Allele-Specific Expression and Gene Methylation in the Control of CYP1A2 Mrna Level in Human Livers

Allele-specific expression and gene methylation in the control of CYP1A2 mRNA level in human livers

Roza Ghotbi1, Alvin Gomez2, The basis for interindividual variation in the CYP1A2 gene expression is not 3 1 fully understood and the known genetic polymorphisms in the gene provide Lili Milani , Gunnel Tybring , no explanation. We investigated whether the CYP1A2 gene expression is 3 Ann-Christine Syva¨nen , regulated by DNA methylation and displays allele-specific expression (ASE) Leif Bertilsson1, Magnus using 65 human livers. Forty-eight percent of the livers displayed ASE not Ingelman-Sundberg2 and associated to the CYP1A2 mRNA levels. The extent of DNA methylation of a 1 CpG island including 17 CpG sites, close to the translation start site, inversely Eleni Aklillu correlated with hepatic CYP1A2 mRNA levels (P ¼ 0.018). The methylation of 1Division of Clinical Pharmacology, Department two separate core CpG sites was strongly associated with the CYP1A2 mRNA of Laboratory Medicine, Karolinska University levels (P ¼ 0.005) and ASE phenotype (P ¼ 0.01), respectively. The CYP1A2 Hospital Huddinge, Karolinska Institutet, expression in hepatoma B16A2 cells was strongly induced by treatment with 2 Stockholm, Sweden; Section of 5-aza-20-deoxycytidine. In conclusion, the CYP1A2 gene expression is Pharmacogenetics, Department of Physiology and Pharmacology, Karolinska Institutet, influenced by the extent of DNA methylation and displays ASE, mechanisms Stockholm, Sweden and 3Molecular Medicine, contributing to the large interindividual differences in CYP1A2 gene Department of Medical Sciences, Uppsala expression. University, Uppsala, Sweden The Pharmacogenomics Journal (2009) 9, 208–217; doi:10.1038/tpj.2009.4; published online 10 March 2009 Correspondence: Dr E Aklillu, Division of Clinical Pharmacology, Department of Laboratory of Medicine, Keywords: CYP1A2; drug-metabolizing enzyme; gene expression; allele-specific expression; Karolinska Institutet, Karolinska University epigenetics; DNA methylation Hospital Huddinge, C-168, SE-141 86 Stockholm, Sweden. E-mail: [email protected]

Introduction

The human cytochrome P450 1A2 (CYP1A2) enzyme is expressed mainly in the liver and accounts for about 13% of the total P450 content in the liver.1 It has an important function in the metabolism of caffeine as well as of several clinically important drugs such as clozapine,2 phenacetin,3 verapmil,4 endogenous substrates such as melatonin,5 estradiol,6 and procarcinogenic substances including heterocyclic amines, arylamines and aflatoxin B1.7,8 The CYP1A2 gene shows large interindividual and interethnic differences in both mRNA expression level and enzyme activity.8–10 The underlying causes for this variation are not known and could be caused by genetic, epigenetic or environmental factors. Indeed, CYP1A2 is genetically polymorphic with more than 16 variant alleles described so far (http://www.cypalleles.ki.se/cyp1a2.htm). However, the low frequency of the functionally different variant alleles in the populations and the results of several genotype–phenotype association studies Received 7 October 2008; revised 13 January 2009; accepted 3 February 2009; published indicate a limited function of CYP1A2 genetic polymorphism as a cause of online 10 March 2009 interindividual variation in CYP1A2 gene expression or activity.11,12 In line of ASE and gene methylation of CYP1A2 R Ghotbi et al 209

this research, we have found one allele CYP1A2*1K causing genes26 causing transcriptional repression by inhibition of lower expression of the gene but it is present in Ethiopians13 binding of transcription factor. and almost absent in whites and Asians.11,12 The interethnic While appreciating the existence of the wide interindivi- differences in CYP1A2 activity is important and by control- dual variation in the constitutive CYP1A2 mRNA expression ling for the effects of smoking, oral contraceptive use and and enzyme activity, the underlying causes remain unclear. genotype, we recently reported the presence of significant It is generally accepted that most of the known SNPs in the differences in enzyme activity between Swedish, Korean and CYP1A2 gene and 50-flanking regions do not determine the Serbian populations, differences not explained by the observed large interindividual variability in constitutive known CYP1A2 single nucleotide polymorphisms (SNPs)/ hepatic CYP1A2 mRNA expression and enzyme activity.11 haplotypes.12,14 Identification of the molecular mechanism behind variation Apart from genetic polymorphisms, phenotypic variations in CYP1A2 enzyme activity may have important implica- are governed by other factors affecting gene expression, tions for drug safety/efficacy and cancer susceptibility. such as gene cis- and trans-acting transcriptional compo- Because the epigenetic factors and ASE can lead to nents, alternative splicing, RNA stability, expression of phenotypic variability in the pharmacodynamic and phar- regulatory RNAs and epigenetics such as histone modifica- macokinetic outcomes of drug therapy, we considered it of tions and DNA methylation.15 Imprinted genes (parent of interest to investigate the function of DNA methylation and origin-specific gene expression in somatic cells) and genes differential allelic expression in CYP1A2 gene regulation as subject to X chromosome inactivation display monoallelic determinants of interindividual variation in total CYP1A2 expression. Differential gene expression between the two mRNA expression and enzyme activity. The results indicate alleles or allele-specific expression (ASE) in nonimprinted the importance of two core methylation CpG sites located genes is also common in the human genome.16–19 The near the transcription start site to be associated with degree of differences in the expression between the two variations in the total CYP1A2 transcript level and differ- alleles varies from individual to individual causing pheno- ential allelic expression phenotype (Figure 1). typic variability. ASE reflects the presence of putative allele- specific cis-acting factors of either genetic or epigenetic Results origin. Consequently, ASE can be used as a promising quantitative phenotyping method for identifying cis-acting Detection of ASE in CYP1A2 factors as well as epigenetic variation influencing gene A total of 65 human liver samples were genotyped for the expression because individuals heterozygous for cis-acting marker SNP rs2470890 and of those 23 were found to be polymorphism show a different level of mRNA expression heterozygous for this SNP. The regions encompassing the 16,17,20 originating from one allele compared to the other. rs2470890 were PCR amplified from both genomic DNA The two major mechanisms for epigenetic modifications (gDNA) and mRNA (after conversion to cDNA), followed by are DNA methylation and histone deacetylation, and minisequencing analysis of each allele. PCR testing for interindividual variations in the degree of these mechan- gDNA and cDNA samples was carried out using the same isms could form a basis for interindividual differences in conditions for all fragments, and the amplicon size was gene expression. Epigenetic modifications and ASE in verified by agarose gel electrophoresis. The averages of the pharmacogenetic-related genes have been shown to affect signal fraction (SAllele1/(SAllele1 þ SAllele2)) from the cDNA expression and activity. ASE for the MDR1, CYP2D6 and were compared with the signal fraction from the respective 21 22 CYP2C9 genes has been reported recently. Hirota et al. gDNA as a reference to determine the presence of reported the occurrence of ASE in the CYP3A4 gene differential allelic expression. Of the 23 heterozygous liver expression that correlated with the total CYP3A4 mRNA samples, 11 (47.8%) displayed significant differences in their levels in liver but no clear association between gene allelic expression (Po0.05, Figure 2). There was no correla- methylation and total mRNA levels was found. Nakajima tion in these samples between the total mRNA expression 23 et al. indicated a function of histone deacetylation and level and allelic expression ratio or ASE phenotype. DNA methylation in cell-specific inducibility of the human CYP1 family. Tissue-specific variation in the degree DNA methylation frequency in human livers and hepatic CYP1A2 of gene methylation in CYP2E1 gene is associated with mRNA expression CYP2E1 mRNA levels and enzyme activity that has been The region covering the CpG island (Figure 3) was PCR reported previously.24 Furthermore, Hammons et al.25 amplified from gDNA after bisulfite treatment. DNA methy- showed an association between interindividual variation in lation frequency was determined by pyrosequencing using human hepatic CYP1A2 expression and degree of methyla- six internal sequencing primers that cover overlapping tion of a CpG site (bp À2759) in the 50-flanking region of the regions. Complete data concerning the DNA methylation CYP1A2 gene, with hypermethylation correlated with frequencies and mRNA quantification were obtained from decreased CYP1A2 expression. However, the CpG site 48 human livers. Analysis of the methylation pattern investigated by Hammons et al. is not located within a revealed two domains with region-specific differences in CYP1A2 CpG island and is located far away from the the distribution and level of CpG methylation. The first transcription start site. Indeed, functional CpG islands domain (domain 1) contained CpG sites 5, 6, 7, 13, 17, 18 typically occur at or near the transcription start site of and 19, and the second domain (domain 2) consisted of

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 210

Figure 1 Schematic view of the region of the CYP1A2 gene analyzed in this study indicating the transcription start site in exon 1 (À832) and the location of the CpG island in relation to the translation start codon ATG site in exon 2.

Figure 3 Sequence of the fragment including the CYP1A2 CpG islands in exon 2 ( þ 72 to 494 bp) investigated in the present study before sodium bisulfite treatment. Each CpG site with the respective number- ing in superscript is indicated with the letter Y. The aryl hydrocarbon receptor (AhR) binding site is underlined. Analysis of the methylation pattern revealed two domains with region-specific differences in the distribution and level of CpG methylation. The first domain (domain 1) contained CpG sites 5, 6, 7, 13, 17, 18 and 19, and the second domain

Figure 2 Comparison of fluorescence signal fractions (SAllele1/(SAlle- (domain 2) consisted of CpG sites 8, 9, 10, 11, 12, 14, 15, 16, 17 and le1 þ SAllele2)) between cDNA and the respective genomic DNA in human 21. CpG site 17 was common for both domains. livers heterozygous for the marker coding single nucleotide polymorphisms (SNP) (rs2470890C/T). Human liver samples showing deferential allelic expression significantly deviated from equal molar expression or significant negative correlation between the overall mean allele-specific expression (Po0.05) are indicated above and below with methylation frequency of the 17 CpG sites and total CYP1A2 the arbitrary dashed lines. The individual human liver samples are P depicted by number. mRNA expression ( ¼ 0.018); thus, a lower extent of methylation associated to a higher CYP1A2 mRNA expression level. The mean methylation frequencies of domain 2 CpG sites 8, 9, 10, 11, 12, 14, 15, 16, 17 and 21. CpG site 17 correlated significantly with the total CYP1A2 mRNA levels was common for both domains (Figure 3). (P ¼ 0.02), but there was no correlation between the mRNA The distribution of the methylation frequency pattern levels and the methylation of domain 1 (P ¼ 0.18). We and level of variation in the whole CpG island, domain 1, further compared the CYP1A2 mRNA levels with the mean domain 2 and CpG sites 14 and 16 in relation to the hepatic methylation frequency of each CpG site within domain 2. CYP1A2 mRNA expression is shown in Table 1. We found a The highest negative correlation was found for the CpG sites

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 211

in CYP1A2 enzyme activity determination between the three probes: caffeine vs EROD (Po0.0001, r2 ¼ 0.75), caffeine vs phenacetin (P ¼ 0.0001, r2 ¼ 0.86) and phenacetin vs EROD (Po0.0001, r2 ¼ 0.86). The enzyme activity using caffeine, EROD and phenacetin was correlated significantly with mean CYP1A2 mRNA level (P ¼ 0.035, P ¼ 0.015 and P ¼ 0.029, respectively) and protein content (Po0.0001 for all three probes). There were no significant correlations between mRNA level and protein content (P ¼ 0.16, statistical power ¼ 28.0%, using a two-sided test and a set at 0.05).

Figure 4 Relative CYP1A2 mRNA expression in human hepatoma Correlations of CYP1A2 genotype with CYP1A2 phenotype, ASE B16A2 cells treated with various concentrations of the hypomethylating and DNA methylation frequency agent 5-aza-20-deoxycytidine (AzaC) for 5 days. CYP1A2 gene expres- To investigate linkage of the marker SNP rs2470890 with sion increased linearly with the concentration of AzaC. other known CYP1A2 SNPs that may have influence on the mRNA expression, we genotyped the livers for the known 14 (P ¼ 0.006) and 16 (P ¼ 0.009), located in middle of CYP1A2 variants (À3858 G4A, À2467 delT, À739T4G, domain 2, whereas no correlation was found with the other À729 C4T and À163 C4A). The SNP frequencies of CpG sites. À3860G4A, À2467delT, À163C4A and rs2470890 were 5.0, 5.0, 67.5 and 62.5%, respectively. CYP1A2*1K Association of methylation frequency with ASE (À729C4T and À739T4G) was not detected. Linkage There was no significant correlation between the overall analysis using Arlequin population genetics software in- mean methylation frequency of the 17 CpG sites with ASE dicated significant linkage of rs2470890 with À163C4A phenotype in the 23 human livers used in this study occurring at a frequency of 62.5% but not with the other (P ¼ 0.11). Considering the methylation frequency of the SNPs. However, none of the individual CYP1A2 SNPs/ two domains separately, no correlation was found with haplotypes including the haplotype comprising the marker domain 1 but a borderline correlation with mean methyla- SNP rs2470890 and À163C4A showed any significant tion frequency of domain 2 was observed (P ¼ 0.054). We correlation with CYP1A2 mRNA expression level, protein further analyzed associations between the mean methyla- content, enzyme activity as determined by three different tion frequencies of each individual CpG site located in probes (caffeine, EROD and phenacetin), ASE or variations domain 2 with the ASE phenotype. The result indicated a in DNA methylation frequencies in human livers. significant correlation between the mean methylation frequency of CpG sites 9 (r2 ¼ 0.21, P ¼ 0.04) and 11 Discussion (r2 ¼ 0.33, P ¼ 0.006) located in the upper region of domain 2 with ASE phenotype in that a lower methylation We investigated the function of gene methylation and frequency predicted a higher variation in the relative differential allelic expression in CYP1A2 gene regulation transcript level derived from the two alleles. Neither the with the aim to understand the molecular basis behind the methylation frequency at CpG sites 9 and 11 nor the ASE observed wide variation in CYP1A2 gene expression in phenotype correlated with the total hepatic CYP1A2 mRNA human liver. The results indicated ASE of the CYP1A2 gene levels. that is independent of the total CYP1A2 mRNA expression. We were furthermore able to identify two independent core DNA hypomethylation and CYP1A2 mRNA level in hepatoma methylation sites located close to the translation start site as B16A2 cells determinants of interindividual variation in the CYP1A2 To clarify whether DNA demethylation increases the mRNA levels and ASE phenotype, respectively, presenting an CYP1A2 gene expression, we treated human hepatoma additional insight into the regulation of CYP1A2 gene to B16A2 cells with varying concentrations of a hypomethylat- cause altered expression and enzyme activity in humans. ing agent 5-aza-20-deoxycytidine (AzaC). The CYP1A2 mRNA We investigated whether the marker SNP (rs2470890) used expression level, determined by reverse transcriptase (RT)– in the ASE and DNA methylation analysis is in linkage PCR, linearly increased (r2 ¼ 0.82, Po0.01) with the amount disequilibrium with other known CYP1A2 SNPs that might of AzaC used, indicating that the gene is silenced by influence the mRNA expression and enzyme activity. methylation in the cells (Figure 4). Indeed, the rs2470890 was linked to À163C4A. However, none of the individual SNPs/haplotypes including the Correlations between CYP1A2 mRNA expression, protein content marker SNP plus À163C4A haplotype had significant and enzyme activity influences on mRNA expression, protein content, enzyme CYP1A2 enzyme activity was measured in the 21 human activity, differential expression of the alleles or variations livers using 3 different probes, caffeine, phenacetin and 7- in DNA methylation frequencies in human livers. Recently, ethoxyresorufin (EROD). There were significant correlations we reported that none of the CYP1A2 SNPs/haplotypes

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 212

investigated for affected enzyme activity using caffeine as a screen could be in linkage disequilibrium to another SNP probe in Swedish and Korean populations.12 Accordingly, that is responsible for the differential expression of allelic sequencing of the CYP1A1_CYP1A2 locus of the highest and transcripts, but that does not show unidirectional ASE due lowest CYP1A2 subjects worldwide, who have previously to noncomplete linkage disequilibrium with the marker been CYP1A2 phenotyped, revealed that no SNP or SNP. haplotype in the CYP1A2 gene has yet been identified that We used the Pyrosequencing technology that is ideally can unequivocally be used to predict the metabolic suited for the simultaneous analysis and quantification of phenotype.11 Lack of CYP1A2 the observed genotype– the methylation degree of several CpG positions in proxi- phenotype relationship was indeed the basis for the present mity. The results indicate that the overall DNA methylation study; that is, searching for additional factors such as frequency of the 17 CpG sites in the CpG island located epigenetic that may influence interindividual variations in close to the translation start site in exon 2 most likely CYP1A2 expression level in human livers. determine individual variation in total CYP1A2 mRNA The CYP1A2 gene is specifically expressed in the liver why expression levels in human livers. analysis of ASE and gene methylation was analyzed in 65 Previous studies indicated that a group of non-X-linked different human liver samples. About half of the livers bona fide promoter CpG islands are densely methylated in investigated by the ASE assay displayed significant variation normal somatic tissues and that DNA methylation has an in the levels of transcript derived from the two alleles. Some important function in silencing germ-cell-specific genes in genes that display ASE are present on chromosomes that somatic cells.34 Thus, we evaluated the effect of DNA contain other imprinted genes,16 which is also warrant for methylation on CYP1A2 gene expression and investigated the CYP1A2 gene on chromosome 15 where other imprinted whether demethylation alleviates the silencing pressure to genes in the region are present.27–29 increase gene expression in human B16A2 cell lines. We found no significant correlation between ASE and Hepatoma B1A2 cells were treated with increasing concen- total CYP1A2 gene expression at the mRNA level. Differ- trations of a hypomethylating agent AzaC and relative ential allelic expression is influenced mainly by cis-acting quantification of CYP1A2 mRNA by RT–PCR revealed a mechanisms affecting gene regulation and/or mRNA proces- linear raise in CYP1A2 gene expression with increasing sing, whereas total gene expression is influenced by both cis- concentration of AzaC, illustrating that DNA methylation is and trans-acting factors. Measuring the relative allelic mRNA involved in the repression of CYP1A2 (Figure 4). expression level selectively detects cis-acting factors and/or Comparison of mean methylation frequency of each CpG epigenetic factors.30,31 Consequently, the observed differ- site within domain 2, where methylation primarily corre- ences between ASE and CYP1A2 gene expression levels lated to CYP1A2 mRNA levels, indicated significant correla- indicate that the CYP1A2 gene is also regulated by trans- tion with CpG sites 14 and 16 (located in the middle of acting mechanisms that differ between individuals. domain 2) whereas no association was found with other As indicated in Figure 2, the preferably expressed CYP1A2 CpG sites. This may indicate that CpG sites 14 and 16, as allele varied between livers and this bidirectional ASE could core methylation regulatory sites, are of importance in result from allelic heterogeneity of one or more cis-acting determining variations in CYP1A2 gene expression. Inter- regulatory polymorphisms present in coding, intronic and/ estingly, the level of and variance in the methylation or other noncoding regulatory sequences,32 or from DNA frequency among the human livers was the highest for methylation, histone acetylation and other epigenetic CpG sites 14 and 16 compared to the other CpG sites in the factors.16,33 Alternatively, the marker SNP used in the ASE island (Table 2).

Table 1 Summary of sequencing primers, number of CpGs, sequencing length and dispensation order used for sequencing of CpG island in CYP1A2 exon 2

Sequencing primers Number of CpGs Sequence length Dispensation order

Primer 1 316TCAGCTGTCACTGACTG TTTGGTATTGTTAAGGATGAGT Primer 2 425GACTCTGTGTGTGCTCTGTCTGGCTG AGTTTGATTTTTAGTATAGATTTTG Primer 3 2 23 TCTGTCAGTGTCGCTG GAGTYGTTTGGATATTAT Primer 4 18TCACTCTG GGTTTAGAATGTTTTTAATATTTT Primer 5 436TCGCTATGCTATCGTCGTGTGTGTGAGTCG GGGAYGTTTTGTAGAT Primer 6 319GTCGATCGACTAGTCG AGGTTTTGGTGYGGTAGG

Cytosine used as internal controls to verify completion of sodium bisulfite treatment is indicted in bold in their dispensation orders. Underlined TCs are CpG sites. Y is either a C or T.

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 213

Table 2 Distribution and level of variation in methylation frequency in the CYP1A2 CpG island, different domains, core methylation sites (CpG 14 and 16) and their correlation with relative hepatic CYP1A2 mRNA level in 48 human livers using regression analysis

CpG sites Mean±s.d. Variance Range Correlation with hepatic CYP1A2 mRNA level F P

Overall 17 CpG sites 25.5±3.3 10.9 19.4–33.8 5.96 0.018 Domain 1 24.8±3.8 14.7 15.6–32.1 1.88 0.177 Domain 2 28.5±4.0 16.4 21.0–39.1 5.70 0.021 CpG 14 30.0±5.8 32.2 18.0–46.0 8.28 0.006 CpG 16 32.0±7.1 50.7 18.0–49.0 7.39 0.009 CpG 14+CpG 16 31±6.1 37.2 18.0–47.0 8.70 0.005

List of CpG sites located in domains 1 and 2 is described in Figure 3.

Previous studies presented evidence for variation in the that both core methylation sites are surrounded by tran- degree of methylation as a cause for differential allelic scription binding sites for transcription factors such as expression.16,17,20,31 In the present study, although ASE BRAC1, TFAP3A, SP1, SPI1, GATA-2 and GATA-3. Given the phenotype did not correlate significantly with the overall fact that CYP1A2 is a hepatic P450, we found that variation methylation frequency of the CpG island, region-specific in the methylation frequency of the binding sites and hence comparison of methylation frequencies indicated a signifi- blocking access to transcription factors might be important cant correlation between the methylation frequency of CpG in controlling the level of hepatic CYP1A2 gene expression. sites 9 and 11 (located in the upstream region of domain 2) It is also of interest to note that the aryl hydrocarbon with the ASE phenotype. Thus, a lower methylation receptor (AhR) binding site is located close to the CpG island frequency indicated a higher ASE phenotype. This indicates (Figure 3) and thus variation in degree of gene methylation that in the absence of a silencing pressure from gene might also result in differential CYP1A2 inducibility. methylation, the cis-acting factor responsible for the ASE Methylation patterns may be cell type and/or tissue phenotype exerts its effect. It is interesting to note that specific and change slowly with age and in response to similar to the ASE phenotype, methylation frequency of environmental effects such as diet.37,38 Variation in methy- CpG sites 9 and 11 was also not associated with the total lation frequency due to dietary related or environmental CYP1A2 mRNA expression level. This indicates that factors may be important in determining the observed wide although methylation frequency at CpG sites 9 and 11 ethnic variation in CYP1A2 enzyme activity between whites may contribute to the presence of differential allelic and Asians.12 Furthermore, previous pedigree studies de- expression, CYP1A2 mRNA expression levels are probably monstrated heritability of ASE being transmitted by Mende- influenced to a major extent by the degree of methylation of lian inheritance.17,39 Gene methylation patterns in somatic the whole CpG island, mainly at the CpG sites 14 and 16. differentiated cells are also generally stable and heritable.35 Hammons et al.25 reported the function of a single site- Inheritance of ASE in the CYP3A4 gene expression has been specific DNA methylation status of the CCGG site (bp previously reported.22 Accordingly the observed ASE in À2759) located adjacent to an AP-1 site to be responsible for CYP1A2 gene expression may be inherited to cause variation the interindividual variation of CYP1A2. This site, not any in enzyme activity similar to genetic polymorphisms. other place 5 kb upstream of the gene, could however be Several studies reported lack of genotype–phenotype rela- identified by us as a CpG site in a CpG island. Instead, we tionship in CYP1A2. Our study explains the molecular found a CpG island containing 17 CpG sites, located close to mechanism behind variations in CYP1A2 gene expression the translation start site, and considered it of importance to and genotype–phenotype concordance to a certain extent. investigate further. The magnitude of ASE as well as methylation frequency Silencing of one of the X chromosomes in females and varies between individuals and possibly affects enzyme allelic imprinting or monoallelic expression, depending on activity and treatment response. parental origin, is highly regulated through gene methyla- In conclusion, our study demonstrates that CYP1A2 gene tion reprogramming that may occur in germ cells and expression displays ASE and is influenced by DNA methyla- during gametogenesis.35,36 However, the function of CpG tion. We showed that only a few CpG sites and quite small methylation in nonimprinted, monoallelic expressing genes regions of the CpG island provide relevant information for is not yet fully explored. The two identified core methyla- differential methylation analysis. The present study has tion sites are located in the same domain close to each other given a new insight into the regulation of CYP1A2. but apparently having two different functions: the upper However, further investigations are needed to explore the region determines the ASE phenotype and the middle region exact mechanisms involved in ASE, transregulatory mechan- determines the total CYP1A2 mRNA expression. Searching isms and DNA methylation in explaining interindividual for transcription binding sites in the CpG island, we found and ethnic variability in CYP1A2 expression. ASE discovery

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 214

coupled with elucidation of the underlying mechanisms previously.30,40 In brief a 164-bp PCR fragment spanning may provide a comprehensive view of transcriptional the marker SNP site was amplified by PCR using a forward regulation as well as functional markers for drug metaboliz- primer (50-ATGAAGCACGCCCGCTGT-30) and a reverse ing genes (for example, CYP1A2 and CYP3A4) where genetic primer (50-TTGGCTAAAGCTGCTATTTTTTA-30) from both polymorphisms have a minor function in explaining the gDNA and cDNA. To remove remaining dNTPs and primers, observed wide interindividual and ethnic variability in we treated the PCR products (7.2 ml) with 1 U mlÀ1 shrimp enzyme activity. alkaline phosphatase and 0.48 U mlÀ1 exonuclease I (USB Corporation, Cleveland, OH, USA). The oligonucleotides 0 Materials and methods complementary to the tag sequence attached to the 5 -end of the minisequencing primer (50-AATACGCTGAATAG 0 Study samples AGCCCTAGGCGCGGCTGCGCTTCTCCATCAA-3 ) were A total of 65 unrelated healthy human livers of Scandina- immobilized on the CodeLink Activated Slides (GE vian origin belonging to a donor liver biobank, established Healthcare, Uppsala, Sweden) in duplicate spots in each at the Division of Clinical Pharmacology, Karolinska Uni- subarray. Cyclic minisequencing of the PCR products from versity Hospital Huddinge, Sweden, were utilized for this both gDNA and cDNA, hybridization of the minisequencing study after approval from the ethics committee at Karolinska reaction products to the captured oligonucleotides (cTags) Institutet. Tissue for preparation of microsomes was only on the slides and washing of the slides were performed as 41 available from 21 human livers. described previously. Each sample (DNA and the corresponding cDNA) was run in triplicates on the same Preparation of genomic DNA and cDNA slide. Genomic DNA and total RNA were isolated from the liver samples using QIAamp DNA Mini Kit and RNeasy Mini Kit Signal detection and data analysis. Fluorescence signals from (Qiagen, Hilden, Germany), respectively. Total RNA was the microarrays were measured using a ScanArray Express treated with RNase-free DNase (Promega, Madison, WI, USA) instrument (PerkinElmer Life Sciences, Boston, MA, USA). to remove any contamination from gDNA during total RNA The laser power was kept constant at 80% and the preparation. cDNA was synthesized using the First Strand photomultiplier tube gain adjusted to obtain equal signal cDNA Synthesis Kit (Fermentas, Burlington, Ontario, Cana- levels from the reaction control spots in all four-laser da), following the manufacturer’s instructions using 2 mgof channels. The excitation lasers were Blue Argon 488 nm total RNA. cDNA was prepared from each liver using random for R110, green HeNe 543.8 nm for Tamra, yellow HeNe hexamer and oligo (dT)18 primer. 594 nm for Texas Red and red HeNe 632.8 nm for Cy5. The signal intensities were determined with the QuantArray Detection of differential allelic expression by Tag array analysis 3.1 software (PerkinElmer Life Sciences). The SNP minisequencing genotype were assigned using the SNPSnapper software version Assay design. Several SNPs in the CYP1A2 gene were 3.0.0.191 (http://www.bioinfo.helsinki.fi/SNPsnapper/) based considered as candidate markers for the ASE assay. on scatter plots with the logarithm of the sum of both

Selection criteria for a marker SNP were high minor allele fluorescence signals (SAllele1 þ SAllele2) on the vertical axis and frequency with no significant effect on enzyme expression the fluorescence signal fractions (SAllele1/(SAllele1 þ SAllele2)) level or enzyme activity and location in the transcribed on the horizontal axis. ASE for each heterozygous individual coding region so that the same set of primers for PCR was determined by calculating the fluorescence signal amplification can be used in both gDNA and mRNA. A between the two alleles (SAllele1/(SAllele1 þ SAllele2)) in both nonfunctional silent SNP (5347T4C, rs2470890) located in cDNA and the respective gDNA as a reference for each exon 7 of the CYP1A2 gene was selected as marker. heterozygous liver sample. The mean signal intensity of the Previously we identified this SNP (rs2470890) in duplicate spots run in triplicates on one subarray was Ethiopians and reported lack of influence on CYP1A2 considered as one replicate assay. enzyme activity using caffeine as a probe.13 To examine linkage of the marker SNP with other CYP1A2 SNPs, we Determination of DNA methylation frequency by pyrosequencing genotyped livers for known CYP1A2 variant SNPs Assay design. Searching for CpG islands in the CYP1A2 gene (À3858G4A, À2467delT, À739T4G, À729C4T, À163C4A) covering up to 5000 bp upstream region from the translation as described previously.12 start codon was carried out using Methyl Primer Express PCR primers and minisequencing primers with 50 tag software (Applied Biosystems, Foster City, CA, USA). We sequences covering the marker SNP site were designed using found one CpG island close to the transcription start site in the Autoprimer (http://www.autoprimer.com; Beckman exon 2. Seventeen of the CpG sites located in the center of Coulter) software. The primers were obtained from Inte- the island were selected for DNA methylation frequency grated DNA Technologies (Coralville, IA, USA). analysis (Figure 1).

PCR minisequencing and hybridization. Genomic DNA and Bisulfite treatment and PCR cDNA were used for genotyping of the marker SNP using The gDNA isolated from human liver samples was treated 4 ng DNA by minisequencing analysis as described with sodium bisulfite using the EZ DNA Methylation Kit

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 215

(Zymo Research Corporation, Orange, CA, USA). PCR (Invitrogen) following manufacturer’s recommendations. primers were designed to amplify the CpG island containing Reverse transcription was carried out from 5 mg total RNA 17 CpG sites located in exon 2. A 309-bp region spanning using the SuperScript II (Invitrogen) system in the presence the island was amplified by PCR from bisulfate-treated of the ribonuclease inhibitor RNaseOUT (Invitrogen). gDNA samples. The PCR mix contained 2.5 ml10Â PCR buffer, 0.5 ml25mM MgCl2 solution, 0.125 ml Smart-Taq Hot Thermostable DNA Polymerase (Naxo Ltd., Tartu, Estonia), Quantitative real-time PCR 0.5 ml10mM of forward (50-TTTGGTATTGTTAAGGATGAGT The quantification of the total CYP1A2 mRNA levels from TAG-30) and biotinylated reverse (50-ACATACTCCTCCAAA human livers and B16A2 cell lines treated with varying TAACAAAAA-30) primers (Invitrogen, Carlsbad, CA, USA), concentrations of AzaC was performed by real-time PCR and 0.5 ml10mM dNTP (GE Healthcare), 1 ml bisulfite-treated using 7500 Fast Real-Time PCR system. The quantification DNA and sterile water in a final volume of 25 ml. The PCR mixture contained 7.5 ml of TaqMan 2 Â PCR Master Mix, conditions were initial activation of the Smart-Taq Hot 0.5 ml of cDNA, 0.75 mlof20Â mix of predesigned TaqMan enzyme at 95 1C for 10 min followed by 50 cycles of 95 1C for Gene Expressions Assay contained primers for CYP1A2 30 s, 53 1C for 1 min, 72 1C for 1 min and a final extension at mRNA spanning sequence boundary between exons 6 and 72 1C for 7 min on a GeneAmp PCR system 2700 (Applied 7 and probe in a final 15 ml volume of reaction. Each sample Biosystems). The PCR products were analyzed on a 1.5% was ran in triplicates and human 18S rRNA 20 Â was used as agarose gel for quality assurance and verification before endogenous control. The real-time PCR instrument, TaqMan pyrosequencing. master mix, primers and probe were purchased from Applied Biosystems. Pyrosequencing The amplified PCR product was analyzed by pyrosequencing using six internal sequencing primers designed to cover all Determination of CYP1A2 enzyme activity in human liver the 17 CpG sites. Sequencing primers, CpG sites and microsomes pyrosequencing dispensation orders including internal con- Microsomes were prepared from 21 human livers bank as 46 trol for bisulfite treatment are listed in Table 3. Pyrosequen- described previously. The pellet containing the microso- cing was performed on Pyro Q-CpG software (Biotage, mal fraction was suspended and homogenized in 50 mM Uppsala, Sweden) using the PyroGold SQA Reagent Kit potassium phosphate buffer, pH 7.4, to give a final (Biotage) as described previously.45 The software reports a concentration of 30–50 mg microsomal protein per ml and peak height directly proportional to the number of mole- stored at À80 1C until use. The protein content was 47 cules incorporated into the growing DNA chain. determined by Lowry method, and the total cytochrome P450 content was quantified.48 The CYP1A2 enzyme activity AzaC treatment and total RNA isolation from cultured cells was estimated by the using caffeine, EROD and phenacetin The human hepatoma cell lines B16A2 were cultured based as marker probes (Sigma-Aldrich, St Louis, MO, USA). Assay on ATCC protocols. All cell culture media and supplements conditions used for CYP1A2 enzyme activity determination were obtained from Invitrogen (Rockville, MD, USA). Cells in human liver microsomes using the three different probes were treated with AzaC (Sigma-Aldrich, Steinheim, are summarized in Table 3. Microsomes were diluted with Germany) at varying concentrations (0, 100, 500 and buffer and preincubated for 3 min at 37 1C before the probe 1000 mM) for 5 days. B16A2 cells were allowed to reach total substrate was added. Multiple control incubations were confluence and maintained at such level for 1 week before performed with boiled microsomes, without microsomes, AzaC treatment was commenced. Subsequent to AzaC without nicotinamide adenine dinucleotide phosphate treatments, total RNA was isolated using Trizol reagent (NADPH) or by adding stop solution before NADPH.

Table 3 Summary of assay conditions used for CYP1A2 enzyme activity determination in human liver microsomes using three different probes: caffeine, phenacetin and EROD

Activity measured Probe Solvent final Protein Buffer Cofactor Final Incubation Analytical assay conc. conc. (mg mlÀ1) final conc. incubation time (min) (mM) volume (ml)

Ethoxyresorufin 5 0.5% DMSO 0.05 100 mM potassium 0.25 mM 0.2 20 Fluorescence recorded O-deethylation phosphate (pH 7.8) NADPH in real time42 Caffeine 100 — 0.5 100 mM potassium 1mM 0.5 15 HPLC43 N-demethylation phosphate (pH 7.4) NADPH Phenacetin 200 2.5% 0.2 100 mM potassium 1mM 0.2 15 LC–MS44 O-deethylation Acetonitrile phosphate (pH 7.4) NADPH

Abbreviations: DMSO, dimethyl sulfoxide; HPLC, high-pressure liquid chromatography; LC–MS, liquid chromatography-mass spectrometry; NADPH, nicotinamide adenine dinucleotide phosphate.

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 216

CYP1A2 protein content determination 3 Tassaneeyakul W, Birkett DJ, Veronese ME, McManus ME, Tukey RH, Determination of CYP1A2 protein content was carried out as Quattrochi LC et al. Specificity of substrate and inhibitor probes for described previously.49 In brief, a total of 5 mg protein was human cytochromes P450 1A1 and 1A2. J Pharmacol Exp Ther 1993; 265: 401–407. separated by SDS–polyacrylamide gel electrophoresis using a 4 Kroemer HK, Gautier JC, Beaune P, Henderson C, Wolf CR, Eichelbaum 10% gel. After transfer, the Hybond-C extra membrane M. Identification of P450 enzymes involved in metabolism of verapamil (Amersham Pharmacia Biotech, Uppsala, Sweden) was in humans. Naunyn Schmiedebergs Arch Pharmacol 1993; 348: 332–337. blocked in Tris-buffered saline containing 0.05% (v/v) 5 Hartter S, Ursing C, Morita S, Tybring G, von Bahr C, Christensen M et al. Orally given melatonin may serve as a probe drug for cytochrome Tween 20 and 5% fat-free milk and incubated with 1:2500 P450 1A2 activity in vivo: a pilot study. Clin Pharmacol Ther 2001; 70: dilution of the primary CYP1A1/CYP1A2 polyclonal anti- 10–16. body (Daiichi Pure Chemicals Co. Ltd., Ibaraki, Japan). This 6 Yamazaki H, Shaw PM, Guengerich FP, Shimada T. Roles of was followed by incubation with a 1:2500 diluted goat anti- cytochromes P450 1A2 and 3A4 in the oxidation of estradiol and estrone in human liver microsomes. Chem Res Toxicol 1998; 11: rabbit horseradish peroxidase-conjugated antibody (Pierce, 659–665. Rockford, IL, USA). Detection was performed using Super- 7 McManus ME, Burgess WM, Veronese ME, Huggett A, Quattrochi LC, Signal West Pico Chemiluminescent Substrate (Pierce). Tukey RH. Metabolism of 2-acetylaminofluorene and benzo(a)pyrene and activation of food-derived heterocyclic amine mutagens by human cytochromes P-450. Cancer Res 1990; 50: 3367–3376. Statistics 8 Eaton DL, Gallagher EP, Bammler TK, Kunze KL. Role of cytochrome A two-tailed Student’s t-test was used to compare the mean P4501A2 in chemical carcinogenesis: implications for human variability in expression and enzyme activity. Pharmacogenetics 1995; 5: 259–274. allelic signal (SAllele1/(SAllele1 þ SAllele2)) between gDNA and 9 Schweikl H, Taylor JA, Kitareewan S, Linko P, Nagorney D, Goldstein JA. cDNA for each heterozygous liver. Analysis of variance was Expression of CYP1A1 and CYP1A2 genes in human liver. Pharmacoge- used to analyze association between CYP1A2 genotype, netics 1993; 3: 239–249. mRNA expression protein content and enzyme activity, 10 Rodriguez-Antona C, Donato MT, Pareja E, Gomez-Lechon MJ, Castell using all three marker probes. Haplotype analysis and JV. Cytochrome P-450 mRNA expression in human liver and its estimation of expected haplotype frequencies from the raw relationship with enzyme activity. Arch Biochem Biophys 2001; 393: 308–315. genotype data were carried out using population genetic 11 Jiang Z, Dragin N, Jorge-Nebert LF, Martin MV, Peter Guengerich F, software program Arlequin version 2.000. The Spearman’s Aklillu E et al. Search for an association between the human CYP1A2 rank correlation test and regression analysis (after convert- genotype and CYP1A2 metabolic phenotype. Pharmacogenet Genomics ing to log values) were used to analyze association between 2006; 16: 359–367. 12 Ghotbi R, Christensen M, Roh HK, Ingelman-Sundberg M, Aklillu E, mean methylation frequency with ASE phenotype and total Bertilsson L. Comparisons of CYP1A2 genetic polymorphisms, enzyme mRNA level. activity and the genotype-phenotype relationship in Swedes and Koreans. Eur J Clin Pharmacol 2007; 63: 537–546. 13 Aklillu E, Carrillo JA, Makonnen E, Hellman K, Pitarque M, Bertilsson L Duality of interest et al. Genetic polymorphism of CYP1A2 in Ethiopians affecting induction and expression: characterization of novel haplotypes with None declared. single-nucleotide polymorphisms in intron 1. Mol Pharmacol 2003; 64: 659–669. 14 Djordjevic N, Ghotbi R, Bertilsson L, Jankovic S, Aklillu E. Induction of CYP1A2 by heavy coffee consumption in Serbs and Swedes. Eur J Clin Abbreviations Pharmacol 2008; 64: 381–385. 15 Wray GA, Hahn MW, Abouheif E, Balhoff JP, Pizer M, Rockman MV et al. CYP1A2 cytochrome P450, family 1, subfamily A, polypeptide 2 The evolution of transcriptional regulation in eukaryotes. Mol Biol Evol ASE allele-specific expression 2003; 20: 1377–1419. CpG cytosine-guanine 16 Lo HS, Wang Z, Hu Y, Yang HH, Gere S, Buetow KH et al. Allelic variation SNP single nucleotide polymorphism in gene expression is common in the human genome. Genome Res EROD 7-ethoxyresorufin 2003; 13: 1855–1862. 17 Pastinen T, Sladek R, Gurd S, Sammak A, Ge B, Lepage P et al. A survey of genetic and epigenetic variation affecting human gene expression. Physiol Genomics 2004; 16: 184–193. 18 Yan H, Zhou W. Allelic variations in gene expression. Curr Opin Oncol Acknowledgments 2004; 16: 39–43. 19 Serre D, Gurd S, Ge B, Sladek R, Sinnett D, Harmsen E et al. Differential This study was financially supported by the Swedish Research allelic expression in the human genome: a robust approach to identify Council for Medicine, the Swedish Research Council for Science and genetic and epigenetic cis-acting mechanisms regulating gene expres- Technology, and by Torsten and Ragnar So¨derbergs Stiftelser. sion. PLoS Genet 2008; 4: e1000006. 20 Tao H, Cox DR, Frazer KA. Allele-specific KRT1 expression is a complex trait. PLoS Genet 2006; 2: e93. References 21 Johnson AD, Zhang Y, Papp AC, Pinsonneault JK, Lim JE, Saffen D et al. Polymorphisms affecting gene transcription and mRNA processing in 1 Shimada T, Yamazaki H, Mimura M, Inui Y, Guengerich FP. Inter- pharmacogenetic candidate genes: detection through allelic expression individual variations in human liver cytochrome P-450 enzymes imbalance in human target tissues. Pharmacogenet Genomics 2008; 18: involved in the oxidation of drugs, carcinogens and toxic chemicals: 781–791. studies with liver microsomes of 30 Japanese and 30 Caucasians. 22 Hirota T, Ieiri I, Takane H, Maegawa S, Hosokawa M, Kobayashi K et al. J Pharmacol Exp Ther 1994; 270: 414–423. Allelic expression imbalance of the human CYP3A4 gene and individual 2 Bertilsson L, Carrillo JA, Dahl ML, Llerena A, Alm C, Bondesson U et al. phenotypic status. Hum Mol Genet 2004; 13: 2959–2969. Clozapine disposition covaries with CYP1A2 activity determined by a 23 Nakajima M, Iwanari M, Yokoi T. Effects of histone deacetylation and caffeine test. Br J Clin Pharmacol 1994; 38: 471–473. DNA methylation on the constitutive and TCDD-inducible expressions

The Pharmacogenomics Journal ASE and gene methylation of CYP1A2 R Ghotbi et al 217

of the human CYP1 family in MCF-7 and HeLa cells. Toxicol Lett 2003; 37 Bird A. DNA methylation patterns and epigenetic memory. Genes Dev 144: 247–256. 2002; 16: 6–21. 24 Botto F, Seree E, el Khyari S, de Sousa G, Massacrier A, Placidi M et al. 38 Waterland RA, Jirtle RL. Early nutrition, epigenetic changes at Tissue-specific expression and methylation of the human CYP2E1 gene. transposons and imprinted genes, and enhanced susceptibility to adult Biochem Pharmacol 1994; 48: 1095–1103. chronic diseases. Nutrition 2004; 20: 63–68. 25 Hammons GJ, Yan-Sanders Y, Jin B, Blann E, Kadlubar FF, Lyn-Cook BD. 39 Yan H, Yuan W, Velculescu VE, Vogelstein B, Kinzler KW. Allelic variation Specific site methylation in the 50-flanking region of CYP1A2 inter- in human gene expression. Science 2002; 297: 1143. individual differences in human livers. Life Sci 2001; 69: 839–845. 40 Lindroos K, Sigurdsson S, Johansson K, Ronnblom L, Syvanen AC. 26 Saxonov S, Berg P, Brutlag DL. A genome-wide analysis of CpG Multiplex SNP genotyping in pooled DNA samples by a four-colour dinucleotides in the human genome distinguishes two distinct classes of microarray system. Nucleic Acids Res 2002; 30: e70. promoters. Proc Natl Acad Sci USA 2006; 103: 1412–1417. 41 Lovmar L, Fredriksson M, Liljedahl U, Sigurdsson S, Syvanen AC. 27 Cassidy SB, Dykens E, Williams CA. Prader-Willi and Angelman Quantitative evaluation by minisequencing and microarrays reveals syndromes: sister imprinted disorders. Am J Med Genet 2000; 97: accurate multiplexed SNP genotyping of whole genome amplified 136–146. DNA. Nucleic Acids Res 2003; 31: e129. 28 Robinson WP, Horsthemke B, Leonard S, Malcolm S, Morton C, Nicholls 42 Burke MD, Mayer RT. Ethoxyresorufin: direct fluorimetric assay of a RD et al. Report of the Third International Workshop on Human microsomal O-dealkylation which is preferentially inducible by 3- Chromosome 15 Mapping 1996. October 25-27, 1996 in Vancouver methylcholanthrene. Drug Metab Dispos 1974; 2: 583–588. B.C., Canada. Cytogenet Cell Genet 1997; 76:1–13. 43 Tassaneeyakul W, Mohamed Z, Birkett DJ, McManus ME, Veronese ME, 29 Nicholls RD, Knoll JH, Butler MG, Karam S, Lalande M. Genetic Tukey RH et al. Caffeine as a probe for human cytochromes imprinting suggested by maternal heterodisomy in nondeletion Prader- P450: validation using cDNA-expression, immunoinhibition and Willi syndrome. Nature 1989; 342: 281–285. microsomal kinetic and inhibitor techniques. Pharmacogenetics 1992; 30 Milani L, Gupta M, Andersen M, Dhar S, Fryknas M, Isaksson A et al. 2: 173–183. Allelic imbalance in gene expression as a guide to cis-acting regulatory 44 Distlerath LM, Reilly PE, Martin MV, Davis GG, Wilkinson GR, single nucleotide polymorphisms in cancer cells. Nucleic Acids Res 2007; Guengerich FP. Purification and characterization of the human liver 35: e34. cytochromes P-450 involved in debrisoquine 4-hydroxylation and 31 Pinsonneault JK, Papp AC, Sadee W. Allelic mRNA expression of X-linked phenacetin O-deethylation, two prototypes for genetic polymorphism monoamine oxidase a (MAOA) in human brain: dissection of epigenetic in oxidative drug metabolism. J Biol Chem 1985; 260: 9057–9067. and genetic factors. Hum Mol Genet 2006; 15: 2636–2649. 45 Tost J, Gut IG. Analysis of gene-specific DNA methylation patterns 32 Hardison RC, Oeltjen J, Miller W. Long human-mouse sequence by pyrosequencing technology. Methods Mol Biol 2007; 373: alignments reveal novel regulatory elements: a reason to sequence 89–102. the mouse genome. Genome Res 1997; 7: 959–966. 46 von Bahr C, Groth CG, Jansson H, Lundgren G, Lind M, Glaumann H. 33 Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the Drug metabolism in human liver in vitro: establishment of a human liver genome integrates intrinsic and environmental signals. Nat Genet 2003; bank. Clin Pharmacol Ther 1980; 27: 711–725. 33: (Suppl) 245–254. 47 Lowry OH, Rosebrough NJ, Farr AL, Randall RJ. Protein measurement 34 Shen L, Kondo Y, Guo Y, Zhang J, Zhang L, Ahmed S et al. Genome- with the Folin phenol reagent. J Biol Chem 1951; 193: 265–275. wide profiling of DNA methylation reveals a class of normally 48 Omura T, Sato R. The carbon monoxide-binding pigment of liver methylated CpG island promoters. PLoS Genet 2007; 3: 2023–2036. microsomes, I. Evidence of its hemoprotein nature. J Biol Chem 1964; 35 Reik W, Dean W, Walter J. Epigenetic reprogramming in mammalian 239: 2370–2378. development. Science 2001; 293: 1089–1093. 49 Karlgren M, Gomez A, Stark K, Svard J, Rodriguez-Antona C, Oliw E et 36 Lees-Murdock DJ, Walsh CP. DNA methylation reprogramming in the al. Tumor-specific expression of the novel cytochrome P450 enzyme, germ line. Epigenetics 2008; 3: 5–13. CYP2W1. Biochem Biophys Res Commun 2006; 341: 451–458.

The Pharmacogenomics Journal