Introduction

Alternariol (AOH) and alternariol-9-methyl ether (AME; chemical structures in Fig. 1) are major toxins produced by fungi of the genus Alternaria and common contaminants of food items such as cereals, fruits, and fruit juices (Chelkowski and Visconti 1992; Scott 2001; Scott et al. 2006). Alternaria toxins have been associated with an increased incidence of esophageal cancer in certain areas of China (Liu et al. 1992). Recent studies have shown that AOH and AME are able to induce DNA strand breaks and gene mutations in cultured human and animal cells (Lehmann et al. 2006; Brugger et al. 2006; Pfeiffer et al. 2007a), and to inhibit topoisomerase I and IIα under cell-free conditions (Fehr et al. 2009). Despite the mounting concern about the health hazards posed by Alternaria toxins, data on the toxicokinetics and metabolism of these important mycotoxins are scarce. We have recently reported that AOH and AME are prone to cytochrome P450-mediated hydroxylation at various positions, leading to several catechol metabolites in vitro (Pfeiffer et al. 2007b). As AOH has three and AME two phenolic hydroxyl groups, both mycotoxins must be expected to form conjugates with glucuronic acid and sulfate. Our laboratory has recently clarified the chemical structures of the monoglucuronides of AOH and AME generated by hepatic and intestinal microsomes from rats, pigs, and humans (Pfeiffer et al. 2009a). Moreover, the activities of ten major isoforms of human uridine diphosphate glucuronosyltransferases (UGTs) for the glucuronidation of AOH and AME were determined (Pfeiffer et al. 2009a).

Fig. 1
figure 1

Chemical structures of AOH and AME

Full size image

In the present study, the metabolism of AOH and AME has been investigated in differentiated Caco-2 cells in vitro. As these cells have many features in common with human intestinal epithelial cells, the Caco-2 cell system is widely accepted as a suitable in vitro model to study intestinal absorption and metabolism of xenobiotic compounds (Press and Di Grandi 2008; van Breemen and Li 2005). Although Caco-2 cells are derived from a human colon carcinoma, they differentiate within 3 weeks when grown on a permeable membrane under suitable culture conditions to form a monolayer with tight junctions. The surface attached to the membrane represents the basolateral side, which corresponds to the blood side of the intestinal epithelium, whereas the opposing surface develops microvilli and corresponds to the apical or intestinal lumen side. In the Millicell® system, the Caco-2 cells grow in inserts of cell culture wells and their apical and basolateral sides are facing different compartments. Uptake of xenobiotics into the cells can thus be studied from either compartment, as can the release of the compounds and their metabolites into both compartments. Caco-2 cells are devoid of cytochrome P450 activity but have active UGTs and sulfotransferases (Biasutto et al. 2009). Therefore, in addition to the in vitro absorption of AOH and AME, the formation of their glucuronides and sulfates as well as the basolateral and apical excretion of the conjugates was determined in the present study.

Materials and methods

Chemicals

AOH was chemically synthesized in the laboratory of J. Podlech, University of Karlsruhe, as previously reported (Koch et al. 2005) and contained 1.1% AME as determined by HPLC. AME was purchased from Sigma/Aldrich/Fluka (Taufkirchen, Germany) and had a purity of >96% according to HPLC analysis, containing 2.2% AOH. 3′-Phosphoadenosine-5′-phosphosulfate (PAPS), β-glucuronidase type B-1 from bovine liver, sulfatase from Aerobacter aerogenes, tetrabutylammonium dihydrogenphosphate (TBAP), and all other chemicals and reagents were also obtained from Sigma/Aldrich/Fluka.

Formation of sulfates in rat liver cytosol

In order to obtain standards for the identification of the sulfate conjugates formed in Caco-2 cells, AOH or AME were incubated with rat liver cytosol prepared as the 100,000 g supernatant of a liver homogenate from untreated male Wistar rats. Incubation mixtures consisted of a total volume of 200 μl 0.1 mol/l potassium phosphate buffer containing 50 μmol/l mycotoxin, 10 mmol/l MgCl2, 0.4 mmol/l PAPS, and 200 μg cytosolic protein. AOH and AME were first dissolved in DMSO and added to the buffer at an amount not to exceed 0.5% DMSO in the final incubation. A mixture containing all constituents except PAPS was preincubated for 5 min at 37°C, PAPS dissolved in phosphate buffer was added, and incubation continued for another 30 min with gentle shaking. Subsequently, the mixture was centrifuged (5 min at 2,000g) and the supernatant directly analyzed by HPLC-UV or LC-DAD-MS.

Caco-2 cell culture and subculture

Caco-2 cells (DSMZ No. ACC169) were from Deutsche Sammlung von Mikroorganismen und Zellkulturen (Braunschweig, Germany). Cells were cultured at 37°C and 5% CO2 in dishes with 15 cm diameter containing 20 ml medium and initially 0.4 × 106 cells. Culture medium consisted of Dulbecco’s Modified Eagle Medium (DMEM/F12) with 10% fetal calf serum, 100 U/ml penicillin, and 100 μg/ml streptomycin, and was replaced every other day. After 7 days, medium was removed and the attached cells washed twice with 10 ml phosphate-buffered saline (PBS) each, followed by treatment with 2.5 ml of a 0.625% aqueous trypsin solution for 40 s. After removal of the trypsin, cells were incubated at 37°C for 5 min and subsequently suspended in 10 ml medium. Cell number was determined in a 1:10 dilution using a Neubauer chamber, and the volume containing 0.4 × 106 cells was added to a new dish.

AOH and AME metabolites in Caco-2 cells

For studying the metabolism of AOH and AME in differentiated Caco-2 cells, 106 cells per well were seeded into 6-well dishes and grown for 21 days with renewal of the medium after 2–3 days. On day 21, medium was removed, cells were washed twice with Hank’s buffered salt solution (HBSS), and HBSS containing the mycotoxin at 20 μmol/l concentration with 1% DMSO was added. After 1, 2, and 3 h incubation, the total fluid of each well was removed for analysis by HPLC-UV and LC-DAD-MS. The attached cells of each well were washed twice with 2 ml PBS, scraped off the support using a plastic spatula, suspended in 1 ml HBSS and transferred into a 2-ml centrifuge vial. The pellet obtained after centrifugation at 200g for 2 min was resuspended in 0.5 ml HBSS and lysed at −80°C for 24 h. After ultrasonic treatment, 20 μl of a 5 mmol/l solution of 4,4′-isopropylidenbis(2,6-dimethylphenol) in DMSO was added to facilitate the quantification of the mycotoxin metabolites as described recently (Pfeiffer et al. 2009b). For the analysis of unconjugated mycotoxins, 100 μl of the lysate was diluted with the same volume of 0.1 mol/l potassium phosphate buffer pH 7.4 and extracted three times with 0.5 ml each of ethyl acetate. For the analysis of conjugates, other aliquots of the lysate were diluted with 0.1 mol/l potassium phosphate buffer pH 7.1 and incubated for 2 h at 37°C with 250 U ß-glucuronidase from bovine liver or 0.2 U sulfatase from Aerobacter aerogenes, followed by extraction with ethyl acetate. All extracts were evaporated to dryness under a stream of nitrogen, and the residues dissolved in 100 μl methanol and analyzed by HPLC-UV and LC-DAD-MS without further purification.

Caco-2 Millicell® system and in vitro absorption of AOH and AME

Each insert (apical compartment) of the 24-well Millicell® plates (Millipore, Billerica, MA, USA) was filled with 0.4 ml medium containing 6 × 104 cells, and each well (basolateral compartment) received 0.8 ml medium alone. Cells were grown into a differentiated monolayer for 21 days, with renewal of the medium in both compartments every other day. Then the medium was removed and the apical and basolateral compartments were washed twice with HBSS. Subsequently, HBSS containing a defined concentration of the mycotoxin was filled into the apical compartment, whereas the basolateral compartment contained only HBSS. For the time-course study with 10, 20, 30 and 40 μmol/l AOH or AME, 200 μl of the basolateral HBSS were removed four times in 30-min intervals and replenished with fresh HBSS. For the calculation of the Papp value, each well contained 20 μmol/l toxin initially and both compartments were analyzed after the respective time. In order to ensure an intact monolayer of the Caco-2 cells, both compartments were washed once with HBSS after each experiment, and 0.4 ml of HBSS containing 100 μg lucifer yellow per ml were filled into the apical well whereas the basolateral compartment contained dye-free HBSS. After 1 h at 37°C, 200 μl of the basolateral HBSS were analyzed for lucifer yellow fluorimetrically (excitation at 485 nm and emission at 535 nm). Only cell layers with less than 1% transfer of the dye from the apical to the basolateral side were considered intact (Grès et al. 1998).

HPLC-UV analysis

A Beckman system equipped with a binary pump, a Beckman 166 UV detector, and 32 Karat 7.0 software for data collection and analysis was used. Separation was carried out on a 250 × 4.6 mm i.d., 5 μm, reversed-phase Luna C8 column (Phenomenex, Torrance, CA, USA) protected by a 3 × 4.0 mm i.d. SecurityGuard C18 (ODS) column (Phenomenex). Solvent A was deionized water containing 5 mmol/l TBAP, and solvent B was acetonitrile with 0.1% formic acid. For the separation of AOH and its sulfates, a linear solvent gradient was used, starting isocratically at 5% B for 2 min, then changing to 50% B in 20 min and subsequently to 100% B in 7 min. After 4 min of eluting the column with 100% B, the initial 17% B were reached in 2 min. The flow rate was 1 ml/min. The gradient used for the separation of AME and its sulfates started isocratically at 30% B for 1 min, then changed to 100% B in 25 min, stayed at 100% B for 5 min, and returned to 30% B within 2 min. The flow rate for this gradient was 0.5 ml/min. The UV detector was set to 254 nm for both toxins.

LC-DAD-MS analysis

A LXQ Linear Ion Trap MSn system (Thermo Fisher Scientific, Waltham, MA, USA) equipped with a Finnigan Surveyor Autosampler Plus and a Finnigan Surveyer PDA Plus Detector was used. This allowed online detection of UV absorption and mass spectrometry. The same column was used as described for HPLC-UV analysis. Solvent A was deionized water containing 5 mmol/l ammonium acetate, and solvent B was acetonitrile with 0.1% formic acid. Elution started with a linear increase from 30% B to 100% B within 25 min. After 3 min at 100% B, the initial 30% B were reached in 1 min and kept for 3 min before the next sample was injected. The flow rate was 0.5 ml/min, the detection wavelength 254 nm for the mycotoxins and 288 nm for 4,4′-isopropylidenbis(2,6-dimethylphenol). UV absorbance at these wavelengths was used for quantification, assuming the same molar coefficient for the free mycotoxins and their conjugated metabolites. The mass spectrometer was operated in the negative electrospray ionization (ESI) mode. Nitrogen was used as sheath gas, auxiliary gas, and sweep gas with flow rates of 30.0, 15.0, and -0.02 l/min, respectively. Spray voltage was 4.47 kV, spray current 3.15 μA, capillary voltage -45.10 V, capillary temperature 350°C, and tube lens voltage 125.58 V. Full scan mass spectra were recorded from m/z 100–600 in order to confirm the identity of the compounds through their M-H ions.

Calculation of Papp

Papp values were calculated according to Artursson and Karlsson (1991) using the formula

$$ {\text{P}}_{\text{app}} \left[ {{\text{cm s}}^{- 1} } \right] = \left( {{\text{V}}_{\text{api}} /{\text{A}} \cdot {\text{t}}} \right)\left( {{\text{C}}_{\text{baso}} /{\text{C}}_{\text{api}} } \right) $$

where Vapi is the volume of apical compartment (0.4 ml), A the surface area of the monolayer (0.7 cm2), t the time (s), Cbaso the concentration (µmol/l) of mycotoxin in the basolateral compartment (either parent compound or sum of parent compound and metabolites), and Capi the initial concentration (µmol/l) of mycotoxin in the apical compartment.

Results and discussion

Metabolism of AOH and AME in Caco-2 cells

Synthesis of reference sulfates

In order to identify the conjugated metabolites of the two mycotoxins in Caco-2 cells, it was desirable to have the authentic standards at hand. AOH-3-O-glucuronide, AOH-9-O-glucuronide, AME-3-O-glucuronide, and AME-7-O-glucuronide were available as reference compounds from our previous glucuronidation study (Pfeiffer et al. 2009a). Various sulfates of AOH and AME were obtained by incubation of the aglycones with rat liver cytosol in the presence of PAPS, followed by analysis of the incubation mixtures using HPLC with UV detection. In order to improve the chromatographic separation of the sulfates, TBAP was added to the eluent for the formation of ion pairs. As depicted in Fig. 2, three peaks not present in control incubations without PAPS were observed in the HPLC profile of the complete incubation of AOH, whereas AME gave rise to two new peaks. Treatment of the incubation mixtures with sulfatase from Aerobacter aerogenes caused a pronounced decrease of the three AOH sulfate and two AME sulfate peaks with a corresponding increase of the aglycone peaks, confirming the sulfate nature of the conjugates (data not shown).

Fig. 2
figure 2

HPLC-UV analysis (using ion pair chromatography) of the sulfates of AOH (top) and AME (bottom) formed upon incubation of the aglycones with rat liver cytosol and PAPS. The assignment of the position of the sulfate group is tentative and based on the retention time as explained in the text. S Sulfate

Full size image

Although the exact positions of the sulfate group in the three AOH sulfates and two AME sulfates has not yet been unequivocally elucidated, a tentative assignment is proposed based on analogy with the respective glucuronides, the structures of which were derived through chemical modifications (Pfeiffer et al. 2009a). Thus, the very polar AOH sulfate eluting after 12.5 min and AME sulfate at 15.9 min are believed to arise from conjugation of the hydroxyl group at position 7, which obliterates the strong hydrogen bonding with the carbonyl group at position 6 present in compounds with a free hydroxyl group at C-7. AME-7-O-glucuronide is also much more polar than AME-3-O-glucuronide (Pfeiffer et al. 2009a). Conversely, the non-polar AOH sulfate at 29.8 min and the AME sulfate at 31.0 min, eluting later than the aglycones from the reversed-phase column, are proposed to represent the 3-O-sulfates, because only this hydroxyl group is available for conjugation in both AOH and AME. By exclusion, AOH sulfate eluting at 22.7 min must then be the 9-O-sulfate.

Identification of AOH and AME conjugates formed in Caco-2 cells

Differentiated Caco-2 cells grown in a normal cell culture dish were incubated with AOH or AME for 2 h and the cell culture medium analyzed by HPLC-UV and LC-DAD-MS. HPLC-UV analysis under the same conditions as used for the separation of the sulfates obtained with rat liver cytosol (Fig. 2), i.e., with an eluent containing TBAP, showed that only the 3-O-sulfates of AOH and AME were formed in Caco-2 cells. Analysis by LC-DAD-MS (Fig. 3) confirmed that one of the peaks in the cell culture medium was a monosulfate of AOH and AME with the correct M-H (337 and 351 for AOH-sulfate and AME-sulfate, respectively) and M-sulfate ions (257 and 271, respectively). The shift in retention times of the 3-O-sulfates of AOH and AME in LC-DAD-MS as compared to HPLC-UV is due to the omission of TBAP from the LC-MS eluent, which was necessary because of the high ion background caused by this ion pair reagent. The other major metabolites present in the culture media of AOH and AME (Fig. 3) were identified as glucuronides by comparison with authentic reference compounds in HPLC-UV and LC-DAD-MS analysis. From AOH, about equal amounts of the 3-O-glucuronide and 9-O-glucuronide, but no 7-O-glucuronide, were formed, whereas AME yielded much 3-O-glucuronide together with a small amount of the 7-O-glucuronide (Fig. 3). Treatment with ß-glucuronidase or sulfatase caused the disappearance of the glucuronides or sulfates, respectively, and only the aglycones were observed after combined treatment of the medium with both ß-glucuronidase and sulfatase (data not shown).

Fig. 3
figure 3

LC-DAD-MS analysis (using reverse-phase chromatography) of the culture media after incubating Caco-2 cells with AOH (top) and AME (bottom) for 2 h. G Glucuronide, S sulfate

Full size image

When the culture media of the incubations of AOH and AME with Caco-2 cells were analyzed by HPLC-UV after 1 and 3 h, the same conjugates were observed as after 2 h, and a time-dependent decrease of the aglycones and increase of the conjugates in the medium was noted, which was somewhat faster for AOH than for AME (Table 1). The apparently faster conjugation of AOH was confirmed when the lysates of the Caco-2 cells were analyzed for the presence of parent aglycone and the conjugate metabolites (Table 1). The unconjugated parent compounds represented the major portion of the intracellular material after exposure to AOH or AME, and the amounts of glucuronides and sulfates in the cells were very small; the aglycone concentrations were highest after 1 and lowest after 3 h of incubation, and, in addition to a faster conjugation, AOH appeared to be faster absorbed than AME (Table 1).

Table 1 Amounts of glucuronide and sulfate conjugates in the cell culture medium and cell lysate of Caco-2 cells after incubation with AOH and AME for 1, 2, and 3 h. Data represent nmol compound and are the mean ± standard deviation of 4 culture media or mean ± half range of 2 cell lysates obtained from independent experiments
Full size table

Absorption and metabolism of AOH and AME in the Caco-2 Millicell® system

Throughout this study, 24-well Millicell® plates were used which contained inserts with monolayers of differentiated Caco-2 cells on a permeable membrane. The apical chamber of each well had a volume of 0.4 ml and the basolateral chamber of 0.8 ml. Incubations with AOH and AME were carried out using HBSS. Following each incubation, the monolayers were tested by apical application of the fluorescent dye lucifer yellow, which is not taken up by the cells and therefore does not reach the basolateral compartment of wells with intact monolayers. Only data obtained with intact cell layers were used.

Figure 4 depicts the time-dependent concentrations of the aglycones and of the glucuronide and sulfate metabolites of AOH and AME in the basolateral compartment observed after apical exposure to four different concentrations of the mycotoxins. With AOH at concentrations of 20 μmol/l and higher, about equal amounts of the 3- and 9-O-glucuronide appeared in the basolateral chamber, whereas the amount of AOH-3-O-sulfate was much lower and its increase delayed. Only with 10 μmol/l AOH was more 3-O-glucuronide than 9-O-glucuronide formed. The basolateral concentration of unconjugated AOH plateaued at 1.5–2 h of incubation and correlated with its apical concentration. In contrast, unconjugated AME did not reach the basolateral compartment even at high apical concentration, and small amounts of AME-3-O-glucuronide together with traces of AME-7-O-glucuronide were the only basolateral metabolites detected.

Fig. 4
figure 4

Time course of the basolateral concentrations of aglycones and conjugated metabolites after apical exposure to various concentrations of AOH and AME. Data represent mean and standard deviation of three independent experiments

Full size image

In addition to the kinetic data shown in Fig. 4, the unconjugated material and the various conjugated metabolites of AOH and AME were measured both in the basolateral and in the apical chamber after 3 h (Fig. 5). Independent of the concentration, 22.7–25.8% of the apically applied AOH reached the basolateral compartment and consisted of both unconjugated AOH and three conjugates. In contrast, only about 3.0–7.1% of the total AME administered to the apical chamber reached the basolateral side, almost exclusively as AME-3-O-glucuronide. This again illustrates the poor absorption of AME as compared to AOH. Interestingly, the apical compartments 3 h after administration of AOH or AME did not only contain the unconjugated parent compounds but also their glucuronides and sulfates. With AOH, the concentration of apical 3-O-sulfate exceeded that of the 3- and 9-O-glucuronide, whereas the AOH glucuronides were the dominating conjugates observed in the basolateral compartment (Figs. 4 and 5). The amounts and patterns of compounds in both compartments clearly indicate that AOH is readily taken up by the cells and metabolized to two glucuronides and one sulfate. Subsequently, both AOH glucuronides are equally well transported into the basolateral and apical compartments, whereas AOH-3-O-sulfate is preferentially transported to the apical side. With AME, uptake into the cells occurs less readily than with AOH. Within the cells, AME is preferentially metabolized to the 3-O-glucuronide and 3-O-sulfate, and to a much lesser extent to the 7-O-glucuronide. Subsequently, the AME-3-O-glucuronide is preferentially, and the 3-O-sulfate and 7-O-glucuronide are completely transported back to the apical compartment.

Fig. 5
figure 5

Apical and basolateral concentrations of aglycones and conjugated metabolites 3 h after the apical exposure to various concentrations of AOH and AME. Data represent mean and standard deviation of three independent experiments

Full size image

Determination of the apparent permeability coefficient Papp of AOH and AME

The transition of a compound from the apical to the basolateral compartment in the Caco-2 Millicell® system is commonly expressed by its Papp value, which correlates well with the absorption of xenobiotic compounds by humans in vivo (Artursson and Karlsson 1991; Yee 1997). In order to calculate the Papp value of AOH and AME, a concentration of 20 μmol/l in the apical compartment was chosen because this does not pose a solubility problem for AME, which is less soluble than AOH. The basolateral amounts of unconjugated material as well as the conjugates were determined after 1, 2, 3, and 6 h (Table 2), and the data were in good agreement with those obtained in the earlier experiment (Fig. 4) with the exception that AME-3-O-sulfate was also detected in the basolateral compartment, in particular after 6 h. From these data, the Papp values listed in Table 2 were calculated for the unconjugated toxins and for the total mycotoxin-associated material, i.e., the sum of aglycone and all conjugated metabolites, according to Artursson and Karlsson (1991).

Table 2 Papp values of AOH and AME obtained in the Caco-2 Millicell® system with an initial apical concentration of 20 μmol/l. Data represent mean and standard deviation of three independent experiments
Full size table

The apparent permeability coefficient Papp indicates the velocity of permeation of the compound across the cell, i.e., across the apical and basolateral cell membrane. For AOH, the Papp values both for unconjugated AOH and the sum of AOH aglycone and conjugates were highest for the first hour and declined continuously when longer incubation times were used (Table 2). This time course suggests that both the AOH aglycone and AOH conjugates are able to readily permeate the apical and basolateral cell membrane, which is supported by the observation that appreciable concentrations of AOH conjugates are observable in the basolateral chamber already after 0.5 and 1 h (Fig. 4), and that sizeable amounts of AOH conjugates are found in the apical chamber (Fig. 5). According to the correlation of Papp values with human absorption in vivo (Yee 1997), AOH must be expected to be well absorbed from the gastrointestinal tract and to reach the portal blood both as aglycone and as glucuronide and sulfate conjugate.

The Papp values obtained for AME were surprisingly different from those of AOH (Table 2). Unconjugated AME was unable to permeate into the basolateral chamber, and AME-3-O-conjugates reached the basolateral side to a significant extent (Table 2). The linear increase of the basolateral concentration of the AME conjugates and the low and time-independent Papp value (Table 2) implies a poor and slow in vivo absorption into the portal blood after conjugation of AME in the intestinal epithelium. According to our in vitro data, the majority of the AME conjugates formed in the intestinal epithelium must be expected to be excreted back into the intestinal lumen (Fig. 5).

Conclusion

The present study on the fate of AOH and AME in differentiated Caco-2 cells in vitro has revealed some similarities but also surprising differences between these closely related mycotoxins. Similarities include the structures of the conjugated metabolites: both toxins are readily conjugated with glucuronic acid and sulfate. AOH is glucuronidated at the hydroxyl groups of C-3 and C-9 but not C-7, and AME, which lacks a hydroxyl at C-9, is extensively glucuronidated at C-3 and only to a very small extent at C-7. This pattern of glucuronidation is in accordance with that observed with hepatic and intestinal human microsomes (Pfeiffer et al. 2009a). With respect to sulfate conjugation, only the hydroxyl group at C-3 is involved with both toxins; AOH-9-O-sulfate, which is readily formed with rat liver cytosol (Fig. 2) was not observed with Caco-2 cells. Once formed in the cells, the further fate of the conjugates differs, in part, between AOH and AME: AOH-3-O-glucuronide, AOH-9-O-glucuronide, and AOH-3-O-sulfate quickly permeate both the basolateral and the apical membranes, although to somewhat different extents; likewise, AME-3-O-glucuronide appears fast on the basolateral and apical side (Figs. 4 and 5). In contrast, AME-3-O-sulfate is preferentially sequestered to the apical side (Fig. 5). The most striking difference between AOH and AME is that sizeable amounts of unconjugated AOH reach the basolateral compartment whereas only AME conjugates, but no AME aglycone could be detected basolaterally. Thus, AOH but not AME may reach the portal blood in unconjugated form upon intestinal absorption in vivo.

Another difference between AOH and AME relates to the rate of absorption. Based on the Papp values and other kinetic data obtained in this study and discussed above, intestinal absorption of AOH in vivo must be expected to be extensive and fast, whereas AME absorption appears to be poor and slow. It will be of interest in future in vivo studies to find out whether the plasma levels of these mycotoxins and their conjugated metabolites after oral administration of equal doses support this notion.