Matrix modulation of the toxicity of alkenylbenzenes, studied by an integrated approach using in vitro, in vivo, and physiologically based biokinetic models

Alkenylbenzenes such as estragole and methyleugenol are common components of spices and herbs such as tarragon, basil, fennel, mace, allspice, star anise and anise and their essential oils (Smithet al., 2002). There is an interest in the safety evaluation of alkenylbenzenes because these compounds can induce hepatic tumours in rodents when dosed orally at high dose levels (Milleret al., 1983; NTP, 2000). Based on the rodent studies with estragole, methyleugenoland structurally related alkenylbenzenes like safrole the hepatocarcinogenicity of alkenylbenzenes is ascribed to their bioactivation by cytochrome P450 enzymes leading to the formation of the proximate carcinogenen, the 1′-hydroxy metabolite, which is further bioactivated to the ultimate carcinogenen, the 1′-sulfooxy metabolite (Milleret al., 1983; Phillipset al., 1984; Randerathet al., 1984; Smithet al., 2010). The 1′-sulfooxy metabolite is unstable and binds via a presumed reactive carbocation intermediate covalently to different endogenous nucleophiles including DNA (Phillipset al., 1981; Boberget al., 1983; Milleret al., 1983; Phillipset al., 1984; Randerathet al., 1984; Fennellet al., 1985; Wisemanet al., 1987; Smithet al., 2002).

Because of their genotoxicity and carcinogenicity, the addition of estragole and methyleugenolas pure substances to foodstuffs has been prohibited within the European Union since September 2008 (European Commission, 2008). In 2008, the Joint FAO/WHO Expert Committee on Food Additives (JECFA) re-evaluated the safety of alkenylbenzenes and indicated that although evidence of carcinogenicity to rodents given high doses of alkenylbenzenes exists, further research is needed to assess the potential risk to human health at relevant dietary exposure levels (JECFA, 2008).

A significant difficulty in evaluating the toxicological data for alkenylbenzenes is that human exposure to these substances results from exposure to a complex mixture of food, spice, and spice oil constituents which may influence the biochemical fate and toxicological risk of the alkenylbenzenes. In this regard, it was shown that a methanolic extract of basil inhibited the formation of estragole DNA adducts in human HepG2 cells exposed to the proximate carcinogen 1′-hydroxyestragole (Jeurissenet al., 2008). This inhibition occurred at the level of sulfotransferase (SULT)-mediated bioactivation of 1′-hydroxyestragole into 1′-sulfooxyestragole (Jeurissenet al., 2008).

The objective of this PhD research was to study the inhibitory action of components in alkenylbenzene-containing herbs and spices on SULT-mediated alkenylbenzene DNA adduct formation and the consequences of this combination effect for risk assessment using estragole and methyleugenol as the model alkenylbenzenes. To achieve this objective, an integrated approach of in vitro, in vivo and physiologically based biokinetic (PBBK) models was applied to investigate how the SULT inhibition influences the bioactivation and thus potentially also the toxicity and risk assessment of estragole and methyleugenol.

Chapter 1of the thesis presents an introduction to the bioactivation, detoxification, genotoxicity and carcinogenicity of the alkenylbenzenes estragole and methyleugenol as well as a short introduction to PBBK modeling and the state-of-the-art knowledge on risk assessment strategies and regulatory status for alkenylbenzenes.

Chapter 2of the thesis identifies nevadensin as a basil constituent able to inhibit SULT-mediated DNA adduct formation in rat hepatocytes exposed to the proximate carcinogen 1′-hydroxyestragole and nevadensin. The type of inhibition by nevadensin was shown to be non-competitive with an inhibition constant (Ki) of 4 nM. Furthermore, nevadensin up to 20 μM did not inhibit 1′-hydroxyestragole detoxification by glucuronidation and oxidation. The inhibition of SULT by nevadensin was incorporated into the PBBK models describing bioactivation and detoxification of estragole in male rat and human. The models thus obtained predict that co-administration of estragole at a level inducing hepatic tumours in vivo (50 mg/kg bw) with nevadensin at a molar ratio to estragole representing the molar ratio of their occurrence in basil, results in more than 83% inhibition of the formation of the carcinogenic metabolite, 1ʹ-sulfooxyestragole, inthe liver of male rat and human even at 1% uptake of nevadensin.

To extend the work to other alkenylbenzene-containing herbs and spices than basil chapter 3  presents data showing that methanolic extracts from different alkenylbenzene-containing herbs and spices such as nutmeg, mace, anise and others are able to inhibit the SULT enzyme activity. Flavonoids including nevadensin, quercetin, kaempferol, myricetin, luteolin and apigenin were the major constituents responsible for this inhibition of SULT activity with Kivalues in the nano to sub-micromolar range. Also, the various flavonoids individually or in mixtures were able to inhibit estragole DNA adduct formation in human HepG2 cells exposed to the proximate carcinogen 1ʹ-hydroxyestragole, and to shift metabolism in favour of detoxification (e.g. glucuronidation) at the cost of bioactivation (e.g. sulfonation).

In a next step, the kinetics for SULT inhibition were incorporated in PBBK models for estragole in rat and human to predict the effect of co-exposure to estragole and (mixtures of) the different flavonoids on the bioactivation in vivo. The PBBK-model-based predictions indicate that the reduction of estragole bioactivation in rat and human by co-administration of the flavonoids is dependent on whether the intracellular liver concentrations of the flavonoids can reach their Ki values. Finally, we concluded that it is expected that this is most easily achieved for nevadensin which has a Kivalue in the nanomolar range and is, due to its methylation, more metabolically stable and bioavailable than the other flavonoids.

Chapter 4of the thesis investigates whether the previous observation that nevadensin is able to inhibit SULT-mediated estragole DNA adduct formation in primary rat hepatocytes could be validated in vivo. Moreover, the previously developed PBBK models to study this inhibition in rat and in human liver was refined by including a sub-model describing nevadensin kinetics. Nevadensin resulted in a significant reduction in the levels of estragole DNA adducts formed in the liver of Sprague–Dawley rats orally dosed with estragole and nevadensin simultaneously at a ratio reflecting their presence in basil. Moreover, the refined PBBK model predicted the formation of estragole DNA adducts in the liver of rat with less than 2-fold difference compared to in vivo data and suggests more potent inhibition in the liver of human compared to rat due to less efficient metabolism of nevadensin in human liver and intestine.

Also, an updated risk assessment for estragole was presented taking into account the matrix effect and this revealed that the BMDL₁₀ and the resulting MOE for estragole increase substantially when they would be derived from rodent bioassays in which the animals would be exposed to estragole in the presence of nevadensin instead of to pure estragole.

To extend the work to other alkenylbenzenes than estragole chapter 5 of the thesis investigates the potential of nevadensin to inhibit the SULT-mediated bioactivation and subsequent DNA adduct formation of methyleugenolusing human HepG2 cells as an in vitro model. Nevadensin was able to inhibit SULT-mediated DNA adduct formation in HepG2 cells exposed to the proximate carcinogen 1′-hydroxymethyleugenol in the presence of nevadensin.To investigate possible in vivo implications for SULT inhibition by nevadensin on methyleugenolbioactivation, the rat PBBK model developed in our previous work to describe the dose-dependent bioactivation and detoxification of methyleugenolin male rat was combined with the recently developed PBBK model describing the dose-dependent kinetics of nevadensin in male rat. Similar to what was presented for estragole in chapter 4, chapter 5 presents an updated risk assessment for methyleugenoltaking the matrix effect into account. This revealed that the BMDL₁₀ and the resulting MOE for methyleugenolincrease substantially when they would be derived from rodent bioassays in which the animals would be exposed to methyleugenolin the presence of nevadensin instead of to pure methyleugenol.

In a next step, we aimed at moving one step forward towards endpoints that are closer to initiation of carcinogenesis than DNA adduct formation, namely, formation of hepatocellular altered foci (HAF). Chapter 6 presents data showing that the potent in vivo inhibitory activity of nevadensin on SULT enzyme activity and on alkenylbenzene DNA adduct formation is accompanied by a potent in vivo reduction in early markers of carcinogenesis such as HAF. This also suggests that a reduction in the incidence of hepatocarcinogenicity is expected in liver of rodents when alkenylbenzenes would be dosed simultaneously with nevadensin.

Chapter 7presents a discussion on the in vitro and in vivo activity of dietary SULT inhibitors and their potential in reducing the cancer risk associated with alkenylbenzene consumption. This chapter also presents some future perspectives based on the major issues raised by our research. 

Altogether, the results of the present thesis indicate that the likelihood of bioactivation and subsequent adverse effects may be lower when alkenylbenzenes are consumed in a matrix containing SULT inhibitors such as nevadensincompared to experiments using pure alkenylbenzenes as single compounds. Also,the consequences of the in vivo matrix effect were shown to be significant when estragole or methyleugenolwas tested in rodent bioassays in the presence of nevadensin at ratios detected in basil, thereby likely increasing BMDL₁₀ and resulting  MOE values substantially in a subsequent risk assessment. However, the results also indicate that matrix effects may be lower at daily human dietary exposure levels of estragole or methyleugenoland nevadensin resulting from basil consumption. Also, matrix effects seem to be limited in the presence of other SULT inhibiting dietary flavonoids even at high exposure levels of these flavonoids coming from supplements. This indicates that the importance of a matrix effect for risk assessment of individual compounds requires analysis of dose dependent effects on the interactions detected, an objective that can be achieved by using PBBK modeling.

Overall, the present study provides an example of an approach that can be used to characterise dose- species- and inter-individual differences as well as matrix effects in the risk assessment of food-borne toxicants present (e.g. alkenylbenzenes). In this approach the most important toxicokinetic interactions are addressed using an integrated strategy of in vitro, in vivo and PBBK modeling approaches.