cDNA Cloning of Human Retinoic Acid-metabolizing Enzyme (hP450RAI) Identifies a Novel Family of Cytochromes P450 (CYP26)*

Retinoids, including all-trans-retinoic acid (RA) and its stereoisomer 9-cis-RA play important roles in regulating gene expression, through interactions with nuclear receptors, during embryonic development and in the maintenance of adult epithelial tissues (Chambon, P. (1995) Rec. Prog. Horm. Res. 50, 317–32; Mangelsdorf, D. J., and Evans, R. M. (1995)Cell 83, 841–850; Petkovich, M. (1992) Annu. Rev. Nutr. 12, 443–471). Evidence suggests that 4-hydroxylation of RA inside the target cell limits its biological activity and initiates a degradative process of RA leading to its eventual elimination. However, 18-hydroxylation and glucuronidation may also be important steps in this process. In this paper, we describe the cloning and characterization of the first mammalian retinoic acid-inducible retinoic acid-metabolizing cytochrome P450 (hP450RAI), which belongs to a novel class of cytochromes (CYP26). We demonstrate that hP450RAI is responsible for generation of several hydroxylated forms of RA, including 4-OH-RA, 4-oxo-RA, and 18-OH-RA. We also show that hP450RAI mRNA expression is highly induced by RA in certain human tumor cell lines and further show that RA-inducible RA metabolism may correlate with P450RAI expression. We conclude that this enzyme plays a key role in RA metabolism, functioning in a feedback loop where RA levels are controlled in an autoregulatory manner.

retinaldehyde, and its catabolism to more polar hydroxylated forms such as 4-OH-RA, 4-oxo-RA, and 18-OH-RA are counterbalanced metabolic pathways that regulate RA levels in RAsensitive tissues (4,5). Cellular retinoic acid-binding proteins may also play a role in establishing this balance by sequestering high levels of RA (6). There is considerable evidence to suggest that 4-OH-, 4-oxo-, and 18-OH-RA are polar intermediates in the catabolism and eventual elimination of RA (5,7,8). Thus both sequestration and metabolism may function to protect RA-sensitive tissues from deleterious concentrations of RA.
We have cloned and characterized cDNAs corresponding to a retinoic acid-inducible gene encoding a human cytochrome P450-related hydroxylase (P450RAI) responsible for generation of multiple hydroxylated products of RA. hP450RAI appears to be the human ortholog of the previously characterized zebrafish P450RAI (zP450RAI) (9), indicating that this important cytochrome is highly conserved structurally and functionally across species. We also demonstrate that hP450RAI is inducible by RA in a number of different cell types. We speculate that this enzyme plays a key role in determining the metabolic fate of endogenous retinoids and may also be implicated in the clearance of exogenous retinoids administered therapeutically.

MATERIALS AND METHODS
cDNA Library Screening-A NTERA2-D1 cDNA library (Stratagene) was screened according to the manufacturer's directions. Briefly, 1.0 ϫ 10 Ϫ6 independent plaques were screened using a random-primed, ␣-[ 32 P]dATP-labeled full-length zP450RAI cDNA. Filters were prehybridized for 4 h at 37°C in 50% formamide, 5 ϫ SSPE, 1 ϫ Denhardt's (without bovine serum albumin), 0.2 mg/ml denatured salmon sperm DNA. Hybridization was performed overnight at 37°C. Filter were washed two times for 20 min in 2 ϫ SSC, 0.05% SDS at room temperature followed by one 10-min wash in 1 ϫ SSC, 0.1% SDS, and then exposed to Kodak XAR film overnight at Ϫ70°C. Positive plaques were rescreened until purified. pBluescript-containing colonies, generated using the in vivo excision protocol (Stratagene) were plated onto LB ϩ ampicillin plates and grown overnight at 37°C. Plasmid DNA was purified using the Qiaprep Mini-Plasmid prep kit (Qiagen) and sequenced using the T7 sequencing kit (Pharmacia Biotech Inc.). Sequence data analyses were performed with the Geneworks software package (Intelligenetics).
Analysis of RA Metabolism in Cultured Cells-Cells were transfected with 3 g of hP450RAI in pTL1 or the empty vector control pTL1, together with 1 g each of ferridoxin and ferridoxin reductase expression vectors (10). Media from transfected cells incubated with 575 pM [11, H]RA (Fig. 2, A and B) or 1 M RA (Fig. 2C) for 24 h were acidified with 0.1% acetic acid. All incubations were performed in the presence of fetal calf serum. For the analysis of RA metabolism, untransfected MCF7 (Fig. 3C) and MCF10A cells (Fig. 3B) were incubated in the presence of 10 Ϫ6 M RA or vehicle (0.1% ethanol) for 24 h, washed with phosphate-buffered saline containing 1% bovine serum albumin five times to remove as much RA as possible, and then analyzed for their ability to metabolize [11,12-3H]RA as described below, except that cells were incubated with [11, H]RA for only 4 h. Lipid-soluble metabolites were separated from aqueous-soluble metabolites using a total lipid extraction of the medium (11). Metabolism of [11, H]RA to total aqueous-soluble metabolites was measured using ␤-scintillation of aliquots of the aqueous phase of media extracts (Figs. 2B and 3D). Lipid-soluble extracts were evaporated to dryness under a stream of * This work was supported by grants from the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF005418.

RESULTS AND DISCUSSION
Cloning Human P450RAI-We probed a panel of mRNAs from human cell lines with cDNA from zebrafish cytochrome P450 (zP450RAI) previously shown to be involved in RA-inducible RA metabolism (9). Strong cross-hybridization was observed with RNA from RA-treated teratocarcinoma cells (NTERA2-D1) (13) (data not shown). A cDNA library from RA-treated NTERA2-D1 (Stratagene) was screened with the zP450RAI cDNA, and a full-length human cDNA was isolated. A high degree of amino acid identity (68%) was observed between hP450RAI and zP450RAI (Fig. 1A). We also have recently cloned the murine P450RAI, which is 89% conserved with the human counterpart (data not shown). In contrast, there was less than 30% amino acid identity between the P450RAIs and other cytochromes listed in GenBank™, suggesting that zP450RAI and hP450RAI comprise a new class of cytochromes designated CYP26. We have identified five conserved regions (CR1-CR4, Fig. 1A), with amino acid identity  1. Amino acid alignment between human P450RAI and zebrafish P450RAI. A, comparisons between the human and zebrafish P450RAIs indicate that these proteins are highly conserved across species, with zP450RAI exhibiting 68% amino acid identity to its human counterpart. B, Kyte-Doolittle hydrophobicity (14) plot comparing human and zebrafish P450RAIs. The profiles are almost identical except for a short nonconserved region between CR3 and CR4. greater than 80%; and the heme binding motif located at the C-terminal end of all cytochrome P450s. The  hydrophobicity plots for hP450RAI and zP450RAI are essentially superimposable (Fig. 1B) with the exception of one short region located in the middle of the protein between CR3 and CR4, suggesting that these homologs share a similar protein structure. Both zP450RAI and hP450RAI sequences have hydrophobic regions at the N termini, characteristic of microsomally located cytochrome P450s (15). This is consistent with the microsomal location of previously characterized RA-inducible RA metabolic activities (16 -18).
Metabolic Activity of hP450RAI-To determine whether hP450RAI could metabolize RA, COS-1 cells were transiently transfected with a hP450RAI expression vector, pTL1-hP450RAI, and cell extracts were analyzed as described previously (9). HPLC analysis of the lipid-soluble extracts from the media of pTL1-hP450RAI-transfected cells indicates a P450RAI cDNA-dependent increase in metabolism of [11, H]RA and the generation of RA metabolites, including those which comigrate with 4-OH-and 4-oxo-RA standards on two HPLC systems ( Fig. 2A). Using picomolar concentrations of RA (575 pM), we found the percent reduction in retinoic acid occurring in control transfections to be maximally 29.6%, whereas in P450RAI-transfected cells it is 69.5% (based on retinoic acid recovered as a percent of total recovered radioactivity; total recoveries averaged 70 -74% of added radioactivity). This percent metabolism is equivalent to a rate of disappearance of retinoic acid of 75 fmol/g of protein/24 h in control transfected cells, compared with 158 fmol/g of protein/24 h in P450RAI transfected cells. The rate of RA metabolism due to P450RAI activity is remarkable, since in our transfection system the percentage of cells transfected is rarely greater than 10% (data not shown). The decreases in the amount of RA substrate and production of 4-OH-and 4-oxo-RA metabolites by hP450RAI is accompanied by the formation of aqueous-soluble radioactive metabolites (Fig. 2B). We calculate the rate of formation of aqueous soluble radioactivity, by analyzing the aqueous-soluble fraction isolated from the experiment in (Fig. 3A), to be 8 fmol/g protein/24 h in control as compared with 39.5 fmol/g of protein/24 h in P450RAI-transfected cells.
When cells were incubated with 1 M RA (Fig. 2C), two major peaks retaining the spectral properties of RA were generated, which comigrated with 4-OH-RA and 4-oxo-RA ( max ϭ 340 nm for 4-OH RA; max ϭ 360 nm for 4-oxo-RA), and were increased in hP450RAI-pTL1-transfected, as compared with control, cells. A third peak with spectral properties and retention time characteristic of RA is observed, but its identity has not been determined. It is likely that hP450RAI may catalyze hydroxylations other than 4-hydroxylation, resulting in production of some of these additional metabolic products. Again the substrate peak of RA is greatly reduced in hP450RAI-pTL1-transfected cells compared with control cells (Fig. 2C).
hP450RAI Expression in Human Cell Lines-We have studied the expression of hP450RAI mRNA in a variety of human cells lines by Northern blot analysis (Fig. 3A). Poly(A) ϩ mRNAs from RA-treated and control cells show several different expression profiles: constitutive expression (embryonic kidney HEK293 and small cell lung carcinoma SK-LC6 (19)), inducible expression (nonsmall cell lung carcinoma LC-T, breast adeno- carcinoma MCF7, acute promyelocytic leukemia-derived NB4 (20), and hepatocarcinoma HepG2), or a complete lack of expression (breast carcinoma-derived EL-E and nontransformed breast MCF10A). Inducible P450RAI expression was also observed in the human keratinocyte-derived cell lines HPK1A and HPK1A-ras (21) (data not shown). These findings suggest that P450RAI expression is regulated by multiple factors in a cell-specific manner. It is notable that P450RAI expression is highly inducible in the acute promyelocytic leukemia-derived cell line NB4 since 1) increased RA metabolic activity of the type demonstrated by P450RAI has been implicated in the acquired clinical resistance to RA (22), and 2) NB4 cells have been shown to have inducible RA metabolism following pretreatment with RA (23). The diversity of cell types exhibiting P450RAI expression suggests that, if P450RAI induction is indeed responsible for conferring clinical RA resistance, induced metabolism in both the leukemic cells and other tissues where P450RAI is induced would contribute collectively to lowering the therapeutic efficacy of RA.
Correlation between hP450RAI mRNA Expression and RA Metabolism-RA-inducible RA metabolism has been previously characterized in a number of cell lines and tissues, including the breast epithelial cell line MCF7 (24). To determine whether the inducibility of P450RAI expression, which we have observed in this cell line (see Fig. 3A), correlates with RA metabolic activity, MCF10A and MCF7 were incubated with or without RA for 24 h and then tested for their ability to metabolize [11, H]RA. Interestingly, in MCF10A, where there is no inducible expression of hP450RAI, there is also no detectible production of 4-oxo-or 4-OH-RA in either control or RA-treated cells (Fig. 3B). In contrast, MCF7 cells, which contain highly inducible P450RAI expression, do produce these metabolites in response to RA (Fig. 3C). Analysis of the aqueous-soluble fractions shows a 3.5-fold increase in aqueous-soluble radioactivity only in the induced MCF7 cells (Fig. 3D), which correlates with the ability of the cells to convert RA to 4-hydroxylated derivatives. 4-Hydroxylation of RA may be an obligatory step in the production of these aqueous soluble products. However, we note that the 4-OH-RA and 4-oxo-RA peaks observed in MCF7 cells are broad, suggesting that they may contain multiple metabolic products. These products may arise from alternate hydroxylations or isomers, suggesting that if P450RAI is responsible for all the metabolism in MCF7 cells that it can perform metabolic steps other than 4-hydroxylation. These experiments are consistent with the possibility that P450RAI plays a major role in RA metabolism.
In summary, we have cloned and characterized a human cytochrome P450, which is very highly conserved between zebrafish and human, constituting a novel family, CYP26, of cytochromes P450. P450RAI is able to metabolize all-trans-RA, and Northern blot analysis of its expression in a number of human cell lines shows that it can be induced in response to RA treatment, expressed constitutively, or not expressed at all, suggesting its regulation is complex and may be tissue typedependent. Several additional findings (not presented in this paper) support our hypothesis that this enzyme is specific for RA and is responsible for RA catabolism. Preliminary enzyme kinetic studies on whole cells indicate a relative K m less than 1 M in accordance with the K m of microsomal cytochrome preparations shown to hydroxylate RA in vitro (5,25) and in the same range as retinoid metabolizing activity defined in T47D cells (18). In addition, we find that 9-cis-RA is a good substrate for hP450RAI, while retinol, even at micromolar concentrations, is not. In our studies we find that expression of the P450RAI message and RA metabolic activity appear to be correlated implicating P450RAI in the observed RA metabolism.
The identification of P450RAI, its inducibility by RA, and its RA metabolic activity define a feedback loop, which may be critical in regulating both normal and therapeutic RA levels. Other autoregulatory feedback loops, such as that described for conversion of retinol to retinoic acid, contribute to controlling overall levels of active retinoid (26). This emphasizes the importance of maintaining stable physiological levels of RA. Inhibitors designed to block P450RAI function may therefore be useful in elevating normal tissue RA levels or maintaining high therapeutic levels of RA (22,27). Since RA has proven useful in the treatment of a number of cancers, premalignancies, and skin disorders, an understanding of the role of hP450RAI in regulating RA levels in normal and disease states will be important clinically.