The involvement of the mesocorticolimbic dopamine system in behaviors that are compromised in patients with mood disorder has led to the investigation of dopamine system genes as candidates for bipolar disorder. In particular, the functional VNTRs in the exon III of the dopamine D4 (DRD4) and in intron I of the tyrosine hydroxylase (TH) genes have been investigated in numerous association studies that have produced contrasting results. Likewise, linkage studies in multiplex bipolar families have shown both positive and negative results for markers in close proximity to DRD4 and TH on 11p15.5. We performed a linkage disequilibrium analysis of the DRD4 and TH VNTRs in a sample of 145 nuclear families comprised of DSM-IV bipolar probands and their biological parents. An excess of transmissions and non transmissions was observed for the DRD4 4- and 2-repeat alleles respectively. The biased transmission showed a parent of origin effect (POE) since it was derived almost exclusively from the maternal meiosis (4-repeat allele maternally transmitted 40 times vs 20 times non-transmitted; χ2 = 6.667; df = 1; P = 0.009; while paternally transmitted 26 times vs 21 times non-transmitted; χ2 = 0.531; df = 1; P = 0.46). The analysis of TH did not reveal biased transmission of intron I VNTR alleles. Although replication of our study is necessary, the fact that DRD4 exhibit POE and is located on 11p15.5, in close proximity to a cluster of imprinted genes, suggests that genomic imprinting may be operating in bipolar disorder.
Family, twin and adoption studies demonstrated evidence for the importance of genetic factors in bipolar disorder.1 The mode of inheritance does not follow Mendelian rules and multiple genes2 with complex gene–gene3 and gene–environment interactions are likely to be involved.
Different linkage studies have investigated the short arm of chromosome 11 in bipolar families since the first evidence for linkage in an Old Order Amish kindred.4 This finding however was not replicated in the reanalysis of the families, after the disease occurred in some individuals who were unaffected at the time of the primary analysis.5 Further linkage studies of 11p in several data sets of bipolar families have failed to show consistent results.6,7,8,9,10,11,12,13,14
Preliminary evidence for the involvement of dopamine pathways in bipolar disorder arose from animal studies, and more recently evidence from pharmacological and behavioral studies in humans continues to support a dopaminergic role in the etiology of bipolar disorder.15 The short arm of chromosome 11 on p15.5 harbors two dopamine system genes, ie the dopamine D4 receptor (DRD4) and the tyrosine hydroxylase (TH) genes, and these genes have been extensively investigated in bipolar disorder.16,17,18,19,20,21,22,23,24,25
The gene for the dopamine D4 receptor is highly polymorphic containing a functional variable number of tandem repeats (VNTR) in the third exon that encodes for the longest intracellular loop of the receptor.26,27 The polymorphism consists of 2–10 imperfect tandem repeats of 48 bp, each repeat encoding for 16 amino acids. Nineteen different repeat sequences have been described that lead to 25 different haplotypes coding for 18 different predicted amino acid sequences.28 In vitro studies have found different pharmacological properties for the D4 receptors coded by the 2, 4, 7 and 10 repeat alleles.26,29 Tyrosine hydroxylase converts tyrosine to DOPA and is the rate-limiting enzyme in the synthesis of catecholamines. Its gene contains a microsatellite in Intron I constituted by a tetranucleotide repeat of a (TCAT)n motif (HUMTH01) that can be repeated five to 10 times.30 The 10 repeat alleles can be present in two different sequence variants.31 A functional in vitro study of the tyrosine hydroxylase gene has suggested that the microsatellite HUMTH01 may be implicated in the regulation of the gene expression.32
The inconsistencies shown across the linkage studies that investigated the short arm of chromosome 11 in bipolar disorder do not allow us to exclude the involvement of genes with a small contribution to the disease (eg DRD4 and TH) because of the low power of linkage analysis to detect genes with partial contribution to disease.33 The majority of the association studies that have investigated the role of DRD4 in bipolar disorder did not show evidence for association.18,19,20,21,22,25 However, a large sample of bipolar disorder patients has shown association with delusional symptoms in bipolar patients,23 although the grouping of the DRD4 exon III alleles obscured the effect of the single alleles. In a similar fashion, the association studies of HUMTH01 in bipolar disorder produced both positive and negative association.14,17 However, because DRD4 and TH association studies have mainly adopted case-control study designs and because of the limited use of family-based strategy in small sample sizes,16,24 the involvement of DRD4 and TH in bipolar disorder remains a question. The case-control approach, in fact, is associated with the risk of producing spurious results because of the difference in allele frequencies across different populations34 and this has been shown for DRD4 exon III VNTR and HUMTH01 alleles.31,35 The family-based association strategy has the advantage of having a lower chance of false positives (type I errors) due to population stratification,36 has good power to detect genes with partial effect when compared to linkage strategies, and allows the separate analysis of maternally and paternally transmitted alleles. The separation of the data into analysis of maternal and paternal meioses may help in distinguishing true association from false positive.37 In the present study we performed a family-based association study of the DRD4 48-bp repeat VNTR and the TH intron I tetranucleotide repeat VNTR in a sample comprised of subjects suffering from DSM-IV bipolar disorder and their biological parents.
The sample consisted of 154 patients with a DSM-IV diagnosis of bipolar disorder (104 Bipolar I; 40 Bipolar II; eight Schizoaffective bipolar type disorder; two Bipolar Disorder Not Otherwise Specified) and their biological relatives that make up 145 nuclear families. Nine of the families have two affected probands. The probands were 103 (66.9%) females and 51 males (33.1%) and were predominantly (97%) of Caucasian origin with a mean age of 36.25 (SD = ± 9.48, range 13–62 years) and with a mean age of onset of the disorder of 18.54 (SD = ± 7.08; range 5–40 years). The patients were recruited from several psychiatric care centers in Toronto and across central Canada. The study was approved by the University of Toronto and The Centre for Addiction and Mental Health Research Ethics Board. After a complete description of the study to the subjects, written informed consent was obtained from each patient and their respective parents who agreed to participate in the study. The diagnoses were determined after assembling available medical records and the Structured Clinical Interview for DSM-IV, Axis I Disorders.38 The family interview for genetic study (FIGS) was part of the assessment of the families and it was administered to the probands and to their biological relatives. The interviews were conducted by an experienced clinical research assistant. The diagnosis was based on consensus of at least two psychiatrists (JLK, EM).
Blood samples were drawn from patients and relatives and DNA samples were extracted from whole blood following standard high salt procedures. The 48-base pair VNTR region in the third exon of DRD4 was amplified using PCR techniques with primers and conditions previously published.28 The PCR products containing the DRD4 48-bp repeat were visualized using gel electrophoresis performed in 3.5% agarose prepared with ethidium bromide and 1 × TBE (Tris, boric acid, EDTA). We used an ABI Prism 310 (Applied Biosystems, Foster City, CA, USA) for the fragment analysis of the TH tetranucleotide VNTR using a PCR product obtained with 5′ fluorescently labeled primer with sequence previously published.39 The TH genotypes were assigned using GENOTYPER 2.5 software that compared allele size with the size standard TAMRA (Applied Biosystems, Warrington, UK). The subjects were genotyped blind to their affection status and family structure. The eighteen 2-repeat alleles of the DRD4 48-bp VNTR heterozygous proband mothers that were informative for the TDT analysis were sequenced. To increase the fidelity of the PCR product for sequencing, the PCRs were performed using the same conditions used for the genotyping assay, however instead of Taq polymerase, native Pfu polymerase (Stratagene) was used.40 Each of the 2-repeat alleles was sequenced twice and the sequences were compared to the available published sequences for the 48-bp DRD4 exon III VNTR.28
The PEDMANAGER program software (v0.9) was used to check genotypes according to the family structure in order to exclude the presence of parent-proband incompatibility and to calculate the allele frequencies. We tested our nuclear families for the presence of deviation from equal transmission of alleles from heterozygous parents using the Transmission Disequilibrium Test for mutiallelic polymorphisms as implemented by Sham and Curtis41 in the ETDT package (E-TDT, v1.8). This approach tests the null hypothesis of no association by the classical McNemar test statistic for transmitted vs non-transmitted parental alleles. We searched for evidence for biased transmission of the alleles using the allele-wise and genotype-wise analysis and thereafter we analyzed the transmission for each allele individually. The alleles with a frequency lower than 3% (alleles 179 and 199 for HUMTH01 and the 3-, 5- and 6 repeat DRD4 alleles) were not considered in the analysis for individual alleles because of the inadequate power. Furthermore, a parent-sex-specific analysis of the transmission of the DRD4 and HUMTH01 common alleles was conducted using the command for parental sex-specific analysis in the ETDT package. In addition to the TDT analysis, the Haplotype Relative Risk (HRR) statistic42 was applied. This is a standard chi-squared statistic that, unlike the TDT, includes all parental genotypes, both homo- and heterozygous. For the homozygous parents the same allele is scored both in the transmitted as well as the non-transmitted category. Because the variance estimate for HRR uses more information than the TDT/McNemar statistic, HRR can lead to a more powerful statistical test for association; although, under certain conditions, HRR may be less powerful than the TDT/McNemar test.43 Furthermore, because HRR uses all the transmitted and non-transmitted alleles, the HRR contingency two by two tables (obtained computing the transmission and non transmission of the alleles) allows calculation of the odds ratios for the ‘risk’ and ‘non-risk’ alleles. The polymorphisms we studied were multiallelic and therefore the two by two contingency tables that were used for the HRR analyses were obtained considering in one column, the transmissions and non transmissions of the ‘risk’ or ‘non-risk’ allele (as obtained through the ETDT analysis), and the transmissions and non transmission count for all the remaining alleles were grouped into the other column. The alleles that the TDT analyses showed to have biased transmission were considered in the HRR tables. The odds ratio (that is approximately equal to the relative risk) provides an estimation of the risk associated with the transmission of a given allele.
The results of the transmission of the individual alleles from heterozygous parents to the affected offspring of the HUMTH01 and DRD4 VNTR alleles are indicated in Table 1. The analysis for HUMTH01 did not reveal preferential transmission for any of the alleles nor for maternal or paternal preferential transmission. TDT analysis for the DRD4 VNTR revealed biased transmission of the 4- and 2-repeat alleles. Despite the fact that 40 parents had the 2-repeat allele, it was transmitted only 12 times to affected offspring (χ2 for genotype-wise = 14.84; df = 8; P = 0.062; χ2 for allele-wise = 10.32; df = 5; P = 0.066). The results from the transmissions for individual alleles of DRD4 showed an excess of 2-repeat allele non-transmission to the affected offspring (12 transmitted vs 28 non-transmitted, χ2 = 6.40; df = 1; P = 0.0114). On the other hand the analysis of the 4-repeat alleles revealed an excess of transmission for this allele (77 transmitted vs 52 non-transmitted, χ2 = 4.845 df = 1; P = 0.0278). The separate analysis of the individual alleles from maternal and paternal meioses revealed that the biased transmission for the 2- and 4-repeat alleles were derived mostly from maternal meioses. In fact, the 2-repeat allele was maternally transmitted only twice while not transmitted 16 times (χ2 = 10.889; df = 1; P = 0.0010), and the 4-repeat was maternally transmitted 40 times and non transmitted 20 times (χ2 = 6.667; df = 1; P = 0.009). The same alleles showed a non biased transmission when the paternal meioses were considered (see Table 1) . In Table 3, Table 4 and Table 5 the contingency tables with the transmitted and non-transmitted DRD4 alleles used for the HRR analysis are shown and in Table 2 the odds ratios (OR) derived from the HRR analyses are summarized. The odds ratios indicate that the 4-repeat alleles increase the risk for bipolar disorder (OR = 1.7; CI (95%) = 1.18–2.46; P = 0.004), while the non transmission of the 2-repeat allele is associated with a protective effect (OR = 0.44; CI (95%) = 0.22–0.86; P = 0.015). The analyses of the maternal and paternal transmissions clearly indicate that the risk associated with the 4-repeat alleles is derived almost exclusively from the maternal meiosis (OR = 2.34; CI (95%) = 1.29–4.26; P = 0.004) and not from the paternal side (OR = 1.25; CI (95%) = 0.69–2.27; P = 0.5). In a similar way, the reduced transmission for the 2-repeat allele, that may play a protective role, is exclusively of maternal origin (OR = 0.15; CI (95%) = 0.04–0.52, P = 0.0001) and not of paternal origin (OR = 0.78, CI (95%) = 0.29–2.08, P = 0.4).
Because multiple testing was performed, an increased number of type I errors should be considered. However, because the methods that appropriately correct for non-independent multiple testing are controversial for genetic studies of complex diseases, we presented our data with non-corrected P values. Nevertheless, it is notable that the results from the parent-sex-specific analysis of the DRD4 2-repeat allele maternal transmission continued to be significant (P = 0.012) after Bonferroni correction for twelve multiple tests (ie eight tests to individually analyze the transmission of the TH and DRD4 VNTR alleles plus four tests conducted to explore the maternal and paternal meiosis independently for each of the two loci).
Because of the occurrence of multiple sequences within the 48-bp repeat of the DRD4 exon III28 we tested for a sequence-specific effect by DNA sequencing the DRD4 2-repeat alleles that were informative from the maternal meioses. No sequence variants beyond the one published were detected.
The Family Interview for Genetic Studies revealed the presence of family history for bipolar disorder in 54 of the 145 families. In 25 families the history for bipolar disorder was from the maternal side, in 23 from the paternal, while in six families the disorder was present on both the maternal and paternal sides. The transmissions of the DRD4 VNTR alleles in the families with paternal and maternal lineage for bipolar disorder revealed a similar trend to that observed in the overall sample. The maternal meiosis of the families with a maternal history for bipolar disorder showed an effect, with the 4-repeat allele transmitted 10 times vs two non-transmissions (χ2 = 5.333; df = 1; P = 0.02). The analysis of the transmission of DRD4 alleles stratified according to bipolar disorder subtypes continues also to show the same trend of transmissions as in the principal analysis with the same POE. For example in the bipolar type I families for the DRD4 4-repeat allele we observed 25 transmissions vs 13 non transmissions, and in the bipolar type II we observed 15 transmissions vs 7 non transmissions.
A significant departure from the expected 50:50 ratio was observed for the transmission of the 2- and 4-repeat DRD4 alleles in our sample of bipolar families. The biased transmission of the 2- and 4-repeat alleles showed a parent of origin effect (POE). The excess transmission of the 4-repeat alleles as well as the increased number of non-transmissions of the 2-repeat allele were derived almost exclusively from the maternal meioses. This result suggests that the 4-repeat allele may increase the risk for bipolar disorder while the 2-repeat may confer a protective role when non transmitted from the maternal side. The tyrosine hydroxylase HUMTH01 alleles conversely did not show biased transmission from the heterozygous parents to the affected offspring and did not show a POE. The majority of the published association studies17 that have investigated HUMTH01 alleles in bipolar families, similarly to our results, have failed to detect a major effect of the HUMTH01 alleles.
Several linkage analyses,6,7,8,9,10,11,12,13 unrelated case-control18,19,20,21,22,25 and a family-based association study16 have investigated the role of DRD4 exon III VNTR in bipolar disorder and none of them has shown an effect similar to that we observed in the present study. The discrepancy of the results between previous studies and our finding may be due to the presence of genetic heterogeneity and/or to the different genetic structure of the populations studied. To our knowledge, the only family-based association study of DRD4 and bipolar disorder present in the literature was conducted in a small sample in a partially isolated population from Sardinia16 and POE was not investigated.
Several lines of evidence from clinical,44,45 and molecular genetic studies46 have suggested a POE in the transmission of bipolar disorder. However, the clinical studies that investigated for POE in bipolar disorder were not consistent. Grigoroiu-Serbanescu and colleagues44 showed that bipolar patients with an affected father had earlier age of onset when compared with the onset of individuals who had a mother affected. Other studies have suggested a prevalent maternal transmission.45,46 The presence of POE in the transmission of bipolar disorder may be determined by the localization of the risk locus in mitochondrial DNA, or on the X chromosome, or by the presence of genomic imprinting. The sequencing of the mitochondrial DNA by McMahon and colleagues,47 in bipolar patients who inherited the disorder from the mother, excluded a major contribution from the mitochondrial DNA. The presence of X-linked risk alleles with a major effect in bipolar disorder can be excluded from the results of linkage studies.48 However, since few association studies of candidate genes located on the X chromosome have been conducted and because the results from markers on the Xq26–28 are still showing inconsistent results, a gene with a small contribution for bipolar disorder cannot be ruled out.
Genomic imprinting defines an epigenetic phenomenon where gene expression is regulated depending on the parental origin of the gene.49,50 Genomic imprinting has been found to be involved in the etiology of neuropsychiatric disorders such as Prader–Willi and Angelman syndromes51 and its contribution has been postulated for several psychiatric diseases including bipolar disorder.37,49,52 Linkage studies in bipolar disorder families have shown evidence for linkage for chromosome 18 markers and some of these markers have shown a POE.53,54 These studies have indicated a paternal specific effect for two markers (D18S41 and D18S541) on 18q2153,54 and a maternal specific effect for another marker (D18S464) on 18p.53 However, other studies were unable to replicate these findings making conclusions elusive.55,56
Interestingly, DRD4 and TH are located on the telomeric region of the short arm of chromosome 11 close to a cluster of imprinted genes. DRD4 is approximately 200 kb telomeric to H19, an imprinted gene located close to DRD4.57 TH is 9 kb centromeric to the imprinted cluster on 11p15.5 that contains the imprinted insulin-like growth factor 2 (IGF2) gene and the insulin gene (INS).58 The expression of DRD4 was tested in the temporal cortex of two brains from patients who underwent a neurosurgery intervention.59 The study showed expression of both maternal and paternal alleles of DRD4. The expression of both DRD4 alleles in this study however does not exclude the presence of imprinting for DRD4. In fact, imprinting can be a developmental stage- and tissue-specific phenomenon60 as well as polymorphic, ie its presence can vary from subject to subject as described for brain expressed imprinted genes.61 Because of these complexities, genomic imprinting is a multifaceted phenomenon described only recently in mammals,62,63 with function and mechanism not fully understood.64
The POE that we observed for DRD4 alleles, together with its chromosomal location close to an imprinted gene cluster, suggests that DRD4 may be imprinted in humans. Replication in independent samples of bipolar families, that will analyze separately paternal and maternal transmission of DRD4 alleles, is warranted. However, further studies will have to consider the difficulties encountered in the investigation of imprinted risk alleles for other complex diseases. An example is represented in the difficulties found in the analyses of the imprinted locus containing the insulin gene (INS) VNTR in diabetes. The INS VNTR consists of numerous alleles that have been grouped into three distinct classes according to the number of repeats (Classes I, II and III). Class I alleles confer risk to diabetes while Class III alleles have a protective effect. For the Class III alleles, a paternal POE has been described for its protective role in a data set from US families and a maternal effect in families from Europe.65 The population-specific POE has been shown to be determined by the presence of an allele-specific phenomenon associated with one Class III allele, ie a different determination of the imprinting status within alleles of the same polymorphism. In other words, one allele of the INS VNTR is maternally imprinted while another allele at the same locus appears to be paternally imprinted.66 The possibility for the imprinting status being an allele-specific phenomenon together with the relatively high degree of sequence variation of the DRD4 48-bp repeat have led us to sequence the 2-repeat alleles. However no variation in the sequence within the 2-repeat alleles was found.
Further molecular studies that will investigate the expression of maternal and paternal DRD4 alleles are needed to rule out whether genomic imprinting is occurring for DRD4 alleles. Several techniques have demonstrated their validity to determine the imprinting status of human genes including the use of somatic cell hybrids containing an individual human chromosome.67,68 The analysis of additional DRD4 polymorphisms as well as markers in the nearby region and the study of these polymorphisms in independent samples will help to better characterize the involvement of this region of 11p15.5 in bipolar disorder.
In conclusion, the genetic phenomena that are occurring on the short arm of chromosome 1169 underline the involvement of complex mechanisms in the transmission of genes that may be involved in the etiology of psychiatric disorders. These complexities must be taken into consideration when performing linkage and linkage disequilibrium studies. Moreover, the separate analyses of the maternal and the paternal transmission in the study of complex traits may have an important heuristic value. In fact, the presence of false positives is expected to occur similarly in the maternal and paternal transmission of the alleles while the presence of a POE strongly supports the presence of a true biological effect. This may help to reduce the elevated rate of false positives associated with genetic association studies of complex diseases.
About this article
Molecular Psychiatry (2005)