Introduction

Attention-deficit/hyperactivity disorder (ADHD) is characterized by impairing symptoms of inattention and/or hyperactivity and impulsivity [1]. This disorder has a heritability of approximately 76% [2], affecting about 5% of school-age children. It is a precursor of behavioral problems in adolescence and adulthood [3]. Among the dopamine-related genes investigated to date, the dopamine receptor D4 (DRD4) gene has been widely studied and it is the candidate gene with the strongest association with ADHD in different meta-analysis studies [4]. The most investigated polymorphism is a 48pb VNTR in its third exon. The alleles vary from two to eleven fold repeat (2R-11R), and the most prevalent are the 4R (65.1%), 7R(19.2%) and 2R(8.8%) [5]. Despite the evidences that 7R allele might be a susceptibility allele for ADHD, different groups have reported mixed results [4].

The VNTR region encodes a portion of the third cytoplasmic loop of Dopamine D4 Receptor which is considered an intrinsically disordered region of proteins. These regions lack a stable 3-D unique conformation and provide large interaction surface, conformational flexibility and exposure of short linear motifs, mediating the interaction with several partners [6]. Some properties described as important for signal transduction and proper protein functionality are usually associated to disordered regions, such as degradation (PEST) motifs, post-translational modifications and linear interaction motifs [7-9]. Dopamine D4 Receptor presents several SH3 motifs recognized in its third cytoplasmic loop [10]. Although their biological function within DRD4-VNTR has not been fully elucidated, this kind of domain-binding motif is widely described as important in cell–cell communication and signal transduction [11].

In addition to repeat number variation, rare variants inside the VNTR region were also described [12-16]. The combination of specific mutations results in different motifs (repeats) that are identified by numbers [12]. The haplotypes are composed by a specific combination of repeats (see 12-15 for more details, and Figure S1). 2R (1-4), 4R (1-2-3-4) and 7R (1-2-6-5-2-5-4) are the most frequent haplotypes on worldwide populations [12-15]. The other haplotypes are considered rare. A growing number of evidence suggests that rare haplotypes are an important risk factor to develop ADHD, but the literature addressing this topic is extremely scarce. Higher frequency of rare variants in ADHD probands compared to both control individuals [13] and children with autism [14] was reported. Recently, it has been documented that ADHD Brazilian subjects from a clinical sample presented an excess of rare variants as well as non-synonymous mutations compared to controls. It is noteworthy that these differences were restricted to 7R allele [16].

General-population studies have important advantages over case–control studies [17,18]. When the main objective is to explore rare variants variability and their contribution to complex diseases, this design represents a unique opportunity due to the large sample size. Another important aspect is that clinical cases do not represent the full spectrum of cases found in the general population for the majority of disease, especially for mental disorders [19]. Considering that, genetic findings on ADHD in clinical samples might not be representative of the general populations since ADHD represents only the most severe part of the phenotype and it might be determined by different genetic mechanisms than other portions of the inattentive and hyperactive/impulsive trait in the population [20]. Thus, it is of interest to assess the role of the DRD4 allele length variants as well as rare variants in ADHD etiology in the general population.

The aims of the current study are to investigate the association of DRD4 VNTR length and rare variants with an epidemiological measure of inattention and hyperactivity/impulsivity symptoms in a Brazilian birth cohort study. We also explore the impact of non-synonymous rare mutations in the ADHD etiology.

Results

Ten different DRD4 VNTR alleles with 2 to 11 tandem repeats were present in this study. The 4/4R was the most observed genotype in both hyperactivity/inattention low-score and high-score groups, followed by 4/7R genotype. The genotypic frequencies observed in the total population as well as in the low-score and high-score groups are shown in Table S1. To evaluate the effect of 7R allele in this sample, DRD4 VNTR genotypes were grouped into two categories: with and without 7-repeat allele. A logistic regression adjusted for the effect of gender and ethnicity was performed. Genotypes containing the 7R allele were not associated with high hyperactivity/inattention scores (Table 1).When an allelic model was tested, again no association was observed with high hyperactivity/inattention scores (Table S2).

As rare variants inside the VNTR are hypothesized as being a possible confounder for usual DRD4 VNTR and ADHD association studies, all 2R(n=31) and 7R(n=175) and many randomly selected 4R(n=521) homozygous had the VNTR region sequenced, comprising a sample size of 727 individuals.

Considering the 1,454 sequenced alleles, we covered 30 different haplotypes: two at 2R allele, 15 at 4R and 13 at 7R alleles (Table 2, Figure S1). As expected, the most frequent haplotypes found in our general population sample were 2R (1-4), 4R (1-2-3-4) and 7R (1-2-6-5-2-5-4). A number of new coding SNPs covered 10 novel motifs (repeats from 39 to 48; Figure 1). Six novel haplotypes were found in 4R allele category and six in 7R. Among new motifs, we identified many unreported deletions. Two of them are1bp deletions at 7R allele. The first is a modification of a repeat previously identified as number 6 [12] where a deletion of a cytosine is observed at the 21^st base of the VNTR motif. This repeat was called 42. The other is a modification of motif number 1 where a deletion of a cytosine located at position 20 of the motif is found (repeat 43). The alignment of deleted motifs as well as their chromatograms can be seen on Figure 1, Figure 2 and Figure S1. These deletions lead to a frameshift with the occurrence of a premature stop codon at the position 393(M393X) instead of the position 468 in the protein. The other alteration comprises a 26bp deletion at the 4R allele (Figure 2). This deletion leads to a frameshift in which the stop codon UGA at position 420 is substituted by the codon for arginine. Translation of the protein thus continues until a stop codon encountered at codon 440, leading to a 20-residue-long C-terminal tail. In both cases, the two last transmembrane domains of D4 receptor would be drastically altered by incorporation of several disordered instead of hydrophobic residues, as shown by transmembrane topology prediction (Figure 2).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Nucleotide and amino acid sequences of the novel VNTR motifs.

The nucleotide and corresponding amino acid (bold) sequences of 10 novel DRD4 exon3 48-bp repeat motifs are shown (repeats 39-48). * indicates 1bp deletion motifs.

https://doi.org/10.1371/journal.pone.0085164.g001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Characterization of DRD4 novel deletions.

(A) PCR analysis of 3.5 allele. Lane 1: DNA 50bp marker, Lane 2: homozygote for 2 repeats; Lane 3 homozygote for 3 repeats; Lane 4: homozygote for 4 repeats; Lane 5: heterozygote for 3 and 4 repeats; Lane 6: heterozygote for 3.5 and 4 repeats. PCR products were run on a 3.5% agarose gel. A heteroduplex band is seen for heterozygotes in lanes 5 and 6. (B) The allele 3.5R alignment. The 3.5R refers to an allele with a length between 3 and 4 repeats. Sequencing of this size allele revealed a deletion of 26bp sequence (11bp immediate 5’ downstream sequence of the VNTR and 15bp inside the first repeat of VNTR) of the most common 4R haplotype. (C) Sequence chromatogram of DRD4-7R nucleotide deletions(arrow) at VNTR positions117 (del117C, above) and 20 (del20C, below). (D) Topology prediction of proteins encoded by VNTR with (below) and without (above) deletions. Schematic diagram of the transmembrane region (yellow). The deletions exclude the prediction of the two last transmembrane domains.

https://doi.org/10.1371/journal.pone.0085164.g002

To evaluate the impact of rare variants in hyperactivity/inattentions scores, two models of multivariate logistic regressions were performed based on different strategies of pooling rare variants. The frequency of rare variants in case and control group is described in Table 2.

As 7R rare variants are suggested to be associated to ADHD, rare variants were classified in specific 4R and 7R and included in the regression analysis as independent variables (first logistic regression model). This analysis revealed that individuals carrying at least one 7R rare haplotype were at a significantly increased risk for developing high hyperactivity/inattention scores compared with those homozygous for most common haplotypes (OR=2.561; P=0.024). A trend for association was observed for 4R rare variants (Table 1).

As mutations that determine amino acid changes might be deleterious and potentially disease causing, we investigated the impact of non-synonymous substitution of rare variants in hyperactivity/inattention symptoms. For this analysis individuals were classified as carrying or not 4R or 7R variants that have at least one non-synonymous mutation. 4R rare haplotypes containing missense variants presented a trend in ADHD risk. On the other hand, carriers of 7R rare haplotypes containing missense variants had a greater than three-fold increase in the odds of developing higher hyperactivity/inattention scores as compared to those homozygous for the common haplotype (OR=3.216; P=0.008; Table 1).

Many functional properties related to protein disorder were assessed in each haplotype. The results of potential alterations due to mutations are shown in table 3 and Figure S2. As expected, this region was predicted as to be highly disordered. Although the estimated mean disorder indexes are high and very similar for all haplotypes, many differences can be observed if the mutations are individually analyzed (Figure S2). Many of them confer disorder to order substitution on protein, such as R>C, G>C, G>V and P>Q. When the phosphorylation of serines, threonines and tyrosines were considered, the alterations were more obvious. Many mutations created potential phosphorylation sites in in two 7R (1-2-6-5-37-5-4) (Serine 74) and (40-2-6-5-2-5-4) (Serine 9) haplotypes. Regarding the potential sites for interaction and/or degradation PEST motifs were predicted in the 4R(1-17-3-4), 7R(1-8-25-5-2-5-4) and 7R(1-2-6-5-37-5-4) haplotypes. The most frequent haplotypes showed no potential sites for phosphorylation or potential PEST motifs. Moreover, many SH3 binding sites were described as disrupted in the following haplotypes: 4R(1-17-3-4) (aa 22-28), 7R(1-2-3-17-2-5-4) (aa 61-67) and 7R(40-2-6-5-2-5-4) (aa 3-9, 6-12).

Discussion

We reported herein the unusual architecture of DRD4 gene as well as an important association between high hyperactivity-inattention scores and rare variants, especially in 7R allele. Our analysis suggests that 7R rare variants might have a contribution to the multifactorial inheritance of ADHD while 7R length allele might not.

Our results are in agreement with previous findings [12-16]. We observed a high variability inside DRD4 VNTR in the general population as reported elsewhere [12-15]. Also, a higher variability in the DRD4-VNTR region was observed in ADHD subjects when compared to controls [13] and to autistic subjects [14]. It is important to emphasize that all published results addressing this subject to date showed a consistent association between DRD4 rare variants, especially inside 7R allele, and ADHD [13,14,16].

An increasing body of evidence suggests that rare variants may play a role in complex diseases, including neuropsychiatric traits. To date the Genome Wide Association (GWA) studies conducted on ADHD samples were not able to find any common variants significantly associated with the disorder. Although the small sample size involving these studies is pointed as a possible limitation to find a significant association at genomic level, “synthetic” signals of association on GWA in other case-control studies are reported [21]. Then, the common-disease-rare-variant (CDRV) hypothesis is an attractive framework to understand the susceptibility of the disorder. This hypothesis proposes that the sum of the effects of a series of low frequency variants, each conferring a moderate but readily detectable increase in relative risk, compose an important proportion of the overall genetic risk for common diseases etiology [22]. Recent findings have drawn attention to the involvement of rare structural variants in the pathophysiology of ADHD [23]. Point mutations inside DRD4 appear to be another important example of this phenomenon.

Our analysis suggests that rare variants may make a contribution to the multifactorial inheritance of common diseases. Thus, even though individual rare variants may not contribute much to the overall inherited tendency of a disease, their discovery is likely to be important for understanding disease etiology. Another important aspect of this study is its focus on the identification of multiple rare variations, since individuals with ADHD might frequently possess any of them. This suggests that all the variations contributing to the pool of variations that were greater in frequency among the individuals with the disease phenotype might perturb D4 receptor activity.

Many studies already addressed the plausibility of studying the impact of mutations in disordered regions [24,25]. The main hypothesis is that altering the disorder status in proteins could affect many important biological processes [25]. In this study we observed that many mutations are able to perturb some disorder-related properties. Differences in disorder profile, PEST motifs recognition, additional phosphorylation sites and disruption of binding sites could affect D4 receptor functionality. Among the possible impaired functions there are reduction of loop flexibility, facilitated recognition for internalization and degradation, unusual desensitization, homo-heterodimerization disparities, and disruption of usual interactions. Further functional studies as well as studies evaluating DRD4 variability in other psychiatric disorders are needed to better understand the role of these mutations in disease etiology.

This is the first study showing potential null alleles due to deletions inside the VNTR. In humans, only one homozygous for the null allele was reported [26]. Although the subject did not present ADHD or other psychiatric disorder, many symptoms possibly related to the absence of DRD4 protein were observed, such as somatic ailments, obesity and disturbances of the autonomic nervous system [26]. Out of four subjects containing deletions presented here, only one had extreme phenotypes. However, all subjects are heterozygotes. Further careful phenotype assessment of subjects carrying null alleles is needed to better clarify the role of the DRD4 gene in normal and pathologic development.

Our results must be interpreted in the context of some limitations. Although the hyperactivity/Inattention subscale of the SDQ shows high diagnostic performance for ADHD, the use of this epidemiological instrument might have created biases in our phenotypic definition. However, it is important to note that our analyses were based on individuals who were classified in the same group – low or high score groups – in both assessments. So, we are confident that we selected those individuals in the population with deviant and stable inattentive/hyperactive profile. Second, no genomic control was performed. Therefore our findings might have been biased by hidden genetic heterogeneity present in our specific sample of the southern Brazilian population. However, our case and control groups contain a similar ethnic distribution and the analysis was adjusted for ethnicity. Moreover, it is important to note that recent studies using different sets of informative markers did not show heterogeneity across main Brazilian populations [27] or significant population structure in southern Brazilians [28]. Finally, since the 10 novel motifs found could be artifacts, we re-sequenced all samples in which they were observed through different PCR products to confirm mutations.

In conclusion, DRD4 7R rare variants are associated with hyperactivity and inattention at the general population level confirming previous studies on ADHD probands. These results indicate that the DRD4 7R rare variants could play an important role in ADHD genetic susceptibility. In silico analysis of putative effect of these mutations on receptor function suggests that they could have an impact on receptor signaling, degradation, internalization and desensitization. In addition, we document that taking in consideration the sequence of DRD4 VNTR is very important to estimate the contribution of this dopaminergic receptor to the pathophysiology of ADHD.

Material and Methods

Supporting Information

Acknowledgments

The authors thank Vanessa Rodrigues Paixão-Côrtes for her help with figures and to Fady Dawoud for the English revision.