Abstract
Background: The dopamine transporter (DAT) is the primary mechanism for dopamine clearance from the synapse in midbrain dopaminergic neurons, and the target of psychostimulant and neurotoxic drugs such as cocaine, amphetamine, and MPTP. Consequently, the gene for DAT (SLC6A3) has been the focus of many population-based case-control association studies using a 40-bp VNTR in the 3′-untranslated region. Results have differed depending on the population studied, suggesting allele frequency effects are involved. For this reason, a global survey of allele frequencies for this VNTR polymorphism was performed.
Methods: Individuals (n = 1528) from 30 populations around the world were typed for this VNTR using PCR and agarose gel electrophoresis.
Results: As with previous studies, the ten-repeat allele is most common, except for a Middle Eastern population in which the nine-repeat allele is most frequent. Frequencies of the nine- and ten-repeat alleles vary widely even among European populations.
Conclusions: Many previous association studies have used “white” or “black” U.S. populations. However, many different ethnic groups have contributed to these populations. The large variation in allele frequencies observed in this study emphasizes the inadequacy of most past studies using the case-control design and the importance of matching patient and control populations in future association studies.
Keywords
Introduction
The nigrostriatal and mesolimbic/mesocortical dopaminergic systems have long been implicated in numerous neurologic and neuropsychiatric diseases, as well as in drug abuse. The dopamine transporter (DAT) is the primary mechanism for dopamine (DA) clearance from the synapse in the midbrain, as demonstrated by the prolonged presence of DA in striatal slices of DAT knockout mice (
Giros et al 1996
). However, in medial prefrontal cortex (PFC) and amygdala (basal lateral nucleus) of rats, DA uptake has been shown to be slower, suggesting a more minor role for DAT in these regions (Garris and Wightman 1994
). DAT protein is present in the neostriatum, nucleus accumbens, and lateral ventral tegmental area (VTA) (Ciliax et al 1995
), making it a candidate for disorders involving these dopaminergic systems.DAT is also the target of a number of psychostimulant and neurotoxic drugs. Cocaine binds to DAT and prevents DA reuptake. Amphetamine enters neurons preferentially through DAT and can release DA to the extracellular space through reverse transport (
Jones et al 1998
). The neurotoxin MPTP exerts its parkinsonism effects via DAT (Gainetdinov et al 1997
).The DAT protein is encoded by the locus SLC6A3 and has been FISH-mapped to chromosome 5p15.3 (
Giros et al 1992
, Vandenbergh et al 1992a
), a localization further confirmed by linkage-mapping (Gelernter et al 1995
). The gene is comprised of 15 exons spanning over 64 kb (Donovan et al 1995
, Kawarai et al 1997
). The protein is a 620 amino acid protein with twelve putative transmembrane domains, and is a member of the Na+/Cl−-dependent transporter family.Three polymorphisms in this gene have been studied: a highly polymorphic TaqI VNTR in intron 8 (
Byerley et al 1993
), a biallelic TaqI RFLP (“Taq 492”) (Vandenbergh et al 1992b
), and a 40-bp VNTR in the 3′-untranslated region (Sano et al 1993
, Vandenbergh et al 1992a
). This final 40-bp VNTR (3′-VNTR) has been the most studied for population variation (summarized in Table 1). Most of the population studies published to date show that the ten-repeat allele (A10) occurs most frequently, with some regional variation (Doucette-Stamm and Mao 1995
, Nakatome et al 1995
, Nakatome et al 1996
, Persico et al 1996
). The next most common allele is the nine-repeat allele, except for an African sample (Mbuti) in which A7, A9, and A10 had almost equal frequencies (Gelernter et al 1998
; present study).- Gelernter J
- Kranzler H
- Lacobelle J
Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)).
Genomics. 1998; 1998: 289-297
Table 1Previous Population Studies of the SLC6A3 3′-VNTR
| Region | Population | 2n | Allele frequency (number of repeats) | Reference | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 13 | ||||
| Africa | Mbuti | 76 | 0.34 | 0.01 | 0.28 | 0.37 | Gelernter et al 1998
Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)). Genomics. 1998; 1998: 289-297 | |||||
| Europe | African American | 892 | 0.03 | 0.02 | 0.03 | 0.17 | 0.73 | 0.01 | 0.00 | |||
| Mixed European | 886 | 0.00 | 0.00 | 0.00 | 0.27 | 0.72 | 0.00 | 0.00 | Doucette-Stamm and Mao 1995 | |||
| Hispanic | 282 | 0.01 | 0.00 | 0.26 | 0.71 | 0.01 | 0.01 | |||||
| Adygei | 102 | 0.01 | 0.21 | 0.77 | 0.01 | |||||||
| African American | 80 | 0.03 | 0.03 | 0.03 | 0.15 | 0.76 | 0.01 | Gelernter et al 1998
Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)). Genomics. 1998; 1998: 289-297 | ||||
| European American | 168 | 0.01 | 0.28 | 0.71 | ||||||||
| African American | 164 | 0.01 | 0.01 | 0.02 | 0.03 | 0.22 | 0.70 | 0.01 | ||||
| European American | 810 | 0.00 | 0.00 | 0.00 | 0.01 | 0.29 | 0.69 | 0.00 | Persico et al 1996 | |||
| Italian | 696 | 0.00 | 0.00 | 0.00 | 0.00 | 0.35 | 0.63 | 0.01 | ||||
| Asia | Japanese | 214 | 0.01 | 0.04 | 0.93 | 0.02 | Sano et al 1993 | |||||
| Japanese | 352 | 0.02 | 0.06 | 0.91 | 0.01 | Nakatome et al 1995 | ||||||
| Mongolian | 156 | 0.03 | 0.05 | 0.90 | 0.01 | 0.01 | Nakatome et al 1996 | |||||
| Australo-Melanesia | Nasioi | 46 | 0.07 | 0.93 | ||||||||
| North and Central America | Maya | 84 | 0.06 | 0.93 | 0.01 | Gelernter et al 1998
Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)). Genomics. 1998; 1998: 289-297 | ||||||
| South America | Rondonian Surui | 64 | 1.00 | |||||||||
In addition to the population variation studies, numerous studies, both population- and family-based, have investigated the role of DAT in various diseases implicating the nigrostriatal dopaminergic system (see Table 2for summary of association studies). Parkinson’s disease (PD) is characterized by loss of dopaminergic neurons in the substantia nigra. Furthermore, MPTP-induced parkinsonism is dependent on DAT (
Gainetdinov et al 1997
). Le Couteur 1997
studied Australian (European ancestry) PD patients and controls and found a greater frequency of A11 in the patients. On the other hand, no association was found in a Chinese sample (Leighton et al 1997
).Table 2Previous Association Studies Using the SLC6A3 3′-VNTR
| Phenotype | Population | 2n | Allele frequency (number of repeats) | Association? | Reference | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 3 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | |||||
| Neurodegenerative/neuropsychiatric disorders | ||||||||||||
| Normal | Italian | 108 | 0.01 | 0.40 | 0.59 | |||||||
| Delusional disorder | 122 | 0.39 | 0.61 | 0.01 | No | Persico and Catalano 1998 | ||||||
| Normal | Caucasian (Australian) | 400 | 0.00 | 0.28 | 0.72 | 0.00 | ||||||
| Parkinson’s disease | 200 | 0.26 | 0.72 | 0.03 | Yes–A11 | Le Couteur 1997 | ||||||
| Normal | Chinese | 460 | 0.01 | 0.08 | 0.90 | 0.02 | ||||||
| Parkinson’s disease | 406 | 0.00 | 0.01 | 0.07 | 0.90 | 0.02 | No | Leighton et al 1997 | ||||
| Normal | French | 182 | 0.34 | 0.65 | ||||||||
| Schizophrenia | 182 | 0.27 | 0.71 | No | Bodeau-Pean et al 1995 | |||||||
| Normal | Chinese (Han) | 196 | 0.01 | 0.06 | 0.93 | 0.01 | ||||||
| Schizophrenia | 210 | 0.01 | 0.06 | 0.89 | 0.03 | No | Li et al 1994 | |||||
| Normal | Japanese | 234 | 0.00 | 0.08 | 0.90 | 0.02 | ||||||
| Schizophrenia | 236 | 0.00 | 0.02 | 0.03 | 0.94 | 0.00 | No | Inada et al 1996 | ||||
| Normal | Italian | 168 | 0.33 | 0.66 | 0.01 | |||||||
| Schizophrenia | 294 | 0.01 | 0.01 | 0.01 | 0.34 | 0.64 | 0.01 | Yes | Persico and Macciardi 1997 | |||
| Substance abuse | ||||||||||||
| Normal | Japanese | 470 | 0.02 | 0.00 | 0.06 | 0.90 | 0.01 | |||||
| Alcoholics (ALDH2∗1 Homozygotes) | 264 | 0.02 | 0.05 | 0.92 | 0.01 | No | ||||||
| Alcoholics (ALDH2∗2 Heterozygotes) | 160 | 0.02 | 0.06 | 0.06 | 0.86 | 0.01 | Yes–A7 | Muramatsu and Higuchi 1995 | ||||
| Normal | German | 186 | 0.19 | 0.79 | 0.02 | |||||||
| Alcoholics | 586 | 0.25 | 0.75 | 0.01 | Yes–A9 carriers | Sander et al 1997 | ||||||
| Alcoholics | German | 96 | 0.23 | 0.77 | Yes–A9 carriers | Schmidt et al 1998 | ||||||
| Cocaine abuse | Black American | 90 | 0.02 | 0.03 | 0.18 | 0.76 | 0.01 | No | ||||
| Cocaine abuse | White American | 116 | 0.01 | 0.28 | 0.71 | 0.01 | Yes–A9 | Gelernter et al 1994 | ||||
| Normal | North American Caucasian | 96 | 0.28 | 0.72 | ||||||||
| Polysubstance abuse | 366 | 0.28 | 0.72 | No | Persico et al 1993 | |||||||
| Normal/smoking | Black American | 156 | 0.03 | 0.01 | 0.02 | 0.19 | 0.73 | 0.01 | ||||
| White American | 888 | 0.02 | 0.00 | 0.00 | 0.32 | 0.65 | 0.01 | Yes–A9 | Lerman et al 1999 | |||
| Normal/smoking | Mixed | 2214 | NR | Yes–A9 | Sabol et al 1999 | |||||||
legend NR, Not reported.
a 6, 11, 12 repeat alleles were reported but exact frequencies were not given.
b Increased frequency of homozygous genotypes in patients.
c Although not specifically stated in the study, these subjects were likely heterozygous for ALDH2∗2.
d Association to delirium and withdrawal seizures.
e More severe withdrawal symptoms in A9 carriers.
f Association to cocaine-induced paranoia.
g A9 carriers less likely to be a smoker. Only combined frequencies (smokers and nonsmokers) were reported.
h A9 carriers more likely to have quit smoking.
Association with schizophrenia was negative in Chinese, French, and Japanese samples (
Bodeau-Pean et al 1995
, Inada et al 1996
, Li et al 1994
); likewise, linkage was excluded in American and Canadian pedigrees (King et al 1997
, Persico et al 1995
). However, a genotypic association (greater frequency of 9/9 and 10/10 homozygotes) to schizophrenia was reported in an Italian sample (Persico and Macciardi 1997
), but no association was demonstrated with delusional disorder in another Italian sample (Persico and Catalano 1998
). Linkage to bipolar disorder has also been suggested (Kelsoe et al 1996
, Waldman et al 1997
).Linkage was excluded in two large Tourette syndrome (TS) families under a dominant model (
Gelernter et al 1995
); however, Comings et al 1996
report a greater frequency of the ten-repeat allele in TS patients. This same allele has been reported to be positively associated with attention deficit hyperactivity disorder (ADHD) using the haplotype-based haplotype relative risk method (- Comings D.E
- Wu S
- Chiu C
- Ring R.H
- Gade R
- Ahn C
- et al.
Polygenic inheritance of Tourette syndrome, stuttering, attention deficit hyperactivity, conduct, and oppositional defiant disorder The additive and subtractive effect of the three dopaminergic genes—DRD2, DβH, and DAT1.
Am J Med Genet (Neuropsychiatr Genet). 1996; 67: 264-288
Cook et al 1995
, Gill et al 1997
) as well as the transmission disequilibrium test (Waldman et al 1997
). The haplotype relative risk method and transmission disequilibrium test are much less likely to produce false-positive results, resulting from allele frequency differences between the affected and unaffected groups, than are the population-based case-control studies, because they are both family-based methods.Since DAT is the major site of action of cocaine, and is present in the mesolimbic/mesocortical dopaminergic systems that are important for reward and reinforcement, it is a candidate for functional variation related to differences in susceptibility to substance abuse. While studies have not shown an association between either the TaqI RFLP or the 3′-VNTR with polysubstance abuse (
Persico et al 1993
), the nine-repeat allele of the 3′-VNTR has been reported to be associated with cocaine-induced paranoia in one study of North Americans of European ancestry (Gelernter et al 1994
).A study on alcoholism in a Japanese sample presumably heterozygous for the inactive aldehyde dehydrogenase-2 allele (ALDH2∗2) showed a greater frequency of A7 than controls and alcoholics homozygous for ALDH2∗1, the functional allele (
Muramatsu and Higuchi 1995
). In a mixed European sample, no association to alcoholism itself was demonstrated (Parsian and Zhang 1997
) but there have been associations of A9 to the severity of alcohol-withdrawal symptoms in mixed European samples (Sander et al 1997
, Schmidt et al 1998
).Most recently, two studies examined alleles at the 3′-VNTR and the effects on smoking behavior. One of the studies reported a protective effect of the nine-repeat allele, such that A9 carriers were less likely to be smokers (
Lerman et al 1999
). The second study reported that A9 carriers were more likely to have quit smoking (Sabol et al 1999
).The wide variation of results in population-based case-control association studies of SLC6A3 may partly be due to unappreciated population differences in the frequencies of alleles. For this reason, we have undertaken a world-wide survey of allele frequencies at the 3′-VNTR.
Methods and materials
PCR amplification was performed using the primers (DATVNTRF: 5′-TGTGGTGTAGGGAACGGCCTGAG-3′ and DATVNTRR: 5′-CTTCCTGGAGGTCACGGCTCAAGG-3′) reported previously (
Vandenbergh et al 1992a
). Each 25 μL reaction contained 100 ng genomic DNA, 1.0 U Taq polymerase (Perkin-Elmer, Foster City, CA), 0.20 mM each of dNTPs and both primers, 5% DMSO (v/v), and 1× buffer B (Boehringer-Mannheim, Indianapolis, IN—50 mmol/L KCl, 1.5 mmol/L MgCl2, 10 mmol/L Tris pH 8.3). After an initial 5-min denaturation at 94°C, the following profile was used for 30 cycles: 94°C × 30 sec; 62°C × 30 sec; 72°C × 30 sec. This was followed by a final 10-min extension at 72°C. The PCR products, ranging in size from 198 bp to 563 bp, were electrophoresed on a 2% agarose gel, stained with ethidium bromide, and scanned on a 575 FluorImager (Molecular Dynamics, Sunnyvale, CA—excitation 488 nm, detection 590 nm).Samples tested are from the cell line collection established by J. R. Kidd and K. K. Kidd. Most of the numerous populations represented, as well as the cell line and DNA purification procedures used, are described in
Kidd et al 1998
, Palmatier and co-workers (in press), and on the Kidd Lab Web page (<http://info.med.yale.edu/genetics/kkidd>).Results
SLC6A3 3′-VNTR global allele frequencies and heterozygosities are shown in Table 3. Genotype frequencies were not significantly different from Hardy-Weinberg equilibrium values as determined by χ2 tests. Observed alleles ranged from three to twelve repeats (198–563 bp); however, no four-, five-, or six-repeat alleles were observed in this sample of 3056 alleles. In most populations, the ten-repeat allele was the most frequent, in accord with previous studies. In some populations (Kachari, Arizona Pima, Karitiana), only a single allele (A10) is present in our samples. This observation does not mean that these populations have gone to fixation however. It may simply be a chance result due to small samples (especially in the Kachari) coupled with very low frequencies of the other alleles, a scenario made more plausible by the presence of other alleles in other populations from those same regions. The two Middle Eastern populations (Druze, Yemenite Jews) have nearly equal frequencies of the nine- and ten-repeat alleles. Heterozygosities range from 0 (in the three populations noted previously) to a high of 0.63 in the Biaka.
| Region | Population | 2n | Allele frequency (number of repeats) | Heterozygosity | |||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
| 3 | 7 | 8 | 9 | 10 | 11 | 12 | Expected | Observed | |||
| Africa | Biaka | 130 | 0.02 | 0.15 | 0.05 | 0.22 | 0.56 | 0.61 | 0.63 | ||
| Ethiopian Jews | 66 | 0.02 | 0.20 | 0.79 | 0.34 | 0.24 | |||||
| Mbuti | 78 | 0.33 | 0.01 | 0.28 | 0.36 | 0.01 | 0.68 | 0.62 | |||
| Yoruba | 98 | 0.05 | 0.01 | 0.05 | 0.13 | 0.73 | 0.02 | 0.44 | 0.39 | ||
| Middle East | Druze | 186 | 0.02 | 0.42 | 0.55 | 0.51 | 0.45 | ||||
| Yemenite Jews | 74 | 0.03 | 0.50 | 0.45 | 0.03 | 0.55 | 0.59 | ||||
| Europe | Adygei | 108 | 0.02 | 0.23 | 0.74 | 0.01 | 0.40 | 0.41 | |||
| Danes | 102 | 0.22 | 0.76 | 0.02 | 0.37 | 0.47 | |||||
| Finns | 70 | 0.10 | 0.90 | 0.18 | 0.20 | ||||||
| Irish | 204 | 0.00 | 0.30 | 0.69 | 0.43 | 0.50 | |||||
| Mixed Europeans | 176 | 0.31 | 0.69 | 0.43 | 0.42 | ||||||
| Russians | 92 | 0.20 | 0.80 | 0.31 | 0.30 | ||||||
| East Asia | Ami | 76 | 0.13 | 0.87 | 0.23 | 0.26 | |||||
| Atayal | 84 | 0.06 | 0.94 | 0.11 | 0.07 | ||||||
| Cambodians | 50 | 0.20 | 0.80 | 0.32 | 0.32 | ||||||
| Chinese, SF | 116 | 0.02 | 0.01 | 0.06 | 0.91 | 0.16 | 0.14 | ||||
| Chinese, Taiwan | 96 | 0.10 | 0.84 | 0.05 | 0.27 | 0.29 | |||||
| Hakka | 68 | 0.01 | 0.09 | 0.87 | 0.03 | 0.24 | 0.26 | ||||
| Japanese | 90 | 0.01 | 0.01 | 0.98 | 0.04 | 0.04 | |||||
| Kachari (Assam) | 28 | 1.00 | 0.00 | 0.00 | |||||||
| Yakut | 102 | 0.01 | 0.05 | 0.94 | 0.11 | 0.12 | |||||
| North and Central America | Cheyenne | 112 | 0.02 | 0.98 | 0.04 | 0.04 | |||||
| Jemez Pueblo | 84 | 0.01 | 0.99 | 0.02 | 0.02 | ||||||
| Maya | 104 | 0.07 | 0.92 | 0.01 | 0.14 | 0.15 | |||||
| Pima, Arizona | 90 | 1.00 | 0.00 | 0.00 | |||||||
| Pima, Mexico | 192 | 0.05 | 0.95 | 0.10 | 0.08 | ||||||
| South America | Karitiana | 108 | 1.00 | 0.00 | 0.00 | ||||||
| Rondonian Surui | 94 | 0.01 | 0.99 | 0.02 | 0.02 | ||||||
| Ticuna | 130 | 0.07 | 0.93 | 0.13 | 0.14 | ||||||
| Australo-Melanesia | Nasioi | 48 | 0.06 | 0.94 | 0.12 | 0.13 | |||||
a The population includes individuals previously reported by
Gelernter et al 1998
; all individuals were retyped for this study.- Gelernter J
- Kranzler H
- Lacobelle J
Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)).
Genomics. 1998; 1998: 289-297
legend Observed allele frequencies for 1528 individuals from 30 populations around the world.
legend 2n = number of chromosomes.
Frequencies of the nine-repeat allele, which has been implicated in a number of association studies, have considerable variation. Even within the six European populations studied, the A9 frequencies are significantly different (χ2 = 17.45, df = 5, p = .0037). These are also significantly different from the two Middle Eastern populations (χ2 = 51.12, df = 7, p = .0001). A9 frequencies range from 0.10 in the Finns, to 0.50 in the Yemenites. The African populations show a range of 0.13 in the Yoruba, to 0.28 in the Mbuti; while the East Asian populations show a similarly large range of 0.01 in the Japanese to 0.20 in the Cambodians (excluding the small sample of Kachari).
Whenever possible (as in some of the more isolated populations), genotypes were checked within families to ensure consistent and accurate typings. This procedure identified errors with initial amplifications from four individuals (all Druze) who appeared homozygous for A3, suggesting preferential amplification of this short allele. Reamplification showed that all of these individuals were actually 3/9 or 3/10 heterozygotes. Based on this finding, the three other A3 homozygotes (all Yoruba) were reamplified and two were identified as 3/10 heterozygotes.
In order to measure variation among the populations, the standardized variance (FST) was calculated for this polymorphism as σ2p/(p̄(1 − p̄)) (
Wright 1969
). The overall FST at this VNTR is 0.17. Since this value depends on the populations being included, comparisons are most meaningful when values are derived from similar populations. The FST value was 0.10 for the same 30 populations at a functional single nucleotide polymorphism (SNP) in the catechol-O-methyltransferase (COMT) gene (Palmatier et al in press
). FST values for RFLPs in 12 of the populations included in this study were reported to be between 0.06 and 0.30 (Kidd et al 1993
). The VNTR in this study has an FST of 0.21 for these same populations. A smaller range (0.03 to 0.12) was observed in a study of STRPs in 10 of the populations included in the present study (Calafell et al 1998
). For these 10 populations, the 3′-VNTR FST is 0.21, indicating a much greater variation. However, none of these other studies were on VNTRs. A comparison to a more similar system, an imperfect 48 base pair repeat in exon 3 of the D4 dopamine receptor gene (DRD4), with alleles ranging from 2 to 10 repeats (Chang 1996
), is appropriate here. Many of the 36 populations typed in that study were the same as those included in the present study. Moreover, the same geographic regions were represented. DRD4 also lies quite telomeric at 11p15.5 (Gelernter et al 1992
); the main difference is that the DRD4 VNTR is functional. Based on the allele frequencies reported by Chang 1996
, the overall FST is 0.18, similar to the value for the SLC6A3 3′-VNTR.Discussion
A critical assumption in population-based case-control association studies is that both patients and control subjects are derived from the same homogeneous population. If this is true, then a significant frequency difference at a genetic marker between patients and control subjects represents either a direct etiologic association of that marker to the phenotype, or an indirect association through linkage disequilibrium to another nearby polymorphism. Significant heterogeneity in SLC6A3 3′-VNTR allele frequencies has now been documented by this study. Especially important is the significant variation among European and other so-called “Caucasian” populations. These differences mean that it is essentially impossible to have cases and control subjects perfectly matched for ethnic background even within the U.S. white population, since many different European and Middle Eastern ethnic groups have contributed to this population. For example, since the A9 frequencies are significantly different within Europe and when compared with the two Middle Eastern populations, a significant difference in the A9 frequency could easily arise if case and control subjects have slightly different proportions of ancestry from Middle Eastern populations, which have among the world’s highest frequencies of this allele, or of Northern European ancestry, which have among the lowest frequencies of A9 in Europe.
As an example of the consequences of inappropriate sampling, consider a study by
Lerman et al 1999
in which cigarette smoking was reported to have an association to the 3′-VNTR. Specifically, using a mixed sample of smokers and control subjects having Western European or African ancestries, the study concluded that A9 gave a protective effect such that fewer smokers had the A9 allele. As demonstrated in the present study, as well as previous studies, allele frequencies between European and African populations are very different. Indeed, the frequencies reported by Lerman and co-workers, which were similar to those obtained by Doucette-Stamm and Mao 1995
, show that the two populations differed (χ2 = 16.4, p < .001). Yet their conclusion that A9 carriers were less likely to be smokers is based on the combined data. While this is an extreme example of a “false positive” association because of a heterogeneous sample, it is not really a falsely significant result because the test for heterogeneity detected that the smokers really did have a lower frequency of the A9 allele; the error was in attributing that to smoking and not to the fact that African Americans constituted a larger proportion of the smoking subjects than of the nonsmoking subjects and African Americans have a lower frequency of the A9 allele.Another way of looking at the problem of heterogeneous samples is suggested by this perspective: consider a power analysis of the situation in which two groups, patients and control subjects, or smokers and nonsmokers, are sampled from larger populations, which for a reason unrelated to the distinction between the two groups, really have a different underlying frequency of the genetic trait (allele) being studied. In this situation, there is a certain probability that a heterogeneity test will give a significant result, i.e., detect the difference. This probability is a function of the average frequency of the allele, the difference in frequency between the two larger populations from which the samples are drawn, and the sample sizes. That is the power of the test. It is also the “false positive” rate if the significance is interpreted as attributable to the characteristic used to label the two groups. Table 4summarizes a small Monte Carlo examination of this question. Using the A9 allele at SLC6A3 as a model, we see from Table 3 that the allele frequency in “Europeans” (considered broadly to include Israeli populations) ranges from 0.10 in Finns to 0.50 in Yemenite Jews. Though we have not sampled the full range of populations contributing to the “Caucasian” category, we could take a value of 0.30 as a representative, roughly average, frequency of A9 in a mixed “European” population. (Indeed, that is the frequency we found for our sample of mixed Europeans.) If we then consider that our cases have a slightly different ethnic mix than our control subjects, we have different underlying frequencies of the two groups. For example, if our cases contained a slightly higher proportion of individuals of Finnish and Danish ancestry (“Scandinavian”) and our control subjects contained a slightly higher proportion of individuals of Israeli ancestry (“Jews”), the actual frequency of the A9 allele would be lower in the universe from which cases were sampled and higher in the universe from which control subjects were sampled.
Table 4Percentage of Significant Heterogeneity Tests (as p ≤.05/p ≤.01) between Case and Control Subjects (Each of Sample Size n) Drawn from Populations Differing by the Amount Indicated around a Mean Allele Frequency of 0.30
| Sample Size (n) | Frequency difference (percent) | ||
|---|---|---|---|
| 1 | 5 | 10 | |
| 50 | 5.1/0.9 | 9.9/2.4 | 25.3/9.1 |
| 75 | 5.4/1.1 | 12.3/3.2 | 38.0/12.3 |
| 100 | 5.1/1.2 | 14.7/4.4 | 47.4/24.4 |
legend Percentages based on 10,000 Monte Carlo simulations for each size-difference combination.
Depending on the proportions of individuals with different ancestries, the real difference could be very small or quite large, even as much as 10% to 15%. Then, the sampling error has to be considered: a small sample will not reflect the true frequency as closely as a large sample. Table 4 summarizes the power, or false positive rate (depending on one’s perspective), for a 1%, 5%, and 10% real difference in the populations with sample sizes of 50, 75, and 100 drawn from those populations, all around a mean allele frequency of 30%. As can be seen, the rate of significant differences, estimated from 10,000 replications of sampling, is about the expected 0.05, or 0.01 for real frequency differences of only 1%. However, with frequency differences of 5% and 10%, the rate of significant differences can reach as high as 0.49. Given the range of over 40% among “European” populations, it would be easy to achieve real differences as large as 10% by failing to precisely match the cases and controls.
A second study on cigarette smoking used an even more heterogeneous sample population, including individuals of East Asian ancestry (
Sabol et al 1999
). They report that A9 carriers were more likely to have quit smoking and have lower novelty seeking scores. This stands in contrast to a study of New Zealand alcoholics that did not find any association to novelty seeking (Sullivan et al 1997
). Furthermore, the allele frequencies between the controls in the Lerman et al 1999
and Sabol et al 1999
studies are more different than the differences reported within each study, raising considerable doubt about the interpretation of their borderline statistics.Gelernter et al 1994
reported an increased frequency of the nine-repeat allele among white cocaine abusers with paranoia when compared to abusers without paranoia (f = .35 and f = .16, respectively; p = .047). In the European populations typed in the present study, frequencies for A9 ranged from 0.10 in the Finns, to 0.31 in the mixed European sample. Persico et al 1996
reported an A9 frequency of 0.35 in an Italian sample. While the cocaine abusers with paranoia have a higher reported A9 frequency than any strictly European sample from this study, A9 frequencies in the Middle East are even greater (Druze, f = .42; Yemenites, f = .50). Thus, this association to cocaine-induced paranoia should be interpreted with caution, and since individuals from southwest Asia have contributed to the “white” U.S. population, the observation may not apply to any population groups, but be an artifact of a stratified sample. This is especially a problem since the sample size in that study was small (n = 58).A number of association studies have investigated the link between this VNTR and alcoholism or severity of alcohol-withdrawal. In a study of Japanese alcoholics, those presumably heterozygous for the inactive form of aldehyde dehydrogenase-2 (ALDH2∗2) had a significantly greater frequency of A7 (f = .056, χ2 = 3.87, p < .05) when compared to control subjects (f = .021) (
Muramatsu and Higuchi 1995
). Patients heterozygous for ALDH2∗2 were considered to have overcome the protection against developing alcoholism provided by this genotype and thus assumed to represent a more homogeneous subgroup of alcoholics. In the present study, the only East Asian populations with A7 were the San Francisco Chinese and Japanese (f = .017 and .011, respectively). The only other geographic region in which A7 was found was Africa, where the Mbuti had the greatest frequency of A7 (f = .33).Two studies of alcohol withdrawal symptoms have implicated A9. In the first, an increased prevalence of A9 carriers was observed in Western European (German) alcoholics experiencing delirium or withdrawal seizure (
Sander et al 1997
). In that study, A9 frequencies were 0.31 and 0.34 for subjects with alcohol-related delirium and seizure, respectively, while the A9 frequency for control subjects was 0.19. In the present study, the A9 frequencies ranged from 0.10 (Finns), to 0.31 (mixed Europeans) for the European populations, but were even greater in the Middle Eastern populations (f = .42 and .50 for Druze and Yemenite Jews). The second study, by the same group, showed more severe withdrawal syndromes associated with A9/A10 genotypes when compared to A10/A10 genotypes, assessed by differences in factor scores in a factor analysis indexing withdrawal symptoms (Schmidt et al 1998
). However, a stepwise multiple regression analysis showed that genotype at this locus accounted for only 4% of the variance of withdrawal symptoms, compared to 16% for drinking style. Only two alleles were observed in that study, A9 (f = .23) and A10 (f = .77), both within the range for the European samples in the present study.Since this VNTR polymorphism is in the 3′-untranslated region (UTR), it does not affect the actual protein product, but it may affect mRNA localization, transcript stability, or regulation of protein synthesis. For example, it has recently been shown that cytoplasmic polyadenylation elements (CPE) in the 3′-UTR of α-CaMKII mRNA are important for translation induced by the binding of cytoplasmic polyadenylation element binding protein (CPEB) in neurons (
Wu et al 1998
). Furthermore, the distance of the CPE from the polyadenylation signal AAUAA affects the efficacy of polyadenylation-induced translation, presumably from structural effects. Similarly, a 10 nucleotide cis-element in the 3′-UTR of the glucose transporter (GLUT1) has been shown to increase transcript levels and half-life of luciferase reporter genes (Boado and Pardridge 1998
). Although the 3′-UTR of the DAT mRNA does not have the CPE specific for CPEB (UUUUUAU, consensus) near either of the two possible polyadenylation sites that flank the 3′-VNTR (Kawarai et al 1997
), there may be other such as yet unidentified elements. However, the number of repeats does not significantly affect monoamine metabolite concentrations in cerebrospinal fluid (Jonsson et al 1998
). Unfortunately, that study was performed with a small mixed European sample (n = 65) representing only the nine- and ten-repeat alleles, and only five individuals were homozygous for A9 (compared to 30 homozygous for 10/10), possibly preventing any differences from reaching significance.Alternatively, it is possible that this polymorphism is in linkage disequilibrium with another nearby polymorphism.
Thompson et al 1997
report strong linkage disequilibrium to an exon 15 polymorphism. The extent of linkage disequilibrium is not known, although it does not appear to be in significant linkage disequilibrium with the biallelic TaqI RFLP (Vandenbergh et al 1992b
), at least in the Caucasian and African-American samples reported. Based on linkage analysis, the following order of markers was obtained: pterD5S3923′-VNTRIN08-VNTRD5S406 (Kelsoe et al 1996
). D5S678 has been reported to be approximately 35 kb downstream from the 3′-VNTR (Vandenburgh et al 1995
), placing the gene in the proposed speech delay critical region for cri du chat syndrome (Church et al 1995
).Since other VNTRs, such as the 48-bp VNTR at DRD4 mentioned previously, show sequence variation among alleles with the same number of repeat units (
Lichter et al 1993
), we have also sequenced a number of homozygous individuals selected from different populations. Each of the five A10 homozygotes sequenced (Adygei, Cheyenne, San Francisco, Chinese, Arizona Pima, Yoruba) are identical to the previously published ten-repeat sequence in the repeat region (Vandenbergh et al 1992b
), consistent with the report by Leighton et al 1997
, who also found no difference in two randomly chosen individuals with the ten-repeat allele, one Chinese and one Australian with European ancestry. However, the sequence from an individual homozygous for the nine-repeat allele (Mbuti), shows evidence for heterozygosity at two nucleotides. This latter result is suggestive of a second level of variation for this polymorphism, similar to the findings for DRD4. Future studies will need to take this possibility into account. We found one consistent difference from the published sequence in all individuals sequenced (n = 9): one less nucleotide 3′ of the repeat region [our sequence: GACCA-CACTC; Vandenbergh et al 1992b
: GACCAACACTC].A concern when typing a VNTR such as this is allele dropout, or preferential amplification of shorter alleles when the size difference between the two alleles for an individual are large. This question of preferential amplification of shorter alleles (i.e., less than nine repeats) was raised previously in a study of schizophrenic subjects (
Persico and Macciardi 1997
). Even more recently, an association study reported seven A6 homozygotes out of 444 individuals (f = .016), comprising 78% of the observances of this rare (f = .02) allele (Lerman et al 1999
), a result inconsistent with Hardy-Weinberg. On the most part, we did not find preferential amplification of shorter alleles (i.e., many individuals were heterozygous 3/9, 3/10, 7/10). The only exceptions were seven individuals (four Druze, three Yoruba) who initially amplified only A3, six of whom were found to be heterozygous upon reamplification.To date, there have been numerous studies using this 40-bp VNTR in the 3′-untranslated region of the dopamine transporter gene. Among the population-based association studies, results are inconsistent at best. If one does not precisely match cases and control subjects for ancestry, the default assumption must be that the samples are likely to be drawn from universes with different allele frequencies. However, how different they are likely to be depends on the allele frequencies in the specific ancestral populations, especially those most discrepant in representation between case and control subjects. That can only be determined empirically for each ancestral population because enough different allele frequency distributions exist to know there is no universal pattern. Until such studies are done for the genetic marker being studied, it must be assumed that there is a very high false positive rate for a population-based case-control association study. As the current study of SLC6A3 3′-VNTR allele frequencies demonstrates, “European” is not a sufficient criterion at this locus for matching cases and controls. (We note parenthetically that “Caucasian” is a genetically meaningless term encompassing populations as far east as India but with no precise boundaries that should never be used as a criterion for matching.) The current study does not constitute a sufficient survey of Europe to allow an adequate attempt at setting the significance level that would have to be exceeded to be able to attribute the significance even partially to the diagnostic difference between case and control subjects, but it is sufficient to demonstrate that false positives are highly probable if case and control subjects are not precisely matched for ethnic background.
Acknowledgements
This research was supported by a Yale University Fellowship to AMK, USPHS grants NS01795 to MAP, and GM57672, MH30929, and MH39239 to KKK.
We would like to thank William C. Speed and Michael I. Seaman for assistance with sequencing, and all of the Kidd lab members for helpful discussion and support throughout. We especially thank J. R. Kidd for her work on assembling and maintaining the population DNA resource used in this study.
References
- Ten nucleotide cis element in the 3′-untranslated region of the GLUT1 glucose transporter mRNA increases gene expression via mRNA stabilization.Brain Res Mol Brain Res. 1998; 59: 109-113
- No evidence for linkage or association between the dopamine transporter gene and schizophrenia in a French population.Psychiatry Res. 1995; 59: 1-6
- VNTR polymorphism for the human dopamine transporter gene (DAT1).Hum Mol Genet. 1993; 2: 335
- Short tandem repeat polymorphism evolution in humans.Eur J Hum Genet. 1998; 6: 38-49
- The world-wide distribution of allele frequencies at the human dopamine D4 receptor locus.Hum Genet. 1996; 98: 91-101
- Molecular definition of deletions of different segments of distal 5p that result in distinct phenotypic features.Am J Hum Genet. 1995; 56: 1162-1172
- The dopamine transporter.J Neurosci. 1995; 15: 1714-1723
- Polygenic inheritance of Tourette syndrome, stuttering, attention deficit hyperactivity, conduct, and oppositional defiant disorder.Am J Med Genet (Neuropsychiatr Genet). 1996; 67: 264-288
- Association of attention-deficit disorder and the dopamine transporter gene.Am J Hum Genet. 1995; 56: 993-998
- Human and mouse dopamine transporter genes.Brain Res Mol Brain Res. 1995; 30: 327-335
- Population genetic study of the human dopamine transporter gene (DAT1).Genet Epidemiol. 1995; 12: 303-308
- Dopamine transporter is required for in vivo MPTP neurotoxicity.J Neurochem. 1997; 69: 1322-1325
- Different kinetics govern dopaminergic transmission in the amygdala, prefrontal cortex, and striatum.J Neurosci. 1994; 14: 442-450
- The D4 dopamine receptor (DRD4) maps to distal 11p close to HRAS.Genomics. 1992; 13: 208-210
- Population studies of polymorphisms at loci of neuropsychiatric interest (tryptophan hydroxylase (TPH), dopamine transporter protein (SLC6A3), D3 dopamine receptor (DRD3), apolipoprotein E (APOE), μ opioid receptor (OPRM1), and ciliary neurotrophic factor (CNTF)).Genomics. 1998; 1998: 289-297
- Genetic association between dopamine transporter protein alleles and cocaine-induced paranoia.Neuropsychopharmacology. 1994; 11: 195-200
- The dopamine transporter protein gene (SLC6A3).Genomics. 1995; 30: 459-463
- Confirmation of association between attention deficit hyperactivity disorder and a dopamine transporter polymorphism.Mol Psychiatry. 1997; 2: 311-313
- Cloning, pharmacological characterization, and chromosome assignment of the human dopamine transporter.Mol Pharmacol. 1992; 42: 383-390
- Hyperlocomotion and indifference to cocaine and amphetamine in mice lacking the dopamine transporter.Nature. 1996; 379: 606-612
- Dopamine transporter gene polymorphism and psychiatric symptoms seen in schizophrenic patients at their first episode.Am J Med Genet (Neuropsychiatr Genet). 1996; 67: 406-408
- Mechanisms of amphetamine action revealed in mice lacking the dopamine transporter.J Neurosci. 1998; 18: 1979-1986
- Polymorphisms in the dopamine, serotonin, and norepinephrine transporter genes and their relationships to monoamine metabolite concentrations in CSF of healthy volunteers.Psychiatry Res. 1998; 79: 1-9
- Structure and organization of the gene encoding human dopamine transporter.Gene. 1997; 195: 11-18
- Possible locus for bipolar disorder near the dopamine transporter on chromosome 5.Am J Med Genet. 1996; 67: 533-540
Kidd JR, Pakstis AJ, Kidd KK (1993): Global levels of DNA variation. The Fourth International Symposium on Human Identification. Madison, WI: Promega Corporation, pp 21–30.
- A global survey of haplotype frequencies and linkage disequilibrium at the DRD2 locus.Hum Genet. 1998; 103: 211-227
- Absence of linkage for schizophrenia on the short arm of chromosome 5 in multiplex Canadian families.Am J Med Genet (Neuropsychiatr Genet). 1997; 74: 472-474
- Association of a polymorphism in the dopamine-transporter gene with Parkinson’s disease.Mov Disord. 1997; 12: 760-763
- The dopamine transporter gene and Parkinson’s disease in a Chinese population.Neurology. 1997; 49: 1577-1579
- Evidence suggesting the role of specific genetic factors in cigarette smoking.Health Psychol. 1999; 18: 14-20
- No association between alleles or genotypes at the dopamine transporter gene and schizophrenia.Psychiatry Res. 1994; 52: 17-23
- A hypervariable segment in the human dopamine receptor D4 (DRD4) gene.Hum Mol Genet. 1993; 2: 767-773
- Dopamine transporter gene polymorphism and alcoholism.Biochem Biophys Res Commun. 1995; 211: 28-32
- Amplification of DAT1 (human dopamine transporter gene) 3′ variable region in the Japanese population.Hum Hered. 1995; 45: 262-265
- Genetic polymorphism of the 3′ VNTR region of the human dopaminergic function gene DAT1 (human dopamine transporter gene) in the Mongolian population.Hum Biol. 1996; 68: 509-515
Palmatier MA, Kang AM, Kidd KK (in press): Global variation in the frequencies of functionally different catechol-O-methyltransferase alleles. Biol Psychiatry.
- Human dopamine transporter gene polymorphism (VNTR) and alcoholism.Am J Med Genet (Neuropsychiatr Genet). 1997; 74: 480-482
- D2 dopamine receptor gene TaqI A1 and B1 restriction fragment length polymorphisms.Biol Psychiatry. 1996; 40: 776-784
- Lack of association between dopamine transporter gene polymorphisms and delusional disorder.Am J Med Genet (Neuropsychiatr Genet). 1998; 81: 163-165
- Genotypic association between dopamine transporter gene polymorphisms and schizophrenia.Am J Med Genet (Neuropsychiatr Genet). 1997; 74: 53-57
- Dopamine transporter gene polymorphisms are not associated with polysubstance abuse.Biol Psychiatry. 1993; 34: 265-267
- Exclusion of close linkage of the dopamine transporter gene with schizophrenia spectrum disorders.Am J Psychiatr. 1995; 152: 134-136
- A genetic association for cigarette smoking behavior.Health Psychol. 1999; 18: 7-13
- Allelic association of a dopamine transporter gene polymorphism in alcohol dependence with withdrawal seizures or delirium.Biol Psychiatry. 1997; 41: 299-304
- A 40-nucleotide repeat polymorphism in the human dopamine transporter gene.Hum Genet. 1993; 91: 405-406
- Modification of alcohol withdrawal by the A9 allele of the dopamine transporter gene.Am J Psychiatr. 1998; 155: 474-478
- Novelty seeking and a dopamine transporter gene polymorphism (DAT1).Biol Psychiatry. 1997; 42: 1070-1072
- Association study of Tourette syndrome and alcoholism with novel polymorphisms of the dopamine transporter.Abstr Soc Neurosci. 1997; 23: 408
- Human dopamine transporter gene 5′ and 3′ structures and molecular genetics.Abstr Soc Neurosci. 1995; 21: 374
- Human dopamine transporter gene (DAT1) maps to chromosome 5p15.3 and displays a VNTR.Genomics. 1992; 14: 1104-1106
- A human dopamine transporter cDNA predicts reduced glycosylation, displays a novel repetitive element and provides racially-dimorphic TaqI RFLPs.Brain Res Mol Brain Res. 1992; 15: 161-166
- Linkage disequilibrium between the dopamine transporter gene (DAT1) and bipolar disorder.Genet Epidemiol. 1997; 14: 699-704
- Evolution and the Genetics of Populations. vol. 2. The University of Chicago Press, Chicago1969
- CPEB-mediated cytoplasmic polyadenylation and the regulation of experience-dependent translation of α-CaMKII mRNA at synapses.Neuron. 1998; 21: 1129-1139
Article info
Publication history
Accepted:
April 21,
1999
Received in revised form:
April 19,
1999
Received:
February 5,
1999
Identification
Copyright
© 1999 Society of Biological Psychiatry. Published by Elsevier Inc.
User license
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0) | How you can reuse 
Elsevier's open access license policy
Creative Commons Attribution – NonCommercial – NoDerivs (CC BY-NC-ND 4.0)
Permitted
For non-commercial purposes:
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article (private use only, not for distribution)
- Reuse portions or extracts from the article in other works
Not Permitted
- Sell or re-use for commercial purposes
- Distribute translations or adaptations of the article
Elsevier's open access license policy
