Experimental Procedures

Materials and Reagents.

All-trans-retinol and all-trans-RA were purchased from Sigma (Deisenhofen, Germany) and 13-cis-4-oxo- and all-trans-4-oxo-RA were kindly provided by Hoffmann-La Roche (Basel, Switzerland). Lyophilized analytical grade BSA was purchased from Sigma (St. Louis, MO). Isocitric acid, NADP, and isocitrate dehydrogenase came from Boehringer Mannheim (Mannheim, Germany). All other chemicals were purchased from Merck (Darmstadt, Germany) or Sigma in the highest available purity. Water for HPLC was purified using a Milli-Q water purification system (Millipore, Eschborn, Germany). Liarozole was a generous gift from Janssen Research Foundation (Beerse, Belgium). Stock solutions of retinoids were prepared in ethanol (or dimethyl sulfoxide) at a theoretical concentration of 0.1 mg/ml, which were subsequently photometrically corrected. These solutions were kept in glassware at −20°C. All laboratory manipulations involving the retinoids (preparation of dosing solutions, drug treatment of cells, collection of samples, and analytical procedures) were performed in dark rooms under dim yellow light to prevent photoisomerization. The Caco-2-TC7 cell line (TC7) was a generous gift from Dr. Alain Zweibaum (Institut National de la Santé et de la Recherche Médicale U178, Villejuif Cedex, France).

Cell Culture.

Caco-2 and Caco-2-TC7 cells were kept in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum (Boehringer, Ingelheim, Germany) at 37°C in a humidified atmosphere of 5% CO₂-air.

Retinoid Analyses.

Retinoids in supernatants and cells were analyzed using a reversed-phase HPLC method with gradient elution after sample enrichment with liquid-liquid and solid-phase extraction (Collins et al., 1992). According to this method, samples were extracted with isopropanol, and the supernatants were then extracted on solid-phase extraction cartridges before introduction into the HPLC system. Further processing was performed as described previously byCollins et al. (1992). UV detection of the HPLC eluate was performed at 340 and 356 nm by use of a two-channel SPD-10AV detector (Shimadzu, Duisburg, Germany).

Recovery and reproducibility of the assay and calibration with an external method were reported previously (Collins et al., 1992). Identification of HPLC peaks was based on comigration with authentic retinoids and on coincidence of the absorbance ratios (i.e., ratio of peak areas) at the two wavelengths of detection with that of the standard retinoids.

Preparation of Cytosolic and Microsomal Fraction of Human Intestinal Enterocytes.

Human small intestinal samples for the isolation of cytosol and microsomes were obtained from the Klinik für Abdominal und Transplantationschirurgie (Medizinische Hochschule Hannover, Hannover, Germany), and the collection of human small intestinal samples for research was approved by the Ethical Committee of the Medizinische Hochschule Hannover.

Human enterocytes were isolated according to the method described byPinkus (1981). Cytosolic fractions and microsomes were isolated using a differential centrifugation procedure described by Guengerich (1982)with the modifications that a 0.1 M phosphate buffer (pH 7.4) be used instead of a Tris buffer and that the ultracentrifugation steps (100,000g) be reduced to 45 min. Cytosol was prepared by centrifuging the supernant obtained from the second spin at 105,000g for 1 h. Protein concentrations were measured using the bicinchoninic acid method (Smith et al., 1985) and a BSA standard curve. The protein concentrations of the microsomal suspensions were adjusted with 0.1 M phosphate buffer (pH 7.4).

In Vitro Metabolism of All-trans-ROH and Sample Preparation.

Cytosolic fractions of human enterocytes were incubated for 25 min at 37°C. The incubations contained 1 mg of protein, 2 mM NAD, 2 mM dithiothreitol, and all-trans-ROH (added in 5 μl of ethanol) at various concentrations and were brought to a final volume of 0.5 ml by adding 150 mM KCl or 20 mM HEPES, respectively.

In Vitro Metabolism of All-trans-RA and Sample Preparation.

All-trans-RA (10 μM) in 10 μl of ethanol was added to 1 ml of small intestinal microsomes (1.5 g of protein/l). The reaction was started by adding 0.5 ml of an NADPH-generating system consisting of 2 mM EDTA, 10 mM MgCl₂, 0.84 mM NADP, 18 mM isocitric acid, and 667 U/l of isocitrate dehydrogenase. The reaction mixture was incubated for 30 min at 37°C under aerobic conditions. The reaction was stopped by the addition of a 3-fold volume of isopropanol.

Metabolism of All-trans-RA in Caco-2 Cells.

Caco-2 cells were cultured for 18 to 21 days (maximum of differentiation). All-trans-RA (10 μM) was added to the fresh culture medium and incubated for 6 h. A cell pellet and cell medium volume of 150 μl were extracted with a 3-fold volume of isopropanol, followed by short centrifugation and solid-phase extraction (see above).

Chemical Inhibition of All-trans-RA Metabolism in Caco-2 Cells.

α-Naphthoflavone, a selective CYP1A1 inhibitor (Chang et al., 1994), and liarozole, a inhibitor of the human RA 4-hydroxylase activity (CYP26), were dissolved in dimethyl sulfoxide. Ketoconazole, a selective noncompetitive CYP3A inhibitor at low concentrations (Chang et al., 1994), was dissolved in ethanol. The inhibitor solution (10 μl) was added to 20 ml of medium. The final concentrations ranged from 0 to 750 μM. The samples were preincubated with the inhibitor for 60 min to inactivate the CYP enzymes. Then the substrate all-trans-RA was added and after an incubation period of 3.5 h at 37°C, the reaction was stopped by the addition of isopropanol (see above).

Preparation of RNA.

Total RNA was prepared from freshly isolated cells according to the method of Chomczynski and Sacchi (1987). RNA concentrations were determined spectrally with a UV-visible spectrophotometer (Perkin-Elmer Life Sciences, Foster City, CA), and the integrity as well the concentrations of RNA was checked in an agarose test gel using an RNA-standard (Eurogenitic, Seraing, Belgium).

RT-PCR Analysis.

RT-PCR was carried out using Superscript II reverse transcriptase (Life Technologies, Karlsruhe, Germany) andTaq-polymerase (Qiagen, Hilden, Germany). The samples were then heated for 3 min at 94°C to terminate the reverse transcription reaction. The PCR was performed on 1 μl of a 1:10 dilution in water of a prepared cDNA. Primers for human CYP26 were nt 87–111 for sense and nt 343–368 for antisense (GenBank accession number AF005418). Primers for human CYP1A1 were nt 5762–5784 for sense and nt 6689–6709 for antisense (GenBank accession number X02612). Primers for human CYP3A4 were nt 1349–1376 for sense and nt 1702–1731 for antisense (GenBank accession number J04449). Primers for human CYP3A5 were nt 1236–1266 for sense and nt 1517–1546 for antisense (GenBank accession number NM00777). Primers for human β-actin were nt 371–391 for sense and nt 800–820 for antisense (GenBank accession number X00351). PCR was performed with 0.5 U ofTaq-polymerase (Qiagen, Hilden, Germany) in an automatic DNA thermal cycler (MWG, Ebersberg, Germany) by adding 30 μl of a PCR master mixture containing PCR buffer, MgCl₂ (to a final concentration of 1 mM) and 20 pmol of each primer to the cDNA samples. Thirty cycles [1 min at 94°C, 45 s at 60°C for CYP26 (57°C for CYP3A4 and CYP3A5), and 45 s at 72°C] followed by an additional 5 min at 72°C were used. Each PCR included an aliquot of the reaction mixture without cDNA as a negative control. All amplifications were carried out for 30 cycles. Under these conditions, all cDNA amplifications were found to produce single products within a linear range (data not shown). PCR products were judged to be of the size appropriate for amplification of the specific cDNA by comparison with molecular weight standards included on each gel and with amplified plasmid cDNA of each gene. Amplified cDNA products were separated by agarose electrophoresis. Gels were stained with ethidium bromide, photographed using a charge-coupled device camera of Gel Doc 1000 (Bio-Rad, München, Germany), stored, and densitometrically analyzed using Molecular Analysis (Bio-Rad). We thank Prof. M. Petkovich (Queen's University, Kingston, Canada) for providing us with the CYP26 cDNA.

Statistics.

Values for concentrations and concentration ratios were expressed as mean ± S.D. ANOVA was used for the statistical comparison of two means.

Results

Formation of RA in Small Intestinal Cells.

When the cytosolic fraction of human enterocytes was incubated with all-trans-ROH, the active RAR ligand all-trans-RA was formed. The formation was concentration-dependent (Fig.1A), the K_mand V_max values for the metabolism of all-trans-ROH to all-trans-RA were 15.4 μmol/l and 2 pmol/min × mg of protein, respectively. The formation of all-trans-RA was inhibited by 4-methylpyrazole (Fig. 1B). 4-Methylpyrazole (10 mM) inhibited the formation of all-trans-RA by 85%, suggesting the involvement of ADHs. Furthermore, citral inhibited the formation of all-trans-RA. Citral (50 μM) inhibited the formation of all-trans-RA by 75%.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

A, metabolism of all-trans-retinol to all-trans-RA in the cytosolic fraction of human small intestinal enterocytes. Different concentrations of retinol were incubated for 25 min with 1 mg of cytosolic protein, stopped with isopropanol, and extracted and analyzed using HPLC. B, metabolism of all-trans-ROH to all-trans-RA is inhibited by 4-methylpyrazole in human small intestinal cytosol. All-trans-retinol (10 μM) was incubated with different concentrations of 4-methylpyrazole for 25 min with 1 mg of cytosolic protein, stopped with isopropanol, and extracted and analyzed using HPLC. All values represent mean ± S.D. (n = 4).

Metabolism of All-trans-RA with Human Supersomes.

To evaluate which CYP enzymes are capable of metabolizing all-trans-RA, we incubated a supermix (Gentest, Frankfurt, Germany) containing CYP enzymes of human liver (CYP1A2, CYP2C8, CYP2C9, CYP2C19, CYP2D6, and CYP3A4) and single human CYP1A1, CYP3A4, CYP2C8, CYP2E1, and CYP2D6 supersomes with all-trans-RA and measured the metabolites by HPLC.

Using human supermix all-trans-RA was metabolized after a 30-min incubation to all-trans-4-oxo-RA (0.888 pmol/h × pmol of P450) and all-trans-4-hydroxy-RA (45.51 pmol/h × pmol of P450). With human CYP supersomes all-trans-RA was metabolized mainly to all-trans-4-hydroxy-RA. All-trans-RA was metabolized by CYP3A4 supersomes exclusively to all-trans-4-hydroxy-RA (3.15 pmol/h × pmol of P450). CYP1A1 supersomes metabolized all-trans-RA to all-trans-4-oxo-RA (0.12 pmol/h × pmol of P450) and to all-trans-4-hydroxy-RA (3.0 pmol/h × pmol of P450). After incubation with human CYP2E1 all-trans-4-hydroxy-RA (1.1 pmol/h × pmol of P450) was formed. After incubation of CYP2C8 supersomes all-trans-4-hydroxy-RA (2.66 pmol/h × pmol of P450) was formed. No metabolism of all-trans-RA was detectable using CYP2D6-supersomes. Again no metabolism was detectable after incubation of all-trans retinol with CYP supersomes or supermix.

Metabolism of All-trans-RA by Human Small Intestinal Microsomes.

To investigate the further metabolism of all-trans-RA human small intestinal microsomes were used. After incubation of all-trans-RA with human small intestinal microsomes, two major metabolites were identified by HPLC: all-trans-4-hydroxy-RA and 4-oxo-RA (Fig.2A).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

A, metabolism of all-trans-RA in human small intestinal microsomes. Human small intestinal microsomes were incubated with 0.5 ml of NADPH-generating system and 20 μmol/l all-trans-RA (see Experimental Procedures). Incubations were stopped with isopropanol, and samples were extracted and analyzed using HPLC. All values represent mean ± S.D. (n = 3). B, concentration-dependent metabolism of all-trans-RA in Caco-2 cells. Caco-2 cells (2 × 10⁷ cells) were incubated with various concentrations of all-trans-RA. Four representative chromatograms are shown. Incubations were stopped with isopropanol, and samples were extracted and analyzed using HPLC.

Metabolism of All-trans-ROH and All-trans-RA in Caco-2 Cells.

To investigate the formation and the further metabolism of all-trans-RA in more detail, an in vitro model (Caco-2 cells) was used. When all-trans-ROH (7.5 μM) was incubated with Caco-2 cells (2 × 10⁷ cells) for 24 h we measured the formation of 0.33 μM all-trans-RA by HPLC, indicating that the formation of the RAR ligand all-trans-RA also occurs in human Caco-2 cells.

When Caco-2 cells were incubated with different concentrations of all-trans-RA the polar metabolites all-trans-4-oxo-RA, all-trans-4-hydroxy-RA, and 13-cis-4-hydroxy-RA were mainly formed (Fig. 2B). The peak eluting after 13-cis-4-hydroxy-RA (Figs. 2B and3) could be not identified due to the lack of a reference substance for 18-hydroxy-RA, but may represent 18-hydroxy-RA. The isomerization metabolites 13-cis-RA and 9-cis-RA were also formed. The metabolism depended on the concentration of the substrate used (Fig. 2B) and was time-dependent. The K_m was calculated to be 8.3 μmol/l.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Metabolism of all-trans-RA in Caco-2 cells after pretreatment of the cells with the CYP1A1 inducer β-naphthoflavone. Caco-2 cells were pretreated with 50 μM β-naphthoflavone for 72 h and then incubated with 10 μM all-trans-RA for an additional 6 h. Metabolites were extracted and analyzed using HPLC. Typical chromatograms are shown.

Induction and Inhibition of All-trans-RA Metabolism in Caco-2 Cells.

When the Caco-2 cells were treated with β-naphthoflavone, a specific inducer of CYP1A1, the metabolism of all-trans-RA to polar metabolites was induced (Fig. 3). Fifty micromolar β-naphthoflavone induced the metabolism to the metabolites all-trans-4-oxo-RA, all-trans-4-hydroxy-RA, and 13-cis-4-hydroxy-RA by the factor of 10. On the other hand, the specific CYP1A1 inhibitor α-naphthoflavone inhibited the metabolism of all-trans-RA to the polar metabolites by 76%. These results indicate that CYP1A1 was capable of metabolizing all-trans-RA and suggest that if induced, CYP1A1 metabolizes all-trans-RA in intestinal cells.

Ketoconazole inhibited the metabolism of all-trans-RA to the polar metabolites almost completely but only when high concentrations (>50 μM) were used, suggesting a more unspecific CYP inhibition (data not shown). Other CYP inhibitors such as sulfaphenazole (CYP2C inhibitor) or disulfiram (CYP2E1 inhibitor) showed no effect on the metabolism (data not shown). To investigate the role of CYP3A enzymes in the intestinal metabolism of all-trans-RA, we used the Caco-2 clone TC-7, which has a higher catalytic activity of CYP3A than do Caco-2 cells (Raeissi et al., 1999). Pretreatment of the TC-7 cells with rifampicin induced the formation of all-trans-4-hydroxy-RA (Fig.4A) and induced CYP3A4 and CYP3A5 gene expression (Fig. 4B), suggesting that CYP3A enzymes are involved, at least in part, in the metabolism of all-trans-RA.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

A, induction of the intestinal RA metabolism by a CYP3A inducer. Caco-2-TC7 cells were pretreated with 10 μM rifampicin or vehicle (control) for 24 h, washed three times with PBS, and incubated with 10 μM all-trans-RA for an additional 3 h. Metabolites were extracted and analyzed using HPLC. Values represent mean ± S.D. (n = 3). ∗, significant difference (P < .001; ANOVA) compared with pretreatment without rifampicin. B, induction of CYP3A4 and CYP3A5 gene expression in Caco-2 TC7 cells by rifampicin measured using RT-PCR. CYP3A4, CYP3A5, and β-actin gene expression was measured under optimal conditions using RT-PCR after treatment of Caco-2-TC7 cells with 10 μM rifampicin or vehicle (control) for 24 h. C, untreated; RIF, rifampicin-treated, 10 μM; M, marker.

Pretreatment of the Caco-2 cells with all-trans-RA induced the metabolism of all-trans-RA to polar metabolites by a factor of 3, indicating an autoinduction (Fig.5). Liarozole, an imidazole derivative, is a known inhibitor of RA 4-hydroxylase activity (Fujii et al., 1997;Sonneveld et al., 1998), which is associated with the inhibition of CYP26. Liarozole inhibited the metabolism of all-trans-RA to polar metabolites very efficiently (Fig.6). Liarozole (50 μM) inhibited the metabolism of all three polar metabolites >90%, indicating that CYP26 is mainly responsible for the intestinal metabolism.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5

Autoinduction of the intestinal retinoic acid metabolism by all-trans-RA. Caco-2 cells were pretreated with 1 μM all-trans-RA or vehicle (control) for 24 h, washed three times with PBS, and incubated with 10 μM all-trans-RA for an additional 3 h. Metabolites were extracted and analyzed using HPLC. Values represent mean ± S.D. (n = 3). ∗, significant difference (P < .001; ANOVA) compared with untreated cultures.

• Download figure
• Open in new tab
• Download powerpoint

Figure 6

Inhibition of all-trans-RA metabolism in Caco-2 cells by liarozole. Caco-2 cells were incubated with 10 μM all-trans-RA with or without different concentrations of liarozole. Metabolites were extracted and analyzed using HPLC. All values represent mean ± S.D. (n = 4).

Gene Expression of CYP26 in Caco-2 Cells After RA Treatment.

Untreated Caco-2 cells expressed only small amounts of CYP26, almost at the detection limit (Fig.7). When Caco-2 cells were treated with all-trans-RA, the expression of CYP26 increased specifically and in a dose-dependent manner (Fig. 7). This concentration-dependent induction correlates well with the concentration-dependent formation of polar metabolites after incubation of all-trans-RA with Caco-2 cells (Fig. 2B). When Caco-2 cells were treated with retinol (10 μM for 24 h), CYP26 gene expression was also induced (Fig.8), most probably because of the formation of all-trans-RA from retinol in the cell.

• Download figure
• Open in new tab
• Download powerpoint

Figure 7

CYP 26 gene expression in Caco-2 cells measured using RT-PCR. CYP 26 and β-actin gene expression was measured under optimal conditions using RT-PCR after treatment of Caco-2 cells with 0.001, 0.01, 0.1, 1, and 10 μM all-trans-RA. C, untreated; PL, cDNA (CYP26 or β-actin); M, marker.

• Download figure
• Open in new tab
• Download powerpoint

Figure 8

CYP 26 gene expression in Caco-2 cells measured by RT-PCR. CYP 26, CYP1A1, and β-actin gene expression was measured under optimal conditions using RT-PCR after treatment of Caco-2 cells with 50 μM β-naphthoflavone, 10 μM rifampicin, 20 μM all-trans-retinol (in the presence of 6 mM sodium taurocholate), or 10 μM all-trans-RA. C, untreated; NF, β-naphthoflavone; RIF, rifampicin; ROH, all-trans-retinol; RA, all-trans-RA.

The known inducer of CYP1A1 (β-naphthoflavone) failed to induce CYP26 gene expression at all. Furthermore, inducers of CYP3A such as rifampicin had no effect on CYP26 gene expression (in Caco-2 cells), indicating the specificity of the induction of CYP26 by all-trans-RA (Fig. 8). On the other hand, all-trans-RA and all-trans-ROH failed to induce CYP1A1 in Caco-2 cells (Fig. 8). CYP1A1 was also induced neither by 9-cis-RA nor by 13-cis-RA (data not shown). Therefore it seems likely that in addition to CYP1A1 and CYP3A, particularly when they are induced by a typical inducer, CYP26 is the main CYP enzyme responsible for the metabolism of all-trans-RA in Caco-2 cells.

Discussion

It is well known that when vitamin A is absorbed in the small intestine, it is converted by ARAT or LRAT into retinyl esters, which are packed in chylomicrons and released into the lymph. We show here for the first time that in the human enterocyte, retinol is oxidized to all-trans-RA and 13-cis-RA. All-trans-RA is further metabolized mainly by CYP26 to more polar metabolites in Caco-2 cells. It was shown that CYP26 is expressed in human Caco-2 cells and is inducible by all-trans-RA.

The concentration-dependent formation of the RAR ligands indicates an active saturable enzymatic conversion process in the enterocyte. Enzymes that have been identified in other tissues that catalyze the metabolism of retinol to RA were members of the family of cytosolic medium-chain ADHs (classes I, II, and IV), aldehyde dehydrogenases (class I), microsomal short-chain ADHs (Napoli et al., 1996;Duester, 1996), and CYP1A1 and CYP1A2 for the oxidation of retinaldehyde to RA (Tomita et al., 1996). It has also been known that the aldehyde dehydrogen oxidases can convert retinaldehyde to retinoic acid (Duester, 1996; Napoli, 1996). The oxidation of all-trans-retinol to the RAR ligand all-trans-RA by the cytosolic fraction of human small intestinal enterocytes represents a new intestinal pathway of vitamin A metabolism. The inhibition of the formation of RA from dietary ROH by 4-methylpyrazole (an inhibitor of the unspecific ADH) and citral showed that nonspecific ADHs as well as aldehyde dehydrogenases may also be involved.

The formation of active RAR and RXR ligands in the enterocyte is of particularly interest because retinoic acids (especially all-trans-RA) are much more active than retinol. Various studies have shown that all-trans-RA is essential for proliferation and differentiation of epithelial tissues. The human Caco-2 cell line derived from a human colon carcinoma undergoes enterocytic differentiation in culture with morphological (developing microvilli and tight junctions) and biochemical (increase alkaline phosphatase activity) changes typical for a small intestinal enterocyte (Boulenc et al., 1992; Delie and Rubas, 1997). Caco-2 cells are well established in vitro models for the metabolism of CYP1A1 substrates (Boulenc et al., 1992) as well as for drug transport (Delie and Rubas, 1997). These cells express different CYPs, but only enzymatic catalytic activities of CYP1A1 and CYP3A have been reported (Boulenc et al., 1992; Lampen et al., 1998; Raeissi et al., 1999). The nuclear retinoid receptors RXR and RAR have been detected in Caco-2 cells (McCormack et al., 1996). In Caco-2 cells, RA inhibits growth and stimulates differentiation (McCormack et al., 1996). Incubation of Caco-2 cells with β-carotene, retinal, or ROH leads to the formation of retinyl esters. The catalytic activities of retinal reductase, ARAT, and LRAT have been detected in microsomal preparations of Caco-2 cells (Quick and Ong, 1990). Furthermore, it has been shown that RA induces CRBPII mRNA in Caco-2 cells (Suzuki et al., 1998). Therefore, Caco-2 cells seem to be a valuable model to study intestinal metabolism of retinoids.

In this report we show for the first time that in human intestinal cells (Caco-2), all-trans-RA is formed from dietary vitamin A and further metabolized to the polar metabolites all-trans-4-oxo-RA, all-trans-4-hydroxy-RA, and 13-cis-4-hydroxy-RA. This metabolism is not a simple inactivation because 4-oxo-retinoic acid is believed to be highly active (Pijnappel et al., 1993).

It has long been known that the cytochrome P450 system is active in metabolizing RA. However, a detailed identification of the responsible enzymes failed. Our experiments using a specific CYP1A1 inducer (β-naphthoflavone) as well as a CYP1A1 inhibitor (α-naphthoflavone) suggest an involvement of CYP1A1 in the metabolism of all-trans-RA in Caco-2 cells, provided that CYP1A1 is induced. The incubations with single CYP supersomes showed that CYP3A4, CYP1A1, CYP2C, and CYP2E1 are able to metabolize RA. However, the degree of metabolism was low in comparison with that of specific substrates of these cytochromes, suggesting that these cytochromes play a minor role in intestinal RA metabolism. This interpretation is supported by the observations of Inouye et al. (1999) as well as by those of Yamazaki and Shimada (1999) who reported a competitive inhibition of CYP1A1-dependent mono-oxygenase activity by all-trans-RA using recombinant enzyme systems of rat and human CYP1A1, respectively. These effects seem to be of importance to food toxicological in vivo because 2,3,7,8-tetrachlorodibenzo-p-dioxin, a potent CYP1A1 inducer, induces increased mobilization of retinoids from hepatic storage (Brouwer et al., 1989). Furthermore, 2,3,7,8-tetrachlorodibenzo-p-dioxin induces RA metabolism to polar metabolites in liver microsomes of rats (Fiorella et al., 1995). Additionally, aryl hydrocarbon receptor knockout mice (AHR−/−) exhibit liver retinoid accumulation and reduced RA metabolism (Andreola et al., 1997). Food contaminants such as dioxins and other organohalogen food residues, which are potent CYP1A1 inducers, may have a significant influence on RA metabolism and action. On the other hand, all-trans-RA does not induce CYP1A1 in human Caco-2 cells, although a retinoid-responsive element was identified in the promoter region of the human CYP1A1 gene (Vecchini et al., 1994). In our study, CYP1A1 was induced neither by 9-cis-RA nor by all-trans-RA (nor 13-cis-RA), showing that the CYP1A1 metabolism of RA seems to be a side pathway. The induced metabolism of all-trans-RA to all-trans-4-hydroxy-RA after pretreatment with the CYP3A inducer rifampicin showed that CYP3A is involved, at least in part, in the metabolism of all-trans-RA in the enterocytes. However, the CYP3A inhibitor ketoconazole inhibited the metabolism only at high concentrations (>50 μM).

The metabolism of all-trans-RA to polar metabolites was induced by all-trans-RA itself (autoinduction), suggesting that an RA-regulated CYP enzyme is involved. The almost complete inhibition of the RA metabolism by liarozole and the specific concentration-dependent induction of CYP26 by retinoids, which seems to correlate with the concentration-dependent formation of polar metabolites after incubation of all-trans-RA with Caco-2 cells (Fig. 2B), suggested that this could be a main pathway of all-trans-RA. This cytochrome is able to metabolize RA. Both 13-cis- and 9-cis-RA, but not ROH, were found to serve as substrates for CYP26 (Fujii et al., 1997). In adult mice, CYP26 was expressed only in liver (Fujii et al., 1997). When the human cDNA for CYP26 was expressed in COS-1 cells, all-trans-RA was rapidly metabolized to more polar metabolites (all-trans-4-oxo-RA, all-trans-4-hydroxy-RA, and 18-hydroxy-RA). The all-trans-4-oxo-RA and 4-hydroxy-all-trans-RA are also main metabolites of intestinal RA metabolism. In the adult human, CYP26 is expressed in liver, brain, and placenta (Ray et al., 1997). It was also detected recently in human colon HCT 116 carcinoma cells (Sonneveld et al., 1998). We showed that CYP26 is also expressed in human Caco-2 enterocytes and demonstrated its inducibility by all-trans-RA. Known inducers of other CYPs such as β-naphthoflavone and rifampicin did not induce CYP26 gene expression in Caco-2 cells at all. Based on the available data, it seems likely that the oxidative metabolism of retinoic acid plays an important role in regulating tissue levels of retinoic acid and that CYP26 may have an important role in controlling this metabolism.

In summary, we report here that the active RAR ligand all-trans-RA is formed in the small intestine via direct oxidation of vitamin A. In the enterocyte, all-trans-RA is oxidized further to polar metabolites mainly through CYP26, which may be an important regulator of intestinal RA levels. However, CYP1A1 and CYP3A, particularly when induced by β-naphthoflavone or rifampicin, respectively, seem to participate in part in the metabolism of all-trans-RA in Caco-2 cells.