Variation in P2RX7 candidate gene (rs2230912) is not associated with bipolar I disorder and unipolar major depression in four European samples^†

INTRODUCTION

The P2RX7 gene, located on chromosome 12q24, encodes a purinergic brain-expressed receptor and is involved in Ca²⁺-dependent signaling pathways. In three recent studies, the non-synonymous SNP rs2230912 (P2RX7-E13A), located in exon 13 of P2RX7 gene and resulting in a change of the amino acid glutamine to arginine at position 460 (Gln460Arg), provided evidence of association with BP disorder in a French Canadian family-based sample [Barden et al., 2006] and in a British case–control sample [McQuillin et al., 2008] and with Mdd-UP in a German case–control sample [Lucae et al., 2006]. SNP rs2230912 was the only SNP in the P2RX7 gene reported to be associated with major mood disorders across studies. The minor allele (G) of the SNP was found to be slightly more frequent in BPI and Mdd-UP patients [18%, McQuillin et al., 2008; 20%, Barden et al., 2006; 16.9%, Lucae et al., 2006] than in controls [15%, McQuillin et al., 2008; 16.5%, Barden et al., 2006; 14.7%, Lucae et al., 2006]. In Mdd-UP, the association between rs2230912 and the disorder was mainly due to the overrepresentation of AG heterozygotes among patients. McQuillin et al. 2008 conducted an analysis of three combined samples: their own BPI and control sample from UK, the BP and control sample from Quebec [Barden et al., 2006], and the Mdd-UP patient and control sample from Munich [Lucae et al., 2006]. The combined sample analysis showed significant allelic and genotypic association of the SNP rs2230912 with mood disorders due to an increased frequency of the AG heterozygotes in patients. Negative association results for P2RX7-rs2230912 were also reported; no association was found in a Canadian family-based sample by Strauss et al. 2005 and in a Canadian case–control sample by Barden et al. 2006, although the case–control sample was recruited from the same region as the family sample that provided positive results.

Given the fact that rs2230912 (P2RX7-E13A) showed evidence for association with major mood disorders in previous studies, we aimed at replicating this finding and investigated the involvement of SNP rs2230912 in BP disorder type I (BPI) and Mdd-UP in four European samples from Germany, Romania, Poland, and Russia, comprising 1,445 BPI patients, 640 Mdd-UP patients, and 2,006 controls.

PATIENT AND CONTROL SAMPLES

All patients were recruited from consecutive hospital admissions. The ethical committees of all recruitment sites approved the research protocols. After giving written informed consent all patients were directly interviewed with the SCID-I Patient Version. Information provided by medical records and interviews of family members were also used in a best estimate procedure of diagnoses based on DSM-IV-R criteria. The size of the national and combined samples, their gender distribution, patient age of onset (AO), and age at interview (AI) are presented in Table I. The gender breakdown, the AO, and the AI were similar in all national patient samples. The control samples were population based, drawn from the same population as the patients. Only the Romanian control sample was screened for major psychiatric disorders. The ethnicity of the patients and controls was determined by genealogical investigation to the grandparental generation.

GENOTYPING PROCEDURE OF SINGLE NUCLEOTIDE POLYMORPHISMS

Genomic DNA was isolated from whole blood samples using standard methods at the recruitment sites. Genotyping of all national samples was performed using the MassARRAY system on a Sequenom Compact MALDI-TOF device (Sequenom, Inc., San Diego, CA) at the Department of Genomics, Life & Brain Center of the University of Bonn. Primer sequences and PCR/assay conditions can be obtained from the authors upon request. Genotype call rates for the tested SNP were >92.0% in all samples. Two percent of individuals were genotyped in duplicate and there were no replication errors. The genotypes were in Hardy–Weinberg equilibrium (HWE) in all national samples both in controls and patients.

STATISTICAL ANALYSIS

Single-marker association analysis was performed using the FAMHAP software [Becker and Knapp, 2004]. FAMHAP uses the Cochrane-Armitage Trend Test for allelic association and incorporates the HWE when computing the significance of the association with the disease. In the single-marker analysis, the P-values were corrected by permutations in 100,000 simulations. The genotypic association was determined through a likelihood ratio test based on an Expectation–Maximization algorithm. The power of our total European sample to detect association between the SNP rs2230912 and affective disorders was computed using the Genetic Power Calculator [Purcell et al., 2003] (http://pngu.mgh.harvard.edu/purcell/gpc). The power of our BPI sample (N = 1,445) to detect the genotypic relative risk of 1.3 reported by McQuillin et al. 2008 was 91%, and the power of our total mood disorder sample (N = 2,085) was 96% (minor allele frequency (MAF) = 0.15; disease frequency = 1%; genotypic relative risk Aa = 1.3; relative risk AA = 1.3; alpha = 0.05).

RESULTS

We present the results of allelic association analysis in Table II, and the results of genotypic association analysis in Table III. Both allelic and genotypic analyses failed to show an association between rs2230912 (P2RX7) and either BPI or Mdd-UP, or the combined pool of major mood disorders. In order to check for possible population differences between our samples and the samples of the studies reporting positive results, we compared the genotype frequencies in patient and control groups between studies (Table IV). We observed no significant differences between our samples and previously published samples indicating that population differences are unlikely to be the cause of the discrepant results.

DISCUSSION

We have investigated variation in the P2RX7 candidate gene (SNP rs2230912) in a large European sample of 1,445 BPI patients and 2,085 mood disordered patients (BPI + Mdd-UP), and 2,006 controls. Although our sample had more than 90% power to detect the previously reported effect size of a genotypic relative risk of about 1.3, we could not replicate the associations with BPI disorder reported in the British BPI case–control sample [McQuillin et al., 2008] and in the Canadian family-based sample [Barden et al., 2006], as well as the association with Mdd-UP [Lucae et al., 2006].

No association between rs2230912 and major affective disorders was also reported by Barden et al. 2006 in a BPI and BPII case–control sample and by Strauss et al. 2005 in a family-based sample of BP and Mdd-UP patients.

None of our samples displayed significant allelic or genotypic association with rs2230912 and thus, do not support an involvement of this SNP in the etiology of BPI or Mdd-UP in our samples of Central and Eastern European origin. There are different potential explanations for our observation: (a) the original findings were spurious. (b) Variation in P2RX7 is involved in the etiology of major affective disorders, but the genetic effect was largely overestimated by the original studies. The real effect might be smaller than the effect indicated by an OR equal to 1.3. (c) There are differences in the diagnostic uniformity between our study and the Canadian family-based sample which included several diagnoses: BPI and schizoaffective-BP type (N = 105), BPII (N = 42), Mdd-UP recurrent (N = 54), and single depressive episode (N = 57). In our sample, all patients had unitary diagnoses and illness severity; all 1,445 BPI patients were hospitalized and had BP disorder type I, while all Mdd-UP patients were recurrent hospitalized cases. Their ethnic origin was known back to the third generation. (d) Differences in linkage disequilibrium structure at the P2RX7 locus due to differences in ethnic origin would require a more comprehensive investigation of the P2RX7 locus using a dense set of SNPs before final conclusions can be drawn. Although none of our national samples displayed a significant difference in genotype frequency between patients and controls, a tendency to a higher frequency of the heterozygote AG in patients was observed in the Polish and the Russian samples (both Slav populations). This result is in line with previous findings. The other samples behaved differently. While the frequency of the heterozygote AG was nearly identical in patients and controls in the German samples that included controls who were not psychiatrically investigated, in the Romanian sample (Latin population) the heterozygote AG tended to a greater frequency in controls than in patients in spite of the fact that all Romanian controls were screened for psychiatric disorders.

Additional studies are clearly needed to clarify the potential involvement of P2RX7 and of SNP rs2230912 in the etiology of major affective disorders.