A frame shift mutation and alternate splicing in human brain generates a functional form of the pseudogene, cytochrome P4502D7 that demethylates codeine to morphine

A frameshift mutation 138delT generates an open reading frame in the pseudogene, cytochrome P4502D7 (CYP2D7), and an alternate spliced functional transcript of CYP2D7 containing partial inclusion of intron 6 was identified in human brain but not in liver or kidney from the same individual. mRNA and protein of the brain variant CYP2D7 were detected in 6 of 12 human autopsy brains. Genotyping revealed the presence of the frameshift mutation 138delT only in those human subjects who expressed the brain variant CYP2D7. Genomic DNA analysis in normal volunteers revealed the presence of functional CYP2D7 in 4 of 8 individuals. In liver, the major organ involved in drug metabolism, a minor metabolic pathway mediated by CYP2D6 metabolizes codeine (pro-drug) to morphine (active drug), whereas norcodeine is the major metabolite. In contrast, when expressed in Neuro2a cells, brain variant CYP2D7 metabolized codeine to morphine with greater efficiency compared with the corresponding activity in cells expressing CYP2D6. Morphine binds to micro-opioid receptors in certain regions of the central nervous system, such as periaqueductal gray, and produces pain relief. The brain variant CYP2D7 and micro-opioid receptor colocalize in neurons of the periaqueductal gray area in human brain, indicating that metabolism of codeine to morphine could occur at the site of opioid action. Histio-specific isoforms of P450 generated by alternate splicing, which mediate selective metabolism of pro-drugs within tissues, particularly the brain, to generate active drugs may play an important role in drug action and provide newer insights into the genetics of metabolism.


INTRODUCTION
Cytochrome P450 (E.C. 1.14.14.1; P450) and associated monooxygenases, a family of heme proteins, are the principal class of drug metabolizing enzymes. A supergene family encodes them and the member proteins exist as multiple forms having distinct yet overlapping substrate specificities. Multiple forms of P450, which are selectively induced or inhibited by a variety of drugs, are known to exist in liver, the major organ involved in P450 mediated metabolism (1). However, the potential to generate active metabolite(s) at the site of action has generated interest in extrahepatic P450. This has prompted extensive investigations into the xenobiotic metabolizing capability of extrahepatic organs, such as lung, kidney, skin and nasal epithelium and the far-reaching consequences of such metabolism, in situ, within specific cells in target organs have been recognized in laboratory animals (2) and humans (3). The preferential localization of drug metabolizing enzymes within specific cell types in these organs renders such cells significant capability to metabolize drugs (4). Thus, even minor metabolic pathways of xenobiotic metabolism can produce major effects if they take place at the site of action. These observations have prompted investigation into P450 associated monooxygenases in brain with an effort to determine the capability of the brain to metabolize psychoactive drugs (5,6). P450 mediated metabolism of psychoactive drugs directly in the brain can lead to local pharmacological modulation at the site of action and result in variable drug response.
The inter-individual variability in hepatic metabolism of drugs due to genetic polymorphism exhibited by some forms of P450 such as P4502D6 are reflected in the plasma levels of administered drugs. But plasma drug levels often show poor correlation with therapeutic effect (7) suggesting that metabolism within the brain could influence the therapeutic outcome regardless of hepatic clearance and plasma drug levels. A moderate difference in the pharmacokinetics of psychoactive drugs often leads to dramatic pharmacodynamic effects, again suggesting that metabolism in situ within the brain could play a significant role (8).
Over the past decade studies from our laboratory and others have demonstrated the presence of a competent microsomal P450 system in the rodent (9, 2, 10) and human (11,3) brain and its ability to metabolize a variety of xenobiotics. The appearance of multiple forms of P450 in brain and their selective inducibility by a variety of drugs and xenobiotics has also been identified (9,12,13,14). CYP2D is one of the major forms of P450 present in both rat (15) and human brain (16). Significant differences are seen in the regulation and function of the isoforms of brain P450 as compared to the hepatic forms (17,18,19). For example, drugs, such as, alprazolam are metabolized differently in liver and brain wherein relatively larger amount of the active metabolite is generated in the brain compared to liver (19). These observations have indicated the possible existence of unique P450 isoforms in brain that are different from the well-characterized hepatic P450s. We report here the presence of a unique form of CYP2D7 that is generated by alternate splicing in human brain, but not in liver or kidney.

EXPERIMENTAL PROCEDURES
Human brain was obtained from Human Brain Tissue Repository, NIMHANS, India.
The tissue was collected at autopsy from male and female traffic accident victims with no known neurological or psychiatric disorders. The average age of the individuals was 35.4 years (3 to 70 years) and postmortem delay between death and autopsy was 7.3 + 3.7 hr. A region representing exon 6-9 of CYP2D6 (333 bp) was amplified by RT-PCR with forward and reverse primers 5'-TGATGAGAACCTGCGCATAG-3' and 3'-ACCGATGACAGGTTGGTGAT-5', respectively, using human cortex mRNA and used for screening a human cortex cDNA library constructed in our laboratory using Lambda ZAPII library construction kit (Stratagene, La Jolla).
Three positive clones of approximately 1.2, 1.5 and 1.8 Kb were isolated and sequenced. The 1.8 Kb clone referred to as brain variant CYP2D7 (GenBank Accession Number -AY220845) containing an open reading frame was used for further investigations.
Expression of Brain Variant CYP2D7 -In vitro translation was carried out using TNT T3-coupled reticulocyte system (Promega, Madison) and brain variant CYP2D7 cDNA. The translated products were examined by immunoblot using antibody to CYP2D6 (Gentest, San Jose; 19). Luciferase cDNA provided by the manufacturer was used as positive control. cDNA of brain variant CYP2D7 was subcloned into pcDNA3.1 (+/-), sequence verified and transfected into Neuro2a cells using FuGENE 6 (Roche Applied Sciences, Indianapolis). Positive clones were selected using neomycin. Cells were harvested and a membrane preparation containing both mitochondria and microsomes was prepared as described earlier (16). Neuro2a cells were also grown in chamber slides and processed for immunohistochemistry using antibody to CYP2D6.
Immunostaining was visualized by incubation with biotinylated secondary antibody followed by streptavidin-fluorescein. cDNA to brain variant CYP2D7 ligated in the reverse orientation was used as control in transfection experiments.
RT-PCR and Genomic DNA Amplification -RT-PCR was carried out using cDNA synthesized from total RNA of 12 autopsy human brain samples. PCR was performed using the Genomic DNA was extracted from human brain cortex or blood and PCR amplified using Preparation of Antiserum to Brain Variant CYP2D7 -An antigen derived from the 19 amino acid peptide (GRRVSPGCSPIVGTHVCPV) representing 57 bps of intron 6 of brain variant CYP2D7 was conjugated to bovine serum albumin at N-terminal end and used to generate antiserum to brain variant CYP2D7 in rabbits. The antiserum was verified for lack of cross reactivity with P4502D6 and used for immunoblotting.
Immunoblotting with Membrane Preparations from Human Brain -Brain membrane preparations containing microsomal and mitochondrial protein (200 µg each) isolated from the cortex obtained at autopsy from the 12 human subjects were analyzed by SDS-PAGE (19). The blots were incubated sequentially with antiserum to brain variant CYP2D7 (1:200 dilution) overnight at 4 o C and anti-rabbit IgG conjugated to alkaline phosphatase (1:1000 dilution) for 2 hr at room temperature. Color was developed using chromogenic substrates for the alkaline phosphatase.
Assay of P450 and Metabolism of Codeine -Brain variant CYP2D7 was expressed in Neuro2a cells as described above and the membrane preparation containing both mitochondria and microsomes was isolated. Total cytochrome P450 content in membrane preparation was measured from carbon monoxide reduced minus oxidized difference spectrum (9) Localization of Brain Variant P4502D7 and µ-Opiate Receptor in Human Brain-Serial transverse sections (10 µm thick) were cut from paraffin embedded periaqueductal gray region of a human brain and processed for immunohistochemistry as described earlier (16). Sections were incubated with antisera to brain variant CYP2D7, µ -opioid receptor (Sigma Biochemicals, St. Louis) or non-immune serum and immunostaining was detected using ABC Elite Kit (Vector Laboratories, Burlingame).

Identification of Splice Variants of CYP2D7 from a Human Brain cDNA Library-We
investigated the presence of brain-specific P450 by screening a human brain cortex cDNA library using a 333 bp amplicon generated by RT-PCR of human brain mRNA representing exons 6-9 of CYP2D6. Three positive clones were selected and DNA sequencing identified them as splice variants of CYP2D7 having exon 3 deletion, partial inclusion of intron 6 or both Expression and Functional Activity of Brain Variant CYP2D7: The clone, which we named brain variant CYP2D7 translated, in vitro, in a rabbit reticulocyte system into a 58 kD 10 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from protein, which was immunochemically similar to hepatic P4502D6 ( Fig. 2A). The brain variant CYP2D7 was transfected in Neuro2a cells and the expressed protein was assessed for P450 activity. The reduced carbon monoxide difference spectrum of the expressed protein had an absorption maximum at 450 nm characteristic of P450 ( Fig. 2B; specific content = 1.03 nmoles of P450/mg protein). The expressed protein could also be detected by immunoblotting (Fig. 2C) and localized by immunohistochemistry using antibody to P4502D6 (Fig. 2D). Neuro2a cells, per se, and those transfected with the cDNA in the reverse orientation did not have any detectable P450 as observed by immunohistochemistry (Fig. 2D) or by the reduced carbon monoxide difference spectrum (data not shown).

RT-PCR and Immunoblotting show the Presence of Brain Variant CYP2D7 in subset of
Individuals -A 340 bp fragment representing partial sequence of exons 6-7 including the additional sequence of 57 bp of intron 6 representing 843-1182 bp of brain variant CYP2D7 was amplified using RT-PCR in 12 samples of human brain cortex obtained at autopsy. The generation of the 340 bp amplicon indicated the expression of the brain variant CYP2D7.
Expression of CYP2D6 represented by the 282 bp amplicon was seen in all 12 human brain samples examined (Fig. 4A). The primers used for RT-PCR amplified both CYP2D6 and CYP2D7 because of the considerable similarity (98%) between CYP2D6 and CYP2D7 in the exon 6 region. However, the 57 bp sequence of intron 6 present in brain variant CYP2D7 shares only 77% similarity with the corresponding region in CYP2D6. Thus, the 340 bp RT-PCR product represents the brain variant CYP2D7 exclusively and was seen in 6 of the 12 samples examined.
We generated an antiserum to the 19 amino acid peptide representing 57 bp of intron 6 present in the brain variant CYP2D7. The peptide used to generate the antiserum did not share homology with any known P450 enzyme and the antiserum did not cross-react with P4502D6.
Expression of brain variant CYP2D7 protein as assessed by immunoblotting was seen in 6 of the 12 brain samples (Fig. 4B) that were positive for the presence of brain variant CYP2D7 mRNA by RT-PCR (Fig. 4A) indicating the concordance between the RT-PCR and immunoblot results.

Localization of of Brain Variant CYP2D7 by in situ Hybridization -
The 340 bp RT-PCR amplicon generated as described above was ligated into pCRII vector for riboprobe synthesis to localize the brain variant CYP2D7 in human tissues using in situ hybridization.
Brain variant CYP2D7 mRNA was detected in cortical neurons in human brain but not in liver or kidney of the same individual ( Fig. 3 A, B and C respectively). However, CYP2D6 mRNA was detected in brain, liver and kidney when in situ hybridization was performed using the cDNA to CYP2D6 (Fig. 3 D, E and F respectively).
T-Deletion in the Pseudogene, CYP2D7 generates a Functional Transcript-Since the brain variant CYP2D7 is present only in about 50% of the human brains examined, we developed a genotyping assay to detect the 138delT relative to ATG. The genomic DNA isolated from the 12 human brain samples used for RT-PCR and immunoblot experiments was used as template. The region spanning 45-551 bp (506 bp) relative to the ATG start codon of the genomic sequence of CYP2D7 was amplified using PCR. The 138delT (CCTGC) was found only in those samples in which the mRNA and protein of the brain variant CYP2D7 was detected ( Fig. 4C), others had complete sequence similarity with the pseudogene CYP2D7 (CCTTGC).
Brain variant CYP2D7 metabolizes codeine predominantly to morphine -Monooxygenase activity of brain variant CYP2D7 was ascertained by examining the metabolism of codeine to nor-codeine and morphine. Membrane preparations containing both mitochondria and microsomes were prepared from Neuro2a cells transfected with the cDNA of brain variant CYP2D7 since P450 mediated xenobiotic metabolism takes place in both microsomes and mitochondria in rat and human brain (26,27). The expressed brain variant CYP2D7 metabolized codeine (10 µM) to morphine alone and nor-codeine could not be detected (Fig. 5). In membranes from cells transfected with cDNA of liver CYP2D6, nor-codeine was the major metabolite (62%) and morphine, the minor metabolite (38%; Fig. 5B) Metabolism of codeine to morphine could be inhibited by antiserum to P4502D6 and quinidine, a selective inhibitor of CYP2D (Fig. 5C) indicating that this biotransformation was indeed mediated by a P4502D enzyme.

Brain variant CYP2D7 and µ-Opioid Receptor co-localize in neurons -In order to
determine if the metabolism of codeine to morphine takes place at the site of action of the morphine, namely the neurons in periaqueductal gray that contain µ-opiate receptors, we performed immunohistochemistry using antisera to brain variant CYP2D7 and µ-opiate receptor on serial sections of periaqueductal gray from human brain. Brain variant CYP2D7 and µ-opiate receptor co-localized in the neurons of periaqueductal gray (Fig. 6A-D).
Incidence of CYP2D7 genetic polymorphism -DNA isolated from blood samples of 8 volunteers was used a template and the region spanning 45-551 bp (506 bp) relative to the ATG start codon of the genomic sequence of CYP2D7 was amplified using PCR. The 138delT (CCTGC) was found only in 4 out of 8 samples analyzed. The others had complete sequence similarity with the pseudogene CYP2D7 (CCTTGC). The representative electropherogram of 6 samples is depicted in Fig. 7. Thus, the functional CYP2D7 is presumably present in 50% of the samples analyzed.

DISCUSSION
Several P450 enzymes, such as CYP2D (16), CYP3A (19) and CYP2B (11) are present in human brain and localize predominantly in neurons, the site of action of most drugs (19).
However, drug metabolism in human brain is poorly understood and human brain-specific P450 14 by guest on March 24, 2020 http://www.jbc.org/ Downloaded from enzymes remain to be identified. The presence of unique, tissue-specific P450 enzymes generated through alternate splicing provides a mechanism by which active metabolites can be potentially formed at the site of action of drugs within the target organ, such as the brain.
P4502D6 is an important human P450 enzyme that metabolizes a number of substrates (33). It shows high degree of inter-individual variability, which is primarily due to the extensive genetic polymorphism that influences its expression and function. P4502D6 is a constitutive form of hepatic P450 where it mediates metabolism of several commonly used psychoactive drugs, such as, imipramine, amitryptaline and chlorimipramine (34) and haloperidol (35 in the first exon causing a frame shift resulting in premature termination of translation of CYP2D7 (37). It has been speculated that CYP2D7 may be expressed as a functional protein as result of repair, however, no evidence has been presented so far (37). Here we present evidence for the 138delT mutation that converts the CYP2D7 pseudogene to a functional gene in about half the samples analyzed in the present study. The mutation 138delT was seen in DNA isolated from brain tissue as well as blood (Fig. 7) indicating that the presence of functional CYP2D7 in these individuals. It is to be determined if the functional CYP2D7 is indeed present in the liver of individuals with the 138delT mutation of the CYP2D7 gene.
In individuals having the 138delT mutation, brain-specific splicing led to the formation of a functional CYP2D7 brain variant enzyme. Thus, in 6 out of the 12 brains examined from the Indian population genotyping revealed the presence of the 138delT mutation and RT-PCR and immunoblot experiments showed the presence of the brain variant CYP2D7 mRNA and protein in these samples (Fig. 4). It is to be seen if this mutation exists in other population leading to the presence of functional CYP2D7 and if it does, it is to be determined if the autopsy brain samples from these population express the brain variant CYP2D7 that is generated by brain-specific alternate splicing.
The brain variant CYP2D7 has a partial inclusion of intron 6 (57 bp) in the transcribed mRNA sequence. This was observed only in human brain and not in liver or kidney of the same individual indicating that generation of alternate spliced form is a brain-specific event. Earlier studies have shown the presence of an alternate spliced form of flavin-containing monooxygenase (FMO4) with exon 4 deletion in rat brain but not in other tissues (38). Nervous system has a propensity for generating alternate spliced forms and splicing defects observed in individuals are not related to differences in the genomic sequence but may be regulated by mechanisms involving spliceosomal complex and RNA binding proteins, which are poorly understood (39).
The tendency for human brain to generate alternate spliced genes is seen in the present study wherein three alternate spliced variants were identified in the human brain, namely, exon 3 deletion, partial inclusion of intron 6 and a third having both the deletion and inclusion. The alternate spliced variants containing exon 3 deletion have a premature stop codon, which prevents their translation into functional gene products. Thus, estimation of P450 isoforms by examining gene expression using northern blotting, RT-PCR and in situ hybridization (16,40,41) would represent contributions from functional and non-functional genes and would potentially over-estimate the expression of a particular isoform. The well-characterized CYP2D6 exhibits genetic polymorphism and the gene is absent in the "poor metabolizer (PM)" phenotype (42). None of the individuals whose brain samples were studied lacked the CYP2D6 gene as discerned by the presence of the 282 bp amplicon, while the brain variant CYP2D7 form was expressed in 6 but not in others.
Plasma levels of morphine show poor correlation with pain relief provided by codeine (43) since the amount of morphine formed in liver through P4502D6 metabolism of codeine is insufficient to account for the analgesic effect of codeine (44). It has been speculated that human brain specific metabolic pathways, which can metabolize codeine to morphine exist that are yet to be identified (45,46). Thus, even if a very small amount of codeine is metabolized to morphine by brain P450 at the site of action, it could mediate pain relief (47). The results presented herein provide evidence for the formation of morphine from codeine by brain-specific alternate spliced gene product leading to metabolite(s) profile that is different from liver.
In cells transfected with CYP2D6, nor-codeine was the major metabolite and morphine, the minor metabolite. Earlier studies had indicated that the metabolism of codeine to norcodeine is mediated by P4503A4 (48) however, in the present study it is seen that P4502D6 also metabolizes codeine principally to nor-codeine. However, brain variant CYP2D7 metabolizes codeine largely to morphine and nor-codeine is formed in lesser amounts and only at higher substrate concentration. Morphine is exclusively formed at lower concentration of codeine (10µM) which is closer the physiological concentration achieved after administration of a pharmacological dose of codeine (28).
For effective pain relief it would be ideal if morphine were formed in the brain where it can directly bind to µ-opioid receptors. Opiates mediate their central analgesic effects by acting on neurons within brain regions such as the midbrain periaqueductal gray (49). We localized brain variant P4502D7 and µ-opiate receptor by immunohistochemistry in the periaqueductal gray region of a human brain obtained from at autopsy. The brain variant CYP2D7 and µ-opiate receptor are co-localized in the neurons of periaqueductal gray indicating that morphine could potentially be formed from codeine in pain centers of the CNS. Biochemical estimation of enzyme activity represents the average in a tissue containing several distinct cell types. In heterogeneous tissues, such as the brain low levels of enzyme activity may be observed.
However, since enzymes such as P450 are localized in selective cell population ( Fig. 2D and 7),    Fig. 3A and B). In 1 and 2 brain variant CYP2D7 is not expressed and the genomic sequence revealed presence of the pseudogene, CYP2D7 as represented by the sequence CCTTGC. In 3 and 4 brain variant CYP2D7 is expressed and the genomic sequence CCTGC was seen.