Introduction

Although we are all genetically individual humans share 99.9% sequence similarity between their genomes. It is estimated that the human genome has about 10 million polymorphisms or allelic variations. It is this variation between our genomes, and the mechanisms employed to regulate expression of our genes, that lead in part to individuality between us. A different combination of these polymorphisms creates the “genetic fingerprint”, which is used to distinguish one person’s DNA from another.

Genetic factors have been implicated in the aetiology of mental illness and many studies have determined that changes in protein structure correlate with a predisposition to specific conditions. There are, however, a number of polymorphisms found in non-coding sequences, which do not affect protein structure, that have been identified as risk factors for behavioural and affective disorders including Alzheimer’s disease [1], schizophrenia [2], anxiety [3], obsessive compulsive disorder (OCD) [4], unipolar depression [5], bipolar depression [6–8], migraine [9, 10], and Parkinson’s disease [11]. Having a genetic variant described as predisposing for a certain disorder does not mean that an individual with that variant will develop that disorder, however these polymorphisms might act as “markers” indicating a predisposition to a disorder. We have demonstrated that polymorphisms have functional effects on gene expression, displaying both tissue-specific and stimulus-inducible regulation [12–16]. The functional significance of a polymorphism may be relevant to an individual’s response to pharmacological treatment or for modulation in response to environmental factors. This suggests that individuals with a particular combination of polymorphisms may respond differently to the same medications or environmental stresses.

Identification of specific polymorphisms which associate with specific disorders, or respond to specific environmental stimuli, could lead to tailored treatments or bespoke medication for individuals based on genome variation. This review briefly describes types of allelic variation that exist, with particular emphasis on the Variable Number Tandem Repeats (VNTR) found within non-coding sequences in the serotonin transporter (5-HTT also termed SLC6A4) and dopamine transporter (DAT1 also termed SLC6A3) genes. It will also consider the possibility that these VNTR represent a class of cis-regulatory element to which a number of trans-acting factors may bind to modulate gene expression and it will focus on the modulatory nature of VNTR function in response to illicit drugs.

Allelic variation: Single Nucleotide Polymorphism (SNP)

Genetic variation is manifest in a number of modes. The most widely investigated polymorphisms and the first type that were identified, are single nucleotide polymorphisms (SNPs). SNPs constitute a major source of variation in the genome and occur about every 1,000 bp in the entire human sequence. There is even variability in the type of SNP that can occur. The simplest form of SNP is a base pair substitution where one nucleotide in the DNA sequence is exchanged for another, for example, an adenine (A) may be substituted for one of the other nucleotides such as a guanine (G). Other SNPs include; omissions, where a nucleotide is absent in one sequence but not in another or insertions, where an extra nucleotide is found in a DNA sequence. The SNP Consortium’s Allele Frequency project or the International HapMap project is a collaboration of numerous international research groups working together to produce a freely available, extensive analysis of the variation in the human genome across different populations (http://www.snp.cshl.org/). HapMap, which estimates the occurrence of 10 million SNPs within the genome, has currently identified more than 1.5 million SNPs. Many of these SNPs occur together, which is termed linked, for example as in Fig. 1a, a SNP at locus A might always be present with a SNP at locus B, or even with another polymorphism at loci C or D. This linkage occurs as a result of inheritance and SNPs which are always inherited linked together are known as haplotypes (Fig. 2), so a particular haplotype, rather than an individual SNP, may be associated with predisposition to a specific condition. Haplotypes may be identified by a common tagging SNP (tSNP) reducing the need to examine all individual SNP loci when performing an association study. A tSNP is defined as a change in a particular nucleotide at a given position associated with a specific haplotype (Fig. 2).

Fig. 1
figure 1

Definition of polymorphisms. The polymorphisms at locus A and locus B represent Single Nucleotide Polymorphism(s) or SNP(s), where one nucleotide has been substituted for another. Deletion or insertion of a single nucleotide is another example of a SNP. Locus C shows a micro-satellite polymorphism, or Short Tandem Repeat (STR). Common STRs are composed of di-, tri- or tetra-nucleotide repeats. Locus D is representative of a Variable Number Tandem Repeat (VNTR). These are a subclass of mini-satellite composed of repeats that are 10–100 nucleotides in length. They are variable as each allele may have a different copy number of these repeats. For example, here one allele has 3.5 copies of the repeat while the other has only two copies

Full size image
Fig. 2
figure 2

Haplotypes and tagging SNPs. Linked polymorphisms that are inherited together are termed haplotypes. Here, for example, the first bold polymorphic position is always present with the other two bold polymorphic positions. At any of these positions the polymorphism is a substitution SNP where one nucleotide is substituted for another e.g., g for a, or t for c. In this example this would lead to the possible existence of nine different haplotypes. As genotyping these SNPs allows identification of each haplotype without having to analyse the whole sequence they are know as tagging SNPs or tSNPs. tSNP are particularly useful if one is always linked with a larger polymorphism as genotyping a SNP is more cost effective and less time consuming than genotyping a mini- or micro-satellite

Full size image

Allelic variation: micro-, mini-satellites, repeats and Variable Number Tandem Repeats (VNTR)

There are over 500,000 identified micro- or mini-satellites in the human genome. Mini-satellites consist of 10–100 nucleotides repeated several times in tandem, which are bordered by unique DNA sequences. Micro-satellites also called Short Tandem Repeats (STRs), are similar to mini-satellites but the sequence repeats are smaller e.g., tetra- and di-nucleotide sequence repeats (Fig. 1). Most repeats are not comprised of “perfect” repeating units and show some degree of degeneration, where one repeat may be slightly different to the next but overall the core consensus sequence is maintained. Genotyping is employed to determine if a repeating element found within the genome is variable i.e., exists as a different allelic copy number between individuals in a population, for example at the same allelic loci, one individual may have 10 copies of a repeat, whilst another may have 12 copies. In this review we will concentrate on a subclass of mini-satellite repeats commonly referred to in the literature as Variable Number Tandem Repeats (VNTR). We define these VNTR as having sufficient DNA sequence in the repeat, for example greater than 6 bp, to act as a sequence specific DNA binding site for proteins such as transcription factors, and therefore have the potential to act as transcriptional regulatory domains (Fig. 1). This situation is similar to the regulatory repeat domains seen in retroviruses or herpesviruses [17–19].

These potential regulatory classes of VNTR are many and varied but can share some common features. Firstly, the majority of these VNTRs are located in non-coding regions of the genome. Secondly, many are present in the genome as a feature of an emerging evolution. VNTRs display evolutionary conservation between humans and non-human primates but are often not found in lower mammals [20, 21]. Indeed, many appear to be constantly evolving in modern primates by dramatic changes in copy number or large variation in primary sequence of the repeat (personal observations). Thirdly, many repeats contain potential cis-regulatory elements or transcription factor binding sites producing localised clusters of one or more particular binding site which may have implications for gene expression [14, 15, 21]. Lastly, in support of the prediction that these VNTR may act as modulators of gene expression many are found at higher density in gene-enriched areas compared to non-genic regions (G. Breen SDGP, IOP, Kings College, London, personal communication). Potential regulatory VNTRs can be identified by analysis of variation using a genome browser such as the UCSC Genome Bioinformatics Site, http://www.genome.ucsc.edu, and have been identified for example, in the following human genes: the serotonin transporter (SERT/5HTT), dopamine transporter (DAT1), N-methyl-d-aspartame receptor 1 (GRIN1/NMDAR), dopamine D4 receptor (DRD4), and monoamine oxidase A (MAOA).

Since they do not act to alter protein structure, these VNTRs would be predicted to act as endogenous modulators of transcription or alter post transcriptional properties (of the gene) such as mRNA stability. We propose, based on our published data, that this subclass of VNTR would function in both a tissue-specific and stimulus-inducible manner to fine-tune gene expression. This fine tuning could be correlated, mechanistically, not only with normal physiological function and variation between individuals, but also with a predisposition to behavioural disorders by altering neurotransmitter signalling in response to challenges and stress. Furthermore, if stimulus-inducible expression varies dependant on a specific polymorphism associated with a disorder then that may have similar implications in the response of an individual to pharmacological treatment of that disorder [12–15, 22].

Serotonin transporter

5-HT and the 5-HT transporter (sert/5-HTT/SLC6A4)

Clinical abnormalities in serotonin (5-HT) metabolism have been implicated in the pathophysiology of many CNS-related disorders. Early indications that 5-HT played a role in depression were based on the biogenic amine hypothesis which states that depression is caused by a deficiency of monoamines [23]. It was demonstrated that drugs which increased monoamines alleviated depression and visa-versa. It is now known that the monoamine 5-HT (5-hydroxytryptamine) is a major neuromodulator of cognitive and emotional behaviours involved in regulating diverse processes such as appetite [24, 25], memory and learning [26], aggression and antisocial behaviour [27], anxiety and depression [28, 29].

In the CNS, 5-HT and 5-HTT are mostly localized in the neurons of the raphe nuclei whose processes innervate many areas of the brain thought to be involved in cognition and behaviour regulation [30, 31]. Decreased serotonergic signaling as a result of low levels of 5-HT in the synaptic cleft, desensitization of 5-HT receptors, decreased expression of the 5-HT receptor, or changes in the expression of monoaminergic transporters can lead to the development of mood disorders such as depression. Conversely, increasing the concentration of 5-HT in the synaptic cleft can lead to the development of affective disorders such as schizophrenia. It is not surprising then, that the serotonergic signaling pathway is often a target for treating these complex disorders. In particular, changing the bioavailability of 5-HT in the synaptic cleft is a major mechanism of regulating 5-HT signaling. Abundance of 5-HT in the cleft is regulated principally via re-uptake by the monoaminergic, sodium and chloride dependant, 5-HT transporter (SERT/5HTT/SLC6A4) [32]. Following re-uptake, 5-HT is degraded by monoamine oxidases for recycling [33].

Since 5-HTT is a key regulator of the bioavailability of 5-HT, any modulation in the expression or action of 5-HTT would be expected to have consequences on behaviour and mood and has been strongly linked with depression [34, 35]. As such 5-HTT is a major target for pharmaceutical intervention in mood disorders. The family of antidepressants known as selective 5-HT reuptake inhibitors or SSRIs (such as Prozac) target only 5-HTT [36]. SSRIs block the transporter effectively raising, or at least maintaining, the concentration of 5-HT in the synaptic cleft. Tricyclic antidepressants (TCA) such as amtriptyline also target the 5-HTT and are still in widespread use despite having a number of undesirable side effects because they also target other norepinephrin transporters [36]. 5-HTT is also a target for illicit drugs such as ecstasy (MDMA) [37, 38] and cocaine [39–41].

Variation in the clinical response of individuals to various antidepressant treatments including the SSRIs, in part, will correlate with variation in their ability to process the drug and in their “normal” physiological control of the serotonergic system. For the latter, polymorphisms in the coding regions of the 5-HTT gene cannot account for these differences, and have not been associated with any psychiatric conditions. The identification of a number of VNTR in the non-coding regions of the 5-HTT gene that may have functional effects on the expression of 5-HTT are currently being investigated by a number of research groups, including our own, for their correlation with behavioral and psychiatric conditions.

5-HTTLPR VNTR: association studies and functionality

The human 5-HTT gene consists of 14 exons and spans 37.8 kb on chromosome 17q11.2. It encodes a 630 amino acid protein, containing 12 putative transmembrane domains and it shares 92% sequence homology with the rat 5-HTT gene [42, 43]. Mutation or inappropriate expression of the gene has been postulated as a possible cause of affective disorders. Although there are SNPs in the coding sequence of 5-HTT, thus far, no differences in amino acid sequence have been found between affective disorder patients and healthy controls, this finding therefore does not support a role for alteration of the primary structure of the coding region of the 5-HTT gene in the pathogenesis of affective disorders. However, possible associations between two VNTR polymorphisms in non-coding regions of the 5-HTT gene and susceptibility to affective disorder have been analysed.

The first of these to be identified was a biallelic insertion/deletion found in the 5′ promoter region of the gene 1.2 kb upstream of the transcriptional start site [44, 45]. This VNTR, termed, 5-HTTLPR was initially identified as two variants containing either, 14 (short/deletion) or 16 (long/insertion) copies of a 22 bp repeat (Fig. 3). More recently, Nakamura et al., verified the existence of subgroups within these variants to uncover as many as 14 allelic variations of this locus [46]. One particularly interesting variant of the long polymorphism contains an A to G (la or lg) substitution, tSNP (rs25531) (Fig. 3) [47–49]. Inclusion of this tSNP (lg) has conferred clinical properties normally associated with the short allele onto the long allele. Unless otherwise stated this review examines the common l and s variants referred to elsewhere as 14A (s) and 16A (l) [46]. Individuals can be heterozygous (l/s) or homozygous (l/l or s/s) for the VNTR. The distribution of these genotypes varies, with heterozygous and homozygous l/l individuals appearing with considerably more frequency than homozygous s/s individuals in the Caucasian population [44, 45]. However, allelic frequencies vary between global populations or different ethnic groups. In a review of the literature, Smits et al. determined that the frequency of the alleles for the Caucasian population was 27.4–28.3% (l/l), 43.4–51.0% (s/l) and 21.6–28.3% (s/s), whilst in Asian populations the allelic frequency was 4.2–10.0% (l/l), 30.0–39.2% (s/l) and 55.6–60.0% (s/s) [50].

Fig. 3
figure 3

Sequence and repeat alignment of the 5-HTTLPR VNTR. Schematic representation of the human 5-HTT gene and position of the LPR VNTR (top). On the left is the sequence of the long variant of the VNTR also known as the insertion VNTR due to the presence of four extra copies of the repeat (underlined). The VNTR is not composed of perfect repeating units as can be seen by the designation of Greek symbols to each different repeat (we have arranged the repeat to highlight this core). The bold underlined nucleotide “a” in the first ϕ repeat represents the a/g SNP (rs25531) which confers the functional properties associated with the short variant onto the long variant. On the right is the sequence of the short variant also known as the deletion

Full size image

Association studies compare the genotype or haplotype of loci between patients and healthy individuals and analyse the data obtained to identify possible correlations between the occurrence of a specific variant and a disorder or disease. Many association studies have indicated that the 5-HTTLPR variants have been linked with a number of affective disorders. The homozygous s/s genotype has been associated with an increased risk to depression [51], unipolar depression [7], bipolar depression [52], anxiety [53, 54], substance abuse [55], and a predisposition to suicide or depression following stressful life events [56, 57]. The effect of stress may be important in these association studies as would be suggested by recent evidence demonstrating an altered amygdala response to stress depending on variable 5-HT signaling resulting from 5-HTTLPR genotype [54, 58]. Conversely, the homozygous l/l genotype is associated with a predisposition to OCD [59] and increased intensity of hallucinations in individuals with schizophrenia [60].

Although one can correlate the function of a specific VNTR with endogenous gene expression, more often the ability of a VNTR to modulate gene expression is analysed in a recombinant DNA construct termed a reporter gene construct (Fig. 4). In such analysis the isolated VNTR is tested for its ability to modulate the activity of a minimal promoter in supporting expression of a heterologous gene such as luciferase which can be easily quantified. The 5-HTTLPR VNTR have been shown to be functional transcriptional regulators that will affect gene expression in vitro. Differential expression supported by the 5-HTTLPR variants in the context of their own promoter has been demonstrated in the human placental choriocarcinoma cell line—JAr [45] and in lymphoblast cell lines [53], with the level of expression supported by the long allele being twice that supported by the short allele. In contrast, in a neuronal cell line, RN46A, Sakai et al., did not find any significant difference in activity between the l and s variants when isolated from the promoter and cloned into a luciferase reporter vector [61]. Conversely, Mortensen et al., did not find any difference in the activity of the l and s variants in JAr cells, but did see differential activity in RN46A cells [62]. In lymphoblast cell lines, the 5-HTT promoter fragments containing the long or short variant demonstrated differential stimulus-inducible response to stimuli such as forskolin or phorbol ester treatment, again with the long allele displaying proportionally greater responses [53]. In addition, having an s allele, as either s/l or s/s resulted in 40–70% less mRNA production than when genotype for the allele is l/l [53, 63, 64]. The same study also found greater uptake rates and an increased binding for 5-HT in lymphoblast cells with the l/l genotype compared with either s/l or s/s [53]. In contrast, a study employing allelic expression imbalance (AEI) techniques did not observe a correlation between 5-HTTLPR variants and 5-HTT mRNA levels in B-lymphocytes [65].

Fig. 4
figure 4

Schematic example of a reporter gene construct. The isolated VNTR sequence is cloned upstream of a minimal promoter such as the SV40 minimal promoter, driving a reporter gene such as firefly or renilla luciferase, or a marker gene, for example, green fluorescent protein (GFP). These constructs are transfected into cells by varying methods. The activity measured as reporter or marker gene read out is compared a construct with only the minimal promoter driving expression the reporter/marker gene (no VNTR control). This comparison provides a measure of the modulatory, and therefore functional, ability of the VNTR to regulate gene expression

Full size image

Investigations into the functionality of the 5-HTT VNTR in vivo have also been undertaken. Heinz et al., examined in vivo, human 5-HTT binding in the midbrain using single photon emission computed tomography (SPECT) [66]. They found a difference in binding capacity between l/l and −s individuals which was consistent with the in vitro data of Lesch et al. [53], which states that individuals with the short allele show less binding [66]. Contrastingly, later SPECT studies did not show any correlation between genotype and functional regulation of the 5-HTT in the midbrain [67]. Supporting a lack of function in vivo, quantitative autoradiography in the prefrontal cortex of post-mortem brains of suicide victims did not show any correlation between 5-HTT binding and 5-HTTLPR genotype [68]. Further in vivo evidence against a role for this polymorphism regulating 5-HTT function exists from recent positron emission topography (PET) studies which fail to show any association between genotype and 5-HTT binding in the brain [69, 70]. To further compound matters, analysis of post-mortem brains found no difference in binding across the dorsal raphe, median raphe or substantia nigra in homozygous samples (of either s/s or l/l) but found lower levels of binding in heterozygous individuals [40]. As the authors of this study point out, the differences in binding between this study and that of Lesch et al. [53] may be due to the relatively small sample size employed and the same may be the case in other in vivo studies. However in the same study, a correlation between decreased mRNA expression and individuals with at least a single copy of the s allele was reported, which is in agreement with Lesch et al. [53]. More recently, a study investigating the affects of this polymorphism in cognitive error processing in the prefrontal cortex has demonstrated that the 5-HTTLPR short variant is associated with enhanced responsiveness of the brain [71].

Transgenic animal studies have also hinted at the possible functionality of the 5-HTTLPR by over-expressing human 5-HTT in mice to mimic the higher mRNA expression levels, increased uptake and binding observed with the l/l genotype. The l/l genotype is associated with non-anxious controls in healthy human volunteers, and in the transgenic mice results in a low-anxiety phenotype [72].

STIN2 VNTR association studies and functionality

A second widely studied VNTR found in the non-coding region of the 5-HTT gene is located in intron 2 and comprises 9, 10 or 12 copies of a 16–17 bp repeat termed, STin2.9, STin2.10 and STin2.12 respectively [5, 73] (Fig. 5). Individuals can be hetero- or homozygous for the STin2 VNTR giving rise to six possible genotypes 9/9, 10/10, 12/12, 9/10, 9/12 and 10/12. Association studies have also linked this polymorphism with a number of cognitive and affective disorders including; unipolar depression and bipolar disorder to which the STin2.9 allele seems to be a predisposing factor [5, 6]; OCD in which the STin2.12 was seen at a higher frequency in affected individuals compared to healthy controls [4, 74]; migraine, where an increased incidence of the STin2.12 homozygous genotype was observed in migraine without aura, and an increased incidence of the STin2.9 allele was observed in migraine with aura, when compared to non-migraine controls [9]; anxiety which correlated with the presence of at least one copy of STin2.9 [3]; the STin2.12 allele has been associated with varying aspects of schizophrenia [2, 63, 75]; individuals homozygous for the 10 allele of STin2 have recently been shown to be predisposed to suicide [76, 77].

Fig. 5
figure 5

Sequence and repeat alignment of the 5-HTT STin2 VNTR. Schematic representation of the human 5-HTT gene and position of the STin2 VNTR (top). The STin2 repeats exist as 9–12 copies of a 16–17 bp repeat. The repeats are not perfect repeating units. Variation in repeat sequence is represented by a Greek symbol. As can be seen the 10 repeat lacks a γ and η repeats present in the 12 copy while the nine lacks a further ϕ copy

Full size image

While association of the STin2 VNTR with affective disorders has been noted, the results of meta-analysis of a number of published studies cast doubt on the solidity of these findings [78–80]. For every study reporting a positive association between a genotype and a disorder there are contradictory papers demonstrating a lack of association.

Nevertheless, the STin2 VNTR can display functionality both in vitro [12, 14, 15] and in vivo [13]. In cell lines, we demonstrated that the STin2 VNTR can support differential levels of reporter gene expression, acting as enhancers of transcription, with STin2.12 showing the most robust activity [12]. In addition, we verified that copy number of the repeat unit is not the only factor functioning in the observed regulation of transcription, but that primary sequence variation between the individual repeating units can also affect enhancer activity [14]. We have also demonstrated that in vivo, in transient transgenic mice expressing the human 5-HTT STin2 VNTRs driving a β-galactosidase marker gene, that the STin2 VNTRs may have an important regulatory role in development of the serotonergic system with genotype affecting 5-HTT expression in a temporal and tissue-specific manner [13]. This developmental role for the VNTRs may be particularly significant considering recent implications that transient alterations in 5-HT homeostasis in embryogenesis modify the fine wiring of brain connections leading to permanent changes in adult behaviour [81, 82]. In light of these findings it is possible that the effects of the 5-HTT polymorphisms are more pronounced during embryogenesis and development, a view also expressed more recently by Parsey et al. [70].

5-HTT variants and anti-depressant efficacy

The 5-HTT polymorphic genotypes have been functionally associated with differential efficacy of SSRI treatment, with individuals varying in their response. For different global populations the same response to treatment with SSRI is seen across different genotypes, suggesting that combinatorial interactions of polymorphisms unique to each population may be producing distinct effects e.g., Caucasians with 5-HTTLPR s/s genotype were found to display a poor response to SSRI, but in contrast, Asian individuals with the 5-HTTLPR l/l genotype responded poorly to treatment (for detailed review and meta-analysis see [50, 83–85]). As is the case with the 5-HTTLPR polymorphisms, individuals with different STin2 VNTRs may respond differently to treatments for depression such as TCA or SSRI. A systematic review of the current literature indicated that while Caucasians with the 5-HTTLPR s/s genotype showed the least favourable response to SSRI, in Asian populations those with the STin2 10/12 genotype showed the least favourable response [50].

Dopamine (DA) and the DAT1 transporter (DAT1/SLC6A3)

The dopaminergic system is well known to be physiologically involved in motivation and reward (reviewed in [86]), and memory and learning (reviewed in [87]). In addition an involvement of DA has been strongly implicated in a number of addictive behaviours such as drug-taking (reviewed in [88]), smoking (reviewed in [89]), overeating [90] and gambling [91]. A number of studies utilizing DA agonists and antagonists have demonstrated that DA is involved in long term potentiation (LTP); the process by which behaviour is learnt and memories are stored [92–95]. In the CNS, DA acts to fine tune neuronal firing by altering signal-to-noise ratios to produce more focused activity [96, 97]. In the CNS, DA is localised at high synaptic concentrations in the striatum and lower concentrations in the cortex, and is involved in modulating thalmo-cortical signalling pathways [98, 99]. Unlike 5-HT, which mediates its effect by acting solely on receptors located synaptically, DA also activates receptors which are found in extrasynaptic locations by DA spillover [87]. Elevation of DA concentration and the duration of DA exposure in the synaptic and extrasynaptic space are major determinants of DA receptor activation. In common with the 5-HT system, DA availability is regulated by uptake via a member of the sodium and chloride dependant NET family of proteins, the human DA transporter, DAT1 [100]. DA availability is also regulated via methylation by catechol-O-methyltransferase (COMT), which itself contains a VNTR but this is beyond the scope of this review (for review [101]). The DAT1 is a target for illicit drugs such as cocaine and methamphetamine, but also for therapeutic agents such as those used to treat Attention Deficit Hyperactivity Disorder (ADHD) e.g., Ritalin [102, 103].

3′ UTR VNTR: association and functionality

The human DAT1 gene consists of 15 exons and spans 60 kb on chromosome 5p15.3 [104] (Fig. 6). The human DAT1 protein is 620 amino acids in length and has 12 membrane spanning domains and shares over 90% homology with rodent DAT [103, 105]. Although there are SNPs in the coding sequence of DAT1 they do not affect the protein structure of the transporter [106]. Like 5-HTT, the DAT1 is a member of the Na/Cl-dependant NET family of transporters [107] and like 5-HTT, DAT1 also has a number of VNTRs in non-coding sequences. The first of these to be identified and by far the most extensively studied is located in the 3′ untranslated region (3′UTR) of the DAT1 gene [104, 108]. This VNTR, termed the 3′UTR VNTR, comprises 3–11 repeats of a 40 bp unit, with the 9 and 10 repeats being the most common forms. Individuals can be heterozygous or homozygous for the VNTR. As with all VNTRs the distribution of these genotypes varies in the population with the 10 allele appearing more frequently [109]. However, again as with the polymorphisms described earlier in this review differences in allelic frequency are observed between global populations or different ethnic groups [109].

Fig. 6
figure 6

Sequence and repeat alignment of the DAT 3′UTR VNTR. Schematic representation of the human DAT1 gene and position of the 3′UTR VNTR (top). The common 3′UTR repeats exist as 9 and 10 copies of a 40 bp repeat. The repeats are not perfect repeating units. Bold type lettering in the sequence represents degeneration between repeats and variation is denoted by defining each repeat with a Greek symbol. The nine repeat allele VNTR lacks the γ repeat seen in the 10 repeat allele, and in this example a SNP can be seen in the ε repeat of the 10 allele which produces the ι repeat in the nine allele

Full size image

Association studies have linked the 3′UTR VNTR not only with a number of disorders or affective behaviours but also with differences in normal physiological functioning between individuals e.g., focused neuronal activity [110]. Bertolino et al., demonstrated that individuals with the homozygous 10 allele (10/10) perform with a more focused response to memory related tasks when compared to heterozygous individuals (9/10) [110]. One of the most extensively studied associations for the 3′UTR VNTR is with ADHD. Similarly to the association of 5-HTT polymorphisms with depression, the data is conflicting. For example, a recent study has demonstrated an association of the 9/10 genotype with more severe symptoms of ADHD when compared to the 10/10 genotype [111], whilst others pinpoint association with susceptibility to ADHD with the presence of a 10 repeat allele in family transmission studies [112–117]. In contrast, some studies found no association between DAT1 polymorphisms and ADHD using either family based [118, 119] or population based studies [120, 121]. However, a role for the DAT1 transporter in ADHD is strongly supported by the facts that DAT1 knockout mice are extremely hyperactive [122] and that current therapies for ADHD include drugs that act on the transporter e.g., methylphenidate (Ritalin). Strong conclusive evidence that polymorphisms in DAT1 are associated with ADHD has not yet been found and once again conflicting evidence may have arisen due to differences in the types of association being made e.g., familial versus population studies, ethnic variation, different test parameters, etc. However, it is possible that when acting in concert with other polymorphisms, via an epistatic interaction, the 3′UTR VNTR of DAT1 may display stronger associations with disorders. For example, Carrasco et al., have very recently demonstrated, in a Chilean population, individuals that are heterozygous for a seven repeat VNTR in the dopamine 4 receptor (DRD4) and homozygous for the DAT1 3′UTR VNTR 10 allele polymorphism are likely to suffer from ADHD, but in the presence of only one of these specific polymorphisms the same association can not be made [123].

In addition the DAT1 gene contains several other VNTRs that meet our criteria for acting as potential regulatory domains (personal observations and preliminary data not shown). They could therefore, act synergistically to modulate transporter expression and studies should consider multiple polymorphisms which are both intra- and intergeneic if they target the same neural transmission pathway.

The 3′UTR VNTR has also been linked with a number of other disorders or addictive behaviours. The presence of a nine allele has been linked with a reduced risk for addiction to smoking [124–126] which may be due to decreased DA release in individuals with this genotype [127]. The 3′UTR has also been associated with susceptibility to Parkinson’s disease although the evidence is highly variable. The rare 11 allele has been associated with Parkinson’s in a limited number of studies using different populations [128–130] but not in others [131]. A further study examined the nine allele in Parkinson’s and found an age-related association: those over the age of 60 with a 9-repeat show greater incidence than those younger than 60 with a 9-repeat [132]. In contrast, a further study found no association between DAT1 3UTR polymorphisms and Parkinson’s disease in a Chinese population [133].

Most studies disprove an association of the 3′UTR VNTR and alcoholism [134, 135] although genotype may play a part in severity of withdrawal symptoms [135]. However, one study implies that an A/G SNP in the 10 allele may be associated with vulnerability to alcoholism [136]. To a lesser extent other associations with this VNTR have been examined but most find no correlation, these include; body weight, body mass index or obesity [137], Tourette’s [138], personality traits (novelty seeking, harm avoidance, reward dependence and persistence [139]), cocaine dependence [140]; and schizophrenia [141].

A number of functional studies on the common DAT1 3′UTR VNTR have been conducted in vitro and in vivo with varying results. Transient transfections of recombinant reporter and marker gene plasmids containing a VNTR show conflicting results. However, when we examined the isolated nine allele 3′UTR VNTR linked to a green fluorescent protein marker gene, it could act as an enhancer of transcription in dopaminergic neurons within a murine midbrain slice and in a murine dopaminergic cell line, SN4741 [142]. Furthermore, higher levels of reporter gene expression have been demonstrated for the 3′UTR nine allele in human neuroblastoma SK-N-SH cells [143] and in HEK293 cells when transfected as part of a larger 3′UTR fragment of DNA [144]. In contrast, in a non-human primate cell line, cos-7, a fragment of the 3′UTR containing the 10 repeat was demonstrated to support higher transcription levels than that of the 9 [145]. Furthermore, another study found no difference in transcriptional activity supported by either the 9 or the 10 allele when plasmids were transiently transfected into the human neuroblastoma cell line, SHSY-5Y, or HEK293 cells [146]. Nevertheless, QPCR on cerebellum neurons and lymphocytes demonstrates that the 10 allele is associated with higher levels of endogenous DAT mRNA [147]. In addition, in HEK293 cells with targeted stable integration of recombinant DNA containing the DAT1 coding sequence with different polymorphisms, the 10 allele has been associated with increased DAT1 protein density [148]. In our opinion, the contrasting reporter gene activity demonstrated for this and other VNTRs when tested is that these domains demonstrate tissue specificity as exemplified by our transgenic data for the 5HTT intron 2 VNTRs [13] and that variation in tissue culture reflects this property.

In vivo SPECT studies have demonstrated that having a 9/10 genotype results in lower levels of DAT1 expression in the striatal putamen when compared to individuals homozygous for the 10 allele [149]. In complete contrast, two other studies report higher levels of striatal DAT1 availability in individuals with at least one 9 allele [150, 151]. Three more in vivo studies found no association between 3′UTR polymorphism genotype and DAT1 density [152] or protein availability [153] and function [154].

Intron 8 VNTR: association and functionality

We have identified a 5–6 copy 30 bp VNTR located in intron 8 (Int8) of the human DAT1 gene and were able to demonstrate that it had transcriptional properties [16] (Fig. 7). This VNTR is termed either DAT1Int8.2 (five copies) or DAT1Int8.3 (six copies). As this VNTR has only recently been characterised much less work has focused on association studies with the DATInt8 VNTRs than for instance with that of the 3′UTR. We have described an association between the DATInt8.3 VNTR and predisposition towards cocaine addiction in a Brazilian population [16]. Recently, Brookes et al., have identified an association between the DATInt8 VNTRs and predisposition to ADHD in both English and Taiwanese populations [155]. O’Gara et al., have very recently demonstrated an association between the DATInt8 polymorphism and the ability to quit smoking, but only in the early stages of the attempt [156].

Fig. 7
figure 7

Sequence and repeat alignment of the DAT Int8 VNTR. Schematic representation of the human DAT1 gene and position of the Int8 VNTR (top). The common repeats exist as five (termed two allele) and six (termed three allele) copies of a 25–30 bp repeat. The repeats are not perfect repeating units. Bold type lettering in the sequence represents degeneration between repeats and variation is denoted by defining each repeat with a Greek symbol. The two repeat allele lacks the χ repeat seen in the three repeat allele

Full size image

Transient transfection of the dopaminergic murine cell line, SN4741, with renillin luciferase reporter plasmids containing an isolated VNTR cloned into an intronic position, demonstrated the ability of the Int8.2 VNTR to support higher basal levels of gene expression than the Int8.3 VNTR [16]. In addition, these Int8 VNTR demonstrated differential responses to stimuli. Namely, the Int8.3 allele displayed more pronounced sensitivity to stimuli such as cocaine than the Int8.2 allele [16].

DAT1 variants and therapeutics

DAT1 is a site of action for illicit drugs but also for therapeutic drugs. The DAT1Int8 VNTRs have been associated with differential response and susceptibility to illicit drugs such as cocaine and nicotine as outlined above. In addition there have been a number of reports indicating an association between the response of individuals to treatment for ADHD (namely methylphenidate), and genotype at the DAT1 3′UTR VNTR. Methylphenidate is a DAT1 re-uptake inhibitor that acts to prolong the duration of action of DA at it’s receptors. In children with ADHD undergoing treatment with methylphenidate homozygosity for the 3′UTR 10 repeat was associated with poorer performance in tests and a different pattern of EEG response when compared to children with a copy of the nine repeat [157]. Furthermore, in a sample of African-American children undergoing treatment for ADHD, Winsburg and Comings correlated homozygosity of the 10 repeat allele with non-responsiveness to the drug [158]. In contrast, others found that poor response to methylphenidate was evident in those individuals homozygous for the nine repeat allele [159, 160]. Discrepancies between these studies are likely to arise from differences in test parameters; however they do demonstrate that the allelic variation at the 3′UTR can be associated with differential responses to treatment. These clinical data also support our theory that regulatory polymorphic domains must be considered together in assessing the response of a particular genotype. In future studies, attempting to correlate clinical data with genotype, we believe attempts must be made to determine which allele the polymorphisms are located on to fully elucidate their summative action in modulating gene expression in specific tissues or in response to a particular challenge.

Transcription regulation by polymorphisms based on primary sequence similarities

Our analysis of the VNTRs within the transporters, 5HTT and DAT1, has focused on comparison of the functional characteristics of different copy number VNTRs [12–15]. However variation in the sequence of individual repeat elements might be as relevant in furthering our understanding of VNTR function. One or more base changes in the sequence at the binding site for a transcription factor, as found within the VNTR elements, could alter the affinity or specificity of a specific transcription factor binding to that site [14]. The synergistic or additive action of these individual elements in the whole VNTR domain may result in differential regulatory properties. Consistent with this we have analysed the transcriptional characteristics of individual repeat elements of the 5-HTT VNTR and demonstrated that they act as functionally distinct regulatory elements [12]. Whilst copy number variation can have effects at the level of gross human phenotype [5, 7, 161] and in in vitro and in vivo model systems, our data indicates that there is an additional layer of transcriptional complexity based on the primary sequence of the VNTR [14]. For example, as stated previously there is debate as to the correlation of VNTR copy number with predisposition to affective disorders, perhaps reanalysing this data taking into account the primary sequence of the VNTR could resolve the conflicting literature and resolve the debate as to the clinical significance of these VNTR domains.

We have demonstrated that the intron 2 VNTR domain of the 5-HTT gene is bound and regulated by the transcription factor Y box binding protein 1 (YB1). This factor was identified by a yeast one hybrid screen, followed by demonstration of it’s ability to transactivate in a 5-HTT STIn2 VNTR construct in a cell line model [15]. YB1 has been implicated as an important protein during development. Interestingly, a transcription factor termed CTCF can also regulate the function of YB1 directly [15]. We have demonstrated that YB1 will bind specifically to the 5-HTT STin2 VNTR elements and this binding can be antagonised by CTCF [15]. We have also recently demonstrated that CTCF can bind the 5-HTT STin2 VNTRs [22]. This action of CTCF is of interest because this protein in addition to direct transcriptional properties, has epigenetic functions such as controlling imprinting and has recently been associated with diseases ranging from Alzheimer’s to cancer [162–165].

In general, transcription of a gene will therefore be determined by the combinatorial action of multiple positive and negative promoter domains that specify the tissue-specific and stimulus-inducible expression of a gene [166]. We postulate that VNTRs could have a direct effect on the expression of a gene in different cell types and in response to various physiological or pharmacological stimuli. This may correlate with the aetiology or progression of a disorder. We therefore predict that studies to better define the function of the VNTRs to modulate gene expression will not only be of importance for normal and abnormal behaviour but be of general importance for understanding gene expression, as related VNTRs are present in many other genes and are often proposed to correlate with susceptibility to various disorders.

Summary

Contradictory reports in the literature linking or associating specific polymorphisms or genes with specific disorders may arise from inter-experimental variation, mixed populations, and inappropriate comparisons between subjects. As most association studies demonstrate a lack of power the conclusion of most studies and of recent meta-analysis is that larger sample sizes and samples that are controlled for the variations above are needed to provide more accurate associations. In addition, variation in function as measured by ability to modulate gene expression (endogenous or reporter gene) can also arise due to differences between experimental paradigms e.g., cell type and growth conditions used, length of “VNTR”; some groups include flanking sequences which may have binding sites for additional transacting factors. The use of different transfection systems, a multitude of different cell types and protocol differences such as time spent in culture may all affect reporter gene expression in transient analysis. One group even reports variation depending on the minimal promoter present in the reporter plasmid used for cloning, possibly due to interaction with a SNP in the repeat under investigation [144]. An interesting point raised by Brookes et al., concerning the contradictory reports surrounding the DAT1 3′UTR VNTR function in ADHD, is that this may be due to the 3′UTR VNTR acting as a marker for an associated polymorphism rather than being involved directly in the aetiology [167]. This could also be true for other VNTR that appear to be weakly associated with a disorder but do not appear to display differential function, or show conflicting associations. What has until recently been overlooked, is the possibility that functional polymorphisms on the same gene (or haplotypes) or even on different genes (epistatic interaction) may increase (or decrease) the propensity towards the development of specific disorders or characteristics. In this case having two or three polymorphisms or haplotypes that might predispose to a specific condition is more of a risk factor for the development of that condition than having a single polymorphism. However, one must take into account that some polymorphisms show a stronger correlation with a disorder than others. A number of haplotype association studies have already been undertaken [64, 115, 132, 155, 167]. In addition, Hranilovic et al., recently examined combinatorial effects of the two VNTR 5-HTT polymorphisms, we have discussed, on mRNA expression by quantitative PCR on lymphoblast cells from schizophrenic patients [168]. They determined that the polymorphisms could be banded, based on expression, into high expressing (l/l and 12/12) and low expressing (l/s, s/s and 10/10 10/12). They found a dominant effect of the low expressing alleles on the high expressing alleles resulting in significantly decreased transporter expression [168]. We are currently investigating the ability of the 5-HTT and DAT1 polymorphisms reviewed here to act in a combinatorial or synergistic manner in regulating transcriptional control of gene expression in stably expressing cell lines, and in response to stimuli such as lithium chloride or cocaine hydrochloride, in both cell lines and primary neuronal cells. The integration of variation in expression of a number of genes with related VNTRs (containing the consensus sequence for binding specific transcription factors) is consistent with recent attempts to take a more global analysis of polymorphic variation and correlation with disease reflected in whole genome analysis [169].