4-Hydroxyretinoic Acid, a Novel Substrate for Human Liver Microsomal UDP-glucuronosyltransferase(s) and Recombinant UGT2B7*

It is suggested that formation of more polar metabolites of all-trans-retinoic acid (atRA) via oxidative pathways limits its biological activity. In this report, we investigated the biotransformation of oxidized products of atRA via glucuronidation. For this purpose, we synthesized 4-hydroxy-RA (4-OH-RA) in radioactive and nonradioactive form, 4-hydroxy-retinyl acetate (4-OH-RAc), and 5,6-epoxy-RA, all of which are major products of atRA oxidation. Glucuronidation of these retinoids by human liver microsomes and human recombinant UDP-glucuronosyltransferases (UGTs) was characterized and compared with the glucuronidation of atRA. The human liver microsomes glucuronidated 4-OH-RA and 4-OH-RAc with 6- and 3-fold higher activity than atRA, respectively. Analysis of the glucuronidation products showed that the hydroxyl-linked glucuronides of 4-OH-RA and 4-OH-RAc were the major products, as opposed to the formation of the carboxyl-linked glucuronide with atRA, 4-oxo-RA, and 5,6-epoxy-RA. We have also determined that human recombinant UGT2B7 can glucuronidate atRA, 4-OH-RA, and 4-OH-RAc with activities similar to those found in human liver microsomes. We therefore postulate that this human isoenzyme, which is expressed in human liver, kidney, and intestine, plays a key role in the biological fate of atRA. We also propose that atRA induces its own oxidative metabolism via a cytochrome P450 (CYP26) and is further biotransformed into glucuronides via UGT-mediated pathways.

The purpose of the present work was to examine phase II metabolism of atRA and identify the human UGTs (EC 2.4.1.17) involved in the glucuronidation of atRA, 4-OH-RA, 4-oxo-RA, 5,6-epoxy-RA, and 4-OH-RAc (structures shown in Fig. 1). UGTs catalyze the conjugation of endogenous and xenobiotic compounds to D-glucuronic acid derived from UDPglucuronic acid (UDP-GlcUA). This converts the metabolites into a more polar form, promoting their excretion in urine and/or bile. However, several examples of bioactivation of parent compounds via glucuronidation have been reported (21)(22)(23)(24)(25). UGTs can be divided into two families, UGT1A and UGT2B, based on their sequences. In this work, human liver microsomal UGTs and recombinant isoforms from both families were evaluated for their ability to catalyze glucuronidation of retinoids in various stages of oxidation.
To obtain the appropriate substrates for these studies, we synthesized 4-OH-RAc, 5,6-epoxy-RA, and both radioactive and nonradioactive forms of 4-OH-RA. The studies in this report center on examining the glucuronidation of atRA, 4-OH-RA, 4-oxo-RA, 5,6-epoxy-RA, and 4-OH-RAc by human liver microsomes and the human recombinant UGTs, 1A3 and 2B7. Human liver UGT2B7 is involved in the glucuronidation of a large variety of endogenous and exogenous substrates including bile acids, steroids, and a variety of xenobiotics (26 -30). This report identifies UGT2B7 as the major isoform catalyzing the formation of carboxyl-and hydroxyl-linked glucuronides of retinoids at high levels of activity.

Synthesis of Radioactive and Nonradioactive Retinoid Substrates
Synthesis of (Ϯ)-5,6-Epoxy-(E)-retinoic Acid (Scheme 1)-Racemic 5,6-epoxy-RA (2) was synthesized by epoxidation of atRA (31) with monoperoxyphthalic acid, which was synthesized by a reaction of phthalic anhydride with alkaline H 2 O 2 as described elsewhere (32) and quantitated iodometrically. 1 (1.0 g, 3.3 mmol) was reacted with a 15-fold molar excess of monoperoxyphthalic acid in diethyl ether (50 ml) for 1 h at room temperature. The formation of the oxirane was quantitative as evidenced by TLC analysis using ethyl acetate-hexane (40:60 or 50:50, v/v) as the mobile phase. The reaction mixture was washed with saturated sodium bicarbonate, water, and saturated NaCl. The ethereal layer was dried over anhydrous MgSO 4 , and solvent was removed in vacuo. The oxirane was recrystallized from diethyl ether/ hexane (84% yield). Mass and 1 H NMR spectra were obtained for the corresponding methyl ester, which was synthesized by reaction of the free acid with ethereal diazomethane prepared as described below: Synthesis of (Ϯ)-4-Hydroxy-(E)-retinyl Acetate (5) (Scheme 2)-4 (0.5 g, 1.5 mmol) was dissolved in 50 ml of methanol, an equal volume of water was added, and the mixture was stirred at 4°C. An excess of NaBH 4 was added, and the reaction was allowed to proceed at 4°C for 15 min and then for 1 h at room temperature. The reaction mixture was neutralized with phosphate buffer and extracted with ethyl acetate. The pooled ethyl acetate extracts were washed with water and subsequently dried over MgSO 4 . 5 was subsequently purified by silica gel chromatography using ethyl acetate as the mobile phase, which afforded a 75% yield (EI-MS m/z 344 (M ϩ ), 326 (M Ϫ H 2 O) ϩ ). In order to verify the presence of a hydroxyl group using mass spectrometry, the trimethylsilyl derivative was prepared by reaction of the acetate with bis(trimethylsilyltrifluoroacetamide): EI-MS m/z 416 (Mϩ).
Synthesis of (E)-Methylretinoate (6) (Scheme 3)-atRA (1.0 g, 3.3 mmol) was dissolved in 25 ml of anhydrous diethyl ether (note: not all of the atRA may dissolve) to which distilled ethereal diazomethane (ϳ50 -100 mmol) was added, and the reaction was allowed to sit for 1 h at room temperature, after which the ether phase was removed under a stream of argon. The ethereal diazomethane was prepared by reaction of Diazald ® with NaOH using a Diazald ® kit (Aldrich). The reaction was shown to be quantitative as evidenced by TLC analysis using ethyl acetate/hexane or benzene (20:80 v/v) as mobile phases. EI-MS of 6 demonstrated a molecular ion of m/z 314 (M ϩ ) (base peak), as well as fragmentation ions of 299 (M Ϫ CH 3 ) ϩ , 282, 267, 159, 119, and 105.
Synthesis of 4-Oxo-(E)-methylretinoate (7) and 4-Oxo-(E)-retinoic Acid (10) (Scheme 3)-6 (1.04 g, 3.3 mmol) was reacted with MnO 2 (30 g) by stirring at room temperature in CH 2 Cl 2 (200 ml) for 72 h. The MnO 2 was removed by filtration, the solvent was removed in vacuo, and the product was purified by silica gel chromatography using ethyl acetate/hexane (10:90 v/v) as the mobile phase. The yield was 68%. EI-MS m/z 328 (Mϩ) (base peak); 1  The corresponding free acid (10) was prepared by hydrolysis of 4-oxomethylretinoate (7). A solution of 7 (0.5 g, 1.5 mmol) in methanol (40 ml) was reacted with aqueous NaOH (6 N, 40 ml) for 4 h at room temperature and then added to aqueous H 2 SO 4 (500 ml, 1 N). The mixture was extracted with ethyl acetate, and the pooled ethyl acetate extracts were washed with water and saturated NaCl, dried over MgSO 4 , and solvent was removed in vacuo. 10 was subsequently purified by recrystallization from diethyl ether/hexane (86% yield), and its purity was verified by analytical reverse phase HPLC analysis and UV spectroscopy as described previously (31).
Synthesis of (Ϯ)-4-Hydroxymethylretinoate (8) and (Ϯ)-4-Hydroxyretinoic Acid (9) (Scheme 3)-7 (0.6 g, 1.8 mmol) was dissolved in 50 ml of methanol, an equal volume of water was added, and the mixture was stirred at 4°C. An excess of NaBH 4 was added, and the reaction was The corresponding free acid (9) was obtained by hydrolysis of 8. A solution of 8 (0.35 g, 1.1 mmol) in methanol (40 ml) was reacted with aqueous NaOH (6 N, 40 ml) for 4 h at room temperature and then acidified to pH 5.0 with citrate buffer (3 M). This buffer was used for acidification because we have shown that higher proton concentrations result in significant acid-catalyzed dehydration of the alcohol (data not shown). The mixture was extracted with ethyl acetate, the pooled ethyl acetate extracts were washed with water and dried over MgSO 4 , and solvent was removed in vacuo. 9 was subsequently purified by recrystallization from diethyl ether/hexane (81% yield). The purity of the latter was verified by analytical reverse phase HPLC analysis and UV spectroscopy as described previously (33). (11) (Scheme 3)-50 mg of 10 (0.159 mmol) was dissolved in 1.5 ml of methanol and 0.5 ml of 2-propanol, followed by the addition of 0.2 ml of 1 M NaOH (0.2 mmol). To this mixture, 20 l of 0.1 M NaBH 4 (2 mol, in 0.01 N NaOH) was added. After 2 h, the mixture was transferred to a vial containing 5 mCi of NaB 3 H 4 (204 mCi/mmol, 24.5 mol). The reaction was allowed to proceed overnight, and the course of the reaction was followed by TLC using ethyl acetate-hexane (40:60, v/v) as the mobile phase. Subsequently, NaBH 4 (30 mg) was added in order to assure reduction of any unreacted ketone. After 2 h, the mixture was acidified to pH 6 using 1 M acetic acid and extracted with ethyl acetate. The organic phase was washed with water and dried over Na 2 SO 4 , and the solvent was removed in vacuo. The yield was 64% (36 mg) with a specific activity of 32 mCi/mmol. The product was shown to be pure by analytical reverse phase HPLC, as described in the previous section.

Synthesis of Racemic 4-Hydroxy-4-[ 3 H]retinoic Acid
Human Liver Microsomes-The human liver microsomes used in the experiments were from a 56-year-old man who had died of cerebral bleeding (HLM15) and from a 13-year-old girl who died from brain damage (HLM18). These samples were obtained from the University of Groningen, Groningen, The Netherlands. The HLM served as a control for the glucuronidation assays by providing a basis of comparison for recombinant UGTs.
Human Recombinant UGTs-Human recombinant UGT1A3 was expressed in a mammalian expression system as described previously (34). UGT2B7 was expressed in human embryonic kidney (HK293) cells as reported previously (28,29). Enriched endoplasmic reticulum membrane fractions were prepared as described previously (35). The membrane fractions were stored at Ϫ80°C in 5 mM HEPES, 0.25 M sucrose, 20 mM MgCl 2 (pH 7.4). The enzymatic activity of the recombinant UGT proteins was sustained for up to 6 months under these conditions.
Enzyme Assays-UGT activity was measured with both radioactive and unlabeled forms of atRA and 4-OH-RA as the aglycons with UDP-GlcUA serving as the sugar donor (retinoid structures shown in Fig. 1). All retinoid substrates were prepared in the form of mixed micelles with Brij 58 (0.12%). The Brij 58 micelles both activated the enzyme and solubilized the retinoids. Human liver microsomes and recombinant UGTs (50 g of protein) were used in the assays. All enzymatic assays were performed under yellow light. The amount of product formed was less than 10% of total substrate added and was linearly proportional to the amount of microsomal protein added. The retinoid derivatives (0.10 mM final concentration) were incubated in 100 mM HEPES-NaOH, pH 7.5, 5 mM MgCl 2 , 5 mM saccharolactone, and 0.05% Brij 58 in a final volume of 60 l. The reaction mixture was preincubated with the proteins at room temperature for 10 min before starting the reaction with the addition of either 50 mM UDP-GlcUA for radioactive retinoids  Fig. 2), two methods were used. First, in addition to TLC in the acidic system described above, samples were chromatographed in an alkaline solvent system (ethanol-ethyl acetate-30% NH 4 OH, 45:45: 10, v/v) to hydrolyze carboxyl glucuronides in the course of TLC. Alternatively, after completion of the incubations, 8 N NaOH was added to the incubation mixtures and the samples were incubated for an additional 30 min at room temperature to hydrolyze the carboxyl glucuronide. Controls without 8 N NaOH were run concurrently. Hydroxyllinked glucuronides are stable under these experimental conditions. This procedure allowed clear determination of the formation of hydroxyl-and/or carboxyl-linked glucuronides (structures shown in Fig. 2). After development in chloroform-methanol-glacial acetic acid-water 65: 25:2:4 (v/v), plates were dried and subjected to autoradiography for 3-7 days at Ϫ80°C. When tritium-labeled substrates were used, plates were sprayed with En 3 Hance (NEN Life Science Products) before autoradiography. Glucuronide bands were localized using the autoradiographs, and silica gel containing labeled metabolites (or from corresponding areas in control lanes) was scraped into vials and radioactivity was determined by scintillation counting (LKB RackBeta 1214; Wallac Inc., Gaithersburg, MD). Specific activities of enzymes are expressed as picomoles/min ϫ mg of protein; where applicable, means Ϯ S.E. are reported.
Determination of Kinetic Constants-To determine K m and V max for HLM15 and UGT2B7 with atRA and 4-OH-RA as substrates, the concentrations of atRA and 4-OH-RA were varied (1-200 M [ 3 H]RA or unlabeled 4-OH-RA), while the concentration of UDP-GlcUA was held constant (4 mM). Data were analyzed and Michaelis-Menten kinetic parameters were determined using EnzymeKinetics software (Trinity Software, Campton, NH).
With human liver microsomes, 5,6-epoxy-RA was the best substrate for generating carboxyl-linked glucuronides. Comparison of the rates of glucuronidation of 5,6-epoxy-RA by microsomes and recombinant UGT2B7, however, indicates that other UGT isoforms might be involved in the glucuronidation of this retinoid.
In separate studies designed to compare the rates of glucuronidation for isomeric forms of atRA, the substrates atRA, 9-cis-RA, and 13-cis-RA were incubated with HLM18. Although, this microsomal preparation had much lower activity toward isomeric forms of atRA than did HLM15, the limited amount of the latter preparation necessitated its use. The specific activities indicated that 9-cis-RA was the best substrate for human liver microsomes, followed by atRA and 13-cis RA. The activities were 590 Ϯ 102, 110 Ϯ 9, and 88 Ϯ 31 pmol/mg protein ϫ min, respectively. All three substrates were glucuronidated at their carboxyl function.
Kinetic Analysis-Kinetic parameters were determined for the glucuronidation of the carboxyl-function of atRA and the hydroxyl-function of 4-OH-RA using HLM15 and recombinant UGT2B7 (Table II). The K m for atRA was 182-fold lower than for 4-OH-atRA with both microsomes and recombinant UGT2B7. This indicated that both microsomal and recombinant UGTs have a much higher affinity for free atRA than for the 4-hydroxylated derivative.
Product Identification- Fig. 3 is an autoradiogram of a TLC plate from a representative assay of glucuronidation of [ 3 H]atRA, and unlabeled 4-OH-RA and 4-OH-RAc by HLM15 and UGT2B7. The metabolites formed by glucuronidation of retinoids containing carboxyl and/or hydroxyl functional groups were either carboxyl-or hydroxyl-linked glucuronides (Fig. 2). Biosynthesis of double glucuronides was not observed. Double development of the TLC plate with chloroform-methanol-glacial acetic acid-water 65:25:2:4 (v/v) solvent effectively separated the mixture of the two glucuronides (Fig. 3). An unidentified 14 C-labeled endogenous product was detected (Fig.  3, lanes 1 and 4) when radiolabeled UDP-GlcUA was used.
To identify which glucuronides were produced with a given retinoid substrate, an alkaline solvent system was used to develop the TLC plate. Under these conditions, carboxyl-linked glucuronides were hydrolyzed as has been described previously (36), whereas the 4-OH-RA glucuronide was resistant to alkaline hydrolysis (data not shown). Additional proof of the identity of the glucuronide biosynthesized was achieved by alkaline hydrolysis (8 N NaOH, 30 min room temperature) performed prior to TLC separation.
The TLC analysis demonstrated that free atRA and the cis-isomers, as expected, exclusively formed carboxyl-linked glucuronide. 4-OH-RA or 4-OH-RAc were glucuronidated exclusively at the hydroxyl function, producing alkaline-resistant ether glucuronides. This indicated that the presence of the  hydroxyl function on the ␤-ionine ring redirects the glucuronidation from the carboxyl to the hydroxyl function. DISCUSSION The important roles that atRA and RAG play in cellular differentiation, proliferation, apoptosis, and other cellular events necessitate studies directed at understanding the metabolism of atRA. Additionally, atRA is an important chemotherapeutic agent for skin, breast, endometrial, and hematopoietic cancers. Therefore, knowledge of atRA metabolism is essential to understand its limitations as an anti-cancer compound.
Isomers of atRA, such as 9-cis-RA and 13-cis-RA, are considered to be as biologically active as atRA (37). Experimental results from in vitro and in vivo systems indicate that atRA oxidative metabolism involves both cytochrome P450 and prostaglandin H synthase-mediated processes. We have previously demonstrated that atRA and 13-cis-RA are hydroxylated at the C-4 position by prostaglandin H synthase (33,38,39). Recent reports suggest that the generation of 4-OH-and 18-OH-RA relies on a cytochrome P450-dependent process. Members of a new cytochrome P450 family, CYP26, can be induced by atRA and can also inactivate biologically active atRA (5)(6)(7)(8)(9).
A major unanswered question is whether oxidized retinoic acid metabolites are only intermediates in the catabolic degradation of atRA, or if these oxidized metabolites possess biological functions similar to those of atRA. If the formation of more polar metabolites of atRA via hydroxylation is indeed deleterious to its biological activity, then the role of atRA and RAG will not be completely understood without identifying the sequence of events in the metabolism of atRA. To gain insight into these questions, two major issues must be addressed: 1) the physiological mechanisms that promote atRA-induced RA hydroxylation by CYP26, and 2) the metabolic fate of these oxidizedderivatives. In this report, we focused our investigation on the glucuronidation by UGTs of oxidized retinoids in comparison to atRA.
Several commercially unavailable compounds needed for these studies, such as 4-OH-RA, 4-OH-RAc, and 5,6-epoxy-RA, were synthesized for use as substrates for human UGTs. Metabolites such as these are involved in mediating atRA function. For example, 4-oxo-RA binds to retinoid receptors with an affinity comparable to that of atRA and can regulate the expression of genes (4). 4-oxo-RA can also bind to cellular atRA binding proteins (40). All of this suggests that this derivative possesses biological activity similar to that of atRA.
Since limited information exists on the biosynthesis and biotransformation of 5,6-epoxy-RA, we were particularly interested in characterizing the metabolism of this important compound. 5,6-Epoxy-RA has been identified as a major in vivo metabolite of atRA (as well as retinol and RAc) in rodents (12,(41)(42)(43). We have recently identified 5,6-epoxy-RA as a major in vivo metabolite of 13-cis RA in skin, 2 and we have detected the glucuronide of 5,6-epoxy-RA in human bile. 2 5,6-Epoxy-RA also exhibits potent activity in various biological assays. For example, 5,6-epoxy-RA was more effective than RAc at producing potent growth effects in vitamin A-deficient rats (44). In addition, the oxirane exhibited inhibitory effects similar to those of atRA on 12-O-tetradecanoylphorbol-13-acetate-dependent tu-2 V. M. Samokyszyn, unpublished data.  mor promotion in the two-stage (initiation-promotion) mouse skin carcinogenesis assay (45). Additionally, 5,6-epoxy-RA is more potent than atRA in opposing the effects of 12-O-tetradecanoylphorbol-13-acetate on the induction of tumor promotional markers in bovine lymphocytes (31) and polymorphonuclear leukocytes (46).
These previous findings imply that the retinoid response may involve metabolic activation of atRA to the 5,6-epoxide with the oxirane representing the pharmacologically active agent. Epoxidation of these retinoids does not appear to occur through a cytochrome P450-catalyzed mechanism (47). We have unequivocally demonstrated that the 5,6-epoxides of atRA and 13-cis-RA are generated via peroxyl radical-dependent mechanisms (33, 38, 39, 48 -50). In addition, we have recently demonstrated that 5,6-epoxy-RA is generated during prostaglandin H synthase-catalyzed oxidation of atRA involving a direct oxene transfer mechanism analogous to cytochrome P450. 2 Table I shows the rates of glucuronidation and identification of the type of glucuronide formed for various retinoids. These data demonstrate that glucuronidation of atRA, 4-oxo-RA, and 5,6-epoxy RA is directed toward the carboxyl function of these substrates, whereas the 4-hydroxylated derivatives are glucuronidated almost exclusively at the 4-OH position. This gives rise to three observations. 1) 4-OH-derivatives of atRA and RAc are glucuronidated at their hydroxyl function by human liver microsomes and UGT2B7 with significantly higher activities than those observed for the formation of carboxyl-linked glucuronides of atRA and the 4-oxo-and 5,6-epoxy derivatives of atRA. A comparison of hydroxylated versus non-hydroxylated compounds suggests that the 4-hydroxy group directs glucuronidation to the hydroxyl function and that the carboxyl group is not required for glucuronidation to occur at the 4-OH position. However, the significantly lower enzymatic activity toward 4-OH-RAc indicates that a free carboxyl function is required for optimal glucuronidation. 2) atRA, 5,6-epoxy RA, and 4-oxo-RA are glucuronidated exclusively at the carboxyl function. 3) Although the glucuronidation of 4-OH substrates by UGT2B7 is similar to that found in microsomes, the significantly lower activity of the recombinant protein toward 4-oxo-RA and 5,6-epoxy-RA might suggest the involvement of other UGT isoforms in the glucuronidation of these retinoids. Other isomeric forms of atRA, such as 9-cis-and 13-cis-RA, were also actively glucuronidated by human liver microsomes. Of the three compounds, atRA, 9-cis-RA, and 13-cis-RA, the 9-cis derivative is the best substrate for microsomal glucuronidation at the carboxyl position and, in comparison to the other retinoids studied, is second only to 5,6-epoxy-RA.
The K m for the formation of the carboxyl-linked glucuronide of atRA is in the low micromolar range (1.3-1.5 M) both for HLM15 and recombinant UGT2B7. The maximal catalytic rates (V max ) for atRA, as determined with HLM15 and recombinant UGT2B7, are 764 and 523 pmol glucuronidated ϫ min Ϫ1 ϫ mg protein Ϫ1 , respectively. These correspond to catalytic efficiencies (V max /K m ) of 509 l x min Ϫ1 ϫ mg -1 for HLM15 and 402 l ϫ min Ϫ1 ϫ mg -1 for recombinant UGT2B7, revealing significant efficiency of formation of atRA carboxyl-linked glucuronide. The K m values for 4-OH-directed glucuronidation are 273 and 221 M for HLM15 and recombinant UGT2B7, respectively. Typical V max values were determined to be in the low nanomolar range (2176 and 1709 pmol ϫ mg Ϫ1 ϫ min Ϫ1 for HLM15 and recombinant 2B7, respectively), leading to a much lower efficiency of formation for the hydroxyl-linked glucuronides, as shown by a V max /K m of 8 l x min Ϫ1 ϫ mg -1 for both HLM15 and recombinant UGT2B7. In general, the presence of the hydroxyl group in the retinoid moiety switches the site of glucuronidation from carboxyl to hydroxyl and reverses the affinity of the UGT(s) involved. The corresponding catalytic efficiencies (V max /K m ) were several hundredfold higher for glucuronidation of the carboxyl function of atRA than for glucuronidation of the 4-OH moiety of the hydroxylated retinoid.
Of the UGT isoforms investigated to date, human recombinant UGT2B7 has the highest capacity to glucuronidate atRA and 4-OH-RA. UGT2B7 activities toward atRA and 4-OH-RA are similar to the reported activities in human liver microsomes, suggesting that UGT2B7 plays a key role in metabolizing atRA and 4-OH-RA to the carboxyl-linked RAG and the hydroxyl-linked 4-OH-RAG. UGT2B7 is capable of catalyzing the biosynthesis of the hydroxyl-linked glucuronide when 4-OH-RA is the substrate or the carboxyl-linked glucuronide when atRA is the substrate. Recent studies on glucuronidation of steroid hormones and fatty acids by UGT2B7 have shown that this isoform is actively involved in the formation of both hydroxyl-and carboxyl-linked glucuronides of those lipophilic substrates (30). Taken collectively, retinoids, steroid hormones, and fatty acids are important ligands involved in initiating cellular signaling events. We postulate that UGT2B7 may be involved in controlling intracellular levels of ligands, such as steroids and atRA. If this is the case, it may also be involved in a feedback loop that controls the amounts of ligands available for steroid and retinoid receptors.
In summary, we speculate that atRA induces its own oxidative metabolism via a cytochrome P450, CYP26, mechanism followed by a UGT-dependent mechanism. The hydroxyl-linked glucuronide of 4-OH-RA is the directed product of atRA metabolism by CYP26. Thus, 4-OH glucuronidation of 4-OH-RA terminates the biological activity of atRA, while the carboxyllinked glucuronide of atRA might be a biologically active compound involved in cellular processes. Thus, CYP26 and UGT2B7 may together play a crucial role in the metabolism and biological fate of atRA.