Abstract
Enniatins are cyclic hexapeptidic mycotoxins produced by fungi growing on field grains, especially in wet climates. They show considerable resistance to food and feed processing technologies and might cause intoxication of humans and animals. Enniatins are also under exploration as anticancer drugs. The observed difference of in vitro and in vivo toxicities suggests low absorption or fast elimination of the enniatins after oral uptake. In the study presented here, in vitro metabolism studies of enniatin B were performed using rat, dog, and human liver microsomes under conditions of linear kinetics to estimate the respective elimination rates. Furthermore, cytochrome P450 reaction phenotyping with chemical inhibitors selective for human enzymes was carried out. Twelve metabolites were separated and characterized by multiple high-performance liquid chromatographic/mass spectrometric analyses as products of oxidation and demethylation reactions. Biotransformation rates and metabolite patterns varied considerably in the three species. The intrinsic clearances determined in assays with rat, dog, and human liver microsomes were 1.16, 8.23, and 1.13 l/(h · kg), respectively. The predicted enniatin B in vivo blood clearances were 1.57 l/(h · kg) in rats, 1.67 l/(h · kg) in dogs, and 0.63 l/(h · kg) in humans. CYP3A4 was important for enniatin B metabolism in human microsomes as shown by 80% inhibition and impaired metabolite formation in the presence of troleandomycin. CYP1A2 and CYP2C19 were additionally involved. Preliminary results showed that CYP3A and CYP1A might also be relevant in rats and dogs. The extensive hepatic metabolism could explain the reduced in vivo potential of enniatin B.
Introduction
Enniatins are secondary fungal metabolites that are mainly produced by Fusarium strains, which belong to the most common cereal contaminants (Uhlig et al., 2007; Jestoi, 2008). Grains in northern Europe were contaminated with high levels of enniatins in recent years, adding up to maximal concentrations of 7.7 and 24.8 mg/kg in Norwegian and Finnish wheat, respectively (Uhlig et al., 2007). In Mediterranean countries, wheat and sorghum containing up to 493 and 696 mg/kg enniatins, respectively, were observed (Mahnine et al., 2011; Queslati et al., 2011). Enniatin B (EnnB), the most prevalent among the 28 homologs identified (Firáková et al., 2007), is a cyclic hexadepsipeptide consisting of three d-2-hydroxyisovaleric acid residues linked alternatively to three N-methyl-l-valines (N-Me-Val), resulting in a 18-membered (N-Me-Val-d-2-hydroxyisovaleric acid)3 molecule with the molecular formula and weight C36H63N3O9 and 639.83 g/mol (Fig. 1) (Blais et al., 1992).
EnnB is considerably resistant to heat (melting point: 173–175°C) (Altomare et al., 1995), acids, and digestion and has been shown to propagate through the feed and food chain reaching 150 μg/kg in a Swiss oat bran bread (Noser et al., 2007) and mean concentrations of 47 and 99 μg/kg, respectively, in baby foods and grain-based food products from Finnish and Italian markets, occurring in 97% of all samples (Jestoi et al., 2004). Breakfast cereals from Tunisian and Moroccan markets contained on average 57.4 and 10.6 mg/kg EnnB (Mahnine et al., 2011; Queslati et al., 2011). In egg yolks from laying hens fed with enniatin-contaminated feed, up to 3.8 μg/kg enniatin B was detected (Jestoi et al., 2009), demonstrating the molecule's tendency to bioaccumulate in lipophilic media.
The cyclopeptidic enniatins form ionophores with hydrophobic groups on the outside and polar groups in the core (Jestoi, 2008; Tedjiotsop Feudijo et al., 2010). They can transport mono- and divalent cations in sandwiched complexes or by creating channels in biological membranes.
The primary toxic action of enniatins is thought to result from the compounds' ionophoric character. Different in vitro toxicity studies have elucidated their antibacterial, antihelmintic, antifungal, herbicidal, and insecticidal potency (Jestoi, 2008; Tedjiotsop Feudijo et al., 2010). Enniatin B in levels up to 100 μM did not show genotoxic activity but demonstrated cytotoxicity at low micromolar concentrations (Ivanova et al., 2006; Jestoi, 2008; Behm et al., 2009). The observed activities included the specific inhibition of acyl-CoA cholesterol acyltransferase, depolarization of mitochondria, inhibition of osteoclastic bone resorption, and induction of apoptosis in cancer cells as well as the interaction with ATP-binding cassette transporters such as P-glycoprotein (Ivanova et al., 2010; Tedjiotsop Feudijo et al., 2010). The treatment of bacterial infections in the upper respiratory tract by a fusafungin-called mixture of enniatins A, A1, B, and B1 in a 1% nasal inhalation solution is the only approved application for humans (Lohmann, 1988; Kroslák, 2002).
Despite the considerable range of enniatin's pharmacological and toxicological properties and the mycotoxin's prevalence in grain-based food and feed, only a few studies determining the in vivo potency have been conducted so far. There are no reports of natural cases of mycotoxicosis in humans or animals, but feeding experiments have shown that chronic exposition may lead to subacute effects such as feed refusal, weight loss, and reduced productivity (Jestoi, 2008). Acute toxicity and death occurred only after intraperitoneal application of 10 to 40 mg/kg per day b.wt. over 6 days to HIV-infected, immune-deficient mice (McKee et al., 1997), whereas p.o. doses of 0.5 to 1 g/kg per day b.wt. over 6 days to mice and single doses of up to 50 mg/kg b.wt. in rats did not produce toxic effects (Gäumann et al., 1950; Lohmann, 1988; Bosch et al., 1989).
The lack of correlation between in vitro and in vivo toxicity is presumably the result of low bioavailability p.o., which may be caused by impaired uptake from the gastrointestinal tract because of low compound water solubility (Blais et al., 1992) and interaction with efflux pumps (Ivanova et al., 2010) or by elimination from the systemic circulation because of metabolism reactions. However, the pharmacokinetic properties and biotransformation products of enniatins have not yet been characterized.
Considering the evaluation of enniatins as potential drug candidates and emerging toxins, the aim of this study was to investigate the metabolism of enniatin B by performing in vitro elimination studies in rat, dog, and human liver microsomes and cytochrome P450 (P450) reaction phenotyping with chemical inhibitors selective for human enzymes.
Materials and Methods
Chemicals.
EnnB was isolated and purified from rice cultures of Fusarium avenaceum (Ivanova et al., 2006). NADP, NADPH, d-glucose 6-phosphate sodium salt, d-glucose 6-phosphate dehydrogenase from baker's yeast (Saccharomyces cerevisiae), (+)-N-3-benzylnirvanol, quinidine, furafylline, sulfaphenazole, and diethyldithiocarbamate were purchased from Sigma-Aldrich (St. Louis, MO). Troleandomycin was supplied by Santa Cruz Biotechnology, Inc. (Santa Cruz, CA). Acetonitrile (Romil, Cambridge, UK; Rathburn, Walkerburn, Scotland) and dimethyl sulfoxide (Sigma-Aldrich) were of high-performance liquid chromatography (HPLC) quality or analytical-reagent grade. Formic acid and ammonium formate were purchased from Merck (Darmstadt, Germany) and Sigma-Aldrich, respectively.
Microsomal Source.
The metabolism and reaction phenotyping studies were performed with commercially available liver microsomes (Celsis In Vitro Technologies, Baltimore, MD). Human liver microsomes (HLM) consisted of a mixed gender pool of 50 donors (No. X008067, Lot GJA) resulting in a preparation with moderate and average P450 activity suitable for stability and clearance studies. Dog liver microsomes (DLM) were prepared from male Beagle dogs (No. M00201, Lot LNY), and rat liver microsomes (RLM) were obtained from male Wistar rats (No. M00021, Lot NYB). All microsomal preparations were stored in liquid nitrogen until use. The total P450 content, protein concentrations, and specific activities of the different P450 isoforms were as supplied by the manufacturer.
Liquid Chromatography-Coupled Mass Spectrometry Analysis of EnnB and EnnB-Metabolites.
EnnB and EnnB metabolites produced by incubation with HLM, DLM, and RLM were determined by high-performance liquid chromatography-coupled mass spectrometry (LC-MS) using a Finnigan Surveyor MS Pump Plus with Autosampler Plus connected via an electrospray interface to a Finnigan LTQ linear ion trap mass spectrometer (all from Thermo Fisher Scientific, Waltham, MA). The mass analyzer was run in the full-scan mode (m/z 200-1000), and the electrospray interface was operated in the positive mode. The parameters of the electrospray ionization interface were adjusted as follows: a spray voltage of 4 kV, a capillary temperature of 300°C, a tube lens offset of 100 V, a sheath gas rate of 55 l/min, and an auxiliary gas rate of 5 l/min. Separation was achieved using a SunFire C18 column (50 × 2.1 mm, 3 μm; Waters, Milford, MA) with a 0.5-μm precolumn filter (Supelco, Bellefonte, PA). Compounds were eluted by linear gradient elution using acetonitrile (A) in water (B) (both containing 2 mM ammonium formate and 2 mM formic acid) at a flow rate of 0.35 ml/min, starting at 30% acetonitrile and rising to 100% within 8 min.
The EnnB molecule produced ion peaks at [M+H]+, [M+NH4]+, and [M+Na]+ with molecular ion masses of m/z 640, 657, and 662, respectively. [M+NH4]+ rendered the most intense and stable signal because of the high amount of ammonium salt in the mobile phase and was therefore selected for EnnB quantitation. In addition, the ratio of the three molecular ions was continuously monitored. EnnB was quantified comparing sample peak areas with an external calibration curve of EnnB in methanol in the concentration range of 3.5 to 465.6 ng/ml. The repeatability and reproducibility of the method were evaluated by analyzing replicates on different days, and accuracy was determined by recovery experiments determining EnnB spiked into the microsome assay reaction mixture.
Incubation aliquots were analyzed, and potential EnnB-metabolites were identified by mass and respective peak retention times relative to EnnB. Metabolite concentrations were estimated by using the external EnnB calibration curve because reference materials for the metabolites were unavailable. The structures of EnnB microsomal metabolites were tentatively identified using ion trap mass spectrometry, multiple-stage MSn fragmentation experiments, and high-resolution mass spectrometry. The instrumental analyses were supplemented with specific derivatization reactions for tagging of functional groups. Detailed data on the structure determination of the EnnB microsomal metabolites can be found elsewhere (Ivanova et al., 2011).
Incubation Conditions.
Substrate depletion assays monitoring the concentration-time course of EnnB were performed in RLM, DLM, and HLM under conditions of assumed first-order kinetics to determine the assay half-life (t1/2,assay). EnnB was incubated in a reaction mixture containing an NADPH-generating system (0.91 mM NADPH, 0.83 mM NADP+, 19.4 mM glucose 6-phosphate, 1 U/ml glucose-6-phosphate dehydrogenase, 9 mM magnesium chloride hexahydrate), incubation buffer (45 mM HEPES pH 7.4), and 2 mg microsomal protein in a total volume of 1 ml. After preincubation at 37°C for 3 min, the reaction was initiated by adding EnnB dissolved in acetonitrile to final concentrations ranging from 0.66 to 1.74 μM. The fraction of acetonitrile in the microsomal incubation system was never higher than 0.3%.
The reaction mixtures were incubated at 37°C in a shaking water bath (OLS 200; Grant, Cambridge, UK) in capped round-bottom glass tubes; 150-μl sample aliquots were drawn after 0, 2.5, 5, 10, 15, and 30 min; and enzyme activities were stopped immediately with 150 μl of ice-cold acetonitrile. Samples were kept on ice until centrifugation (Eppendorf AG, Hamburg, Germany) at 2000g for 5 min to precipitate the proteins. Supernatants were collected, transferred to HPLC sample vials, and stored at −20°C until LC-MS analysis. Reaction samples without the NADPH-regeneration system served as negative controls, and reaction samples without EnnB were used as vehicle controls for background subtraction. All incubations were performed at least three times and in duplicate.
Formation of EnnB Metabolites.
The formation of EnnB metabolites was observed in incubations with 0.66 μM EnnB in RLM, DLM, or HLM, and metabolite concentration/time-curves were generated. Formation rates were calculated by determining the metabolite produced per time assuming initial reaction velocity linearity. In addition, metabolite formation was observed in RLM with 0.88 and 1.74 μM EnnB initial concentrations.
Determination of Kinetic Parameters.
In substrate depletion experiments, EnnB peak heights were determined and normalized to the value obtained at t = 0. The percentage remaining versus time was fitted to a first-order decay function to determine the initial substrate depletion rate constant and the half-life of EnnB in the assay (t1/2,assay = ln2/k) for the different microsomal preparations. If the substrate decline showed nonlinearity in the log percentage remaining versus time curves at later incubation times, only the initial time points with apparent log-linearity were used to determine t1/2,assay.
The assay clearances (CLassay) were calculated from t1/2,assay and the assay volume (Vassay) according to CLassay = Vassay · k = Vassay · ln2/t1/2,assay. Assuming that protein binding of EnnB in the reaction mixture was negligible [fraction unbound in the assay (fu,assay ∼ 1)], the assay clearance was an approximation of the intrinsic assay clearance (CLint,assay), which is a measure of enzyme activity derived from the Michaelis-Menten equation parameters maximal velocity (Vmax,assay) and reaction constant (KM,assay) under the condition that the substrate concentration in the assay is well below the KM value (CLint,assay = Vmax,assay/KM,assay).
KM,assay was determined from depletion experiments with different initial EnnB concentrations by plotting the depletion rate constants to the substrate (S) concentrations. The inflection point of the curve in a linear-log plot represents the KM value, occurring when k is half of the theoretical maximum k at an infinitesimally low substrate concentration (k = k[S]→0 · (1 − [S]/([S] + KM)) (Obach and Reed-Hagen, 2002).
The respective CLint,assay for human, dog, and rat microsomes were upscaled to the assay-independent, intrinsic liver clearances (CLint) by considering the amounts of microsomal protein in the assays (Protassay) and the relative liver weights (RLW) and by using specific microsomal recovery indexes (MRI) (Barter et al., 2007; Smith et al., 2008) for milligrams of microsomal protein to grams of liver (CLint = CLint,assay · MRI · RLW/Protassay).
In vitro-in vivo extrapolation was attempted by applying the well stirred liver model (Obach et al., 1997; Ito and Houston, 2005). Systemic blood clearances (CLb) were calculated from the CLint by considering the hepatic blood flow (Q) of the different species (CLb = Q · CLint · fu,b/(Q + CLint · fu,b)). Because binding data for EnnB were not available, the fraction unbound (fu,b) EnnB in blood was set to 1 for no binding.
The bioavailability (f) after p.o. application was calculated for EnnB from the CLb (f = fa · (1 − CLb/Q) assuming complete absorption from the gastrointestinal tract (fa = 1), which allowed the estimation of the maximal bioavailability (fmax = f(fa → 1)).
By inclusion of data from an older study (Lohmann, 1988) determining the in vivo half-life in blood (t1/2,b) after intranasal application of tritiated EnnB to Wistar rats, the volume of distribution in blood at steady state was calculated (Vss,b = t1/2,b · CLb/ln2) to deliver an estimate on EnnB tissue distribution.
Finally, the CLb values for the three species were examined for linear correlation in the allometric scaling of clearance to body weights (BW) (logCLb ∼ logBW) (Obach et al., 1997). Average body weight values of 70 kg for humans, 11 kg for Beagle dogs, and 0.2 kg for Wistar rats were used in the calculations.
P450 Reaction Phenotyping by Selective Human Enzyme Inhibitors.
The effect of several specific chemical inhibitors of human P450 enzymes on the biotransformation of EnnB was investigated in HLM. Standard assay conditions and a single concentration of EnnB (0.66 μM), lower than the estimated KM value, were used. In this preliminary study, only one concentration close to the published optimum of effectivity and selectivity was used for each inhibitor (Pelkonen et al., 2008). CYP1A2 activity was inhibited with furafylline (Fur; 10 μM), CYP2C9 with sulfaphenazole (20 μM), CYP2C19 with N-3-benzylnirvanol (5 μM), CYP2D6 with quinidine (5 μM), CYP2E1 with diethyldithiocarbamate (50 μM), and CYP3A4 was inhibited with troleandomycin (TAO; 50 μM). In addition, Fur and TAO were used in a screening study in RLM and DLM.
All inhibitors were preincubated for 2 min in the presence of the NADPH-generating system and microsomes at 37°C before the addition of EnnB. Preincubation for 15 min was tested for the mechanism-based inhibitors TAO and Fur and showed no difference in comparison to the 2-min preincubation. All experiments were performed at least three times and in duplicate. A vehicle control for background subtraction, a negative control lacking the NADPH-regenerating system to exclude nonenzymatical EnnB degradation, and a toxin stability test in incubations without microsomes were additionally performed. The final solvent concentrations (dimethyl sulfoxide; CH3CN) in the assays were ≤0.3% (v/v).
The half-life of EnnB depletion was compared in the presence and absence of the chemical inhibitors, and the inhibition of EnnB depletion was estimated as a measure for the interaction of EnnB with the respective P450 enzymes. The magnitudes of inhibition expressed as ratios of inhibitor concentration at the active site of the P450 enzyme ([I]) and inhibition constant (Ki) were estimated by using the percentage increase (R) in area under the concentration-time curve (AUC) in the assays in the presence and absence of the chemical inhibitors (R = AUCinhib/AUC = 1 + [I]/Ki) or equivalently, by using the percentage increase in half-lives r = t1/2,inhib/t1/2 = 1 + [I]/Ki), considering that AUCassay = Dose/CLassay = [EnnBo] · Vassay · t1/2,assay/(ln2 · Vassay) resulting in AUC ∼ t1/2 (Eagling et al., 1998; Zhang et al., 2009).
Results
Analysis of EnnB and EnnB Metabolites.
The LC-MS method allowed the specific and semiquantitative determination of EnnB (Fig. 1) in sample aliquots of microsomal incubations. EnnB was determined with high precision and accuracy without interference of the microsomal matrix. The initial EnnB assay concentration (0.42 μg/μl) was in the middle of the LC-MS method's working range considering dilution in sample preparation.
EnnB metabolites were identified by liquid chromatography retention times and specific masses obtained by high-resolution tandem mass spectrometry experiments (Ivanova et al., 2011). In total, the structures of 12 main metabolites could be described (Table 1). They could be divided into three sets on the basis of the functional groups that had been introduced into the EnnB molecule by different biotransformation reactions: mono-oxygenated metabolites (M1–M5), mono- and didemethylated metabolites (M6 and M7, respectively), and dioxygenated metabolites (M8–M12). Considering the retention times observed by the liquid chromatography gradient, M8 to M12 were the most hydrophilic metabolites, followed by M1 to M5, and M6 and M7.
The metabolites M8 to M12 are products of sequential oxidations as indicated by their mass differences compared with EnnB. This was confirmed in an experiment by isolating M1 and M2 from RLM incubation aliquots and continued metabolization in DLM leading to M9 to M12 (data not shown).
Metabolite Formation in Rat, Dog, and Human Microsomes.
The concentration-time curves of EnnB metabolite formation in RLM, DLM, and HLM showed different profiles (Fig. 2). The occurrence and formation rates of the individual metabolites differed considerably in the three microsomal preparations. M8 was only observed in trace amounts and was not included in the evaluation.
In RLM, only mono-oxygenated M1 to M5 and the two demethylated species M6 and M7 were measured after 30-min incubation. The initial metabolite formation velocities (v) were highest for M6 [0.0026 μg/(ml · min)] and M2 [0.0009 μg/(ml · min)] and lowest for M4 and M7 [both 0.0002 μg/(ml · min)]. M6 was the main metabolite after 30 min.
In DLM, the mono-oxygenated M1, M2, and M3 were rapidly formed [v = 0.0176 μg/(ml · min), 0.0156 μg/(ml · min), and 0.0024 μg/(ml · min), respectively], and M1 showed the highest concentration after 5 min of incubation. However, concentrations declined rapidly after reaching the maximum, probably because of further oxidation to the dioxygenated metabolites. The formation of M9, M10, M11, and M12 [v = 0.024, 0.0004, 0.0053, and 0.0007 μg/(ml · min), respectively] started with a time lag of approximately 2.5 to 10 min after incubation start. M11 was the main biotransformation product of EnnB in DLM after 30 min. The monodemethylated M6 [v = 0.0056 μg/(ml · min)] reached maximum concentration at 5 min; however, the didemethylated M7 was not observed.
In HLM, the concentrations of the mono-oxygenated metabolites M1, M2, M3, and M5 increased rapidly within the first 10 min of incubation [v = 0.0026, 0.0097, 0.0034, and 0.0030 μg/(ml · min), respectively]. M2 was the major metabolite after 10 min, twice as high in concentration as the secondly positioned M3. The dioxygenated metabolites M9, M11, and M12 became observable after a time lag of approximately 5 to 10 min. M11 was formed most rapidly [v = 0.0020 μg/(ml · min)] and was the predominant metabolite after 30-min incubation. The demethylated M6 [v = 0.0033 μg/(ml · min)] reached a concentration maximum after 15-min incubation and then decreased slowly. In contrast, M7 was not present in measurable amounts.
RLM assays were performed using three different EnnB initial concentrations (c), and formation velocities of the major metabolites were estimated. However, the graph v = f(c) did not seem to follow Michaelis-Menten kinetics for the respective metabolites; therefore, formation constants (KM) could not be determined (data not shown).
Kinetics of Enniatin B Depletion.
EnnB was rapidly metabolized in RLM, DLM, and HLM. EnnB assay half-lives were calculated from the initial substrate depletion rate constants of first-order decay functions in the three incubation systems (Fig. 3). In addition, a direct curve fit of the nonlogarithmic graphs was performed, confirming the data obtained. Under the chosen assay conditions, EnnB half-lives in RLM, DLM, and HLM were 32, 4.5, and 12.4 min, respectively.
These data could only be used for the calculation of derived pharmacokinetic parameters if the depletion assays were performed in compliance with Michaels-Menten kinetics requiring that the EnnB concentration was clearly lower than the KM value. In view of the difficulties of obtaining KM from metabolite formation, KM was therefore determined directly from depletion experiments in RLM with three different EnnB initial concentrations (Fig. 4). The KM was found as 1.1 μM, confirming that the initial EnnB concentration of 0.66 μM, generally used in the metabolism and inhibition assays, was well below the KM.
Prediction of in Vivo Pharmacokinetic Parameters.
On the basis of the respective t1/2,assay, the EnnB assay clearances were calculated, and applying published consensus values for in vitro-in vivo extrapolation, intrinsic clearances (CLint) and CLb were estimated using the nonrestrictive well stirred liver model (Table 2). The CLb were 1.57 l/(h · kg) in rats, 1.67 l/(h · kg) in dogs, and 0.63 l/(h · kg) in humans, leading to predicted maximal bioavailabilities of 63, 20, and 55% under the assumption that EnnB is completely absorbed from the gastrointestinal tract after oral application, which is truly a considerable overestimation.
According to an older study applying fusafungin-carried tritiated EnnB in a total dosage of 10 mg/kg in isotonic saline to Wistar rats (Lohmann, 1988), the ratio between renally and fecally excreted radioactivities was approximately 1:12. Roughly, the fraction absorbed could be estimated as fa = 0.08, under the condition that the administered compound was solved in the saline formulation. Applying fa for the prediction of EnnB bioavailability in rats using the data from the microsomal assay resulted in f = fmax · fa = 5%, which was the same value as found in the cited study for the amount of renally excreted total radioactivity.
Protein binding in the microsomal assays (fu,assay) and in blood (fu,b) were ignored in the approach presented here to predict in vivo pharmacokinetic parameters of EnnB. In the equation for the well stirred liver model, drug binding to microsomes and blood would cancel out if the unbound fraction is the same in both matrices (Ito and Houston, 2005).
An estimate for the volume of distribution in blood at steady state (Vss,b) in rats was calculated from the blood clearance [CLb = 1.57 l/(h · kg)] that was predicted in the study presented here, and a value for the maximal half-life in blood (t1/2,b = 0.5 h) derived from the measurement of total radioactivity decline in blood after intranasal application of tritiated EnnB (Lohmann, 1988). Comparison of the value determined (Vss,b = 1.13 l/kg) with the blood volume in rats (Vb = 0.06 l/kg) indicates that EnnB is distributed into body tissues.
The predicted CLb values of rat, dog, and humans were combined to perform interspecies scaling (Fig. 5) with the intention to check the probability of the determined values. Linear regression of the log/log graph of CLb = f(BW) rendered a regression coefficient near unity (r2 = 0.98), confirming that the predicted CLb values for the three species are correlated in accordance with the principles of allometric scaling.
Inhibition of Enniatin B Depletion by Selective Inhibitors of Human P450 Enzymes.
The in vitro determination of biotransformation pathways by P450 reaction phenotyping is a central element in the assessment of a compound's inhibition and induction potential. In guidance documents of drug administration authorities, CYP1A2, CYP2C9, CYP2C19, CYP2D6, CYP2E1, and CYP3A4 are named as the critical enzymes for humans (Tucker et al., 2001). In the study presented here, the inhibitory potential of selective chemical inhibitors on the EnnB depletion in HLM was determined. Half-lives in the presence and absence of the inhibitors were determined and used to calculate ratios of inhibitor concentration to inhibition constant ([I]/Ki values) for each inhibitor/EnnB combination.
The P450 enzyme phenotyping in HLM showed that the most effective inhibition of EnnB biotransformation was achieved by blocking CYP3A4, CYP2C19, and CYP2A1 (Fig. 6). Comparing EnnB concentrations after 30-min incubation with and without selective inhibitors, 62, 51, and 54%, respectively, inhibition were observed. In contrast, the inhibition of CYP2E1, CYP2C9, and CYP2D6 had less effect on EnnB metabolism, showing 35, 9, and 3% inhibition effectivity, respectively. The calculated [I]/Ki values reflected the probabilities of EnnB being a substrate of the individual human P450 enzymes (Table 3). The [I]/Ki value for EnnB's interaction with CYP3A was approximately 8 times higher than for CYP1A2, CYP2C19, and CYP2E1, which could still considerably contribute to EnnB metabolism. In contrast, the involvement of CYP2C9 and CYP2D6 was less probable.
In addition, the mechanism-based inhibitors Fur and TAO were used in a screening study in RLM and DLM, showing that the inhibition potencies for EnnB biotransformation were comparable in HLM and DLM and lower in RLM. In both animal species, enzymes belonging to the CYP1A and CYP3A families could be involved in EnnB metabolism (Fig. 6). The provisional [I]/Ki values, which were generated in an attempt to describe the observations (Table 3), supported the findings.
Discussion
Enniatins have been called emerging mycotoxins (Jestoi, 2008; Tedjiotsop Feudijo et al., 2010), showing cellular toxicity at low micromolar concentrations (Ivanova et al., 2006; Firáková et al., 2007). The extensive low-level exposure of humans and animals to enniatins in food and feed has led to increased awareness of the responsible food authorities. In 2010, the European Food Safety Agency issued a call for enniatin occurrence data (http://www.efsa.europa.eu/en/datexdata/docs/InstructionMycoPhytotoxins.pdf), which aimed to prepare a scientific opinion on the risk to human and animal health.
Although reports on acute mycotoxicosis for enniatins are lacking, there is concern with regard to subacute toxicity. In vivo toxicity data are sparse, and the few data available reveal a missing correlation with in vitro toxicity studies. A Ph.D. thesis dealing with “analytical, pharmacological and microbiological investigations of the ionophoric antibiotic fusafungin” (Lohmann, 1988) included in vivo experiments applying tritiated EnnB p.o. and intranasally to rats and revealed the compound's relative low bioavailability and short half-life. Furthermore, it was shown by in vitro incubations with rat liver cytosol that EnnB was not a substrate for phase II metabolism.
The study presented here used an in vitro approach with liver microsomes from different species for the identification of EnnB's biotransformation products by phase I metabolism and the prediction of pharmacokinetic parameters. In vitro metabolic systems have successfully been used to study drug metabolism, kinetics, and P450 enzyme inhibition, and models and procedures for the prediction of in vivo data from in vitro metabolism data have been developed (Iwatsubo et al., 1997; Obach et al., 1997; Naritomi et al., 2001; Ito and Houston, 2005; Riley et al., 2005; De Buck et al., 2007).
EnnB was metabolized by rat, dog, and human microsomes, and produced metabolites were analyzed by LC-MS. The method used allowed for the quantitative determination of EnnB concentrations in incubation aliquots and the identification and semiquantitation of the EnnB metabolites (Ivanova et al., 2011). The 12 characterized metabolites were products of oxidative reactions (Guengerich, 2001), including carbon hydroxylation (M1, M2, M3, and M4), N-methyl-oxydation (M5), dealkylation (M6, M7), carbon hydroxylation and aldehydation (M8), and carbon hydroxylation and carboxylation (M9, M10, M11, and M12). Consequently, the EnnB metabolites were more hydrophilic than their parent compound, which was confirmed by the respective retention times in reversed-phase chromatography.
The characterization of metabolites is an important step in the safety and risk evaluation of a compound. Reactive metabolites can affect the overall toxic profile and have to be assessed with regard to exposure, half-life, matrix of occurrence, and toxicity mechanism (Smith and Obach, 2009). EnnB metabolite formation data obtained in the study presented here showed slower rates and a different profile for rats compared to dogs and humans. Demethylation was favored over hydroxylation, and products of multiple oxidation reactions were not observed, which dominated in dogs and humans. This implied that different P450 enzymes are involved in the EnnB biotransformation in the three species. Furthermore, the occurrence of isobaric metabolites suggested multiple reaction points for the metabolizing enzymes in the EnnB molecule. KM for metabolite formation could not be determined on the basis of three EnnB concentrations, either because metabolite concentrations were measured incorrectly because of the lack of metabolite reference materials or because concentrations were too high compared with the respective KM. Therefore, metabolite kinetics should be examined further in a follow-up study.
EnnB depletion kinetics revealed dissimilar assay half-lives in RLM, DLM, and HLM, mirroring the differences in metabolite formation. The depletion KM (Obach and Reed-Hagen, 2002) was determined for rats as the slowest EnnB –metabolizer, confirming that the initial EnnB concentration of 0.66 μM used in all subsequent assays was sufficiently low to allow reactions to run under first-order kinetics. The advantage of directly determining KM from EnnB half lives in preference to using metabolite formation rates is that authentic metabolite standards are unnecessary. The approach has been proven to comply with drug screening purposes.
The in vitro-in vivo extrapolation calculations were performed using published consensus factors for MRI, RLW, and Q (Iwatsubo et al., 1997; Barter et al., 2007; Smith et al., 2008) in the nonrestrictive well stirred liver model (Naritomi et al., 2001; Riley et al., 2005).The results indicated that EnnB is a intermediate- to high-clearance drug. Although EnnB is considerably lipophilic and tends to bioaccumulate (Jestoi et al., 2009), binding to microsomes (fu,assay) and binding to plasma proteins (fu,b) was disregarded in the predictions, assuming that they were similar and the unbound fraction terms would therefore cancel out from the equations (Obach et al., 1997; Ito and Houston, 2005). This strategy has been shown to work most successfully in several extensive drug metabolism studies yielding the best agreement (85%) between observed and predicted CLb values (De Buck et al., 2007; Wan et al., 2010).
Probabilistic proof of the obtained results by means of allometric scaling was reassuring because it confirmed species interrelation with regard to clearance and body weight in accordance to the empirically found principle (Obach et al., 1997; Ito and Houston, 2005). Furthermore, when data obtained from a metabolism experiment with primary rat hepatocytes were included in the calculation base, the regression coefficient dropped, with only 0.01 (data not shown), confirming previous results (Lohman, 1988) that conjugative phase II metabolism is most likely irrelevant for EnnB.
Data from a previous p.o. rat study with application of tritiated EnnB allowed for the approximation of additional pharmacokinetic parameters such as fa and t1/2 (Lohmann, 1988). However, the 3H label had been incorporated into the methyl group of the EnnB molecule's N-Me-Val moiety, which is subject to heteroatom oxygenation (M5) and dealkylation (M6, M7) reactions, potentially leading to the loss of the label. Therefore, the total amount of EnnB and EnnB metabolites may have been underestimated by determining total radioactivity. Therefore, the estimates for fa and t1/2 might be too low.
Nevertheless, some valuable information could be drawn by combining the in vivo data with the rat CLb value obtained by the study presented here. The bioavailability of EnnB was predicted to be low, and the distribution into body tissues was predicted to be of considerable size, providing an explanation for the absence of acute toxicity and suspected subacute effectivity. Bioaccumulation in liver, kidney, and brain had also been found in the study with [3H]EnnB.
A preliminary risk assessment was attempted by calculating EnnB exposure on the basis of the available in vivo rat data and the predicted rat and human clearances. The maximal concentration in blood (Cmax,b) was reached in rats after 30 min (tmax). Assuming that the ratio of volume of distribution to blood volume is comparable in humans and rats (Vb,human = 0.07 l/kg; Vss,b human = 1.35 l/kg) and using maximal values of food contamination ([EnnB]) and large-scale consumption (A) to calculate a dose (dose = [EnnB] · A), a theoretical Cmax,b was determined (Cmax,b = fmax,human · fa,rat · dose · e−ln2 · tmax/t1/2/ · Vss,b) for adults (70 kg) and compared with IC50 in in vitro cytotoxicity assays. For example, the consumption of 500 g of Swiss oat bread (Noser et al., 2007) would lead to a Cmax,b of 0.03 μM, delivering a safety factor of approximately 100. However, by eating 300 g of Tunisian breakfast cereals (Queslati et al., 2011) a Cmax,b of 6.3 μM would be reached, which is a level of concern.
The P450 enzyme phenotyping experiments in this study showed that human CYP3A4, CYP2C19, and CYP1A2 likely played major roles in EnnB biotransformation. EnnB's susceptibility to CYP3A4 metabolism and transport by P-glycoprotein (Ivanova et al., 2010) are possibly responsible for the compound's low absorption rate (fa), considering the substantial CYP3A4-activity in the gastrointestinal tract.
The catalytic selectivity of the P450 subfamily enzymes from different species can vary considerably (Bogaards et al., 2000), but selective inhibitors of human P450 enzymes have also been used for reaction phenotyping in other species (Eagling et al., 1998; Martignoni et al., 2006). The mechanism-based inhibitors Fur and TAO have been successfully used for cross-species evaluations of CYP1A- and CYP3A-catalyzed biotransformations, respectively (Chauret et al., 1997; Quintieri et al., 2008; Aueviriyavit et al., 2010). Qualitatively similar results were obtained in human and animal microsomes, although the enzyme reaction rates were different (Quintieri et al., 2008), which was attributed to differences in the formation of the metabolic intermediate complexes (Aueviriyavit et al., 2010). By comparing the percentage increase in t1/2 with and without inhibitors, we have attempted to show the relative magnitudes of inhibition potentials. The derived [I]/Ki values describe a compound's inhibitory potential on P450 model substrates (Zhang et al., 2009); however, in the study presented here, the approach was modified in estimating to which extent EnnB biotransformation could be expected to take course via the different pathways (Pelkonen et al., 2008).
The EnnB-depletion inhibition is equivalent to the decrease in intrinsic enzyme clearances, independently from the mechanism of inhibition. The inhibition of EnnB metabolite formation, the determination of Ki values, and correlation of metabolite structures with the respective P450 enzyme reactions could be characterized in a subsequent study. Furthermore, the inhibitor potential of EnnB could be assessed using isoforms-specific substrate probes.
Authorship Contributions
Participated in research design: Ivanova and Fæste.
Conducted experiments: Ivanova.
Contributed new reagents or analytic tools: Ivanova and Uhlig.
Performed data analysis: Fæste, Ivanova, and Uhlig.
Wrote or contributed to the writing of the manuscript: Fæste, Ivanova, and Uhlig.
Footnotes
This work was supported in part by the Norwegian Research Council [Grant 200349/I10] within the project “Mycotoxin 2—biomarkers for adverse endocrine effects of fungal extrolites and mycotoxins—characterisation of molecular mechanisms using cell assays, chemical analyses and proteomics.”
Article, publication date, and citation information can be found at http://dmd.aspetjournals.org.
doi:10.1124/dmd.111.039529.
-
ABBREVIATIONS:
- EnnB
- enniatin B
- M1
- metabolite 1
- N-Me-Val
- N-methyl-l-valines
- HPLC
- high-performance liquid chromatography
- HLM
- human liver microsomes
- P450
- cytochrome P450
- DLM
- dog liver microsomes
- RLM
- rat liver microsomes
- LC-MS
- high-performance liquid chromatography-coupled mass spectrometry
- RLW
- relative liver weights
- MRI
- microsomal recovery indexes
- Fur
- furafylline
- TAO
- troleandomycin
- AUC
- area under the concentration-time curve.
- Received March 15, 2011.
- Accepted May 26, 2011.
- Copyright © 2011 by The American Society for Pharmacology and Experimental Therapeutics