The serotonin-2A receptor gene locus does not contain ... - Nature

Molecular Psychiatry (2004) 9, 109–114 & 2004 Nature Publishing Group All rights reserved 1359-4184/04 $25.00 www.nature.com/mp

ORIGINAL RESEARCH ARTICLE

The serotonin-2A receptor gene locus does not contain common polymorphism affecting mRNA levels in adult brain NJ Bray1, PR Buckland1, H Hall2, MJ Owen1 and MC O’Donovan1 1 Department of Psychological Medicine, University of Wales College of Medicine, Heath Park, Cardiff, UK; 2Department of Clinical Neuroscience, Karolinska Institute, Karolinska Hospital, Stockholm, Sweden

Keywords: HTR2A; 5HT2A; expression; epigenetic; poly-

morphism The serotonin-2A (HTR2A) receptor is a molecule of particular interest in biological psychiatry, as it is an important target for psychotropic drugs,1,2 and altered HTR2A expression has been found in several neuropsychiatric conditions, including depression3 and schizophrenia.4 Genetic association has been reported between a synonymous 102T/C polymorphism in the gene encoding HTR2A and a number of clinical phenotypes, including schizophrenia,5,6 clozapine response,7 psychotic symptoms in Alzheimer’s disease8 and certain features of depression.9 Given that there are no known effects of the 102T/C polymorphism on the structure of the receptor, attention has switched to the possibility that the observations of both altered expression and genetic association point to functional sequence variants that alter expression of the HTR2A gene.10 Moreover, data have been presented recently suggesting that mRNAs containing the 102T- and C-alleles are differentially expressed.11 This suggests a direct effect of the variant itself on mRNA levels, or the influence of a distinct regulatory variant, such as the 1438A/G promoter polymorphism, with which it is in perfect linkage disequilibrium.12 The present study tested this hypothesis by employing a highly accurate quantitative allele- specific primer extension assay13 to measure the relative expression of brain mRNAs carrying each allele in 23 individuals heterozygous for the 102T/C polymorphism. Comparison between allele ratios derived from genomic DNA and mRNA from several cortical regions revealed that the 102C- and T-alleles are expressed identically. Furthermore, the absence of any interindividual variability in relative mRNA allele ratio suggests that the HTR2A locus is unlikely to contain common polymorphisms or epigenetic modification that alter HTR2A mRNA levels in adult brain, and essentially exclude such phenomena as a potential explanation for the altered expression and genetic associations that have been reported to date. Molecular Psychiatry (2004) 9, 109–114. doi:10.1038/ sj.mp.4001366

In order to estimate the expression levels of allelic forms of a single mRNA species, a number of investigators have previously applied methods to quantify the relative expression of each allele in

individuals who are heterozygous for an expressed SNP (eg Singer-Sam et al14). In heterozygous individuals, each allele can be used as an internal control against which the other allele can be measured within each individual sample. This allows identification of genes showing altered expression as a direct consequence of the marker polymorphism, other cis-acting regulatory polymorphisms, or allele-specific epigenetic modification, while controlling for the random variables that confound mRNA measurements, particularly in postmortem tissue. These include differences in tissue and mRNA quality, environmental influences like drugs or preagonal state, or secondary differential expression as a result of intrinsic transacting agents such as hormones. Owing to these inherent advantages, we have applied this method to survey the HTR2A gene for cis-acting polymorphisms and allele-specific epigenetic phenomena that might alter its expression, and which might therefore explain the observations of altered HTR2A gene expression and/or genetic associations in a number of psychiatric disorders. We chose to use allele-specific SNaPshot primer extension chemistry to measure the relative expression of the HTR2A 102C- and 102T-alleles because this assay has been shown both by ourselves13 and by others15 to give accurate measurements of relative allele quantity in DNA pools, and has most recently been successfully applied to the quantitative measurement of mRNA allelic species.16,17 Relative expression of each allele was measured by comparing the peak height ratios of each allele-specific extension product derived from the SNaPshot analysis of cortical RNA samples obtained from 23 102T/C heterozygotes with the corresponding ratios derived from the genomic DNA from the same individuals. The latter acts as a control for the ratio to be expected since this represents a perfect 0.5 : 0.5 ratio of the two alleles.17 The results of the individual analyses of the mRNA and genomic DNA are given in Figure 1. The mean of the ratios of the C/T alleles derived from genomic DNA was 1.18 (n ¼ 23, SD ¼ 0.07). This provides the correction factor that is required in order to adjust for the fact that equimolar concentrations of alleles are

Allelic expression of HTR2A NJ Bray et al

110

100%

% C allele

75%

50%

25%

0% 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

Genomic DNA 100%

% C allele

75%

50%

25%

0% 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

mRNA

Figure 1 Comparison between relative measures of the HTR2A 102T- and C-alleles in 23 genomic heterozygotes (above) and corresponding mRNA samples (below). Percentage C-allele is calculated by correcting the measured peak heights with the mean peak height ratio derived from SNaPshot analysis of all the genomic heterozygotes (representing a perfect 0.5 : 0.5 ratio of the two alleles), as originally described in Hoogendoorn et al.18

not always equally represented.18 The average peak height ratio for cDNA samples was 1.20 (n ¼ 23, SD ¼ 0.09), which was not significantly different from the value obtained from genomic DNA (paired t ¼ 1.11, P ¼ 0.28). When corrected for differential representation in genomic DNA, estimated values for the relative expression of the C- and T-alleles in cortical mRNA were 0.504 and 0.496, respectively. Molecular Psychiatry

Our findings based upon mean expression levels are therefore not compatible with the hypothesis that there is a significant direct effect of the 102T/C polymorphism on mRNA expression. This is in contrast to a recent report where mRNA allele ratios were measured using an RFLP-based assay.11 In that study, the authors reported that the C-allele was expressed at a level 20% lower than the T-allele, a finding that was mirrored in estimates of the total


amount of HTR2A mRNA and protein in individuals with the C/C, compared with the T/T, genotype. However, previous studies using this method to assess allele frequencies in DNA pools19 have yielded less accurate estimates than are obtained using the present methodology. The small error estimations in absolute allele frequency with SNaPshot are SNP dependent, but in a detailed assessment of the accuracy of SNaPshot, the mean error for the 102T/C SNP was 2%.13 Moreover, this estimate was based upon the inferred allele frequency after samples were pooled, and therefore this estimate includes any error arising from the pooling process. This is not an issue in the present study. In contrast, RFLP analysis is subject to artefacts associated with incomplete digestion of PCR products, resulting in overestimation of the allele represented by the uncut amplicon, which in this case is the T-allele. Moreover, although in the previous investigation of HTR2A expression, the authors attempted to control for differential expression by comparison with the results from genomic DNA, this is potentially inaccurate with an RFLP-based assay as the proportion of uncut material increases with the amount of the PCR product that is formed from heteroduplexes of each allele (which are not cut by the endonuclease). This proportion depends upon whether amplification has strayed beyond the exponential phase of PCR into the plateau phase, which is in turn dependent upon the number of starting template molecules. As this is unlikely to be similar between genomic DNA and mRNA, the former may not provide a suitable control for the latter in an RFLP assay. The present data are therefore likely to represent a more exact estimation of the relative expression of the 102T/C alleles. One difference between the present study and the previous study is that we used anonymized adult subjects, whereas they controlled for neuroleptic treatment history and have stratified their data based on the sample origin (healthy controls vs patients with schizophrenia). In studies where the total mRNA expression of a gene is compared between individuals, this would be a crucial distinction. However, as we have noted above, a major strength of the intraindividual allele estimations is that it controls the noise from trans-acting factors, including drug treatment, thereby revealing the effects of primary regulatory events that may potentially be of aetiological relevance. Our data do not address the findings of that study in relation to protein expression, other than to suggest that if one allele is more highly expressed at the protein level, this is likely to be the result of translational effects. However, there is as yet no consensus in the literature as postmortem receptor binding studies assessing differences in HTR2A protein expression between 102T/C genotypes have yielded inconsistent results.20,21,22 In our attempt to replicate the findings of the previous group, we used identical primers to amplify

HTR2A cDNA. However, this may also cause inaccuracies because these primers are not cDNA-specific, and therefore genomic DNA contamination will tend to normalize the data towards the genomic ratio of 0.5 : 0.5. To minimize this potential effect, all RNA samples were DNAse treated prior to RT-PCR. Samples treated this way did not generate detectable PCR product with these ‘universal’ primers in the absence of an RT cDNA synthesis step, suggesting that genomic contamination of cDNA is not an issue. Nevertheless, as an extra precaution, the assay was repeated using a separate set of cDNA-specific primers. Peak height ratios differed slightly from cDNA/genomic DNA amplified using ‘universal’ primers (n ¼ 23, mean ¼ 1.29, SD ¼ 0.10), translating to a corrected allelic expression of 0.52 : 0.48. This most likely reflects slight differences in the allele representation efficiencies using cDNA-specific primers compared with the universal primers, since, by definition, the precise correction factor for the former cannot be obtained from genomic DNA. Moreover, even if this marginal difference were genuine, the 102C allele would be the more highly expressed, which is the opposite of that found by Polesskaya and Sokolov.11 In order to determine if the findings are the result of sampling different brain regions, we compared peak height ratios derived from frontal cortex (n ¼ 11, mean ¼ 1.29, SD ¼ 0.09), parietal cortex (n ¼ 7, mean ¼ 1.28, SD ¼ 0.12) and temporal cortex (n ¼ 6, mean ¼ 1.30, SD ¼ 0.11). These ratios are effectively identical across all tested brain regions, suggesting no regional differences in 102T/C allele expression within the adult brain (one-way ANOVA, F ¼ 0.07, P ¼ 0.94). Under the hypothesis that there are polymorphisms within cis-acting regulatory elements that alter gene expression, our assay should reveal differences in the relative levels of the alleles between individuals, rather than differences in the average relative expression between alleles across the whole sample. For example, if there is a variant distal to the 102T/C polymorphism that alters its expression, depending upon its phase relative to the 102T/C marker, the T-allele will be overexpressed in some individuals, underexpressed in others, while in others (where the functional variant is absent) the two alleles will be equally represented. In the present study, the minimum expression ratio of the 102C- and T-alleles was 0.47 : 0.53, while the maximum was 0.54 : 0.46 (corrected). Differences of a similar magnitude were also observed in the genomic DNA samples (Figure 1), suggesting that these small differences represent variability intrinsic to the assay. Compatible with this explanation, when the assay was repeated in triplicate (as we have recommended for genomic DNA13) using independent RT preparations for those samples showing the greatest differences in allele ratio, the difference in mean ratio between the samples showing the largest variation in the first assay was only 1.5%.

111

Molecular Psychiatry


112

The fact that no individual sample showed consistent differences in allele ratio suggests that, in our sample, the HTR2A locus does not contain common cis-acting polymorphisms or cis-acting epigenetic mechanisms having a significant impact on mRNA expression in the adult brain. Considering Hardy– Weinberg equilibrium of alleles at the putative regulatory locus, and no linkage disequilibrium (LD) with the 102T/C locus, the subset of 23 heterozygotes yields a power of approximately 90% to detect the effect of a distinct regulatory variant present in the general population at a frequency of 0.05. Where linkage disequilibrium between a putative regulatory element and the 102T/C variant is present, the power to detect a cis-acting effect would be even greater as a higher proportion of T/C heterozygotes than expected by chance would be expected also to be heterozygous for the putative functional variant, and therefore display disturbance to the 0.5 : 0.5 null hypothesis ratio. One possible example is the 1438G/A polymorphism within the predicted promoter region that has been found to show perfect (LD) with the 102T/C variant.12 Genotyping in the present study again showed the two SNPs to be in perfect LD (r2 ¼ 1) and that, consequently, all 23 102T/C heterozygotes were also heterozygous for the 1438G/A base change. The present data therefore show that, like the 102T/C polymorphism, the 1438G/A variant does not have a significant effect on HTR2A mRNA expression in adult brain. The HTR2A gene has been reported to be imprinted in human fibroblasts, expressing only the maternal allele,23 and to show polymorphic imprinting in adult human brain, with monoallelic expression observed in a subset of genomic heterozygotes.24 In this regard, our data are compatible with those of Polesskaya and Sokolov,11 who found no evidence for such an effect in a total of 31 genomic heterozygotes. Since monoallelic expression was not observed in any of the 23 heterozygotes tested in the present study either, the phenomenon, if genuine, is likely to be rare in human brain. The results of this study strongly suggest an absence of common, and even moderately uncommon, cis-acting factors causing variability in HTR2A mRNA expression in adult brain. Consequently, our data suggest that observations of differential HTR2A mRNA expression in different phenotypes (eg Hrdina et al,3 Burnet et al4) do not represent a primary pathogenic event per se in these conditions. Instead, such findings might suggest that alteration in HTR2A mRNA is either a secondary pathophysiological event or that it arises as a consequence of the implicated phenotypes or their treatment. Regardless, our data allow us to reject the hypotheses that genetic associations at the 102T/C locus can be explained functionally by direct effects on mRNA expression, or by LD to other variants with this effect. Caveats to this assertion are that it is theoretically possible that our findings may not be relevant to HRT2A expression in the developing nervous system, since we examined


mRNA from only adult brain, and that they may not be attributable to tissue we have not examined, as effects can be tissue specific. Again, our data do not address the possibility that the 102T/C variant has an intrinsic effect on post-transcriptional events, such as translation efficiency. On a more general note, candidate gene studies to date have been extremely limited in their ability to scan for functional variants in regulatory elements because the human genome is poorly annotated for such sequences, which may be quite distal to the coding sequence itself.25 Since measurement of allelespecific expression indirectly scans for the footprint of all regulatory variants, it offers another important tool for candidate gene analysis.

Methods Samples Postmortem brain tissue derived from frontal, parietal or temporal cortex of 50 unrelated anonymized human adults was obtained from three sources (The MRC London Neurodegenerative Diseases Brain Bank, UK; The Stanley Medical Research Institute Brain Bank, Bethesda, USA; The Karolinska Institute, Stockholm, Sweden). From each individual, approximately 500 mg tissue was processed for genomic DNA using standard phenol–chloroform procedures, and approximately 300–500 mg tissue was processed for total RNA using the RNAwizt isolation reagent (Ambion Ltd, Huntingdon, UK). Total RNA was treated with DNAse prior to reverse trancription using the RETROscriptt kit (Ambion Ltd). Allele expression assay Genotyping of genomic DNA samples revealed 23 individuals who were heterozygous for the 102T/C polymorphism (17 male; average age: 56 years; age range: 20–85 years). Genomic DNA and cDNA from heterozygotes were both PCR amplified using ‘universal’ primers described in a previous study:11 50 GGCTTAGACATGGATATTCT-30 and 50 -CTGCAG CTTTTTCTCTAGGG-30 . cDNA samples were additionally amplified using ‘cDNA-specific’ primers spanning intron–exon boundaries: 50 -TGATGACACCAGGCTCTACAGT-30 and 50 -GCCACCGGTACCCATACAG-30 . PCR thermocycling conditions consisted of an initial denaturation step of 941C for 5 min, followed by 40 cycles of 941C for 30 s, 601C for 30 s and 721C for 40 s, with a final extension step of 721C for 10 min. Amplified samples were incubated with 1 U shrimp alkaline phosphatase (Amersham, Bucks, UK) and 2 U exonuclease I (Amersham) for 45 min at 371C and then for 15 min at 851C prior to primer extension reactions. Primer extension was carried out using the SNaPshot Multiplex Kit (PE Applied Biosystems, Warrington, UK), with the extension primer 50 CTACAGTAATGACTTTAACTC-30 . Reactions were performed in a total volume of 10 ml, containing 2 ml treated PCR product, 4.5 ml SNaPshot kit, 2.5 ml water


and 0.3 pmol extension primer. Primer extension thermocycling conditions consisted of an initial step of 951C for 2 min, followed by 25 cycles of 951C for 5 s, 431C for 5 s and 601C for 5 s. Following primer extension, reaction products were treated with 0.5 U shrimp alkaline phosphatase (Amersham) for 45 min at 371C and then for 15 min at 851C. Aliquots of 1 ml SNaPshot reaction product were combined with 9 ml Hi-Di formamide and loaded onto a 3100 DNA sequencer (PE Applied Biosystems). Products were electrophoresed on a 36 cm capillary array at 601C and data processed using Genescan Analysis version 3.7 software (PE Applied Biosystems). Peak heights of allele-specific extended primers were determined using Genotyper version 2.5 software (PE Applied Biosystems). Genotyping of 1438G/A polymorphism Genomic DNA samples were amplified for the HTR2A–1438G/A polymorphism using the primers 50 -AAGCTGCAAGGTAGCAACAGC-30 and 50 -AACCAACTTATTTCCTACCAC-30 . Thermocycling conditions consisted of an initial denaturation step of 941C for 5 min, followed by 35 cycles of 941C for 30 s, 561C for 30 s and 721C for 40 s, with a final extension step of 721C for 10 min. Genotyping was performed using an RFLP assay with the restriction endonuclease MspI. Power estimation Calculation of power is based upon the binomial distribution, Hardy–Weinberg equilibrium at the regulatory SNP, and no LD with the marker SNP. The probability of an individual being homozygous at a putative regulatory locus with alleles in Hardy– Weinberg Equilibrium is p2 þ q2, where p and q are the two allele frequencies. The probability that of n individuals, all are homozygous (and therefore undetected by our assay) for the regulatory polymorphism is then (p2 þ q2)n. This also applies for n individuals selected for heterozygosity at the marker locus if there is no relationship (ie LD) between the genotypes at each locus. The power to detect one heterozygote is then 1(p2 þ q2)n. If the marker and regulatory SNP are in LD, then a higher proportion of people selected for heterozygosity at the marker will also be heterozygous for the regulatory SNP, and the power will be increased.

Acknowledgements We are grateful to the Stanley Medical Research Institute (Bethesda, USA) and to the MRC London Neurodegenerative Diseases Brain Bank (UK) for donating brain tissue. This work was funded by the Medical Research Council (UK).

References 1 Meltzer HY. The role of serotonin in antipsychotic drug action. Neuropsychopharmacology 1999; 21: 106S–115S.

113 2 Aghajanian GK, Marek GJ. Serotonin and hallucinogens. Neuropsychopharmacology 1999; 21: 16S–23S. 3 Hrdina PD, Bakish D, Ravindran A, Chudzik J, Cavazzoni P, Lapierre YD. Platelet serotonergic indices in major depression: upregulation of 5-HT2A receptors unchanged by antidepressant treatment. Psychiatry Res 1997; 66: 73–85. 4 Burnet PW, Eastwood SL, Harrison PJ. 5-HT1A and 5-HT2A receptor mRNAs and binding site densities are differentially altered in schizophrenia. Neuropsychopharmacology 1996; 15: 442–455. 5 Inayama Y, Yoneda H, Sakai T, Ishida T, Nonomura Y, Kono Y et al. Positive association between a DNA sequence variant in the serotonin 2A receptor gene and schizophrenia. Am J Med Genet 1996; 67: 103–105. 6 Williams J, McGuffin P, Nothen M, Owen MJ. Meta-analysis of association between the 5-HT2a receptor T102C polymorphism and schizophrenia. EMASS Collaborative Group. European Multicentre Association Study of Schizophrenia. Lancet 1997; 349: 1221. 7 Arranz MJ, Munro J, Sham P, Kirov G, Murray RM, Collier DA et al. Meta-analysis of studies on genetic variation in 5-HT2A receptors and clozapine response. Schizophr Res 1998; 32: 93–99. 8 Holmes C, Arranz MJ, Powell JF, Collier DA, Lovestone S. 5-HT2A and 5-HT2C receptor polymorphisms and psychopathology in late onset Alzheimer’s disease. Hum Mol Genet 1998; 7: 1507–1509. 9 Arias B, Gutierrez B, Pintor L, Gasto C, Fananas L. Variability in the 5-HT(2A) receptor gene is associated with seasonal pattern in major depression. Mol Psychiatry 2001; 6: 239–242. 10 O’Donovan MC, Owen MJ. Candidate-gene association studies of schizophrenia. Am J Hum Genet 1999; 65: 587–592. 11 Polesskaya OO, Sokolov BP. Differential expression of the ‘C’ and ‘T’ alleles of the 5-HT2A receptor gene in the temporal cortex of normal individuals and schizophrenics. J Neurosci Res 2002; 67: 812–822. 12 Spurlock G, Heils A, Holmans P, Williams J, D’Souza UM, Cardno A et al. A family based association study of T102C polymorphism in 5HT2A and schizophrenia plus identification of new polymorphisms in the promoter. Mol Psychiatry 1998; 3: 42–49. 13 Norton N, Williams NM, Williams HJ, Spurlock G, Kirov G, Morris DW et al. Universal, robust, highly quantitative SNP allele frequency measurement in DNA pools. Hum Genet 2002;110: 471–478. 14 Singer-Sam J, Chapman V, LeBon JM, Riggs AD. Parental imprinting studied by allele-specific primer extension after PCR: paternal X chromosome-linked genes are transcribed prior to preferential paternal X chromosome inactivation. Proc Natl Acad Sci USA 1992; 89: 10469–10473. 15 Le Hellard S, Ballereau SJ, Visscher PM, Torrance HS, Pinson J, Morris SW et al. SNP genotyping on pooled DNAs: comparison of genotyping technologies and a semi automated method for data storage and analysis. Nucleic Acids Res 2002; 30: 74. 16 Yan H, Yuan W, Velculescu VE, Vogelstein B, Kinzler KW. Allelic variation in human gene expression. Science 2002; 297: 1143. 17 Bray NJ, Buckland PR, Owen MJ, O’Donovan MC. Cis-acting variation in the expression of a high proportion of genes in human brain. Hum Genet 2003; 113: 149–153. 18 Hoogendoorn B, Norton N, Kirov G, Williams N, Hamshere ML, Spurlock G et al. Cheap, accurate and rapid allele frequency estimation of single nucleotide polymorphisms by primer extension and DHPLC in DNA pools. Hum Genet 2000; 107: 488–493. 19 Breen G, Harold D, Ralston S, Shaw D, St Clair D. Determining SNP allele frequencies in DNA pools. Biotechniques 2000; 28: 464–466, 468, 470. 20 Turecki G, Briere R, Dewar K, Antonetti T, Lesage AD, Segwin M et al. Prediction of level of serotonin 2A receptor binding by serotonin receptor 2A genetic variation in postmortem brain samples from subjects who did or did not commit suicide. Am J Psychiatry 1999; 156: 1456–1458. 21 Kouzmenko AP, Hayes WL, Pereira AM, Dean B, Burnet PW, Harrison PJ. 5-HT2A receptor polymorphism and steady state receptor expression in schizophrenia. Lancet 1997; 349: 1815. 22 Kouzmenko AP, Scaffidi A, Pereira AM, Hayes WL, Copolov DL, Dean B. No correlation between A(-1438)G polymorphism in Molecular Psychiatry


114 5-HT2A receptor gene promoter and the density of frontal cortical 5-HT2A receptors in schizophrenia. Hum Hered 1999; 49: 103–105. 23 Kato MV, Shimizu T, Nagayoshi M, Kaneko A, Sasaki MS, Ikawa Y. Genomic imprinting of the human serotonin-receptor (HTR2) gene involved in development of retinoblastoma. Am J Hum Genet 1996; 59: 1084–1090. 24 Bunzel R, Blumcke I, Cichon S, Normann S, Schramm J et al. Polymorphic imprinting of the serotonin-2A (5-HT2A) receptor gene in human adult brain. Brain Res Mol Brain Res 1998; 59: 90–92.


25 Enattah NS, Sahi T, Savilahti E, Terwilliger JD, Peltonen L, Jarvela I. Identification of a variant associated with adult-type hypolactasia. Nat Genet 2002; 30: 233–237.

Correspondence: M O’Donovan, Department of Psychological Medicine, University of Wales College of Medicine, Heath Park, Cardiff CF14 4XN, UK. E-mail: [email protected] Received 4 February 2003; revised 28 March 2003; accepted 2 April 2003