, were similar to those found for PHBH.All together, the research presented in this thesis has made a new contribution to the development and validation of computational models that can be used to address a major challenge in the present field of biochemistry, i.e. to obtain insight into enzymatic reaction mechanisms and enzyme activity on the basis of the structure of enzyme and substrate(s). Special emphasis has been on the application and validation of the QM/MM technique in the context of a QSAR approach. The investigations of this thesis provide a first survey of the possibilities of the QM/MM method with respect to the prediction of biochemical activity, taking explicitly into account the influence of the active site surroundings.

Ozone is a major constituent of photochemical smog. The toxicity of ozone is well documented and has been related to its strong oxidative potential. The principal target organ for ozone toxicity is the respiratory system. Unsaturated fatty acids, which are present in both the lipids of the lung lining fluid and the cell membranes of the cells that line the airways, are thought to be primary target molecules for ozone. Ozone reacts with unsaturated fatty acids via the so-called Criegee mechanism. Along with aldehydes and hydroxyhydroperoxides, Criegee ozonides are the main products of the reaction of ozone with unsaturated fatty acids. It is generally assumed that these lipid ozonation products act as secondary toxins in ozone-induced toxicity. However, very little is known about the reactivity and fate of ozonides and the role they play in ozone toxicity. Therefore, the aim of the present study was to gain insight into the mechanism of the toxic action of ozonides in relation to ozone-induced lung toxicity.For the investigations the ozonide from the polyunsaturated fatty acid methyl linoleate was used. On ozonation of methyl linoleate the major product formed appeared to be the trans -9,10-methyl linoleate ozonide (MLO).It has been suggested that ozonides are a kind of peroxide, and it is, therefore, believed that ozonides are a source of free radicals. This implies that ozonides might be capable of initiating the chain autoxidation of other non-ozonated polyunsaturated fatty acids in the membrane bilayer, and consequently of producing a cascade of damage. This hypothesis was investigated by comparing the in vitro cellular toxic action of MLO with a model peroxidative agent, i.e., cumene hydroperoxide. The ozonide was shown to be three times more toxic towards alveolar macrophages than the peroxide. On the basis of the cellular protection of antioxidants it was clearly shown that the ozonide and the hydroperoxide exert their toxic effects using different mechanisms. Whereas the results with the model hydroperoxide, cumene hydroperoxide, confirmed the mechanism by which peroxides exert their toxic effects, namely by lipid peroxidation and/or depletion of intracellular glutathione (GSH) levels, the investigations regarding the potency of MLO to induce lipid peroxidation revealed that the main toxic mechanism of MLO did not proceed via a radical-mediated mechanism. Nevertheless, suppletion of cells with the lipid-soluble radical scavengerα-tocopherol resulted in a significant protection towards ozonide exposure. In addition, preincubation of MLO withα-tocopherol resulted in a detoxification of the ozonide. This suggests thatα-tocopherol is able to interact directly with the ozonides themselves, thus scavenging these reactive intermediates.Investigations regarding the chemical characteristics of the detoxification of MLO byα-tocopherol revealed that the main products formed were the aldehyde nonanoic 9-oxo methyl ester and the acid nonanedioic acid monomethyl ester. This finding is in agreement with the general opinion that the decomposition of ozonides results in the formation of aldehydes and acids. In general, it is believed that thermal decomposition of ozonides proceeds via homolytic cleavage of the peroxide bond to yield the oxy bi-radical followed by a rearrangement. A modification on this mechanism includes concerted homolysis and intramolecular hydrogen atom abstraction. In contrast to the results obtained at elevated temperatures (≥50 °C), in previous studies ozonides have been shown to be stable compounds at 37 °C. No radical formation could be detected when MLO was incubated for 30 min at 37 °C using spin traps and electron spin resonance (ESR). Taking into account the fact that peroxide bond homolysis is an essential part of ozonide decomposition, one might postulate thatα-tocopherol, being an efficient hydrogen atom donor, facilitates the process of O-O bond homolysis, thereby inducing aldehyde and acid formation already at relatively low (i.e.≤37 °C) temperatures.An additional interesting finding was that the degradation products of MLO, i.e., nonanoic 9-oxo methyl ester and nonanedioic acid monomethyl ester (azelaic acid monomethyl ester), were not toxic towards alveolar macrophages at concentrations where MLO showed complete loss of cell viability. This observation is especially of importance because it is often suggested that Criegee ozonides will be formed in relatively small amounts (ca. 10%) when ozone reacts with unsaturated fatty acids in the lung lining fluids and may, therefore, play a minor role in ozone-induced toxicity. However, the results of the present study clearly indicate that ozonides are far more toxic than their aldehyde and acid type degradation products, which are generally observed as major products resulting from the reaction of ozone with fatty acids under physiological conditions.In addition, the mechanism underlying the reaction of another cellular antioxidant, namely glutathione, with polyunsaturated fatty acid ozonides was investigated. In previous studies it has been shown that preincubation of MLO with glutathione caused a significant detoxification of the ozonide. Furthermore, the detoxification reaction was shown to be catalysed by glutathione S -transferases leading to the formation of oxidised glutathione and aldehydes. The reaction of ozonides and peroxides with glutathione was investigated using molecular orbital calculations and the frontier orbital theory. In addition to the results obtained in the comparative study on the toxicity of ozonides and peroxides, the reaction of ozonides with the nucleophilic agent glutathione appeared also to be different when compared with the reaction with peroxides. On the basis of semi-empirical molecular computer calculations the nucleophilic attack by glutathione on the ozonide is expected to occur at one of the carbon atoms of the ozonide ring instead of at one of the peroxidic oxygen atoms as in the case of peroxides. A mechanism for the glutathione S -transferase-mediated detoxification of ozonides, different from that of the reaction with hydroperoxides has been proposed.Furthermore, the in vivo toxicity of ozonides was investigated. Methyl linoleate ozonide (MLO) (0.07 mmol/100 g body wt) was administered to female Wistar rats either intravenously or intraperitoneally. After 24 h the rats were killed and the effects were examined. MLO was found to be toxic only after intravenous administration. The major effects were observed in the lungs. The lungs became enlarged from edema and showed severe haemorrhages. Furthermore, the total thiol level was depleted in serum and lung tissue, accompanied by a decrease in the activity of thiol-dependent enzymes. The vitamin E levels in serum and lung tissue were also reduced. The malondialdehyde (MDA) concentrations in serum and lung tissue were elevated suggesting that in vivo oxidation had occurred. On intraperitoneal administration of MLO, no effects on enzyme activities, thiol and vitamin E content in lung tissue were observed. In serum, however, as on intravenous administration, an increase in the MDA levels and decreases in total thiol and vitamin E levels were found. In view of the route of administration it is to be expected that the ozonide is partially cleared by the liver, and the ozonide and its potentially toxic products are further detoxicated by vitamin E and thiols in the serum before they reach the lung. The above data show that the main target organ for ozonides is the lung, and that the effects caused by MLO in vivo are in many respects similar to the effects found after acute ozone exposure.In short, the most important conclusions of the present studies are:Ozonides exert their toxic effects by a mechanism different from that from the structurally related peroxides.The protective action ofα-tocopherol against ozonides proceeds via a direct interaction by whichα-tocopherol facilitates ozonide degradation through peroxide bond homolysis already at 37 °C.The involvement of ozonides in ozone-induced toxicity may be more important than currently suggested, since they are far more toxic than their aldehyde-type degradation products, which are generally observed as major products resulting from the reaction of ozone with unsaturated fatty acids under physiological conditions.Overall, the results presented in the thesis provide new insights into the toxic mechanism of ozonides and their implications for ozone-induced lung toxicity.

This PhD project was started in 1995 and was supported by the dutch Ministry of Economic Affairs through the programme "IOP Catalysis". The main goal of "IOP catalysis" is to obtain clean and more efficient technologies, to improve the quality of the Dutch fine chemistry. Biocatalysis provides a way to minimize non desirable side products, which can have a negative impact on the environment. The objective of this thesis was to investigate the potential of heme-containing peroxidases as natural biocatalysts for industrially relevant conversions. Peroxidases were chosen since they are able to operate under mild conditions using cheap and clean oxidants (hydrogen peroxide).In this work especially the natural production of food flavours was of interest. After initial screening of reactions of potential interest the N-dealkylation of methyl-N-methylanthranilate) (ex citrus leaves) to methylanthranilate (MA), a concord grape flavour was chosen as the mode) reaction in this thesis.In order to obtain better insight in the industrial applications of peroxidases, it is important to understand structure, function and operational stability of these catalysts. Therefore, this project was started by investigating the reaction mechanism of peroxidases in more detail ( chapter 3 ). Horseradish peroxidase (HRP) was used as the model peroxidase, as this enzyme is the best studied enzyme of all peroxidases. Many heme-containing biocatalysts, exert their catalytic action through the initial formation of so-called high-valent- iron-oxo porphyrin intermediates. For HRP the initial intermediate formed has been reported as a high-valent-iron-oxo porphyrinπ-radical cation, called compound I. A strongly hold concept in the field of peroxidase-type of catalysis is the irreversible character of the reaction leading to formation of this compound I. The results of chapter 3, however, indicate that formation of the high-valent-iron-oxo porphyrin intermediate for various heme-containing catalysts, including HRP, might be reversible. This reversible compound I formation results in heme-catalysed exchange of the oxygens of H 2 O 2 with those of H 2 O. The existence of this heme-catalysed oxygen exchange followed from the observation that upon incubation of 18 O labeled H 218 O 2 with heme-containing biocatalysts, significant loss of the 18 O label from the H 218 O 2 was observed, accompanied by the formation of unlabeled H 2 O 2 . Thus, for the heme biocatalysts studied, exchange of the oxygen of their high-valent-iron-oxo intermediate with that of water occurs rapidly. This observation implied the need for an update of the kinetic model for peroxidases. Revaluation and extension of the previous kinetic model for HRP showed the necessity to include several additional reaction steps, taking both reversible compound I formation and the formation of enzyme-substrate complexes into account As a consequence reactions with HRP are saturable, implying that V max and k cat can be measured. This provides possibilities for investigation of structure-activity relationships.Investigation of quantitative structure-activity relationships (QSARs) is a way to obtain more insight in the influence of the structure of a substrate on its conversion by an enzyme of interest. Moreover, QSARs could be of interest for industrial applications. With QSARs the outcomes of conversions may be predicted by simply calculating chemical parameters of structurally related substrates. In this way QSARs could be helpful in facilitating screening procedures for biocatalytic productions, saving time and money.In chapter 4 predictive computer calculation-based QSARs were defined by comparing second order rate constants for the oxidation of a series of model compounds by HRP compound II to computer calculated chemical parameters characteristic for this reaction step. The model compounds studied were a series of structurally related phenols. For the calculation of the chemical parameters characteristic for the reaction step two approaches were used. In the first approach a frontier orbital parameter of the substrate was calculated being the ionisation potential (i.e. minus the energy of their highest occupied molecular orbital (E(HOMO)). In the second approach the relative heat of formation (ΔΔHF) was calculated for the process of one electron abstraction as well as for H·-abstraction from the phenol derivatives. Assuming a reaction of the phenotic substrates in their non- dissociated, uncharged forms, clear correlations were obtained between the natural logarithm of the second order rate constants (ln k app and ln k 2 respectively ) for their oxidation by compound II and their calculated parameters.The computer calculation-based QSARs thus obtained for the oxidation of the various phenol derivatives by compound II from HRP indicate the validity of the approaches investigated, i.e. the frontier orbital approach but also the approach io which the process is described by calculated relative heats of formation. The results also indicated that outcomes from computer calculations on related phenol derivatives can be reliably compared to one another. Since both mechanisms considered, i.e. initial electron abstraction versus initial H·-abstraction provided clear molecular orbital QSARs the results could not be used to discriminate between these two possible mechanisms for phenol oxidation by HRP compound II. Furthermore, as the actual oxidation of peroxidase substrates by compound II is known to be the rate-limiting step in the overall catalysis by HRP, the QSARs described in chapter 4 may have implications for the differences in the overall rate of oxidation of the phenol derivatives by HRP. As a matter of fact similar QSARs should be obtained when the overall rate of oxidation of the respective compounds by HRP is determined.As mentioned above, reactions with HRP are saturable, implying that V max and k cat can be measured. Thus, in chapter 5 the overall conversion of phenols by HRP was investigated, resulting in overall k cat values. These saturating overall k cat values indeed correlated quantitatively with calculated ionisation potentials of the substrates. The observation that the rates and QSARs obtained for the overall rate of conversion in chapter 5 are similar to those described for the rate limiting reaction step in chapter 4, corroborates that phenol oxidation by compound II is the rate limiting step in the reaction cycle of HRP, but even more important, it also illustrates that QSARs for oxidation of substrates by compound II can be used for prediction of the overall rate of oxidation of phenol derivatives by HRP and vice versa. Moreover, QSARs for overall k cat , instead of for individual rate constants, eliminate the need for extensive rapid kinetic analysis to make predictions for HRP-based substrate conversions.Saturating overall k cat values for HRP catalysed conversion of a second series of substrates, i.e. a series of substituted anilines, described in chapter 5, also correlated quantitatively with calculated ionisation potentials of the substrates. However, in the QSAR plots, the correlations for the anilines were shifted to lower k cat values at similar ionisation potentials as compared to those for the phenols. To investigate whether differences in orientation of phenols and anilines within the active site of HRP may be a factor underlying the higher reactivity of the phenols than expected on the basis of their ionisation potential, 1 H NMR T 1 relaxation studies were performed, using 3-methylphenol aod 3-methylaniline as the model substrates. The 1 H NMR T 1 relaxation studies revealed consistently smaller average distances of the phenol than of the aniline protons to the paramagnetic Fe 3+ centre in HRP, may be resulting in differences io the electron transfer process from the aromatic donor substrate to the Compound II of HRP. A shorter distance between the phenol and the heme than between the aniline and the heme may be a factor contributing to the faster rate of electron transfer with phenol as compared to aniline substrates. However, the actual differences in orientation seem small when the difference in oxidation rate at similar calculated ionisation potential phenols or anilines is considered. Since (partial) deprotonation will largely influence, i.e. decrease the ionisation potential of the aromatic substrate, the relatively higher oxidation rates of phenols may be related to their larger extent of deprotonation upon binding to the substrate pocket of HRP, resulting in lower ionisation potentials than actually expected on the basis of calculations on their non-ionised form. Based on the 1 H NMR T 1 relaxation data of chapter 4 and literature data of Hendriksen et al. [1], we put forward the hypothesis that the differential substrate behaviour of phenols and anilines may be due to subtle differences in their binding to the active site substrate pocket of HRP, resulting in i) closer proximity to the heme and ii) larger extent of deprotonation for the phenols than for the aniline substrates. An important conclusion following from the results of chapter 5 is that for each type of reaction and substrate, different QSARs have to be obtained. It is also clear that further investigations are nessecary for succesful application of QSARs to industrial prediction of biocatalysis.Heme-based peroxidases are enzymes with a broad substrate specificity and are capable of catalysing a variety of reactions. However, operational (in)stability limits the use of peroxidases in industrial processes. Chapter 6 describes the possibillities and limitations for using commercially available SP preparations (delivered by QUEST) and other peroxidases (like HRP and microperoxidase-8 (MPS)) for the production of the food flavour methylanthranilate (MA) from methyl-N-methylanthranilate (MNMA). Because W from citrus leaves is a relatively cheap source and MA is more expensive than MNMA, the investigated reaction provides an industrially relevant route for the natural production of an important topnote flavour in concord grape.The tested soybean preparations varied widely with respect to their heme-content and, as a result, their activity for this reaction. Furthermore, the operational stability of purified soybean peroxidase (SP) was at least 25-fold lower than that of HRP and only 5-fold higher than that of MP8, a small peroxidase with a polypeptide chain of only eight amino acids covalently linked to the protoporphyrin IX cofactor. Therefore, the results with SP indicate that the presence of a large protein chain around a porphyrin cofactor in a peroxidase is, by itself, insufficient to explain the observed differences in operational stabifity, and that the inactivation mechanism could be intramolecular. In order to find an explanation for the low operational stability of SP and MP8, it would therefore be interesting to investigate, for example, the shielding of the meso-positions of the heme cofactors of the different peroxidases.All tested peroxidase preparations were able to catalyse the requested N-dealkylation. However, SP proved to be a very efficient biocatalyst for the production of MA with high yield and purity, in spite of its relatively low operational stability. This potential of SP to catalyse the N-demethylation of MNMA to MA more efficiently than HRP and MP8, was especially observed at high temperature and low pH values at which SP appeared to be optimally active. Unfortunately, the prices for MA had dropped by the time this efficient SP catalysed production of MA was developed. As a result the now defined way to produce MA using a peroxidase biocatalyst was no longer of commercial interest for the industrial IOP partner. Nevertheless, the results in chapter 6 clearly define the answer to the initial industrial aim of the preseot IOP project.Biotechnological applicationsPeroxidases, in principle, have remarkable synthetic possibilities, but commerciai processes based on these enzymes have not yet been developed. The inactivation of heme- enzymes by peroxides through oxidation of the porphyrin ring is one of the prime reasoos. Attempts to improve the stability have not been very successful yet. The enzyme stability can be improved by stepwise or continuous addition of the oxidaot, maintaining a low peroxide concentration [2,3]. Hiner et al. [4] tried to increase the resistance of HRP to H 2 O 2 through genetic engineering without satisfying results. Other ways to deal with the problem of poor stability were chemical modification, screening and immobilisation [5].A second reason why commercial processes based on peroxidases have not yet been developed is that the oxidation of some substrates by peroxidases is in competition with their spontaneous chemical oxidation by peroxides. As a consequence, reduction of the purity of products occurs in case enantioselectivity is the target [6]. Maintaming a low peroxide concentration throughout the reaction period, reducing background oxidation, has also been tried here as a solution.Furthermore, in the process of one-electron oxidation by peroxidases free radicals are produced which are difficult to control, thereby reducing the purity of products as well.A fourth reason for the limited scaiing up of peroxidase-catalysed reactions is the low water solubility of most of 8e substrates of synthetic interest. Designing strategies that enhance enzymatic activity in organic solvents [7,8], or the use of hydrophobic matrices that act as a reservoir for both substrates and products are solutions to this general problem.From the above it can be concluded that the problem of (operational) stability of peroxidases is a severe problem for industrial processes and that this problem is difficult to so)ve. On the other hand this thesis shows that a peroxidase exists that, in spite of its low operational stability, can be an efficient biocatalyst for the production of a industrially relevant compound with high yield and purity. Altogether, it can be concluded that peroxidases could represent an interesting tool for industrially relevant reactions, making future research on possibilities and limitations worthwhile.ReferencesHendriksen, A., Schuller, D.J., Meno, K., Welinder, K.G., Smith, A.T., Gaijhede, M. (1998) Biochem.37 : 8054-8060.Colonna, S., Gaggero, N., Casella, L., Carrea, G., Pasta, P., (1992) Tetrahedron asymmetry3 : 95-106.Van Deurzen, M.P.J., Van Rantwijk, F., Sheldon, R.A. (1996) J. Mol. Catal. B. Enzym.2 : 33-42Hiner, A.N.P., Hernandez-Ruiz, J., Arnao, M.B., Garcia-Canovas, F., Acosta, M. (1996) Biotechnol. Bioeng.50 : 655-662Bakker, M., Van de Velde, F., Van Rantwijk, F., Sheldon, R.A. (in preperation) Biotechnol. Bioeng.Colonna, S., Gaggero, N., Richelmi, C., Pasta, P. (1999) Tibtech.17 : 163-168Klibanov, A.M., (1997) Trends Biotechnol.15 : 97-101Osman, A.M., Boeren, S., Boersma, M.G., Veeger, C., Rietjens, I.M.C.M. (1997) Proc. Natl. Acad. Sci. USA94

The amount of newly developed chemicals such as agrochemicals, drugs and food additives in our modem society is ever increasing. The industrial production, use and also the release in the environment of these chemicals expose organisms to these xenobiotics. Due to the often hydrophobic character of these xenobiotics they can be easily absorbed and accumulated in the body. However, organisms defend themselves against these agents by converting them enzymatically to more hydrophilic easily excretable products. Sometimes these biotransformation reactions can lead to reactive intermediates which cause cell damage, toxicity, malignancy and/or ultimately carcinogenicity. This indicates why a lot of toxicological screening and safety tests have to be performed during the development of these agents.However, modern society still requires new compounds with better properties. As it is unfeasible to test the biological activity and toxic effects of all newly developed chemicals in experimental animals, many in vitro systems have been developed in which these chemicals can be quickly tested for various biological and/or toxicological effects. To diminish the use of experimental test systems the QSAR concept has been developed, describing so-called Quantitative StructureActivity Relationships. This QSAR approach should make it possible to predict the biological effect, the biotransformation pattern and/or the toxicity of a chemical based on its physical and chemical characteristics. In this respect the application of MO (Molecular Orbital) computer calculations to describe at least some of these chemical characteristics appears to become more and more useful.The present thesis focuses on the use of these computer calculated MO parameters for the prediction of biotransformation characteristics of benzene derivatives. Benzene derivatives play an important role as precursors for the synthesis of many xenobiotics and industrially relevant chemicals. The cytochrome P450-catalysed biotransformation of various substituted benzenes was studied and the results were evaluated to determine whether (MO-based) computer calculated chemical parameters of benzenes could explain the results obtained. Where results could not be described well by the chemical reactivity of the substrate itself, additional experimental work was performed to investigate to which factors and/or interactions the observed deviations could be ascribed. This information can then be used to adopt and improve the predictive model.Biotransformation of xenobiotics by higher organisms is an enzymatic process in which cytochromes P450 play a major role. This system is able to handle an enormous amount of very structurally different chemicals. The cytochrome P450 enzymes differ in their substrate specificity, metabolic activity and their extent of contribution to the production of toxic metabolites. As outlined in the introduction of this thesis (Chapter 1 ), there are several factors which control the cytochrome P450-catalysed biotransformation. These factors include for example, electron donation to cytochrome P450, the accessibility of the active (FeO) 3+species in the active site of the enzyme, the interaction between the substrate and the active site and the chemical reactivity of the substrate itself. In cases where the latter factor controls the outcomes of the cytochrome P450-catalysed biotransformation, one could describe a QSARs for the reaction. Based on these QSARs the cytochrome P450-catalysed biotransformation of other chemicals might be predicted. Furthermore, these QSARs might give more insight in the mechanistic background of the biotransformation reaction for which it has been developed.An example of this approach is described in Chapter 2 of this thesis. In this chapter in vivo cytochrome P450-catalysed aromatic ring hydroxylation of various substituted benzene derivatives was determined using 19F NMR. Previous reported results on the regioselectivity of cytochrome P450-catalysed aromatic ring hydroxylation of a series (poly)fluorobenzenes demonstrated a clear QSAR (correlation 0.96) relating the site of hydroxylation of the site of the reactive HOMOπ-electrons in the benzene substrate. Thus, in Chapter 2 the same parameter was tested for its ability to predict the regioselectivity of aromatic hydroxylation of a series C4-substituted fluorobenzenes. It turned out that the MO-QSAR approach could well predict the regioselectivity of the series of 4-X-substituted fluorobenzenes when X was a H, F, Cl, and CN group. However, this approach of predicting the regioselectivity of aromatic hydroxylation ran into problems when the para substituent was a bromine or iodine. The extent to which the hydroxylation adjacent to a bromine and iodine substituent was reduced, correlated qualitatively with the size of the substituents as given by their Van der Waals radius, i.e. the larger the substituent the greater the deviation between the predicted and observed hydroxylation at the adjacent aromatic carbon centre. Additional experiments showed that the observed deviations could not be ascribed to a stereoselective orientation of these substrates in the active sites of cytochromes P450. These results led to the hypothesis that bromine and iodine substituents sterically hamper the electrophilic attack of the cytochrome P450 high-valent iron-oxo species on the carbon atoms adjacent to the substituted carbon centres. This hypothesis was supported by calculating the steric effects of bromine and iodine substituents, thus defining a steric correction factor for prediction of the aromatic ring hydroxylation at sites ortho with respect to a bromine or iodine substituent. In a second series of regioselectivity experiments using a series of tri- substituted benzenes containing fluorine, chlorine, bromine, iodine and cyano substituents, these steric correction factors were validated. Without using the steric correction factors for bromine and iodine substituents no correlation between the predicted and observed regioselectivity was obtained. However, upon application of the correction factors a correlation factor of 0.91 was obtained for prediction and actually observed regioselectivity of cytochrome P450-catalysed aromatic hydroxylation of the series tri- substituted benzenes. These results show that the QSAR which was initially developed for a series of (poly)fluorobenzenes was still valid for other substituted benzenes taking into account the steric hindrance by relatively large substituents such as bromine and iodine. Furthermore, from a mechanistic point of view these MO-QSARs suggest that the cytochrome P450-catalysed aromatic ring hydroxylation of at least the benzene derivatives tested proceeds by an initial attack of the iron-oxo species on the ring carbon atom, without the formation of arene oxide intermediates as a major route.The question remains why going from a chlorine to a bromine substituent, steric hindrance is observed rather abrupt. A possible explanation might be found in the observation that when two molecules approach each other the potential energy goes through a minimum and thereafter increases rapidly with decreasing distance.Surprisingly, it was found that steric hindrance by the cyano group was not observed although this group is definitely larger than a bromine. However, with respect to steric hindrance of the electrophilic attack of the (FeO) 3+species, only the atom closest to the aromatic ring might have to be taken into account. Steric hindrance of this atom, a carbon, is not expected since a carbon, has a size similar to that of a chlorine.More complicated substituents than H, F, Cl, Br, I and CN were investigated in Chapter 3 where the regioselectivity of aromatic ring hydroxylation of 3-fluoro(methyl)anilines was investigated. Little is known about the cytochrome P450- catalysed biotransformation of such methylanilines. Results from the in vivo and in vitro cytochrome P450-catalysed regioselectivity of aromatic hydroxylation showed that the frontier orbital density distribution in the aromatic ring could only qualitatively predict the observed regioselectivity. Again, as for the bromine and iodine substituents the observed regioselectivity of aromatic ring hydroxylation for the various amino and methyl substituted substrates showed systematic deviations from the predicted regioselectivity, i.e. hydroxylation at the ring carbon C4 para with respect to the amino moiety was always higher than predicted while hydroxylation at the carbon atoms ortho with respect to the amino group (C2 and C6) was lower than expected. Further research was performed to investigate why for this series of amino and/or methyl substituted benzene derivatives the intrinsic chemical reactivity did not predict the observed regioselectivity of aromatic hydroxylation quantitatively. Three different experimental approaches were used to investigate possible reasons underlying the systematic deviations between predicted and observed regioselectivity for the cytochrome P450-catalysed aromatic hydroxylation of amino-containing benzenes. These results are discussed in some more detail hereafter.1. As a first approach, incubations of 3-fluoro-2-methylaniline with different microsomal preparations containing various cytochrome P450 enzyme patterns as well as with purified cytochrome P4502B1 were performed. The regioselectivity for aromatic hydroxylation of 3- fluoro-2-methylaniline observed in all these incubations were similar. This result indicates that cytochromes P450 with different active sites give similar regioselectivities and, thus, in case of 3-fluoro-2-methylaniline similar deviations for predicted values. This suggests that the observed discrepancy between predicted and observed aromatic hydroxylation patterns can not be ascribed to a stereoselective orientation of the 3-fluoro(methyl)anilines imposed by the active sites of the cytochromes P450 catalysing its conversion.2. Further support for this conclusion was obtained from 1H NMR T 1 relaxation studies on fluoromethylanilines. These studies are described in Chapter 4. It appears that the orientation of the tested fluoromethylanilines in the active sites of cytochromes P4501AI and 2B1 is similar. Based on the observation that all aromatic protons are at about the same average distance from the catalytic Fe 3+centre, these results are most compatible with a time-averaged orientation of the substrates with the Fe 3+above the aromatic ring and theπ-orbitals of the aromatic ring and those of the porphyrin rings in a parallel position. This latter aspect would provide possibilities for energetically favourableπ-πinteractions. Possibilities for a flip-flop rotation around an axis in the plane of the aromatic ring of the substrate can be included in this picture, as such rotations would still result in a similar average distance of all aromatic protons to the Fe 3+centre. Altogether, the data strongly suggest that, independent of the cytochrome P450 enzyme these substrates can rotate freely in the active site and are not hindered and/or specifically orientated by interactions with specific amino acid residues of the active site.3. Finally, the possible influence of the active site on the cytochrome P450-catalysed aromatic hydroxylation was tested by using microperoxidase-8 as a model system for cytochrome P450-catalysed aromatic ring hydroxylation. Microperoxidase-8 is a so-called mini-enzyme prepared by proteolytic digestion of horse-heart cytochrome c. MP-8 contains a heme covalently attached to a peptide chain of only eight amino acids and thus it lacks an active site. MP-8 is well known to perform peroxidase-like reactions. However, experimental evidence has been obtained described in Chapter 5 of this thesis that the MP-8-catalysed aromatic ring hydroxylation of aniline and phenol derivatives in a H 2 O 2 -driven system proceeds by a cytochrome P450-like oxygen-transfer mechanism. This could be concluded from the fact that 18O incorporation from H218O 2 into the 4-aminophenol formed from aniline was 100%. Thus, the H 2 O 2 -driven MP-8 system was used as a model to study regioselectivity of oxygen-transfer in a heme-based catalyst without an active site.The results of such experiments demonstrated that the MP-8-catalysed aromatic ring hydroxylation of 3-fluoro(methyl)anilines (Chapter 3) results in similar regioselectivities of aromatic ring hydroxylation as those observed for cytochrome P450- catalysis. As a specific substrate binding site in MP-8 can be excluded, these results, together with those obtained from different microsomal preparations (Chapter 3) and the 1H NMR T 1 relaxation measurements (Chapter 4), clearly eliminates the possibility that the regioselectivity of aromatic ring hydroxylation is predominantly dictated by a stereoselective orientation of the substrate in the active site of cytochromes P450. Thus for these relatively small and non-polar substrates specific interactions between molecular sides in the substrate and amino acid side chains present in the active site are not dictating a stereoselective orientation of these substrates.Since the MP-8-catalysed reactions showed similar deviations between predicted and observed regioselectivity of aromatic hydroxylation of 3-fluoro(methyl)anilines, it was concluded that the deviations must result from an orientating interaction between the substrate and the high-valent iron-oxo species (FeO) 3+itself, resulting in a stereoselective orientation of the substrate towards the catalytically active species. This interaction can be expected to be quite similar for different cytochromes P450 and might also be comparable in the active MP-8 system. This implies that, in contrast to what was observed by 1H NMR for substrates bound to the Fe 3+resting state of the enzyme, the (FeO) 3+form of the protein would impose a preferential orientation of the amino-containing substrate in such a way that on the average C4 ( para to the amino) is closer to whereas C2/6 (ortho to the amino) are further away from the (FeO) 3+species. Support for a different substrate orientation in the Fe 3+resting state and the (FeO) 3+activated form respectively can be found in a study of Paulsen and Ornstein. They showed that the orientation of camphor in the presence and absence of the high-valent iron-oxo species (FeO) 3+was different and that orientation in the presence of the catalytically active form of the enzyme (FeO) 3+was in accordance with the observed regioselectivity of oxidation.Altogether, results from all studies on the regioselectivity of aromatic hydroxylation of the various benzene derivatives show that the regioselectivity of the H, F, Cl, Br, I, CN, CH 3 and NH 2 substituted benzene derivatives investigated is predominantly determined by their chemical reactivity. However, especially for amino-containing benzene derivatives interactions between the substrate and the high-valent iron-oxo species, resulting in a stereoselective orientation of the substrate, might also play a role. This means that the QSAR for aromatic ring hydroxylation in the case of amino-containing benzenes predicts only qualitatively the regioselectivity of aromatic ring hydroxylation for these series of substrates.An additional result obtained in all regioselectivity studies was the absence of formation of NIH shifted phenolic metabolites. The formation of such NIH shifted phenolic metabolites has been well documented especially for the biotransformation of chlorinated and brominated benzene derivatives. As the benzenes in our studies were mainly fluorinated, we decided to investigate whether there was a (chemical) reason for this discrepancy.In Chapter 6 the possibility of a cytochrome P450-catalysed fluorine NIH shift in a series of polyfluorobenzenes was investigated and compared with the literature data of the chlorinated analogues. The in vivo biotransformation of a series polyfluorobenzenes showed that formation of NIH shifted metabolites was not a significant biotransformation route. As outlined above, this is in contrast to the results reported in the literature for the chlorinated analogues. However, in contrast to the in vivo data, in in vitro microsomal studies with 1,4-difluorobenzene formation of a significant amount of fluorine NIH shifted phenolic metabolite was observed. Unfortunately, the in vitro microsomal conversion of the other used polyfluorobenzenes could not be detected, possibly due to the fact that the HOMO energy of these substrates was too low for an efficient conversion by cytochromes P450 [5]. It is generally accepted that formation of NIH shifted metabolites proceeds via arene oxide intermediates and that these arene oxides, especially in vivo, but not in vitro in a microsomal system, might react in competing conjugation pathways such as GSH conjugation. Additional in vivo and in vitro experiments showed that indeed GSH conjugation was a competing pathway for the arene oxide intermediates, its presence or absence resulting in an influence on the amount of NIH shifted phenolic metabolites observed.Nevertheless, the results thus obtained also clearly illustrated and confirmed the reduced capacity for formation of NIH shifted phenolic metabolites for fluoro- as compared to chlorobenzenes. This differences between the NIH shift for fluorine and chlorine containing benzenes was further investigated and could be explained by MO computer calculations on the proposed reaction intermediates in the two biotransformation pathways. Epoxide ring opening, supposed to be the rate-limiting step in the formation of NIH shifted metabolites, appeared easier for chlorinated benzenes compared to their fluorinated analogues. Furthermore, these MO calculations in combination with a MO-QSAR for the rate of GS-conjugation of electrophilic model compounds were indicative for higher rates of GSH conjugation for the fluorinated benzene arene oxides than the chlorinated analogues. The detoxifying pathway of GSH conjugation is very important for assessing the risks of these kind of substrates as arene oxides are known to react with protein and DNA provoking toxic and/or mutagenic /carcinogenic responses.Altogether, the results from this study show that MO calculations can not only explain and thus predict the regiospecificity of biotransformation in a molecule but also explain the relative rates, i.e. chances, on different biotransformation routes.Finally, in addition to all studies on model benzene derivatives, the last chapter of this thesis (Chapter 7) describes studies on the biotransformation of a halogenated benzene derived compound that is used as an insecticide, i.e. teflubenzuron. This may provide some insight in the possibilities to use outcomes from studies on model compounds to obtain insights also in the biotransformation of other compounds. Teflubenzuron, which is used for the protection of fruit and vegetables against larves, contains two aromatic benzene-derived rings connected via an urea bridge. These aromatic rings are differently substituted and are a suitable starting point to evaluate the influence of various substituents on the metabolic fate of the chemical. Basically, they are an aniline and a benzoate derived moiety. Results of this study on the biotransformation of teflubenzuron demonstrate that upon an oral dose of the insecticide to male Wistar rats the metabolic fate of the two aromatic rings is significantly different. Though about 90% of the dose was excreted unchanged in the faeces, the remainder of the dose was absorbed from the gastrointestinal tract and -in part- excreted in the urine mainly after hydrolysis of the urea bridge. Interestingly, the urinary recovery of the benzoate moiety was about 8 times higher than that of the aniline moiety of teflubenzuron. Dose-recovery studies on the scission products of teflubenzuron confirmed these results as the urinary recovery of the metabolites from the benzoate moiety were nearly 100% while the recovery of the aniline derivatives was only about 50%. The benzoate moiety is excreted unchanged or after a Phase IIreaction, while the aniline moiety has to undergo both a Phase I and II reaction before it can be excreted.The significant difference between the recovery of the benzoate and aniline moiety can be best explained by the formation of a cytochrome P450-catalysed reactive benzoquinoneimine from the C4 halogenated aniline derivatives. Swift binding of these reactive benzoquinoneimines to tissue macromolecules might cause the compound to be withheld in the body. This is also important from a toxicological point of view as binding of these reactive benzoquinoneimines might provoke toxic and/or mutagenic /carcinogenic effects in the body.Altogether, the results described in this thesis show that QSARs are very useful in explaining the biotransformation pattern of relatively small and non-polar benzene derivatives. Their biotransformation pattern was demonstrated to be predominantly determined by the chemical reactivity of the benzenes and - although to a much lesser extent for bromine, iodine and amino containing derivatives- more than by steric: factors (Br,I) and/or electronic dipole-dipole interactions with the activated (FeO) 3+cofactor (NH 2 ). However, the further extension of the presently developed QSARs to more complicated and especially to larger molecules might require the incorporation of computer techniques that take into account preferential stereoselective substrate orientations imposed by the active sites. Although the molecular modelling computer techniques to do these type of additional orientation analyses are available, such studies also require the availability of the enzyme crystal structures. Since such 3-D structures for cytochromes P450 are at present restricted to some bacterial cytochromes P450 and not available for mammalian cytochromes P450 the studies may have to await the elucidation of the mammalian cytochrome P450 structures. Nevertheless, the concept as such might be developed using either molecular modelling combined with quantum chemical calculations for the bacterial cytochrome P450 system as done recently by Zakharieva et al. and Freutel et al., or by performing similar studies for other biotransformation enzymes, such as for example the glutathione S-transferases for which several 3-D structures have been described.In general, QSARs can be used to predict the biological activity of non-tested chemicals but they can also give more insight in the mechanistic rules of biotransformation of molecules. This knowledge can be very helpful for example when the technical ability of cloning cytochromes P450, and thus producing these enzymes in large quantities, is increasing. Cytochromes P450 could then be used for the synthesis of chemical products (P450 biotechnology). Especially cytochromes P450 are suitable as they are able to convert many structural diverse substrates. The advantage of enzymatic synthesis compared to chemical synthesis is that it causes among others less waste problems.

Organisms are frequently exposed to low molecular weight xenobiotic compounds. An advanced enzymatic machinery modifies these compounds into more hydrophilic metabolites which are subsequently excreted from the body. This process of biotransformation aims to detoxify bodyforeign compounds. Ironically, reactive intermediates may also be formed during the biotransformation process and can interact with macromolecules or receptors, with possible toxicological consequences (bioactivation). Toxicological testing of all new and existing compounds is a moneyand time-consuming problem and therefore alternatives are urgently needed. The main objective of the studies decribed in this thesis, which is outlined in chapter 1 , was to describe QSARs (quantitative structure-activity relationships) for the biotransformation and toxicity of halogenated aminoand nitrobenzene derivatives, Special attention was focussed on the most important phase I biotransformation enzyme involved, the cytochrome P450 system. In addition, in the course of the investigations some attention was paid to the glutathione/glutathione S-transferase dependent phase 2 biotransformation pathway.Chapter 2 gives an overview of the biotransformation enzymes primarily involved in the metabolism of halogenated amino- and nitrobenzene derivatives with respect to their function, regulation, occurrence, molecular/biochemical mechanisms and role in bioactivation and detoxication of xenobiotics. Biotransformation of amino- and nitrobenzene derivatives plays a crucial role in the generation of reactive intermediates assumed to contribute to the toxicity of this class of compounds. Since biotransformation of nitro- and aminobenzene derivatives is known to include pathways leading to both detoxication and bioactivation, insight into factors that direct the rates and regioselectivities of biotransformation processes or interactions with nucleophiles, like macromolecules or the tripeptide glutathione, will help to gain insight in factors that direct processes of, and chances on, bioactivation or detoxication of these compounds. Regioselective enzymatic conversion of relatively small aromatic substrates in the relatively large and aspecific active sites of cytochromes P450 is assumed to depend predominantly on chemical characteristics of the substrates, whereas for relatively large substrates stereoselective positioning through steric constraints of the protein core in the active site becomes more important (see also chapter 2 ). Therefore, the relative differences in reactivity of various sites in a small aromatic compound can be expected to affect the possibilities for enzymatic conversion of the molecule. These considerations prompted us to investigate whether computer calculated molecular orbital characteristics of halogenated benzene derivatives, in combination with insight into the molecular mechanisms of their enzymatic conversion, could provide a basis for the prediction of their metabolic fate.Chapter 3 presents a study on the regioselectivity and underlying mechanisms for the cytochrome P450-catalyzed aromatic hydroxylation of monofluoroanilines. This study provides insight into the molecular mechanism of the cytochrome P450-catalyzed aromatic hydroxylation of molecules containing a heteroatom as well as factors that influence the regioselectivity of hydroxylation. Three mechanisms for aromatic hydroxylation can be proposed; hydrogen abstraction (I), or electron abstraction followed by proton release (II) both leading to formation of a NH . radical. Upon rearrangement of the radical and OH- rebound from the (FeOH) 3+species, the aminophenol product is formed. Aromatic hydroxylation might also proceed by a direct interaction of the high-valent iron-oxo cytochrome (FeO) 3+intermediate with theπ-electrons of the aromatic ring resulting in a so-calledσ-adduct (III), which rearranges to the aminophenol, either directly or through formation of epoxides and/or ketones as intermediates. First, it was demonstrated that the regioselectivity of the aromatic hydroxylation was influenced by the position of fluoro-substituents at the aniline-ring and that the observed regioselectivity for hydroxylation of these small aromatics was not influenced by the relatively aspecific and large hydrophobic active sites of the cytochromes P450. As expected, a fluorine-substituent induces an effect on the electronic characteristics of the aniline-molecule and this effect was quantified using molecular orbital calculations. Considering the possible molecular mechanisms for aromatic hydroxylation, it appeared that the observed in vitro and in vivo regioselectivity correlated best with the frontier orbitals of importance for a direct interaction of the (FeO) 3+species with theπ-electrons of the aromatic molecule e g. the density distribution of the HOMO/HOMO-1. The spin density distribution of the NH . radicals -a parameter of importance for the hydrogen abstraction as well as electron abstraction plus proton release mechanisms- could not explain the observed regioselectivities, indicating that the regarding mechanisms are less likely. In a later study it was demonstrated that for a series of fluorobenzenes the regioselectivity for aromatic hydroxylation could even be predicted on the basis of the frontier orbital density distribution for electrophilic attack within 6% accuracy (r=0.96) [Rietjens et al., 1993]. For the fluoroanilines the C4 position is hydroxylated to higher extent, whereas the C2/C6 positions are hydroxylated to an extent lower than their chemical reactivity predicts. This deviation might result from steric hindrance of the amino moiety for electrophilic attack, a stereoselective positioning of the substrate through an interaction of the amino moiety with amino acid residues in the active site as has been described for P450 debrisoquine 4-hydroxylase, or a dipole-dipole or electronic interaction between the substrate and the activated cytochrome P450 (FeO) 3+species. It is stressed here that the juxtaposition of these small substrates in the active site with respect to the Fe 3+resting state as determined on the basis of 1H-NMR T1 relaxation or cristallography studies, might not represent the actual orientation with respect to the reactive (FeO) 3+intermediate actually performing the hydroxylation step [Koerts et al., 1995].In chapter 4 the aromatic hydroxylation of anilines was further investigated with special emphasis on possible relationships between kinetic parameters and both the physicochemical and electronic substrate characteristics. This was done in order to provide a basis for molecular orbital based quantitative structure activity relationships (MO-QSARs) for kinetic characteristics of the cytochrome P450-mediated aromatic hydroxylation of a homologous series of aniline-derivatives. It was demonstrated that the k cat for C4-hydroxylation in a series of substituted anilines strongly correlates with the HOMO energy of the anilines for the iodosobenzene supported P450 reaction in isosafrole induced microsomes. This observation is in accordance with a mechanism that proceeds by an initial electrophilic interaction of the (FeO) 3+intermediate with the frontier ic electron of the aniline-substrate. In a NADPH/oxygen-supported P450 system, however, it was demonstrated that the interaction of the (FeO) 3+species on the aniline-substrate is no longer rate-limiting, and therefore cannot be described by this QSAR. However, when the electrophilic reactivity of the substrates becomes too low, as is the case for 2,3,5,6-tetrafluoroaniline (E HOMO of -9.24 eV) the initial attack of the cytochrome (FeO) 3+on the substrate might become the rate-limiting step in the overall catalysis. The relatively low conversion rate observed for aromatic hydroxylation of fluorobenzenes for example [Rietjens et al., 1993], might be explained by the relatively low electrophilic reactivity of these compounds. If the resulting metabolite is less toxic than its parent compound, a decreased conversion might have implications for the toxicity and the other way around.To fully describe the cytochrome P450-mediated biotransformation of these aniline- derivatives, a sensitive and efficient analytical technique was developed for the detection and quantification of 2-aminophenols ( chapter 5 ). The principle of the method was based on a dimerization reaction of 2-aminophenols to an intensively colored 2- hydroxy-isophenoxazin-3-one in an acidic environment using ferric ions as the catalyst. It was demonstrated that this method was also applicable for the detection and quantification of halogenated 2-aminophenol derivatives.Besides aromatic hydroxylation reactions, also oxidative dehalogenation reactions mediated by cytochromes P450 were investigated. Halogen substituents are often introduced into molecules to block positions at the aromatic ring of drugs or agrochemicals for bioactivation or biodegradation. Fluorine substituents are frequently used for this purpose, due to the strong C-F bond, and a Van der Waals radius that almost equals that of a hydrogen. In order to study the molecular mechanism of oxidative aromatic dehalogenation as well as the consequences of halogen substitution for regioselective hydroxylation and the formation of reactive intermediates, the study described in chapter 6 was undertaken. Using halogenated anilines as the model compound, the effect of a varying halogen substituent patterns on the cytochrome P450-catalyzed dehalogenation of 4- halogenated anilines to 4-aminophenols was investigated. In the case of C4-fluorinated aniline derivatives, the cytochrome P450-mediated metabolic pathway has been unequivocally demonstrated to result in direct formation of a reactive 1,4- benzoquinoneimine and fluoride anions the primary reaction products [Rietjens et al., 1993]. These reactive benzoquinoneimine metabolites may interact with cellular macromolecules and hence can lead to destruction of molecules essential to living cells.The study described in chapter 6 clearly demonstrated that upon the cytochrome P450-mediated oxidative dehalogenation to the primary reactive 1,4-benzoquinoneimine a fluorine substituent at C4 position of the aromatic aniline-ring was more easily eliminated than a chloro-, bromo- or iodo-substituent. A similar decrease in dehalogenation was observed in a NADPH/0 2 supported microsomal P450 system, as well as in a tBuOOH supported microsomal P450 system, or a system with purified reconstituted P4502B1, or in a system with the heme-based mini-enzyme microperoxidase 8. This indicates that the decrease in dehalogenation was not a result of a change in rate-limiting steps in the P450 catalysis or a change in the contribution of P450 enzymes with a change in the halogen substituent. Structureactivity relationship principles were applied to investigate the reaction mechanism of dehalogenation. The results obtained strongly indicate that the possibilities for the cytochrome P450-mediated dehalogenation of 4-halogenated anilines to 4-aminophenol metabolites are dependent on i) the characteristics of the halogen that has to be eliminated, the most electronegative and smallest halogen (fluorine) being the one most easily eliminated and ii) the electron-withdrawing capacities of the other substituents in the aromatic ring, electron-withdrawing substituents decreasing the relative rate of the reaction. The conclusion that not only a fluoro-, but also a chloro-, bromo- and iodo- substituent, is eliminated as a halogen anion serves as the best explanation for the observations in this study. In addition, it was demonstrated that blocking the C4-position for aromatic hydroxylation by cytochromes P450 resulted in a metabolic switch from dehalogenation and 4-aminophenol formation, to formation of 2-aminophenol- and N- hydroxyaniline-derivatives. Although halogen substitution at the C4-position of an aniline indeed leads to a decreased metabolism at that site, the formation of the reactive 1,4- benzoquinoneimine instead of 4-aminophenol as the primary metabolite, as well as the switch to N-hydroxylation, giving rise to reactive hydroxylamino- and nitroso-derivatives may have considerable toxicological implications.The biological activity of numerous aniline derivatives has been shown to be closely related to the cytochrome P450 mediated oxidative attack at their nitrogen center, yielding products with increasing toxic properties, namely N-hydroxyaniline and nitrosobenzene derivatives. Reactive N-hydroxyanilines cause for example ferrihemoglobin formation (methemoglobinemia) with concomittant co-oxidation to nitrosobenzenes. The resulting nitrosobenzenes in turn are able to interact with cysteine residues of hemoglobin or with the tripeptide glutathione [Eyer, 1988]. On the other hand, the cytochrome P450 mediated aromatic hydroxylation of anilines may represent a detoxification pathway due to the efficient conjugation of the resulting phenolic metabolites and subsequent excretion from the body. However, the cytochrome P450 mediated formation of aminophenol metabolites has been proposed to play a role in the nephrotoxicity of halogenated anilines, occuring predominantly at the proximal tubule and to a lesser extent at the distal tubule [Lo et al., 1990 & 1991; Rankin et al., 1986a & b; Valentovic et al., 1992]. Chapter 7 presents a clear example of a metabolism-toxicity relationship study. Since the cytochrome P450 mediated regioselective hydroxylation seems to direct the toxicity of aniline compounds, a study was performed on the relationships between the regioselectivity of the hydroxylation of C4-substituted 2-fluoroaniline derivatives and their toxic endpoints nephrotoxicity and/or methemoglobinernia. Depending on the derivative, nephrotoxicity at the tubular site and/or methemoglobinernia was shown to occur. The extent of nephrotoxicity induced by C4-substituted 2-fluoroanilines was shown to be related to the extent of C4-hydroxylation, and the extent of methemoglobinernia was shown to be related to the extent of N-hydroxylation. Consequently, a change in the regioselective hydroxylation from the cytochrome P450-mediated C4-hydroxylation to N-hydroxylation resulted in a change of toxic endpoint from nephrotoxicity to methemoglobinemia.In chapter 8 a study was directed at understanding the marked differences in biotransformation pathways of anilines and their chemically oxidized analogues nitrobenzenes. In vivo and in vitro, 2,5-difluoroaniline was demonstrated to become predominantly metabolized through a cytochrome P450-mediated pathway, whereas 2,5- dihuoronitrobenzene is predominantly converted through glutathione conjugation, and the a minor extent through nitroreduction and cytochrome P450-mediated aromatic hydroxylation. On the basis of computer calculations a hypothesis was presented that might explain the differences in metabolic pathways on the basis of their molecular orbital substrate characteristics. It was suggested that the HOMO of the thiolate anion of glutathione will interact more efficiently with the LUMO of a nitro compound (E LUMO = -1.57 eV) than with the LUMO of a less electrophilic amino compound (E LUMO = -0.08 eV). Concerning the aromatic hydroxylation on the other hand, it was suggested that the SOMO of the cytochrome P450 (FeO) 3+intermediate can interact more efficiently with the amino compound (E HOMO = -8.83 eV) than with the nitro compound (E HOMO = -10.29 eV). These considerations are in accordance with the observation in chapter 4, that the reaction rate of the cytochrome P450-mediated C4-hydroxylation decreases with decreasing electrophilic reactivity of an aniline substrate. Substrates with an E HOMO value below -9.2 eV were converted at a relatively low rate by the cytochromes P450. The considerations are also in accordance with the recent observation that the rate of both the chemical and glutathione S-transferase catalyzed glutathione conjugation of fluoronitrobenzenes increases with decreasing E LUMO values [Rietjens et al., 1995].Halogenated nitrobenzenes are known to become metabolized by nitroreduction, glutathione conjugation and aromatic hydroxylation. For some of the toxic effects of nitrobenzenes, the formation of reactive metabolites is a prerequisite. Reduction of nitrobenzenes to the aminobenzene analogues, via the formation of methemoglobinemia-inducing intermediates (nitrosobenzene, N-hydroxyaniline or radical intermediates), is a clear example of bioactivation of nitrobenzenes. The different susceptibility for nitroreduction, and also the presence of competing biotransformation pathways, like glutathione conjugation and aromatic hydroxylation, have been proposed to set the chances for methemoglobinernia exerted by nitrobenzene-derivatives. Applying QSAR approaches, a study was performed in order to investigate whether a metabolism-toxicity relationship exists for a series of fluoronitrobenzenes providing insight in the relative importance of these two suggestions. Chapter 9 describes a combined study concerning the influence of the substituent pattern of fluoronitrobenzenes on their biotransformation and their capacity to induce methemoglobinemia. It was demonstrated that increased possibilities for the conjugation of fluoronitrobenzenes to cellular nucleophiles was accompanied by decreased contributions of aromatic hydroxylation and nitroreduction. A QSAR could be described for the rate of conjugation with the tripeptide glutathione or with bovine serum alburnine (a model for cellular nucleophiles) and calculated parameters for electrophilicity of the fluoronitrobenzenes, showing that with increasing number of fluoro-substituents the conjugation pathway will become more important. In addition, the intrinsic reactivity of the fluoronitrobenzenes for nitroreduction by cecal microflora and rat liver preparations was not related to the in vivo methemoglobin-forming capacity. This observation, in combination with the QSAR for conjugation, led to the conclusion that the different methemoglobinemic capacity must rather result from differences in the inherent direct methemoglobinemic reactivity of the various toxic metabolites, and/or from the difference in reactivity of the fluoronitrobenzenes with glutathione or other cellular nucleophiles. As a result of an increased electrophilic reactivity of fluoronitrobenzenes, another toxic endpoint than methemoglobinemia can be expected.In the final study described in chapter 10 , comparative MO-QSAR studies clearly demonstrated that outcomes on both the regioselectivity and the rate of the cytochrome P450-catalyzed aromatic hydroxylation could be described on the basis of frontier orbital characteristics of the substrates. The MO-QSARs obtained for rats, were also valid for other species, including man. These results strongly support the conclusion that the conversion of the relatively small halogenated benzene derivatives in the relatively large and aspecific active sites of the mammalian cytochromes P450 -even when derived from various species- are mainly dependent on chemical reactivity parameters of the substrates. This importance of the substrate characteristics also dominates over the influence of the specific P450 enzymes in determining the regioselectivity of biotransformation both in vitro and in vivo.In conclusion, the studies described in this thesis cleary demonstrate that molecular orbital calculations in combination with the frontier orbital theory are a usefull additional tool to study the mechanism of cytochrome P450 enzyme catalysis. In addition to the theoretical considerations on the molecular reaction mechanisms and enzyme catalysis, empirical data on the metabolism are essential to construct valid rules or even quantitative structure activity relationships for the biotransformation and toxicity of series of related compounds. Insight into the possibilities for biotransformation have been demonstrated to provide insight into the formation of toxicity determining reactive intermediates. Defining QSARs for biotransformation have been shown to offer an approach to explain the type and extent of the toxic effects exerted by the series of halogenated amino- and nitrobenzene derivatives.Refined models, which do not only focus on the chemical reactivity parameters of the substrates, but also take into account possible stereoselective positioning or interactions with specific amino acid residues in the active site, will contribute to a better prediction of metabolic profiles of existing and new drugs, agrochemicals and other industrially relevant compounds.Finally, it is challenging to apply the approaches described in this thesis to other enzymes with more specific active sites than the relatively large aspecific active sites of the biotransformation enzyme cytochrome P450.

The cytochrome P450 enzyme system is involved in the metabolism and elimination of an almost unlimited number of endogenous and exogenous substrates. Biotransformation by cytochromes P450 plays a role in the conversion xenobiotics into more hydrophilic products. Generally, this process of biotransformation in which cytochrome P450 reactions take part, leads to elimination of the xenobiotic through urine and / or faeces although in some cases this process can also lead to the formation of more toxic metabolites. Because of its significant role in the conversion of numerous endogenous and exogenous compounds, a complete understanding of the cytochrome P450 enzyme system is of importance in toxicology, pharmacology, anesthesiology, pathology and other related biomedical fields.Twenty-five years ago the phospholipids of the membrane of the endoplasmatic reticulum, to which the enzyme system is bound, appeared to play an important role in the in vitro cytochrome P450 enzyme system. Although the role of the membrane(phospholipids) in the cytochrome P450 system has been intensively studied since then, it has not resulted in a unanimous conclusion. The goal of this thesis was therefore, to gain further insight into the role of the membrane and membrane phospholipids in the cytochrome P450 system. Attention is thereby not only paid to the effects of the membrane and the phospholipids on cytochrome P450 enzymes but also to the effects on another important protein component of the enzyme system, namely NADPH-cytochrome reductase.In the first two chapters the cytochrome P450 enzyme system and the membrane(phospholipids) of the endoplasmatic reticulum are described respectively. In chapter 1A the structural and catalytical properties of the cytochrome P450 enzyme system are briefly discussed. Special attention is paid to the occurence, multiplicity, induction, the structure of the individual protein components of the enzyme system and the catalytic cycle of cytochromes P450. The individual steps in this catalytic cycle are discussed in detail.Chapter 1B deals with the structural aspects of the membrane phospholipids and the membrane of the endoplasmatic reticulum. In addition a brief description of the different types of reconstituted systems used in this thesis is given. Finally, the effects of phospholipids on the cytochrome P450 enzyme system, reported up to now in the literature, and various current hypotheses for the stimulating effect of phospholipids are discussed briefly.In chapter 2 results are described that characterise the sensitivity of microsomal and isolated cytochrome P450 IA1 and IIB1 for an organic hydroperoxide; cumene hydroperoxide (CuOOH). These data provide information on the difference in the way of membrane incorporation of these two cytochrome P450 enzymes. Up to now, very little attention has been paid to possible differences in sensitivity of cytochromes P450 to conditions of oxidative stress. A difference in sensitivity for (hydro)peroxides between different forms of cytochrome P450, as demonstrated with CuOOH in the present study, can be of importance from a toxicological point of view. Especially in cases in which conversion by one cytochrome P450 enzyme results in detoxification of the substrate whereas another cytochrome P450 enzyme causes bioactivation of the substrate, a difference in sensitivity for conditions of oxidative stress can then result in a shift in the metabolite pattern.Cytochrome P45 0 IIB1, embedded in the microsomal membrane is more sensitive towards CuOOH treatment than microsomal cytochrome P450 IA1. Purification of these enzymes and reconstitution in a system in which the proteins remain soluble results in a disappearance of the difference in CuOOH sensitivity between the two cytochrome P450 enzymes. Upon incorporation of cytochrome P450 IA1 and IIB1 into an artificial membrane, cytochrome P450 IIB1 again appears to be more sensitive towards CuOOH than cytochrome P450 IA1. Furthermore, the EC-50 values (effective cumene hydroperoxide concentration which causes 50% inhibition of cytochrome P450 dependent activities) in the microsomal and membrane incorporated reconstituted systems are comparable. Based on these results it is concluded that (1) the difference in sensitivity between cytochrome P450 IA1 and IIB1 towards treatment with CuOOH originates from a difference in the way these cytochrome P450 enzymes are incorporated into the membrane, that (2) the purification procedure does not affect the parameters determining the way of incorporation of the protein into the membrane and that (3) the way of membrane incorporation of cytochrome P450 enzymes in microsomal and reconstituted systems is comparable.In chapter 3 the effects of changes in the fatty acyl moiety of phosphatidylcholine (PC) from dilauroyl (di 12:0) to distearoyl (dil8:0) on the kinetics of reconstituted cytochrome P450 IA1 and IIB1 were investigated. So far, studies on the effect of phospholipids on the kinetics of cytochrome P450 dependent reactions have focussed on one P450 enzyme or one phospholipid. Furthermore, mainly the effect on the activity at non-saturating substrate concentrations was investigated without paying much attention to the effect on the kinetic parameters K m and V max , of a cytochrome P450 catalysed reaction. An effect of phospholipids on for example the K m could be of considerable importance especially because in living organism the substrate concentrations will generally be low. Furthermore, a different effect of phospholipids on the substrate's apparent Km of different cytochrome P450 enzymes - converting the substrate to different metabolites - might not only affect substrate conversion rates but also the metabolite pattern.The results presented in chapter 3 demonstrate that the V max of the cytochrome P450 dependent O-dealkylation of alkoxyresorufins and ethoxycoumarin for both cytochrome P450 IA1 and IIB1 is two times higher in the PC di 12:0 system compared to the PC di 18:0 system. The effect of a change in the fatty acyl moieties on the Km of the xenobiotic substrate however, appeared to be different for the two cytochrome P450 enzymes. For cytochrome P450 IA1 the K m appeared to be lower in the PC di 12:0 system compared to the PC dil8:0 system whereas for cytochrome P450 IIB1 the K m was lowest in the PC di l8:0 system. Additional results demonstrated that the kinetic parameters were dependent on the PC : P450 ratio and that changing this ratio affected the kinetic parameters of cytochrome P450 IA1 and IIB1 in a different way.The reason for the differential effect on the substrate apparent K m was further investigated in a series of experiments in which the effect of PC di 12:0 and PC di l8:0 on individual steps of the catalytic cycle of cytochrome P450, like substrate binding, oxygen binding and rate of electron transfer, was studied. From these experiments it was concluded that the higher V max in the PC di 12:0 system, observed for both cytochrome P450 enzymes, was at least in part due to the higher affinity of cytochrome P450 for NADPH-cytochrome reductase in the PC di 12:0 system. Furthermore, these experiments demonstrated that the different effect of a change in the fatty acyl moieties of PC on the K m of cytochrome P450 IA1 and IIB1 did not result from a different effect on substrate binding, oxygen binding and rate of electron transfer. This means that the differential effect on the K m must result from an effect on one or more of the other steps in the catalytic cycle such as reductive oxygen splitting, substrate conversion and / or product release.The results show that the effect of a change in the type of PC and or the PC : P450 ratio on the kinetic parameters, K m and V max , is dependent on the cytochrome P450 enzyme used in the reconstitution. Furthermore, in contrast to what is generally assumed and based on results under non- saturating substrate conditions [1 -4], the addition of PC appears to result for some cytochrome P450 enzymes in a decrease of the V max in the reconstituted system. The results in chapter 3 also demonstrate that this is not reflected in lower but - in contrast - in higher conversion rates at non- saturating substrate concentrations (Figure 9.1), because the K m is decreased simultaneously.In chapter 4 the existence of a preference - with respect to binding - of cytochrome P450 IIB1 for phospholipids with certain headgroups or fatty acyl moieties was investigated. The existence of "boundary" phospholipids (phospholipids which bind to cytochrome P450 with specificity and high affinity) for microsomal cytochrome P450 has been a topic for several studies. Nevertheless, little is known about this subject and unanimous conclusions have not been reached. It has been suggested that the composition of the membrane in the direct vicinity of cytochromes P450 is different from the rest of the membrane [5] and that specific interactions exist between cytochrome P450 and phosphatidylethanolamine (PE) [61 and phosphatidic acid (PA) [7].The results in chapter 4 show that the apparent binding constant (K d ) of a cytochrome P-450 IIB1 - phospholipid complex is dependent on the degree of unsaturation of the phospholipid side chains; demonstrating a decrease in the Kd with increasing degree of unsaturation, but independent of the length of the acyl chains. In addition, the apparent Kd appeared to be dependent on the headgroup of the phospholipid molecule, showing a significantly higher K d for PE di 16:0 compared to PC di 16:0, PS di 16:0 and PI 16:0/18:1.Translation of these results to the in vivo situation has to be done with caution because the results were obtained in a reconstituted system with isolated, solubilised cytochrome P450. In the membrane of the endoplasmatic reticulum other factors such as for example the presence of other proteins can play an additional important role in the interaction of cytochrome P450 with phospholipids. Furthermore, the effect of the length and the degree of unsaturation of the fatty acyl chains on the Kd was determined for PC. It remains to be established whether for phospholipids with different headgroups similar influences of the length an degree of unsaturation of the acyl chains are observed. Investigations in this direction are however, seriously hampered by the fact that series of pure molecular species of PE, PS and PI are not commercially available.In chapter 5 the existence of specific phospholipid : protein interactions for NADPH -cytochrome reductase was investigated. Compared to cytochrome P450 very little attention has been paid to possible interactions between phospholipids and NADPH-cytochrome reductase and possible consequences of such interactions for the cytochrome P450 system. NADPH-cytochrome reductase is a very important component of the cytochrome P450 enzyme system and the stimulating effect of phospholipids on the rate of cytochrome P450 dependent reactions may in part originate from an effect on NADPH-cytochrome reductase resulting in a more efficient electron transfer to the cytochromes P450.Based on the results from 31P-NMR experiments and chemical analysis, it was concluded that NADPH-cytochrome reductase exhibits a preference for the negatively charged phospholipids phosphatidylserine (PS) and phosphatidylinositol (PI). In addition, experiments investigating the possible consequences of a special interaction of NADPH-cytochrome reductase with PS and PI demonstrated that (1) PS and PI had a significantly different effect on the DPH-PC dependent quenching of tryptophan fluorescence of NADPH-cytochrome redcutase compared to PE and PC and that (2) the V max of cytochrome P450 IIB1 dependent O-dealkylation of pentoxyresorufin in the presence of 1:1 mixtures of PS:PC and PI:PC were respectively higher and lower compared to the Vmax in the presence of a 1:1 mixture of PE:PC or PC alone. These phenomena might best be explained by a PS and PI induced specific change in the conformation of NADPH-cytochrome reductase. Regarding the fact that the specific interaction in both cases involves a negatively charged phospholipid suggest a possible role of the phospholipid charge. However, the fact that the effects of PS and PI on the V max of the cytochrome P450 catalysed reaction are different demonstrates that phospholipid charge cannot be the only factor.In chapter 6 the redox cycling of 7-alkoxyresorufins and the product of their metabolism by cytochrome P450, resorufin, by NADPH-cytochrome reductase is investigated. Redox cycling is a process in which a substrate is I-electron reduced, in this case by NADPH-cytochrome reductase. The electron is transfered to molecular oxygen and the substrate is returned to its initial state and can enter a new cycle. During this process reactive oxygen species are formed which can initiate lipidperoxidation and / or inactivate cytochrome P450. Especially in systems in which the NADPH-cytochrome reductase concentration is relatively high this process is a disturbing side-reaction, because it uses up reduction equivalents resulting only in the formation of reactive oxygen species which may cause protein inactivation and lipidperoxidation. In the reconstituted systems used in this thesis redox cycling can play an important role because the NADPH-cytochrome reductase : cytochrome P450 ratio is 15 to 30 times higher than in the in vivo situation. Furthermore, cytochrome P450 has been demonstrated to be very sensitive to lipidhydroperoxides - formed during lipidperoxidation - and reactive oxygen species.The results of the present chapter demonstrate that at physiological pH alkoxyresorufins are much better substrates for redox cycling than resorufin. The inability of resorufin to stimulate redox cycling originates from the fact that at physiological pH resorufin exists mainly in its deprotonated form and this form is a much worse substrate for redox cycling than its protonated form. AM1 molecular orbital computer calculations demonstrated that the energy (E) of the lowest unoccupied molecular orbital (LUMO), i.e. the orbital into which the electron will be placed during redox cycling, of the deprotonated form is higher compared to the E LUMO of the protonated form. Furhermore, one-electron reduction of the protonated form appeared to be energetically favorable by 363.5 kJ/mol over one-electron reduction of the deprotonated form. In addition, the computer calculations demonstrated that the one electron reduced resorufin is most likely to become protonated at the O-atom of the intramolecular semiquinone imine moiety before reduction by a second electron. Finally, it was demonstrated that incorporation of NADPH-cytochrome reductase into an artificial membrane results in an increased redox cycling activity of resorufin compared to solubilized NADPH-cytochrome reductase. This was explained by an increase in the protonated form in the membrane either by (1) favored partitioning of the protonated form into the membrane or by (2) an effect of the membrane on the protonation equilibrium of resorufin in favor of the protonated form. This result points at the role of the membrane in concentrating apolar substrates of the cytochroom P450 : NADPH-cytochrome reductase svstem [8-10].In chapter 7 the use of AM1 MO calculations in predicting the ability of compounds to stimulate redox cycling, as demonstrated in chapter 6, was further investigated. Therefore, in addition to resorufins, the redox cycling ability of 1,4-benzoquinones was investigated. Quinones are toxic compounds that are often used in chemistry. They are also used in pharmacology, for example as anticancer drugs because of their toxic character. The mechanism through which quinones exert their toxic effects is believed to involve the covalent binding of quinones to cellular nucleophilic macromolecules and /or the quinone catalysed process of redox cycling. 1,4- Benzoquinone has been demonstrated to redox cycle very poorly. In the literature, the poor redox cycling of 1,4-benzoquinone has been ascribed to a very high 1-electron reduction potential. The results from chapter 7 demonstrate however, that, at physiological pH, 1,4-benzoquinone is quickly 2-electron reduced by NADPH-cytochrome reductase to form 1,4- hydroquinone. Instead of transfering its electron on to molecular oxygen, the 1 -electron reduced semiquinone is protonated and subsequently reduced by a second electron. However, at pH>9 the 1,4-benzoquinone appears to be capable of stimulating redox cycling. Furthermore, the pH- and concentration-dependencies of redox cycling in a system with NADPH- cytochrome reductase between 1,4-benzoquinone and 1,4-hydroquinone are demonstrated to be similar. Based on these observations it was concluded that 1,4-benzoquinone is capable of redox cycling from its deprotonated, 2-electron reduced form at relatively high pH levels which ensures an adequate concentration of the deprotonated form The results from chapters 6 and 7 demonstrate the importance of the protonation / deprotonation equilibrium of the I- and 2-electron reduced forms in the redox cycling process.In the paragraphs above the results of the study on the role of phospholipids in the cytochrome P450 : NADPH-cytochrome reductase system as it was executed for this PhD thesis are presented. Ever since 1968 [11] the role of phospholipids in this system has been a topic for numerous studies in which many questions concerning this role have been answered but many new questions have also been raised. Unfortunately, this thesis does not provide the answers to all these new questions because the cytochrome P450 system is too complex and the number of different phospholipids and the effects they induce is too large. Answering these questions will require many, many years of additional research. The goal of this thesis was to investigate certain aspects of phospholipids and the cytochrome P450 enzyme system to which up to now little attention has been paid in order to gain further insight into the role of phospholipids in the cytochrome P450 enzyme system.Altogether, the results of the experiments in this thesis present some new insightsin the role of phospholipids on the cytochrome P450 enzyme system. Furthermore, additional evidence for already existing hypotheses of the stimulating effect of phospholipids are also presented. The conclusions of the present thesis can be summarized as follows.(1) Cytochromes P450 can differ in the way they are incorporated in the membrane which might result in differences in their sensitivity towards cumene hydroperoxide.(2) The way of incorporation of cytochrome P450 enzymes in microsomal and reconstituted systems is comparable.(3) Phospholipids influence the kinetic parameters, Km and Vmax, of cytochrome P450 catalysed reactions in reconstituted systems.(4) The result of the effect phospholipids on the kinetic parameters of cytochrome P450 dependent reactions is not the same for all P450 enzymes.(5) The affinity of cytochrome P450 for NADPH-cytochrome reductase in a reconstituted system is dependent on the fatty acyl moiety of the phospholipid added to the system.(6) Phospholipids can decrease the apparent Kd of cytochromes P450 for their xenobiotic substrates.(7) The apparent K d of cytochromes P450 for phospholipids is dependent on the headgroup of the phospholipid and the degree of unsaturation of the fatty acyl chains but independent of the length of the acyl chains of the PC molecule.(8) There is a specific interaction between NADPH-cytochrome reductase and the negatively charged phospholipids PS and PI.(9) The membrane functions as a place where apolar substrates can accumulate thereby decreasing the apparent K m .(10) The membrane causes a shift in the overall protonation equilibrium of the substrate towards the protonated form.In addition to these conclusions a number of other interesting phenomena have been observed that have no bearing on the role of phopsholipids in the cytochrome P450 enzyme system but are also worth mentioning again.(1) The parameters determining the way of incorporation of cytochromes P450 in the membrane are not affected by the isolation procedure.(2) AM1 molecular orbital calculations is a useful additional tool in investigating the redox cycling capacity of chemicals.(3) The protonationequilibria of I- and 2-electron reduced compounds play an important role in their redox cycling ability.Finally, regarding the results of the experiments presented in this thesis one final conclusion must be added. The effect of phospholipids on the cytochrome P450 enzyme system is dependent on many factors such as the cytochrome P450 form, the fatty acyl moiety and headgroup of the phospholipid, the P450 : reductase ratio and the phospholipid : P450 ratio. Therefore, for a complete understanding of the mechanism(s) of action of phospholipids in the cytochrome P450 enzyme system, a detailed investigation of the effects of all phospholipids (and mixtures of phospholipids) on all P-450 forms at several P-450 : reductase and phospholipid : P-450 ratio's is necessary. This requires a vast amount of work although some of this work has already been done in these last twenty- five years. Comparison of these results is , however, difficult because of the different conditions used in these studies. It is therefore, advisable to come, in analogy to the nomenclature of cytochrome P-450, to standardized conditions for research in order for the results of differentlaboratories to be compared.