The aims of this study were twofold: first to characterise the genomic structure of the DAT1
VNTR; and second to investigate the effect of the 9- and 10-repeat VNTR alleles on levels of transcription. It is possible that any effects, both on behavioural phenotype or levels of DAT1
expression, commonly associated with the VNTR are due to polymorphisms within
the repeat rather than the actual length of the VNTR itself. This theory has often been postulated to explain cases of non-replication of association studies with complex disorders. Recent work on the DRD4
exon III VNTR polymorphism has highlighted the existence of both within individual and between individual variation in repeat motif sequence [24
]. In other words the 1st
repeat motifs within
a certain length DRD4
VNTR allele may differ in sequence in a single individual, but there is also variation within the 1st
motif between different individuals. We sequenced the DAT1
VNTR in individuals homozygous for either the 9- or 10-repeat allele. Differences were found between repeat motifs in both the 9- and 10-repeat alleles, and nine variants were detected in total. Interestingly, although differences were found between the 9- and 10-repeat alleles, the order of the motifs was identical for all examples of each allele size. Given that not one variation in these sequences was observed in the 60 chromosomes sequenced, it can be concluded that any between individual variations that do occur are extremely rare and unlikely to be the cause of common disorders such as ADHD.
Previously, Ueno et al screened the DAT1
3'UTR for novel polymorphisms in a Japanese population, but only detected a G>A SNP located upstream of the VNTR at position 2319 [25
]. Fuke et al have also sequenced the DAT1
VNTR in a small number of Japanese subjects with results identical to those we describe [20
]. Miller et al report a SNP that abolishes a Dra
I restriction site, which they claim to be within the 10-repeat allele [26
]. Closer inspection of their paper, however, shows that this polymorphism is located outside of the repeat region and thus their conclusion that they have found a 'novel variant of the 10-repeat allele' is technically incorrect. It is possible, however, that either this polymorphism or that discovered by Ueno et al [25
] may be the real risk variant, and associations reported for the VNTR polymorphism may be a result of LD relationships with either of these SNPs.
The lack of between-individual variation in the sequence of the DAT1
VNTR is perhaps surprising given the number of polymorphisms seen in other large VNTRs (e.g. [24
]). VNTRs are often mutation hotspots with a high level of genetic recombination due to misaligned repeat units that can cause variations in sequence as well as length [27
]. Furthermore, according to several estimates of SNP frequency across the genome (e.g. [30
]), the average total repeat length of ~400 bp might be expected to contain at least one SNP, especially as it is not located in a highly-conserved coding-region of the gene. The fact that there appears to be no between-individual variations within the 9- and 10-repeat VNTR sequences suggests that it may be subjected to some form of selective pressure, and perhaps have an important functional role. A clue to the importance of the DAT1
VNTR may come from its location within the 3'UTR of the gene. Sequence motifs in the 3'UTR have been shown to have important roles in translation, mRNA stability, subcellular localization, and polyadenylation [6
]. Alternatively, given that the between-motif variations seen within the VNTR are ubiquitous, it is possible that they are relatively ancient and have been pushed towards fixation within the population via demographic and random genetic drift.
The second aim of this study was to examine in vitro
possible functional consequences of the DAT1
VNTR polymorphism. We made constructs containing the 9- and 10-repeat alleles of the DAT1 VNTR and the luciferase reporter gene, and transiently transfected them into SH-SY5Y and HEK-293 cells. Constructs containing the VNTR alleles gave slightly lower levels of luciferase activity compared to vectors without the inserts. Although these differences were not statistically significant, they appear to go against the conclusions of a previous study in which the VNTR sequence acted as a strong enhancer of transcription [18
]. However, the trend of our results do agree with data presented by Fuke et al who found that vectors containing the VNTR polymorphism gave lower luciferase expression levels compared to positive control vectors having no VNTR inserted [20
]. Furthermore, Greenwood & Kelsoe found no evidence to suggest that the VNTR sequence acts as a transcriptional enhancer [23
]. The discrepancies between these results may be explained by the different methodological strategies employed. Michelhaugh et al utilised GFP as a reporter gene [18
], which does not have the sensitivity or specificity associated with the luciferase reporter system, and is generally used in qualitative detection assays. Furthermore, they cloned the VNTR sequence upstream of the SV40 promoter, in a location not analogous to the 3'UTR location of the polymorphism in the DAT1 gene. The importance of the location of regulatory elements has been well-documented [19
] and so it is possible that the enhancing effect reported by Michelhaugh et al [18
] is specific to the location of the VNTR in their particular experimental design. Additionally, it is possible that the VNTR polymorphism has an important role in mediating processes such as mRNA stability – such effects will have been missed by Michelhaugh et al, as their insert is not transcribed.
We found no significant differences in luciferase activity between constructs containing the 9- or 10-repeat DAT
1 VNTR alleles. Other transient expression studies that have compared the VNTR alleles have provided mixed results. Fuke et al found that luciferase expression was significantly higher in cells transfected with vectors containing the 10-repeat allele compared to the 7-repeat or 9-repeat alleles [20
]. Their study used a different cell-line, COS-7, which is derived from African Green Monkey kidney. Miller and Madras, on the other hand, concluded that the 9-repeat allele was correlated with increased expression in HEK-293 cells, but that expression was further mediated by a SNP also located in the 3'UTR of DAT1
]. Finally, our data is in agreement with that of Greenwood & Kelsoe, who also found no effect on transcription of the 9- and 10-repeat alleles in SN4741 cells, but found that introns 9, 12, and 14 may contain enhancer elements capable of increasing expression ~2-fold [23
Therefore, our data suggest that the VNTR polymorphism itself may not
be functional. Unlike the studies of Fuke et al and Miller & Madras [20
], our inserts contained no flanking sequence and were restricted to specifically the VNTR itself. There is considerable evidence that there is a functional polymorphism in the vicinity of the 3'UTR of the DAT1
gene. Genetic association studies with ADHD, SPECT brain imaging studies, and correlations with levels of DAT protein and DAT1
mRNA all suggest that a variant associated with DAT1
expression is present in this region. Given that in each of these associations, the VNTR has been nominated as the causative polymorphism, it is likely that the real risk variant is in strong LD with it. Ueno et al and Miller et al both report novel SNPs located within the 3'UTR and close to the VNTR [25
]. It is possible that either of these SNPs, or another polymorphism yet to be characterised, is mediating expression of DAT1
and is the real risk variant.
There are several obvious limitations to this study. First, it is not known how well in vitro studies of gene expression reflect patterns seen in vivo. Future work could employ animal models to characterise more realistically the effect of the VNTR on DAT1 expression. Second, while we ensured that we used cell-lines that naturally express DAT1, and inserted the VNTR into the correct 3'UTR location of the luciferase reporter gene, our constructs could have been improved by using a homologous DAT promoter. Third, we only cloned a very small portion of the DAT1 gene. Even though this was necessary to examine functional effects specific to the VNTR, the fact that the majority of the DAT1 gene was absent means that it is likely that several cis-acting regulatory elements were not present in the constructs and thus the observed expression may not reflect the actual regulation of the gene. Finally these studies have only analysed DAT1 gene expression in the basal state, and complexities such as the induction of expression by factors such as cellular signals would thus be missed. Future work should focus on systematically characterising the remainder of the 3'UTR to discover the functional effects of other polymorphisms in this candidate region.