Genomic cloning and protein expression of a novel rat brain cytochrome P-450 CYP2D18* catalyzing imipramine N-demethylation.

We have previously reported the isolation of two cDNA clones, designated 2d-29 and 2d-35, which have identical open reading frames and code for a novel brain cytochrome P-450 (P-450) belonging to the CYP2D subfamily, and noted that the mRNA of clone 2d-35 seems to be expressed in the brain but not in the liver (1). Although the deduced amino acid sequence of these clones differs from that of the liver CYP2D4 by only 5 amino acids distributed in the C-terminal region, this new P-450 cDNA clone contained a unique 5′-extension, and we posit in this report by analysis of a genomic clone that this 5′-untranslated sequence is derived from a gene distinct from that of CYP2D4. Thus, this novel P-450 was named P-450 2D18 according to the recommended nomenclature (2). The expressibility of this cDNA was confirmed by in vitro translation using a reticulocyte system, and protein expression was performed using COS-M6 cells. Immunoblot analysis showed a cross-reacting band of the predicted size range with anti-P-450 2D6 antiserum, which was not seen in control cells. Furthermore, the CYP2D18-expressed COS cell lysate showed N-demethylation activity toward imipramine, whereas another brain P-450 CYP4F6-expressed COS cell lysate showed 10-hydroxylation activity. This is the first report that associates an individual P-450 isozyme in brain with a particular metabolic alteration of the antidepressant imipramine.

Cytochrome P-450 1 is a general term for heme-thiolate enzymes, which exist in large quantities primarily in the liver and catalyze the metabolism of xenobiotics, carcinogens, steroids, and fatty acids. Among the P-450 superfamily members, much attention is paid to subfamily 2D P-450, which shows polymorphisms (3)(4)(5), because possible relationships between members of this subfamily of isozymes and disorders of the central nervous system have been suggested (5,6). For instance, it has been shown that the polymorphism in the 4-hydroxylation of debrisoquine, a substrate of CYP2D, 2 is associated with susceptibility to Parkinson's disease (5) and that 4-hydroxylation of 1,2,3,4-tetrahydroisoquinoline, a possible candidate for causing Parkinson's disease, is catalyzed by subfamily 2D P-450 in rat liver (7). It was also demonstrated that the human isoform CYP2D6 participates in the metabolism of imipramine, a commonly used antidepressant (8).
However, the vast majority of these studies were performed using liver microsomes or liver isoforms of P-450. Thus, whether subfamily 2D P-450 catalyzes these metabolisms in brain is not clear. In our laboratory, we have been interested in the brain P-450 monooxygenase system, and our studies have revealed evidence for the existence and functionality of multiple forms of P-450 (9 -13). Recently, Sequeira and Strobel demonstrated brain microsomal catalytic activities toward imipramine and showed differences in the effects of the inducers on liver and brain imipramine metabolism (14). Thereafter, we cloned from a rat brain cDNA library two cDNA clones, termed 2d-29 and 2d-35, which have an identical open reading frame and code for a novel brain P-450 belonging to the CYP2D subfamily. The difference between the two clones is that cDNA clone 2d-29 has a foreshortened 3Ј-untranslated sequence with a poly(A) tail. We further showed that the mRNA of the longer clone, 2d-35, was expressed in the brain but not in the liver (1). Although they code for a P-450 protein, which differs from CYP2D4 (15,16) by only 5 amino acids in the C-terminal region, clones 2d-29 and 2d-35 have unique 5Ј-untranslated sequences.
In order to investigate whether or not this brain CYP2D is a distinct form, we screened a rat genomic library resulting in the demonstration shown in this report that the 5Ј-untranslated sequence of this novel brain P-450 2D is derived from a gene distinct from CYP2D4, suggesting that the expression of this P-450 is regulated differently. Thus, this P-450 was named CYP2D18 according to the recommended nomenclature (2). Subsequently, we expressed the protein product of CYP2D18 using COS-M6 cells and performed an analysis of the catalytic capabilities of the expressed recombinant protein.
T3-coupled reticulocyte lysate system was from Promega (Madison, WI); DNA labeling kit (strand-specific DNA probe synthesis kit) and S1 nuclease protection assay kit were from Ambion (Austin, TX). All restriction enzymes were purchased from Stratagene (La Jolla, CA) or Promega. Imipramine hydrochloride was purchased from Research Biochemicals International (Natick, MA); desipramine hydrochloride was from Sigma. 2-hydroxyimipramine and 10-hydroxyimipramine were gifts from Ciba-Geigy (Basel, Switzerland). Anti-P-450 2D6 polyclonal antibody was purchased from Gentest (Woburn, MA). The expression vector pcDNAI and COS-M6 cells were generous gifts from Dr. Tetsu Kamitani (University of Texas Medical School).
Screening of Rat Genomic Library--A probe for screening the rat genomic library was prepared by PCR using P-450 2D18 cDNA as a template with a pair of primers as follows; 5Ј-AGTGGATCCTCCTCT-GAGTTT, designated 29S1, and 5Ј-AAAGCCCGACTGGTCATTGAA, designated 29A2. As shown in Fig. 1C, the PCR product covered the N-terminal coding sequence and the unique 5Ј-untranslated sequence. The probe was radiolabeled with [␣-32 P]dCTP using the random primer method, and hybridization was carried out at 65°C in 50 mM Tris buffer (pH 7.5) containing 1 M NaCl, 10 mM EDTA, 0.1% (w/v) sodium Nlauroylsarcosinate, 0.2% (w/v) polyvinylpyrrolidone, 0.2% (w/v) Ficoll, 0.2% (w/v) bovine serum albumin, and 100 g/ml salmon sperm DNA. The filters were washed twice at 65°C for 30 min in 0.2 ϫ SSC (1 ϫ SSC ϭ 0.15 M NaCl and 0.015 M sodium citrate) and 0.1% (w/v) SDS. The positive plaques were purified through one more round of screening.
Southern Blot Analysis--Phage DNA of positive plaques was digested with SacI (the site for which is located just outside of the cloning site of the DNA insert to EMBL3 SP6/T7), electrophoresed on 0.6% agarose gel, transferred to a Nytran filter, and hybridized with the same probe under the same conditions as used in screening. The filter was washed twice at 65°C for 30 min in 0.2 ϫ SSC and 0.1% (w/v) SDS solution. In order to detect the unique 5Ј-untranslated sequence of CYP2D18 in genomic DNA, the same filter was prepared and hybridized at 42°C with 32 P-labeled oligonucleotide, 5Ј-AGGCTCCAGACT-TCTCGACTT, which is designated 29S2 and located in the unique 5Ј-untranslated sequence of CYP2D18. The filter was then washed in 6 ϫ SSC and 0.1% sodium dodecylsulfate solution for 40 min with a gradual increase of temperature from 42 to 56°C.
DNA Sequence Analysis--Restriction enzyme fragments of phage DNA were subcloned into pBluescript SK Ϫ , and the sequence analysis was performed using deletion clones prepared using Exo nuclease III and Mung bean nuclease (Stratagene). Sequence reactions were performed using Taq FS dye primer cycle sequence system (Applied Biosystems) and analyzed by an automated sequencer model 377 (Applied Biosystems). All the sequencings were performed in both directions.
RNA Mapping with S1 Nuclease--A strand-specific DNA probe internally labeled with [␣-32 P]dATP was synthesized using an oligonucleotide, 5Ј-AAGTCGAGAAGTCTGGAGCCT, as a primer which is an antisense sequence located in the 5Ј-untranslated sequence of CYP2D18. The template used was a linearized deletion clone of the EcoRI fragment of the CYP2D18 gene, which generated a 0.4-kb DNA probe. The strand-specific DNA probe was then hybridized with rat brain mRNA and total RNA, rat liver mRNA and total RNA, and yeast RNA as a control in a hybridization solution containing 80% deionized formamide, 100 mM sodium citrate (pH 6.4), 300 mM sodium acetate (pH 6.4), and 1 mM EDTA at 42°C overnight, digested with 15 units of S1 nuclease at 37°C for 30 min, and analyzed by electrophoresis on 6% polyacrylamide/8 M urea gel followed by autoradiography.
In Vitro Translation--In vitro translation was carried out using the TNT T3-coupled reticulocyte system (Promega) in accordance with the manufacturer's instructions. The cDNA of CYP2D18 cloned into pBluescript SK Ϫ was used without modification as a template. The 35 Slabeled translation products were analyzed by 10% SDS-PAGE followed by autoradiography.
Expression of CYP2D18 Protein Product in COS-M6 Cells--The CYP2D18 cDNA insert was subcloned into a transient mammalian expression vector pcDNAI between the EcoRI and XhoI sites of the vector without further modification. Transfection of the host cells was performed by the DEAE-dextran method (17). The cells from 4 culture dishes (100 mm) were harvested 72 h after transfection, suspended in 150 l of lysate buffer (20 mM potassium phosphate (pH 7.4), 1 mM EDTA, 0.25 M sucrose, 0.1 g/ml leupeptin, 0.04 units/ml aprotinin), and sonicated three times for 5 s.
Analytical Methods and Activity Assay--SDS-PAGE was performed by the method of Laemmli (18), and immunoblot analysis was carried out according to the method of Towbin et al. (19) with a slight modification. The metabolism of imipramine was determined as described previously (14). In short, the reaction was carried out in 100 mM KH 2 PO 4 (pH 7.25) using 1 mM imipramine, 1.4 mg of cell lysate, 4 units/ml NADPH-cytochrome P-450 reductase and an NADPH-generating system containing 5 mM glucose 6-phosphate, 0.5 mM NADP ϩ , and glucose 6-phosphate dehydrogenase in a total volume of 250 l. For extraction of N-demethylation products, the reaction mixture was alkalinized to pH 12 and extracted using 3 volumes of heptane containing 1.5% isoamyl alcohol.
To the heptane extract, 0.5 volume of 0.2 M phosphate buffer (pH 5.9) was added and mixed, and the aqueous layer was separated. The aqueous layer was again extracted with heptane as above. For hydroxylation products, the aqueous layer from the first extraction with heptane was collected, acidified to pH 9.0, and extracted with ethylene dichloride. Finally, the extracts were evaporated to dryness, and the residues were dissolved in 0.1 ml of methanol and subjected to reverse phase high performance liquid chromatography using a 25 cm ϫ 4.6 mm cyanopropylmethylsilyl column (Supelco, Bellefonte, PA) with continuous monitoring at 214 nm. The mobile phase was composed of acetonitrile/methanol/0.01 M K 2 HPO 4 , pH 7.0 (40:30:30, v/v/v), and the flow rate was 1.4 ml/min. The elution times for N-demethylated metabolite (desipramine), substrate, and 10-hydroxyimipramine were 14.1, 8.9, and 7.9 min., respectively.

Isolation of the CYP2D18 Gene and Southern Blot
Analysis-In order to determine whether our brain subfamily 2D P-450 is derived from a gene distinct from liver CYP2D4, 1,800,000 plaques from a rat genomic library were screened with a 503-bp PCR product designated 29S1A2 as a probe. The PCR product probe covers the 5Ј-untranslated sequence and the N-terminal region of the novel clone (Fig. 1C). After two rounds of screening, 6 positive clones were obtained. Fig. 1 shows the Southern blot analysis of these 6 positive clones hybridized with the common probe, 29S1A2 (Fig. 1A) and a CYP2D18-specific oligonucleotide, 29S2 (Fig. 1B). In lanes 2 and 6 of Fig. 1 (A and B), bands of approximately 3.5 kb that hybridized with both the common probe 29S1A2 and the CYP2D18-specific oligonucleotide probe were observed, whereas in the other lanes, only hybridization bands with the common probe were detected. The clones that hybridized only with the common probe probably reflect the CYP2D4 gene or another CYP2D gene of high similarity with 2D4. These results reveal that the unique 5Ј-untranslated sequence of the novel brain P-450 is derived from a gene distinct from CYP2D4. Thus, this P-450 was named CYP2D18 according to the recommended nomenclature (2).
Sequence Analysis of the 5Ј-Upstream Region of the CYP2D18 Gene--The nucleotide sequence of the 5Ј-upstream region of CYP2D18 gene is shown in Fig. 2. The putative transcription start site, indicated as ϩ1 in Fig. 2, was determined by S1 mapping analysis (Fig. 3). The 5Ј-flanking sequence of CYP2D18 is quite different from that of CYP2D4, which was reported to have "a slightly atypical TATA sequence and no CCAAT sequence" (16). The 5Ј-flanking sequence of CYP2D18 has no putative TATA sequence nor CAAT box close to the transcription start site, and it has typical TATA-less features such as having a very GC-rich sequence close to the transcription start site and other putative transcription start sites shown by weaker protection bands seen in Fig. 3B. Two glucocorticoid responsive element "half-sites" (TGTTCT) (20) were found separated by 200 bp at nucleotides Ϫ564 and Ϫ766, although further study is needed to show whether these sequences are related to the regulation of CYP2D18 expression or not.
RNA Mapping with S1 Nuclease--A strand-specific DNA probe was synthesized using an antisense sequence within the unique 5Ј-untranslated sequence of CYP2D18 as a primer. The template used was a deletion clone of the EcoRI fragment of the CYP2D18 gene generating a 0.4-kb DNA probe (Fig. 3A). As shown in Fig. 3B, the lanes containing brain mRNA, liver mRNA, and liver total RNA showed protection bands of 128 bp in size. Weaker and broader bands were also detected 4 -5 bp above the 128-bp bands. Due to a low expression level of 2D18 in brain as previously shown in Northern blot analysis (1), we used 5 g of mRNA and 50 g of total RNA for hybridization and a longer exposure time for autoradiography to obtain visible protection bands. The observation that the intensity of bands in the lanes containing liver samples is the same or rather weaker than that of brain mRNA suggests the existence of CYP2D18 in liver at a low expression level.
In Vitro Translation of CYP2D18 cDNA Clones-Prior to expressing the recombinant protein product in mammalian cells, in vitro transcription/translation reactions were performed using the TNT T3-coupled reticulocyte system with CYP2D18 cDNA, 2d-29, and 2d-35, as templates without further modification. The data of Fig. 4A show bands of the predicted size range (50 kDa) on SDS-PAGE in both lanes containing cDNA 2d-29 and 2d-35.
Immunoblot Analysis of CYP2D18-expressed COS-M6 Cell Lysate-In order to show that the protein product expressed in COS cells is a member of the P-450 2D subfamily, immunoblot analysis was performed using anti-P-450 2D6 (21) polyclonal antibody. Transfection of COS-M6 cells was performed with the expression vector pcDNAI carrying the CYP2D18 cDNA insert. Cells transfected with expression vector only were prepared as a control, and cells expressing CYP4F4 and CYP4F6, also isolated from the brain cDNA library (22), were likewise prepared for comparison. As shown in Fig. 4B, protein immunoblot analysis of the P-450 2D18-expressed cell lysate with anti-P-450 2D6 (21) antibody gave a cross-reacting band with a molecular mass of 50 kDa, which was identical to the size shown in the in vitro translation study, whereas neither control cells nor CYP4F4-expressed cells showed reaction bands with anti-P-450 2D6 antibody.
Catalytic Activity of P-450 2D18 toward Imipramine Metabolism-The catalytic activity toward imipramine (14) was determined using control cells and CYP2D18-and CYP4F6-expressed cell lysates as described above. As shown in Fig. 5B, CYP2D18-expressed cell lysate showed significant catalytic activity toward imipramine N-demethylation. On the other hand, we were able to detect significant imipramine 10-hydroxylation activity using a CYP4F6-expressed cell lysate (Fig. 5C). The calculated catalytic activities obtained by subtracting the background values of control COS cells from the values of CYP2D18-or CYP4F6-expressed COS cell lysate were 2.1 pmol/ min/mg protein for N-demethylation by CYP2D-expressed COS cells and 2.7 pmol/min/mg protein for 10-hydroxylation by CYP4F6-expressed COS cell lysate. These data show a metabolite specificity of the two isoforms resulting in either N-demethylation or hydroxylation.

DISCUSSION
Because P-450 isoenzymes are intensively studied in liver, antibodies and protein or nucleotide sequence data from liver FIG. 1. Southern blot analysis of positive clones (A and B) and schematic sequences of CYP2D18 cDNA and CYP2D4 gene (C). 1,800,000 plaques from a rat genomic library were screened, and 6 positive clones were obtained. 1.5 g of each phage DNA was digested with SacI, electrophoresed on 0.6% agarose gel, transferred to a Nytran filter, and hybridized with the common probe termed 29S1A2 (A) and a CYP2D18-specific probe termed 29S2 (B). Lanes 1-6 show positive clones 1-6, respectively. C shows diagrams of CYP2D18 cDNA and the CYP2D4 gene. The diagram of CYP2D4 gene was made according to the sequence reported by Matsunaga et al. (16). The small arrows indicate each primer. Open boxes show exons in the CYP2D4 gene. Each SacI site is also shown.
FIG. 2. Nucleotide sequence of the 5-upstream region of the CYP2D18 gene. The sequence underlined represents the longest cDNA clone sequence identified containing the CYP2D18-specific 5Ј-untranslated sequence. The putative transcription start site indicated as ϩ1 was determined by S1 mapping using rat brain mRNA.
isoforms are usually employed to investigate the P-450 monooxygenase systems in extrahepatic tissues such as brain, in which P-450 content and mRNA expression level seem to be lower. In our laboratory we have cloned from a rat brain cDNA library new P-450 isozymes including CYP4F4, 4F5, 4F6 (22), and CYP2D18 (1). Among them, the family 2D P-450 is a point of focus due to a possible correlation between this subfamily of isoenzymes and disorders of the central nervous system (5-7).
Komori has reported the isolation from a rat brain cDNA library of a partial cDNA clone of P-450 2D4 (23). Wyss et al. have recently published that they have cloned CYP2D4 cDNA from rat brain mRNA by reverse transcription PCR and that they have detected CYP2D4 protein in a brain P-450 fraction by immunoblot analysis (24). In our hands, all the positive clones sequenced during the cloning of CYP2D18 cDNA from a rat brain library had the 2D18-specific 5Ј-extension except one partial clone without an N-terminal sequence, suggesting that reliable distinction between CYP2D4 and 2D18 is best obtained using the 5Ј-or 3Ј-untranslated region unique sequence. Thus, the weight of the evidence from isolation of these clones reflects the expression of CYP2D18 in brain. Furthermore, our Northern blot analysis showed that a strong hybridization band in the lane containing the liver sample was shorter in size than a weak hybridization band in brain (data not shown), indicating FIG. 3. RNA mapping with S1 nuclease. A, a strand-specific DNA probe internally labeled with [␣-32 P]dATP was synthesized using a oligonucleotide bearing an antisense sequence in the 5Ј-unique extension of CYP2D18 cDNA as a primer and a deletion clone of CYP2D18 gene as a template, generating a 0.4-kb probe. B, the probe was then hybridized with rat brain mRNA and total RNA, and rat liver mRNA and total RNA, followed by S1 nuclease digestion. The protected bands of 128 bp in size are indicated by the arrow. the major expression of CYP2D4 or other members of the CYP2D subfamily in liver, because 2D18 cDNA is longer by a 0.15-kb unique extension than those forms as reported by Mastunaga et al. (16). However, as shown in the S1 mapping results, CYP2D18 is expressed in brain and at a very low level in liver as well. Also the results of our Northern blot analysis are quite consistent with those published by Wyss et al. (24) in which two hybridization bands corresponding in size to cDNA 2d-29 and 2d-35 were shown in lanes containing brain mRNA.
The sequence of the 5Ј-flanking region and exon 1 as shown in Fig. 2 demonstrates that the 5Ј-unique extension of CYP2D18 is derived from a gene distinct from the CYP2D4 gene whose sequence was reported by Matsunaga et al. (16). That the Matsunaga sequence for CYP2D4 is correct was confirmed by a partial cDNA sequence for CYP2D4 reported earlier by Ishida et al. (15). On the other hand, our previous report (1) demonstrated that two cDNA clones of CYP2D18 (i.e. 2d-29 and 2d-35) have identical coding sequences that are distinct and separate from that of CYP2D4. These points confirm the sequence differences between CYP2D4 and CYP2D18. If those differences were those of allelic variants, it would not be very likely that the two genes would have such different 5Ј-upstream sequences.
In order to investigate the differential expression of CYP2D18 and 2D4, another S1 nuclease protection assay using a probe bearing a unique 5Ј-extension of 2D18 and a common sequence was performed, demonstrating the existence of 2D18 both in liver and brain. On the other hand, it seems that the expression level of CYP2D4 was too low to detect the protection bands in either brain or liver (data not shown). These data are consistent with the low expression levels of CYP2D4 in liver reported by Matsunaga et al. (16).
CYP2D4 was first cloned as a partial cDNA from rat liver (15), and the intron-exon structure was determined by the cloning of genomic DNA (16). Neither the purification of CYP2D4 protein nor a full functional study of this form has been performed previously. Therefore, further studies on the characterization of this particular form of P-450 would seem in order to define further the role of 2D4 in liver.
Recently, several studies have defined the aromatase cytochrome P-450 gene structure as consisting of nine exons, II-X, encoding the protein sequence and multiple exons I containing the 5Ј-untranslated sequences (25)(26)(27)(28). Each exon I with a putative promoter region is transcribed in a tissue-specific manner, and alternative splicing results in forming mRNA with the same protein encoding sequence. A possible explanation of the CYP2D18 unique 5Ј-untranslated sequence is that this unique 5Ј-extension exists upstream of the reported 2D4 gene (16) and is spliced to the 5Ј-end of the conserved protein encoding sequence. However, the alternative splicing hypothesis cannot explain the nucleotide differences in the C-terminal region.
A CYP2D18-protein product expressed in COS-M6 cells showed both a cross-reacting band with anti-P-450 2D6 antibody and a significant catalytic activity toward imipramine N-demethylation. Inhibition studies of brain microsomal catalytic activity toward imipramine metabolism using quinidine, a subfamily 2D P-450 inhibitor, provide evidence that both hydroxylation and N-demethylation can be mediated by subfamily 2D P-450 or some other quinidine-inhibitable enzyme (29). In order to elucidate these discrepancies, we are now expressing a protein product in Escherichia coli for large scale purification of this unique form of brain P-450.
In summary, we have demonstrated that CYP2D18 cloned from a brain library is a unique member of the 2D P-450 subfamily expressed in rat brain and is a form distinct from CYP2D4 by genomic cloning. Moreover our data provide a basis for an explanation of the relation between gene structure and regulation of CYP2D18 in relation to CYP2D4. It is possible that CYP2D18 and CYP2D4 may have a tissue-specific distribution. The CYP2D18 protein product expressed in COS-M6 cells cross-reacts with anti-P-450 2D6 antibody, demonstrating its membership in the 2D subfamily, and shows a significant catalytic activity toward imipramine N-demethylation. This report constitutes the first direct evidence that a novel specific form of P-450 participates in the metabolism of tricyclic antidepressants in brain.