Molecular Psychiatry (2004) 9, 151–160 & 2004 Nature Publishing Group All rights reserved 1359-4184/04 $25.00 www.nature.com/mp ORIGINAL RESEARCH ARTICLE Human catechol O-methyltransferase genetic variation: gene resequencing and functional characterization of variant allozymes AJ Shield1, BA Thomae1, BW Eckloff2, ED Wieben2 and RM Weinshilboum1 1Department of Molecular Pharmacology and Experimental Therapeutics, Mayo Medical School, Mayo Clinic, Mayo Foundation, Rochester, MN, USA; 2Department of Biochemistry and Molecular Biology, Mayo Medical School, Mayo Clinic, Mayo Foundation, Rochester, MN, USA

Catechol O-methyltransferase (COMT) plays an important role in the metabolism of , catecholestrogens and catechol drugs. A common COMT G472A genetic polymorphism (Val108/158Met) that was identified previously is associated with decreased levels of activity and has been implicated as a possible risk factor for neuropsychiatric disease. We set out to ‘resequence’ the human COMT gene using DNA samples from 60 African-American and 60 Caucasian-American subjects. A total of 23 single nucleotide polymorphisms (SNPs), including a novel nonsynonymous cSNP present only in DNA from African-American subjects, and one insertion/deletion were observed. The wild type (WT) and two variant allozymes, Thr52 and Met108, were transiently expressed in COS-1 and HEK293 cells. There was no significant change in level of COMT activity for the Thr52 variant allozyme, but there was a 40% decrease in the level of activity in cells transfected with the Met108

construct. Apparent Km values of the WT and variant allozymes for the two reaction cosubstrates differed slightly, but significantly, for 3,4-dihydroxybenzoic acid but not for S-adenosyl-L-. The Met108 allozyme displayed a 70–90% decrease in immunor- eactive protein when compared with WT, but there was no significant change in the level of immunoreactive protein for Thr52. A significant decrease in the level of immunoreactive protein was also observed in hepatic biopsy samples from patients homozygous for the allele encoding Met108. These observations represent steps toward an understanding of molecular genetic mechanisms responsible for variation in COMT level and/or properties, variation that may contribute to the pathophysiology of neuropsychiatric disease. Molecular Psychiatry (2004) 9, 151–160. doi:10.1038/sj.mp.4001386 Keywords: catechol O-methyltransferase; COMT; ; genetic polymorphisms; SNPs; pharmacogenetics

Introduction lian cells as both membrane-bound (MB) and soluble (S) cytosolic forms.3 The enzyme is ubiquitously Catechol O-methyltransferase (COMT, EC 2.1.1.6) expressed, although S-COMT is expressed at higher plays an important role in the metabolism of levels in most tissues than MB-COMT.4 neurotransmitters.1 COMT catalyzes Family studies of red blood cell (RBC) COMT the O-methylation of catechol compounds, with activity showed a bimodal distribution, S-adenosyl-L-methionine (AdoMet) as the methyl and the results of segregation analysis were consistent donor. COMT substrates include not only neurotrans- with the inheritance of level of activity as a result of mitters such as , epinephrine and two autosomal codominant alleles.5–8 Levels of RBC , but also catecholestrogens and catechol COMT activity were correlated with levels of activity drugs that are used to treat hypertension, asthma and in other tissues such as kidney, lung and liver.9,10 In Parkinson’s disease.1,2 COMT is present in mamma- addition, the trait of a low level of COMT activity was associated with decreased enzyme thermal stability, suggesting that the COMT polymorphism might result Correspondence: Dr R Weinshilboum, Department of Molecular 10–12 Pharmacology and Experimental Therapeutics, Mayo Medical from an alteration in amino-acid sequence. Sub- School, Mayo Clinic, Mayo Foundation, Rochester, MN 55905, sequently, the human COMT cDNA and gene were USA. E-mail: [email protected] cloned.4,13,14 COMT is encoded by a single gene with The DNA resequencing data described in this manuscript have six exons that maps to chromosome 22q11.21.15 The been deposited in the NIH-sponsored Pharmacogenetics Research Network database PharmGKB with Accession Number PA117. phenotype of low levels of COMT activity has been Received 10 February 2003; revised 6 May 2003; accepted 6 May shown to be due mainly to a single nucleotide 2003 polymorphism (SNP) that changes the amino acid at COMT genetic variation AJ Shield et al 152 codon 108 in S-COMT (codon 158 in MB-COMT) from promoter region (intron 2) for S-COMT (PCR primer Val to Met.16 The frequency of the low-activity allele sequences and detailed reaction conditions are avail- that encodes Met 108/158 ranges from approximately able online in Supplementary Information Table 1). 50% in Caucasian subjects to 20–30% in East Asians, Locations of primers were chosen to avoid repetitive with some populations having even lower allele sequence and to ensure amplification specificity. M13 frequencies, for example, 6% in Ghana.17,18 tags were added to the 50-ends of each forward and The Val108/158Met genetic polymorphism has reverse primer to make it possible to use dye-primer been the subject of intense molecular epidemiologic sequencing chemistry to enhance our ability to study because of the important role of COMT identify heterozygous bases.28 The G472A SNP re- in catecholamine metabolism, and has been reported sponsible for the Val108/158Met polymorphism was to be associated with risk for schizophrenia, obses- also determined in 33 DNA samples from human liver sive–compulsive disorder, bipolar disorder, Parkin- biopsies by amplifying and sequencing COMT exon 4 son’s disease and breast cancer.19–23 Many of only. Amplicons were sequenced on both strands in these associations have proven difficult to con- the Mayo Clinic Molecular Biology Core Facility with firm,24–26 but, based on this type of data, a study an ABI 377 DNA sequencer using BigDyet dye- of COMT inhibitors in schizophrenic patients who primer chemistry. Samples with ambiguous chroma- have been stratified for the codon 108/158 poly- tograms or those with SNPs that were observed in morphism is currently underway.27 However, not all only a single DNA sample were subjected to an COMT phenotypic variance can be attributed to independent amplification, followed by sequencing. the Val108/158Met polymorphism, indicating that DNA sequencing chromatograms were analyzed using additional genetic variants may contribute to this the PolyPhred 3.029 and Consed 8.0 and GCG phenotype.16 Although several additional COMT programs.30,31 The consensus sequence used was the gene sequence variants have been identified, none COMT gene sequence in the RefFeq chromosome 22 of them has been reported to be of functional draft sequence (NT_011519). significance. In the present study, we set out to systematically resequence the exons, splice junctions COMT expression and promoter for the portion of this gene encoding S- A cDNA encoding S-COMT was amplified from a COMT using DNA samples from both African-Amer- human liver Marathon-Ready cDNA library (BD ican (AA) and Caucasian-American (CA) subjects. Biosciences Clontech, Palo Alto, CA, USA). S-COMT Functional genomic studies were then performed for cDNA sequences encoding polymorphic amino acids all nonsynonymous cSNPs observedFincluding that were created using site-directed mutagenesis per- responsible for the Val108/158Met polymorphism. In formed with overlap extension as described by Ho the course of these experiments, we discovered and et al.32 The wild type (WT) and variant S-COMT characterized a novel, common nonsynonymous cDNAs were then cloned into the eukaryotic expres- cSNP that was observed only in DNA from AA sion vector pCR3.1 (Invitrogen, Carlsbad, CA, USA) subjects. We also determined that the mechanism and were verified by DNA sequencing. These expres- responsible for the functional effects of the common- sion constructs were used to transfect COS-1 and Fand intensely studiedFVal108/158Met COMT HEK293 cells with the TransFastt transfection re- polymorphism was a decrease in the quantity of agent as suggested by the manufacturer (Promega, COMT protein. Madison, WI, USA). pSV-b-galactosidase (Promega) was cotransfected with each COMT expression con- struct in a 1 : 1 ratio to make it possible to correct for Materials and methods transfection efficiency. Transfected cells were har- DNA samples vested after 48 h, and high-speed supernatant (HSS) DNA samples from 60 AA and 60 CA subjects (sample cytosol preparations were prepared as described sets HD100AA and HD100CAU) were obtained from previously.33 Hepatic cytosol was also prepared from the Coriell Institute Cell Repository (Camden, NJ, the 33 human liver biopsy samples as described by USA). These samples had been anonymized prior to Lachman et al.16 Aliquots of these preparations were deposit, and all subjects had provided written con- stored at À801C prior to assay. sent for the use of their DNA for experimental purposes. DNA was also extracted from 33 anon- Enzyme activity and protein assays ymized liver biopsy samples studied previously by COMT enzyme activity was measured using the assay Lachman et al.16 using the QIAamp DNA mini kit of Raymond and Weinshilboum34 as modified by (QIAGEN, Valencia, CA, USA). Our experiments had Boudı´kova´ et al.10 This assay is based on the transfer been reviewed and approved by the Mayo Clinic of a radioactive methyl group from [14C-methyl]-S- Institutional Review Board. adenosyl-L-methionine (AdoMet) to the catechol substrate 3,4-dihydroxybenzoic acid (DBA). Blank COMT resequencing samples did not contain DBA. Endogenous COMT Seven PCR amplifications were performed for each of activity was never more than 1% of the activity the 120 human DNA samples using primers that observed after transfection with the WT construct.

flanked COMT exons 2–6, as well as the proximal Apparent Km values for DBA were determined with

Molecular Psychiatry COMT genetic variation AJ Shield et al 153 substrate concentrations that varied from 5 to 300 mM. statistically using ANOVA (StatView program version A series of AdoMet concentrations (1.3–88 mM) was 4.5; Abacus Concepts, Inc., Berkeley, CA, USA). also tested with each allozyme to determine apparent Linear regression analysis was used to determine T50 Km values for the cosubstrate. Enzyme activity was values (GraphPad Prism, version 3.03, GraphPad expressed as nanomoles of methylated product Software, San Diego, CA, USA). Linkage analysis formed per minute of incubation per microgram of was performed by calculating D 0 values for all protein (ie, U/mg protein). Enzyme thermal stability possible pairwise combinations of polymorphisms. was tested by preincubating each recombinant allo- D 0 is a method for calculating linkage analysis that is zyme for 15 min at temperatures that ranged from 37 independent of allele frequency.38,39 Haplotypes were to 651C. b-Galactosidase activity was measured inferred using a program based on the EM algo- spectrophotometrically using the b-Galactosidase En- rithm.40–42 Values for p, the average heterozygosity per zyme Assay System (Promega). Protein concentration site and y, the population parameter, as well was determined with the dye-binding method of as Tajima’s D-test for neutrality were determined Bradford35 using BIORAD Protein Assay Dye reagent using the Arlequin (2.0) software developed by (Bio-Rad Laboratories, Hercules, CA, USA) with Schneider et al.43 bovine serum albumin (BSA) as the standard.

COMT antibodies and Western blot analysis Results Polyclonal antibodies to COMT were produced in COMT gene resequencing rabbits. As a first step, two peptides were synthesized Human COMT was resequenced by amplifying exons that corresponded to S-COMT amino acids 1–25 and 2–6, intron–exon splice junctions and the entire 125–148, with an additional cysteine residue at the ‘proximal promoter’ (intron 2). Unfortunately, exon 1 amino terminus of each peptide. These peptides proved difficult to amplify, probably because of its differed from comparable regions of other known high GC content, so it was not resequenced. A total of methyltransferase and did not correspond to approximately 816 kb of DNA sequence was analyzed any other known human proteins. The synthetic for the 120 human DNA samples studied. All samples peptides were conjugated to keyhole limpet hemo- were sequenced on both strands. A total of 24 cyanin, and these conjugates were used to generate polymorphisms were observed in the 240 alleles rabbit polyclonal antibodies (ResGen, Huntsville, AL, analyzed, including eight ORF SNPs, 15 SNPs located USA). The rabbit polyclonal anti-COMT antibodies within intronic and flanking regions and one inser- were tested against pooled human liver cytosol and tion/deletion in the 30-UTR (Figure 1 and Table 1). The were found to react with a 25 kDa proteinFthe numbering scheme for polymorphisms located within F 0 0 anticipated Mr value for COMT with little cross- exons and in 5 -and3-untranslated regions is based reactivity. Furthermore, this 25 kDa protein was on the MB-COMT cDNA sequence, with the ‘A’ in the detected in COS-1 and HEK293 cells only after initiation codon designated ( þ 1). Nucleo- transfection of the cells with a COMT expression construct. The antibodies were used to perform quantitative Western blot analyses with recombinant COMT allozymes. In those experiments, the quantity of COS-1 or HEK293 cytosol loaded on a 12% acrylamide gel was adjusted so that each lane had been corrected for transfection efficiency. After transfer to a PVDF membrane, the blots were probed with rabbit antiserum, followed by secondary anti- body (goat anti-rabbit horseradish peroxidase) and bound antibody was detected using the ECL Western Blotting system (Amersham Pharmacia, Piscataway, NJ, USA). The AMBIS Radioanalytic Imaging System, Quant Probe Version 4.31 (Ambis Inc., San Diego, CA, USA) was used to quantitate the immunoreactive protein in each lane, and the data were expressed as a percentage of the intensity of WT COMT protein bands on the same gel. When human hepatic cytosol was used to perform Western analysis, the quantity of Figure 1 Human COMT polymorphisms. Schematic repre- cytosol loaded in each lane was adjusted to achieve sentation of genetic variations within exons 2–6 of the equivalent quantities of total cytosol protein. COMT gene that were observed during our resequencing studies. The diagram is not to scale. Locations of poly- morphisms are indicated by arrows. Red arrows represents Data analysis frequencies greater than 10%, blue frequencies between 1 Apparent Km and Vmax values were calculated using and 10%, and black frequencies below 1%. ‘CA’ indicates the method of Wilkinson36 with a computer program data for the DNA samples from Caucasian-American and written by Cleland.37 These data were compared ‘AA’ data from African-American subjects.

Molecular Psychiatry COMT genetic variation AJ Shield et al 154 Table 1 Human COMT polymorphisms

Location Polymorphism WT nucleotide Variant nucleotide Amino-acid Variant allele frequency in gene position alteration AA CA

Exon 2 À71 C G 0.033 0.000 Intron 2 I2(51) G C 0.400 0.450 Intron 2 I2(56) G A 0.033 0.000 Intron 2 I2(201) T C 0.458 0.558 Intron 2 I2(401) G A 0.000 0.008 Intron 2 I2(520) G A 0.008 0.000 Intron 2 I2(602) G A 0.000 0.008 Intron 2 I2(614) C A 0.017 0.000 Intron 2 I2(832) G C 0.383 0.500 Intron 2 I2(935) T C 0.075 0.017 Intron 2 I2(1140) A G 0.380 0.508 Exon 3 186 C T 0.333 0.417 Exon 3 219 G A 0.150 0.017

Exon 4 304 G A Ala52/102Thr 0.025 0.000 Exon 4 408 C G 0.258 0.492 Exon 4 438 C T 0.050 0.000 Exon 4 472 G A Val108/158Met 0.300 0.425 Intron 4 I4(69) T G 0.000 0.008 Exon 5 597 G A 0.125 0.025 Exon 5 609 C T 0.000 0.008 Intron 5 I5(75) G C 0.133 0.125 Exon 6 820 C CC 0.433 0.775 Exon 6 975 G A 0.042 0.000 30-FR D23 C G 0.283 0.025

Locations of polymorphisms, nucleotide sequence variation, changes in encoded amino acids and observed frequencies for variant alleles in the 60 samples from AA and the 60 samples from CA subjects are listed. The WT nucleotide was the nucleotide with the highest frequency in the AA subjects. The numbering scheme for nucleotide positions is described in the text. Data for exons are ‘boxed’, with information for the two nonsynonymous cSNPs in bold type.

tides located 50 to that position were assigned negative COMT amino acid sequence with those of chimpan- numbers, while those located 30 were assigned positive zee, gorilla, orangutan, bonobo, dog, cat, rat, mouse numbers. Positions within introns were numbered and pig COMT showed that the amino acid encoded relative to splice junctions, with the initial 50 nucleo- by codon 52/102 was Ala in all of these species except tide in an intron designated ( þ 1). In all, 13 of the the pig. Pigs had the variant human amino acid, Thr, SNPs were present in samples from both AA and CA at this position. No other species among those studied subjects, but allele frequencies often differed in the had Met, the amino acid encoded by the variant two populations. Most of the polymorphisms had human allele at codon 108/158, at that position. frequencies greater than 1% and, as a result, would be Neither of these nonsynonymous cSNPs altered considered ‘common’ in these populations. Nine of the amino acids which are thought to be involved in 24 polymorphisms that we identified had been substrate binding on the basis of X-ray crystallogra- described previously in publicly available databases phy data,45 and neither was located within the such as LocusLink, dbSNP and the EST database. methyltransferase ‘signature sequences’ that have Two of the 24 COMT polymorphisms were non- been designated regions I, II and III46Fobservations synonymous cSNPs, that is, they altered the encoded relevant to our subsequent substrate kinetic studies. A amino acid, resulting in two variant COMT allozyme third COMT nonsynonymous cSNP at amino acid 22/ sequences. One was the Val108/158Met polymorph- 72 that had been reported previously47,48 was not ism that had been described previously, with allele observed in either of our sample sets. That SNP had frequencies in these two population samples similar been observed in population samples of 48 Japanese to those described previously.18,44 The other, a novel subjects48 and 51 individuals of mixed ethnicity.47 nonsynonymous cSNP, resulted in an Ala52/102Thr Unfortunately, the frequency of the codon 22/72 SNP change in the encoded amino acid and was observed was not reported for either of these two population only in AA subjects. A comparison of the human samples.

Molecular Psychiatry COMT genetic variation AJ Shield et al 155 The region containing the proximal promoter for 1.0 Â 10À4 for CA subjects, while p was 5.472.6 Â 10À4 S-COMTFintron 2Fwas also of particular interest. for DNA from AA and 4.372.1 Â 10À4 for samples Ten SNPs were observed within this region, one of from CA subjects. These results are similar to data which was located within the S-COMT core promo- reported by Stephens et al.52 for 292 autosomal genes ter.4,49 The TRANSFAC database50 was used to (average y ¼ 9.6 Â 10À4 and average p ¼ 5.8 Â 10À4). determine whether any of these intron 2 proximal The difference between p and y is the basis for promoter SNPs might be present within putative Tajima’s D statistic. Under conditions of neutrality, factor recognition sequences. SNPs it is assumed that p ¼ y, with D ¼ 0. D was positive I2(521), I2(602) and I2(832) were located within for both of our study populations, but neither putative recognition sequences value differed significantly from zero (data not for TCF11 (also KCR-F1 and Nrf1), Sp1 and AP1 (Fos/ shown). Jun), respectively. In addition, the I2(521) polymorph- ism was located within an estrogen response element Linkage and haplotype analysis that had been characterized by Xie et al.49 The SNP Linkage analysis performed for all possible pairwise at I2(832) was present in both populations, while combinations of COMT polymorphisms demonstrated the other two polymorphisms were observed at that many of them were tightly linked (see Supple- low frequencies in only one ethnic group (Table 1). mentary Information Table 2, which is available The SNP at I2(1140), within the S-COMT core online). Patterns of linkage, like those for SNP promoter, was present at high frequency in both frequencies, differed between the two populations populations. studied. Tight linkage between SNPs located in the Of the 24 polymorphisms, 20 were observed in the proximal promoter region and the common Val108/ DNA samples from AA subjects, and 17 were present 158Met polymorphism was more common in samples in samples from CA subjects. The average number of for the CA subjects (three SNPs) than in samples from polymorphisms in COMT was 6.7 per kb resequenced AA (no SNPs). in the AA samples and 5.0 per kb in samples from CA Haplotype analysis was also performed. Unambig- subjects. These numbers are similar to a value of 4.6 uous haplotypes, those with only a single hetero- SNPs per kb for 75 human genes reported by zygous nucleotide, accounted for 28 and 86% of Halushka et al.51 ‘Nucleotide diversity’ is a measure all alleles studied in the AA and CA population of genetic variationFadjusted for the number of samples, respectively. There were also 45 inferred alleles studied. Two standard measures of nucleotide haplotypes for the AA and 15 for the CA subjects. diversity are p, average heterozygosity per site, and y, Haplotypes, both inferred and unambiguous, that had a population mutation measure that is theoretically frequencies of 2% or greater are listed in Table 2. The equal to the neutral mutation parameter.52,53 In our data in the table illustrate the differences in haplotype populations, y ¼ 3.571.1 Â 10À4 for AA and y ¼ 3.07 patterns between the two populations studiedFmost

Table 2 Human COMT haplotype analysis

Allele Frequency I2 I2 I2 I2 I2 I5 AA CA (51) (201) (832) (935) (1140) 186 219 408 438 472 597 (82) 820 D23 *1A -- 0.306 G C C T G C G G C G G G CC C *1B 0.111 0.101 G C C T G C G G C G G G C C *1C 0.092 -- G T G T A C G C C G G G C C *1D 0.068 0.013 G C G T A C G C C G G G CC C *1E -- 0.040 G C G T A C G C C G G G C C *1F 0.035 -- C C C C G C A C C G G G C G *1G -- 0.035 C C C T G C G G C G G G CC C *1H 0.030 -- G C C T G C G G C G A G C G *1I 0.028 -- G C C T G C A C C G G C C C *1J 0.025 -- G T G T A C G C T G G G C C *1K 0.020 -- G T G T A C G C C G G G CC C

*2A 0.199 0.234 C T G T A T G C C A G G CC C *2B -- 0.088 C T G T A T G C C A G C CC C *2C -- 0.036 G T G T A T G C C A G G CC C

COMT haplotypes, both unambiguous and inferred, with frequencies greater than 2%, are listed. Nucleotide positions that are not listed had WT sequence. The numbering scheme for nucleotide positions is listed in the text. Variant nucleotides as compared with the ‘reference sequence’ (haplotype *1C) are highlighted as white on black. Initial haplotype designations (*1 or *2) were made on the basis of amino acids that vary, with the WT sequence at nucleotide 472 assigned *1 (ie Val108/ 158) and those with the variant sequence (Met108/158) assigned *2. Letter designations were then assigned based on descending frequencies. Haplotypes with ‘bold’ frequency values were unambiguous in that population.

Molecular Psychiatry COMT genetic variation AJ Shield et al 156 striking for *1A, the most common haplotype for WT allele as well as alleles encoding the two COMT CA subjects. However, this haplotype was neither variant allozymes (Thr52/102 and Met108/158), and observed nor inferred among the AA subjects. these constructs were used to transfect COS-1 and The data in Table 2 also serve to highlight the tight HEK293 cells transiently. Mammalian cells were used linkage observed between Met108/158 and SNPs to ensure appropriate post-translational modification located in the proximal promoter for DNA samples as well as the presence of mammalian protein from CA subjects. Allele designations shown in degradation systems. Since S-COMT was expressed, the table were made on the basis of encoded amino the nomenclature for codon number used subse- acids, with the WT sequence at nucleotide 472 quently will be that which applies to S-COMT. assigned *1 (ie Val108/158), while those with Met at Expression of the Met108 allozyme in both COS-1 codon 108/158 were assigned the designation *2. and HEK293 cells resulted in a 35–40% decrease in Letter designations were then assigned on the basis of activity (Po0.04) as compared to cells expressing the descending allele frequencies. Although, as a result of WT allozyme (Figure 2 and Supplementary Informa- its low frequency, we did not observe unequivocal tion Table 3, which is available online). These haplotypes that contained Thr52/102, we assigned observations correlate with results anticipated on the designation *3 to alleles that included this the basis of the association of this genotype with genotype. significantly decreased COMT activity in human tissue samples.16 The Thr52 allozyme displayed a Recombinant COMT activity slight, but not significant, increase in the level of The next series of experiments was designed to test activity when compared with the WT allozyme the functional implications of the two nonsynon- (Figure 2). ymous cSNPs observed during the gene resequencing The changes in the level of COMT activity that we studies. Expression constructs were created for the observed might have resulted from alterations in the kinetic properties of the enzyme, as has been observed for other genetically polymorphic en- 54–56 zymes. However, apparent Km values after expres- sion in COS-1 cellsFthe only cell type in which the substrate kinetic studies were performedFshowed only small differences for both cosubstrates, DBA and

AdoMet (Table 3). Although the apparent Km values did differ statistically for some of the comparisons, these small differences seemed unlikely to explain the differences in levels of activity observed with the saturating concentrations of substrate used to mea- sure enzyme activity. Finally, thermal stability has often been used as an indirect indicator of alterations in the amino-acid sequence, and it was reported over 20 years ago that enzyme activity in subjects homo- zygous for the common genetic variant controlling low COMT activity was more thermolabile than Figure 2 Recombinant human COMT allozyme activity. activity in subjects homozygous for the WT (high Levels of activity for COMT allozymes expressed in COS-1 11 (N ¼ 12) or HEK293 (N ¼ 8) cells were determined with activity) phenotype. Therefore, we used recombi- 300 mM DBA as substrate. Activities (mean7SEM) are nant protein to confirm that the Met108 allozyme had shown relative to the activity of the WT allozyme after a significantly lower T50 (temperature resulting in correction for transfection efficiency. *Po0.05 compared 50% inactivation) than either the WT or Thr52 with cells transfected with the WT construct. allozymes (Po0.005; Table 3).

Table 3 Recombinant human COMT allozyme substrate kinetics when expressed in COS-1 cells

Construct DBA Ado-Met T50 (1C)

Km Vmax Km Vmax

WT 28.070.47 0.8170.01 14.270.21 0.6070.01 53.770.10 Met108 25.670.64* 0.6170.01* 13.670.26+ 0.7370.01 49.770.28** Thr52 22.770.67** 0.4870.01** 15.370.47 0.5470.02 52.770.34

Apparent Km (mM) and Vmax (U/mg protein) values plus T50 values, the temperature resulting in 50% inactivation, are listed as average 7 SEM (N=3). *Po0.05 and **Po0.001 as compared with values for the WT allozyme. +Po0.01 when compared with values for the Thr52 allozyme.

Molecular Psychiatry COMT genetic variation AJ Shield et al 157 Recombinant COMT Western blot analyses Alterations in the amino-acid sequence as a result of genetic polymorphisms have often been found to be associated with changes in the levels of immunoreactive protein.54–58 Therefore, levels of COMT immunoreactive protein were also measured for the recombinant human COMT allozymes by Western blot analysis. A representative Western blot for COS-1 cell preparations is shown in Figure 3. Differences in the levels of immunoreactive protein for the three allozymes paralleled variations in the level of enzyme activity (Figure 4a). Specifically, average levels of COMT protein in both cell lines were decreased significantly in cells expressing Met108 (Figure 4a). Conversely, cells expressing Thr52 had increased level of COMT protein, significantly increased for COS-1 cells (Po0.05), when compared with those expressing the WT allozyme. It should be pointed out that the peptide used to generate the rabbit polyclonal antibody used to perform the Western blot analyses did not include the polymorphic amino acids. These observa- tions were compatible with a growing body of evidence that alteration of only a single amino acid as a result of genetic polymorphisms can result in Figure 4 Recombinant human COMT immunoreactive significant changes in levels of immunoreactive protein. (a) COMT immunoreactive protein expressed in protein.54–58 COS-1 and HEK293 cells (N ¼ 6 independent transfections in each case) corrected for variation in transfection efficiency. Values shown are mean7SEM relative to WT Human liver COMT genotype–phenotype correlation allozyme. *Po0.05 and **Po0.005 compared with cells analysis transfected with the WT allozyme. (b) COMT immunoreac- Our observation of a significant decrease in the level tive protein in human liver biopsy samples. The gels were of immunoreactive protein for the recombinant loaded with equal quantities of hepatic cytosol protein. A Met108 allozyme after the transfection of both COS- standard curve of WT COMT expressed in COS-1 cells was used to quantitate the data. *Po0.05 and **Po0.005 for the 1 and HEK293 cells suggested that a similar mechan- comparison indicated. ism might occur in vivo. Therefore, we also performed genotype–phenotype correlation analysis for the

level of COMT immunoreactive protein using 33 liver biopsy samples that had previously been genotyped for the codon 108/158 polymorphism and had been phenotyped for the level of enzyme activity.16 This group included 14 samples from heterozygous individuals, seven from subjects homozygous for the WT allele encoding Val108 and 12 samples from subjects homozygous for the Met108 allele. Equiva- lent concentrations of hepatic cytosol were loaded on the gels on the basis of protein concentration, and these samples were subjected to quantitative Western blot analysis. There was a significant decrease in the level of COMT protein in samples from subjects homozygous for the allele encoding Met108 when compared with heterozygous samples or samples from subjects homozygous for Val108 Figure 3 COMT Western blot analysis. Western blots were (P 0.002 by ANOVA; Figure 4b). These observations used to analyze levels of COMT allozymes expressed in o mammalian cells in culture as well as in 33 human liver demonstrated that the major mechanism responsible biopsy samples. Representative Western blot data are shown for decreased COMT activity in COS-1 cells and for each allozyme expressed in COS-1 cells (left) and for HEK293Fa decreased level of enzyme protein F was livers from three individuals with the genotypes indicated also associated with decreased enzyme activity at S-COMT codon 108 (right). in vivo.

Molecular Psychiatry COMT genetic variation AJ Shield et al 158 Discussion resulted primarily from a decreased level of COMT immunoreactive protein. To determine whether that COMT plays an important role in the metabolism of same phenomenon might occur in vivo, 33 liver endogenous catecholamines, catecholestrogens and biopsy samples that had previously been phenotyped catechol drugs. We previously identified a common for the level of activity and genotyped for the genetic polymorphism that regulated the level of Val108Met polymorphism16 were used to perform COMT enzyme activity,5 and later demonstrated that a quantitative Western blot analysis. The level of common nonsynonymous cSNP that results in a immunoreactive protein in these samples was also Val108/158Met alteration in the encoded amino acid found to be significantly decreased in subjects was associated with this polymorphism.16 However, homozygous for Met108. These observations were that SNP did not explain all variation in the level of compatible with a growing body of evidence which enzyme activity in 33 liver biopsy samples.16 We have shows that a change in only a single amino acid as a now applied a genotype-to-phenotype research strat- result of a genetic polymorphism can often result in a egy in an attempt to identify additional DNA decreased level of immunoreactive protein.54–58 sequence variation that might contribute to individual The most common underlying mechanism for this differences in COMT activity or properties. As a first phenomenon is an increased rate of protein degrada- step, we resequenced COMT using DNA from 60 AA tionFas a result of a ubiquitin-proteasome-mediated and 60 CA subjects. Regions of the gene sequenced processFwith the involvement of chaperone pro- included exons 2–6, exon–intron splice junctions and teins.62–64 Obviously, genetic polymorphisms in gene the promoter for S-COMT (intron 2). There were 23 promoter and regulatory regions can also result in SNPs in the 120 DNA samples (240 alleles) studied, alterations in transcription and, thus, the level of including eight in the ORF and 15 in intronic and expression.65 flanking regions. In addition, there was a single In summary, we have identified a series of common insertion/deletion located in the 30-UTR. Two of the genetic polymorphisms and haplotypes for the hu- cSNPs were nonsynonymous: the common Val108/ man COMT gene. Functional studies of variant 158Met polymorphism that we had described pre- allozymes confirmed in vivo data and demonstrated viously16 and a novel cSNP found only in the AA that the common Met108 allozyme results in de- population that resulted in an Ala52/102Thr change creased enzyme activity, primarily as a result of a in the encoded amino acid. An additional nonsynon- decrease in the quantity of immunoreactive proteinF ymous cSNP within codon 22/72 that had been both in human tissue and after transient expression reported by Cargill et al47 and Saito et al48 was not in cultured mammalian cells. We found no evidence observed in either of our population samples. In all, that the novel Thr52 allozyme which we observed in 20 of the 24 polymorphisms were present in samples AA subjects greatly alters functionFalthough both from AA subjects and 17 were observed in DNA from the level of activity and immunoreactive protein were CA subjects. Many of the polymorphisms were slightly elevated (Figures 2 and 4). Future studies will linked, with different linkage patterns in the two be required to elucidate mechanism(s) responsible for populations. Analysis of our data also resulted in the the decreased level of this important neurotransmit- identification of a series of COMT haplotypes which ter-metabolizing protein in response to the change in also differed between the two populations (Table 2). the amino acid at codon 108 from Val to Met. Finally, We then performed experiments designed to study the present results represent a step towards under- the possible functional implications of the two standing mechanisms that might contribute to inter- nonsynonymous COMT cSNPs that we had observed. individual differences in the biotransformation of Neither of these polymorphisms altered an amino catecholamine neurotransmitters and catechol drugs acid located within substrate-binding sites on the and, perhaps, in risk for the development of neurop- basis of the crystal structure of the protein.45 Perhaps sychiatric disease. for that reason, neither had major effects on the

apparent Km values of recombinant allozymes for either DBA or AdoMet (Table 3). However, thermal Acknowledgements stability was decreased for the Met108 allozyme, as reported previously for both tissue preparations and We thank Dr Daniel Schaid for performing the linkage bacterially expressed protein.11,59 The thermal stabi- and haplotype analyses and Ms Luanne Wussow for lity of the Thr52 allozyme did not differ significantly her assistance with the preparation of this manu- from that of the WT enzyme. script. This work was supported in part by NIH Grants The decreased level of COMT activity that occurs in RO1 GM28157 (RMW), RO1 GM35720 (RMW), PO1 association with the codon 108/158 polymorphism CA82267 (RWM), UO1 GM61388 (RMW and EDW) has been well described over the past two dec- and R25T CA92049 (AJS). ades.5,11,16,44,60 However, the underlying mechanism has remained elusive.61 After transient expression in COS-1 and HEK293 cells, we found that the level of References activity for the Met108 allozyme was decreased by 1 Axelrod J, Tomchick R. Enzymatic O-methylation of epinephrine 35–40% compared to the WT allozyme. This effect and other catechols. J Biol Chem 1958; 233: 702–705.

Molecular Psychiatry COMT genetic variation AJ Shield et al 159 2 Guldberg HC, Marsden CA. Catechol-O-methyltransferase: phar- 23 Lavigne JA, Helzlsouer KJ, Huang H-Y, Strickland PT, Bell DA, macological aspects and physiological role. Pharmacol Rev 1975; Selmin O et al. An association between the allele coding for a low 27: 135–206. activity variant of catechol O-methyltransferase and the risk for 3 Roth JA. Membrane-bound catechol O-methyltransferase: a reeva- breast cancer. Cancer Res 1997; 57: 5493–5497. luation of its role in the O-methylation of the catecholamine 24 Craddock N, Dave S, Greening J. Association studies of bipolar neurotransmitter. Rev Physiol Biochem Pharmacol 1992; 120: disorder. Bipolar Disord 2001; 3: 284–298. 1–29. 25 Schindler KM, Richter MA, Kennedy JL, Pato MT, Pato CN. 4 Tenhunen J, Salminen M, Lundstrom K, Kiviluoto T, Savolainen R, Association between homozygosity at the COMT gene locus and Ulmanen I. Genomic organization of the human catechol-O- obsessive compulsive disorder. Am J Med Genet 2000; 96: 721– methyltransferase gene and its expression from two distinct 724. promoters. Eur J Biochem 1994; 223: 1049–1059. 26 Mizuta I, Mizuta E, Yamasaki S, Kuno S, Yasuda M, Tanaka C. 5 Weinshilboum RM, Raymond FA. Inheritance of low erythrocyte Meta-analysis of polymorphism of the catechol-O-methyltransfer- catechol-O-methyltransferase activity in man. Am J Hum Genet ase gene in relation to the etiology of Parkinson’s disease in Japan. 1977; 29: 125–135. Movement Disord 2000; 15: 1013–1014. 6 Spielman RS, Weinshilboum RM. Family studies of low red cell 27 Holden C. Deconstructing schizophrenia. Science 2003; 299: 333– COMT activity. Am J Hum Genet 1979; 31: 63A. 335. 7 Floderus Y, Wetterberg L. The inheritance of human erythrocyte 28 Chadwick RB, Conrad MP, McGinnis MD, Johnston-Dow L, catechol-O-methyltransferase activity. Clin Genet 1981; 19: Spurgeon SL, Kronick MN. Heterozygote and mutation detection 392–395. by direct automated fluorescent DNA sequencing using a mutant 8 Siervogel RM, Weinshilboum R, Wilson AF, Elston RC. Major Taq DNA polymerase. BioTechniques 1996; 20: 676–683. gene model for the inheritance of catechol-O-methyltransferase 29 Nickerson DA, Tobe VO, Taylor SL. PolyPhred: automating the activity in five large families. Am J Med Genet 1984; 19: detection and genotyping of single nucleotide substitutions using 315–323. fluorescence-based resequencing. Nucleic Acids Res 1997; 25: 9 Weinshilboum RM. Human erythrocyte catechol-O-methyltrans- 2745–2751. ferase: correlation with lung and kidney activity. Life Sci 1978; 22: 30 Gordon D, Abajian C, Green P. Consed: a graphical tool for 625–630. sequence finishing. Genome Res 1998; 8: 195–202. 10 Boudı´kova´ B, Szumlanski C, Maidak B, Weinshilboum R. Catechol 31 Devereux J, Haeberli P, Smithies O. A comprehensive set of O-methyltransferase (COMT) pharmacogenetics. Clin Pharmacol sequence analysis programs for the VAX. Nucleic Acids Res 1984; Ther 1990; 48: 381–389. 12: 387–395. 11 Scanlon PD, Raymond FA, Weinshilboum RM. Catechol-O- 32 Ho SN, Hunt HD, Horton RM, Pullen JK, Pease LR. Site-directed methyltransferase: thermolabile enzyme in erythrocytes of sub- mutagenesis by overlap extension using the polymerase chain jects homozygous for the allele for low activity. Science 1979; 203: reaction. Gene 1989; 77: 51–59. 63–65. 33 Wood TC, Aksoy IA, Aksoy S, Weinshilboum RM. Human liver 12 Spielman RS, Weinshilboum RM. Genetics of red cell COMT thermolabile phenol sulfotransferase: cDNA cloning, expression activity: analysis of thermal stability and family data. Am J Med and characterization. Biochem Biophys Res Commun 1994; 198: Genet 1981; 10: 279–290. 1119–1127. 13 Bertocci B, Miggiano V, Da Prada M, Dembic Z, Lahm H-W, 34 Raymond FA, Weinshilboum RM. Microassay of human erythro- Malherbe P. Human catechol O-methyltransferase: cloning and cyte catechol-O-methyltransferase: removal of inhibitory calcium expression of membrane-associated form. Proc Natl Acad Sci USA ion with chelating resin. Clin Chim Acta 1975; 58: 185–194. 1991; 88: 1416–1420. 35 Bradford MM. A rapid and sensitive method for the quantitation of 14 Lundstro¨m K, Salminen M, Jalanko A, Savolainen R, Ulmanen I. microgram quantities of protein utilizing the principle of protein– Cloning and characterization of human placental catechol-O- dye binding. Anal Biochem 1976; 72: 248–254. methyltransferase cDNA. DNA Cell Biol 1991; 10: 181–189. 36 Wilkinson GN. Statistical estimations in . Bio- 15 Grossman MH, Emanuel BS, Budaf ML. Chromosomal mapping of chem J 1961; 80: 324–332. the human catechol-O-methyltransferase gene to 22q11.1–q11.2. 37 Cleland WW. Computer programmes for processing enzyme Genomics 1992; 12: 822–825. kinetic data. Nature 1963; 198: 463–465. 16 Lachman HM, Papolos DF, Saito T, Yu Y-M, Szumlanski CL, 38 Hartl DL, Clark AG. Principles of Population Genetics, 2nd edn. Weinshilboum RM. Human catechol O-methyltransferase pharma- Sinauer Associates, Inc.: Sunderland, MA, 1989. cogenetics: description of a functional polymorphism and its 39 Hendrick PW. Genetics of Populations, 2nd edn. Jones and Bartlett potential application to neuropsychiatric disorders. Pharmacoge- Publ.: Sudbury, MA, 2000, pp 396–405. netics 1996; 6: 243–250. 40 Long JC, Williams RC, Urbanek M. An E-M algorithm and testing 17 Ameyaw M, Syvanen A, Ulmanen I, Ofori-Adjei D, McLeod HL. strategy for multiple-locus haplotypes. Am J Hum Genet 1995; 56: Pharmacogenetics of catechol-O-methyltransferase: frequency of 799–810. low activity allele in a Ghanaian population. Hum Mutat 2000; 16: 41 Excoffier L, Slatkin M. Maximum-likelihood estimation of mole- 445–446. cular haplotype frequencies in a diploid population. Mol Biol Evol 18 Palmatier MA, Kang AM, Kidd KK. Global variation in the 1995; 12: 921–927. frequencies of functionally different catechol O-methyltransferase 42 Schaid DJ, Rowland CM, Tines DE, Jacobson RM, Poland GA. alleles. Biol Psychiatry 1999; 46: 557–567. Score tests for association between traits and haplotypes 19 Strous RD, Bark N, Parsia SS, Volavka J, Lachman HM. Analysis of when linkage phase is ambiguous. Am J Hum Genet 2002; 70: a functional catechol-O-methyltransferase gene polymorphism in 425–434. schizophrenia: evidence for association with aggressive and 43 Schneider S, Roessli D, Excofier L. Arlequin: a software for antisocial behavior. Psychiatr Res 1997; 69: 71–77. population genetics data analysis, Version 2.000. Genetics and 20 Karayiorgou M, Altemus M, Galke BL, Goldman D, Murphy DL, Biometry Lab, Department of Anthropology, University of Geneva, Ott J et al. Genotype determining low catechol-O-methyltransfer- 2000. ase activity as a risk factor for obsessive–compulsive disorder. Proc 44 McLeod HL, Fang L, Luo X, Scott EP, Evans WE. Ethnic differences Natl Acad Sci USA 1997; 94: 4572–4575. in erythrocyte catechol O-methyltransferase activity in black and 21 Mynett-Johnson LA, Murphy VE, Claffey E, Shields DC, McKeon P. white Americans. J Pharmacol Exp Ther 1994; 270: 26–29. Preliminary evidence of an association between bipolar disorder 45 Vidgren J, Svensson LA, Liljas A. Crystal structure of catechol O- in females and the catechol-O-methyltransferase gene. Psychiatr methyltransferase. Nature 1994; 368: 354–358. Genet 1998; 8: 221–225. 46 Ingrosso D, Fowler AV, Bleibaum J, Clarke S. Sequence of the 22 Kunugi H, Nanko S, Ueki A, Otsuka E, Hattori M, Hoda F et al. D-aspartyl/L-isoaspartyl protein methyltransferase from human High and low activity alleles of catechol O-methyltransferase gene: erythrocytes: common sequence motifs for protein, DNA, RNA ethnic difference and possible association with Parkinson’s and small molecule S-adenosylmethionine-dependent methyl- disease. Neurosci Lett 1997; 221: 202–204. . J Biol Chem 1989; 264: 20131–20139.

Molecular Psychiatry COMT genetic variation AJ Shield et al 160 47 Cargill M, Altshuler D, Ireland J, Sklar P, Ardlie K, Patil N et al. pharmacogenetics: common genetic polymorphisms that alter Characterization of single-nucleotide polymorphisms in coding activity. Mol Pharmacol 1998; 53: 708–717. regions of human genes. Nat Genet 1999; 22: 231–238. 58 Szumlanski C, Otterness D, Her C, Lee D, Brandriff B, Kelsell D et 48 Saito S, Iida A, Sekine A, Miura Y, Sakamoto T, Ogawa C et al. al. Thiopurine methyltransferase pharmacogenetics: human gene Identification of 197 genetic variations in six human methyltrans- cloning and characterization of a common polymorphism. DNA ferase genes in the Japanese population. J Hum Genet 2001; 46: Cell Biol 1996; 15: 17–30. 529–537. 59 Lotta T, Vidgren J, Tilgmann C, Ulmanen I, Melen K, Julkunen I et 49 Xie T, Ho SL, Ramsden D. Characterization and implications of al. Kinetics of human soluble and membrane-bound catechol O- estrogenic down-regulation of human catechol O-methyltransfer- methyltransferase: a revised mechanism and description of the ase gene transcription. Mol Pharmacol 1999; 56: 31–38. thermolabile variant of the enzyme. Biochemistry 1995; 34: 4202– 50 Wingender E, Chen X, Hehl R, Karas H, Liebich I, Matys V et al. 4210. TRANSFAC: an integrated system for gene expression regulation. 60 Klemetsdal B, Staume B, Giverhaug T, Aarbakke J. Low catechol- Nucleic Acids Res 2000; 28: 316–319. O-methyltransferase activity in a Saami population. Eur J Clin 51 Halushka MK, Fan J-B, Bentley K, Hsie L, Shen N, Weder A et al. Pharmacol 1994; 46: 231–235. Patterns of single-nucleotide polymorphisms in candidate genes 61 Goodman JE, Jensen LT, He P, Yager JD. Characterization of human for blood-pressure homeostasis. Nat Genet 1999; 22: 239–247. soluble high and low activity catechol-O-methyltransferase cata- 52 Stephens JC, Schneider JA, Tanguay DA, Choi J, Acharya T, lyzed catechol estrogen methylation. Pharmacogenetics 2002; 12: Stanley SE et al. Haplotype variation and linkage disquilibrium in 517–528. 313 human genes. Science 2001; 293: 489–493. 62 Tai H-L, Fessing MY, Bonten EJ, Yanishevsky Y, d’Azzo A, 53 Fullerton SM, Clark AG, Weiss KM, Nickerson DA, Taylor SL, Krynetski EY et al. Enhanced proteasomal degradation of mutant Stengard JH et al. Apolipoprotein E variation at the sequence human thiopurine S-methyltransferase (TPMT) in mammalian haplotype level: implications for the origin and maintenance of a cells: mechanism for TPMT protein deficiency inherited by major human polymorphism. Am J Hum Genet 2000; 67: 881–900. TPMT*2, TPMT*3A, TPMT*3B or TPMT*3C. Pharmacogenetics 54 Freimuth RR, Eckloff B, Wieben ED, Weinshilboum RM. Human 1999; 9: 641–650. sulfotransferase SULT1C1 pharmacogenetics: gene resequencing and 63 Siegel D, Anwar A, Winski SL, Kepa JK, Zolman KL, Ross D. Rapid functional genomic studies. Pharmacogenetics 2001; 11: 747–756. polyubiquitination and proteosomal degradation of a mutant form 55 Thomae BA, Eckloff BW, Freimuth RR, Wieben ED, Weinshilboum of NAD(P)H:quinone oxioreductase 1. Mol Pharmacol 2001; 59: RM. Human sulfotransferase SULT2A1 pharmacogenetics: geno- 263–268. type-to-phenotype studies. Pharmacogenomics J 2002; 2: 48–56. 64 Wang L, Sullivan W, Toft D, Weinshilboum R. Thiopurine S- 56 Xu Z-H, Freimuth RR, Eckloff B, Wieben E, Weinshilboum RM. methyltransferase (TPMT) pharmacogenetics: role of chaperone Human 30-phosphoadenosine 50-phosphosulfate synthetase 2 proteins in variant allozyme degradation. Clin Pharmacol Ther (PAPSS2) pharmacogenetics: gene resequencing, genetic poly- 2003; 73: P29. morphisms and functional characterization of variant allozymes. 65 Beutler E, Gelbart T, Demina A. Racial variability in the UDP- Pharmacogenetics 2002; 12: 11–21. glucuronosyltransferase 1 (UGT1A1) promoter: a balanced poly- 57 Preuss CV, Wood TC, Szumlanski CL, Raftogianis RB, Otterness morphism for regulation of bilirubin metabolism? Proc Natl Acad DM, Girard B et al. Human N-methyltransferase Sci USA 1998; 95: 8170–8174.

Molecular Psychiatry