Association of Impulsivity and Polymorphic MicroRNA-641 Target Sites in the SNAP-25 Gene

Impulsivity is a personality trait of high impact and is connected with several types of maladaptive behavior and psychiatric diseases, such as attention deficit hyperactivity disorder, alcohol and drug abuse, as well as pathological gambling and mood disorders. Polymorphic variants of the SNAP-25 gene emerged as putative genetic components of impulsivity, as SNAP-25 protein plays an important role in the central nervous system, and its SNPs are associated with several psychiatric disorders. In this study we aimed to investigate if polymorphisms in the regulatory regions of the SNAP-25 gene are in association with normal variability of impulsivity. Genotypes and haplotypes of two polymorphisms in the promoter (rs6077690 and rs6039769) and two SNPs in the 3′ UTR (rs3746544 and rs1051312) of the SNAP-25 gene were determined in a healthy Hungarian population (N = 901) using PCR–RFLP or real-time PCR in combination with sequence specific probes. Significant association was found between the T–T 3′ UTR haplotype and impulsivity, whereas no association could be detected with genotypes or haplotypes of the promoter loci. According to sequence alignment, the polymorphisms in the 3′ UTR of the gene alter the binding site of microRNA-641, which was analyzed by luciferase reporter system. It was observed that haplotypes altering one or two nucleotides in the binding site of the seed region of microRNA-641 significantly increased the amount of generated protein in vitro. These findings support the role of polymorphic SNAP-25 variants both at psychogenetic and molecular biological levels.


Introduction
Impulsivity is a multidimensional personality trait characterized by action without planning and lack of consciousness to guide acts and behavior [1]. The phenotype is based on a complex neurochemical background, and genetic factors play a crucial role in its development, twin studies suggested that inheritance plays approximately 45% role in its determination [2]. There is a relatively broad spectrum of individual differences in healthy subjects regarding their impulsivity, therefore to summarize this trait in one specific definition has been a challenge for many decades. There is an agreement however that major deflection of impulsivity is one important component of Attention Deficit Hyperactivity Disorder (ADHD) together with impaired attention and/or hyperactivity [3].
Recent studies have focused mainly on the serotonergic [4] and dopaminergic [5] system investigating the genetic background of impulsivity, and our group has also previously shown the importance of the interaction between these two systems [6]. Further studies showed an association between impulsivity and a functional dinucleotide repeat polymorphism in the promoter region of NOS1 gene (exon 1f-VNTR) [7,8], and the rs11624704 of the neurexin gene (NRXN) was also suggested to contribute to the background of impulsivity [9]. Impulsivity is one of the core features characterizing ADHD, thus, candidate genes shown to associate with ADHD might emerge as putative genetic components of impulsivity. A number of association studies as well as meta-analyses suggested the role of DRD4, DRD5, HTR1B and SLC6A3 genes, reviewed by Faraone [10]. Moreover, several studies pointed out an association between two SNPs (rs3746544 and rs1051312) in the 39 UTR of the SNAP-25 gene and ADHD [11,12,13], and the role of the former polymorphism was also confirmed in a comprehensive metaanalysis [14]. Genetic variants of SNAP-25 including promoter polymorphisms [10,12,15], have already been investigated as putative risk factors of other psychiatric disorders as well [16,17], but it has not yet been studied in the background of impulsivity in a non-clinical sample.
SNAP-25 (synaptosomal-associated protein, 25 kDa) plays a crucial role in the central nervous system, being one essential component of the SNARE (soluble N-ethylmaleimide-sensitive factor attachment protein receptors) complex and contributes to exocytosis by targeting and fusion of vesicles to the cell membrane [18]. There are two SNPs (rs3746544 and rs1051312) in the 39 UTR of the SNAP-25 gene with special interest, we previously demonstrated by in silico sequence analysis that they are localized in the putative target site of miR-641 [19].
As microRNAs are known to play a role as translational regulators of protein synthesis, SNPs located either in the coding sequences of microRNAs or in their binding sites might cause an imbalance of this regulation. SNPs have been described in genomic regions encoding miRNAs [20], the rs1625579 polymorphism is located in the coding region of miR-137, which has been shown to target the ZNF804A, a candidate gene of schizophrenia [21]. However, more attention have been focused on the significantly higher number of SNPs that are present in the 39 UTR of target genes altering the binding region of a miRNA to its specific target mRNA [20]. Polymorphisms in the microRNAbinding sites have been suggested to contribute to the genetic background of various diseases, such as coronary heart disease [22], different types of cancer [23,24] or ADHD and other comorbid psychiatric illnesses [12,25]. A polymorphic variant (rs13212041) in the HTR1B gene 39 UTR was demonstrated to influence miR-96 binding and to be associated with conductdisorder phenotype in a sample of 359 students [26].
Here we present data on the functional effect of rs3746544 and rs1051312 SNPs and their haplotypes on miR-641 regulated reporter gene expression in cell culture. We also aimed to carry out an association study to analyze whether either these loci or the polymorphic variants of the SNAP-25 promoter (rs6077690 AT and rs6039769 AC), studied so far in psychiatric disorders, contribute to the genetic background of impulsivity in a nonclinical sample.

Methods
Ethics Statement, Participants 901 healthy Hungarian young adults participated in the study. Selection criteria included no past or present psychiatric history (based on self-report). Before providing buccal samples for genetic analysis, participants signed written informed consent. The study protocol was approved by the Scientific and Research Ethics Committee of the Medical Research Council (ETT TUKEB). The mean age of the investigated population was 21.3 (63.3) years, 45.1% were males and 54.9% were females.

Phenotype analysis
Hungarian version of the 11 th revised Barratt Impulsiveness Scale [27] was used to measure impulsivity. The questionnaire was originally published by Barratt. Translation into Hungarian was carried out by a ''forward-backward'' procedure [6], Cronbach alpha value for the total score of the scale was 0.808. The Barratt Scale is a self-reporting measure widely used both by clinicians and in research settings. It consists of 30 items regarding acting and thinking in different situations. These statements are asked to be rated by the participant on a fourpoint-scale, reaching from ''occasionally'' to ''always''. The highest theoretically possible total score is 120 and impulsivity is considered ''normal'' in the range from 52 to 71.

DNA isolation
DNA purification was initiated by incubating the buccal samples at 56uC overnight in 0.2 mg/ml Proteinase K cell lysis buffer. It was followed by protein denaturation using saturated NaCl solution. Finally, DNA was precipitated using isopropanol and ethanol by standard procedures and DNA pellet was resuspended in 100 ml 0.56 TE (16 TE: 10 mM Tris pH = 8, 1 mM EDTA) buffer. Concentration of each DNA-sample was measured by Varioscan Flash spectral scanning multimode reader.

Genotyping of SNAP-25 SNPs
The promoter SNP rs6039769 was genotyped by the C__29497348_10 (Life Technologies) pre-designed primer-and TaqMan probe-set. A 7300 Real-Time PCR System (Life Technologies) was employed to detect the FAM and VIC signals corresponding to the C and A alleles, respectively.
Genotypes of rs6077690 promoter SNP were determined by PCR-RFLP. Flanking region of the polymorphic site was amplified using the 59 ATG TCA GTG TGG GGC ATC 39 sense and 59 AGG CAT GTT GCT GAA ATT TGT T 39 antisense primers. The Qiagen HotStarTaq DNA-polymerase system was applied for PCR amplification, the reactions were carried out in a total volume of 10 mL containing 1 mM sense and antisense primers, 0.2 mM of each deoxyribonucleotide-triphosphate, 0.25 U Hot-StarTaq DNA-polymerase together with 16 buffer and 16 Qsolution and approximately 4 ng genomic DNA sample. Thermocycle was initiated by 15 min at 95uC initial denaturation and polymerase activation. It was followed by 40 cycles of 1 min denaturation at 94uC, 30 sec annealing at 63uC and 1 min extension at 72uC. The last step of the PCR was a final extension at 72uC for 10 minutes, amplicons were then kept at 8uC for downstream processing. In the next step PCR-products were digested by TspI 509 restriction endonuclease. Two non-specific recognition sites were incorporated in the amplicons to verify optimal conditions of digestion. Reactions were carried out accoriding to manufacturer's instructions. A 301-and a 110-bplong product (together with the two control fragments) were generated in the presence of the A allele, whereas the 411-bp long fragment could be seen in case of the T allele. Digestion pattern was analyzed by traditional submarine gel electrophoresis. Call rate of genotyping was 97%.

Direct haplotyping of SNAP-25 39 UTR SNPs
As the two SNPs (rs3746544 and rs1051312) in the 39 UTR are separated by only 3 basepairs from each other, it was possible to identify haplotypes directly in each individual sample using the published ''double-tube'' method [19]. In summary, the method is based on the application of haplotype-specific TaqMan probes in a real-time polymerase chain reaction. Two simultaneous analyses contained the four different probes corresponding to the four haplotypes labeled by FAM and VIC, respectively. Amplification and data-collection were done by a 7300 Real-Time PCR System (Life Technologies). Call rate was 98%. Individual genotypes of rs3746544 and rs1051312 were deduced from haplotype data.

Haplotype analysis of SNAP-25 promoter SNPs
SNPs in the promoter were about 1.5 kb apart from each other which provided rather limited possibilities for simultaneous analysis and thus direct haplotype determination. Therefore individual haplotypes for these SNPs were calculated from genotype data. Haplotype was ambiguous only in case of double heterozygote individuals (rs6077690 AT and rs6039769 AC; i.e. either A-A/T-C or A-C/T-A). Linkage disequilibrium analysis, however, revealed that the frequency of the T-A haplotype was as low as 0.6%. As a total of 103 double heterozygotes were identified in our sample, approximately 1 participant was mathematically expected to possess the rare T-A haplotype (i.e. A-C/T-A), which was neglected. Using this assumption, the haplotypes of double heterozygotes were also able to be identified from the genotypes, as these haplotypes were considered to be A-A/T-C. To confirm this approach, haplotypes were also reconstructed by Phase 2.1, which provided the same result for each sample [28].

Plasmid constructs
The entire 39 UTR region of the human SNAP-25 gene was cloned behind the firefly luciferase gene at the multicloning site of the pMIR-REPORT Luciferase miRNA Expression Reporter Vector (Life Technologies), using the 59 TGT AAT GAG CTC CTG GGA AGT GGT TAA GTG T 39 sense and antisense 59 CCC GAC AAG CTT AAA CTA GCT ACA AAA TGT CAA TCA 39 primers. Bold italic letters show the recognition sites of Sac I and Hind III restriction endonucleases in the sense and antisense primers, respectively. Genomic DNA possessing a TT haplotype was applied to amplify the SNAP-25 39 UTR, constructs with the GC, GT and TC haplotypes were subcloned by QuickChange Lightning Site-Directed Mutagenesis kit (Agilent Technologies). All four constructs were verified by direct sequencing. Another pMIR construct was used as internal control, which contained an insert with same length but different sequence and most importantly missing the binding site of the analyzed miR-641.

Cell culture and transient transfections
Human embryonic kidney (HEK) 293 cells (purchased from Sigma-Aldrich Ltd. Budapest, Hungary) were cultured in 24-well tissue culture plates in DMEM supplemented with 10% fetal bovine serum and 1% penicillin solution at 37uC in a humidified atmosphere containing 5% CO 2 . After optimization experiments 0.05 mg of the pMIR reporter constructs, 5 pmol miR641 and 0.2 mg b-galactosidase constructs were co-transfected in a reaction mixture containing 2.5 ml Lipofectamin and 60 ml OptiMem.
Cells were incubated at 37uC after transfection.

Luciferase and b-galactosidase assay
Medium was removed 36 hours after transfection and cells were washed twice in phosphate buffered saline, and the cell extracts were suspended in 100 mL 250 mM Tris-HCl buffer.
Cell lysis was carried out by three consecutive freeze-thaw cycles, lysate was centrifuged (13000 rpm for 15 minutes at 4uC) and the supernatans were collected for luciferase and bgalactosidase activity measurements.
Luciferase activity was detected by adding 60 mL of luciferin reagent to 12 mL sample, b-galactosidase activity was measured by adding a reaction mixture containing 33 mL ONPG solution (ortho-nitrophenil-b-galactoside) to 20 mL sample, respectively. Fluorescence and luminescence values were measured using the Varioscan Flash spectral scanning multimode reader (Thermo-Scientific). Analyses were carried out in triplicates.

Statistical analysis
Raw data of luciferase enzyme activities were normalized to the b-galactosidase levels for each sample. Statistical analysis was performed with the Tukey-Kramer multiple comparisons test. P values lower than 0.05 were considered to be statistically significant. D9 and R 2 measures of linkage disequilibrium were assessed using HaploView v4.2.
Genetic association analyses were carried out using SPSS 19.0 for Windows. Chi-square analyses were applied to test if genotype frequency distributions corresponded to Hardy-Weinberg-equilibrium. Independent-Samples t-test was used to test genetic associations by one way analyses of covariance (ANCOVA) assuming a bi-allelic inheritance model and co-dominant inheritance.

Results
Genotype and haplotype frequencies of SNAP-25 promoter and 39 UTR region in healthy participants of European descent Allele-and genotype distribution together with haplotype frequencies of two promoter polymorphisms (rs6039769 and rs6077690) as well as those of the two SNPs in the 39 UTR (rs3746544 and rs1051312) of the SNAP25 gene were determined in a healthy Hungarian population (N = 901). Genotype distributions were in Hardy-Weinberg-equilibrium for each SNP, minor allele frequencies varied between 0.255 and 0.417 ( Table 1). As the two SNPs of the 39 UTR were in close proximity (see Table 1 for genomic localization), a direct haplotype determination was possible to perform for each individual DNA samples using haplotype-specific TaqMan probes, and genotypes were deduced from haplotype data. On the other hand, the two SNPs of the promoter region were genotyped individually as their greater distance did not allow a direct haplotype analysis, and haplotype frequencies were estimated by calculation. Analysis of the 39 UTR haplotypes confirmed our previous result [19], that the G-C haplotype did not occur at all in the population. Interestingly a similar situation was observed regarding the promoter haplotypes. Although the T-A haplotype was not absent completely, its frequency (0.006) was much lower than expected (0.161). This resulted in a special type of linkage disequilibrium characterized by a high Lewontin's D9 value, together with a relatively low R 2 . On the other hand absolutely no linkage disequilibrium could be observed between the promoter and the 39 UTR region (Figure 1).

Association analysis of SNAP-25 SNPs and impulsivity
As a first step, a single marker analysis was used by comparing the average impulsivity scores of participants in the different genotype categories (Table 2). No association could be detected with the promoter SNPs. On the other hand, one of the 39 UTR (rs1051312) polymorphism showed a nominal association with Barratt-scores of impulsivity (p = 0.042) which disappeared after Bonferroni correction for multiple test (4 SNPs: p,0.0125).
As a second step, we applied a haplotype-based allele-wise analysis for the promoter and the 39 UTR SNPs, separately. No association was found between the estimated promoter haplotypes and total impulsivity scores (data not shown). It is important to note that the 39 UTR haplotypes were not estimated but determined in each subject individually by molecular methods (see the ''Methods'' section). It was observed, that individuals possessing the T-T haplotype of the rs3746544 and rs1051312 SNPs achieved a lower total Barratt-score (58.24) compared to participants without this haplotype (59.63), and this effect was significant either calculating for the three haplotype categories (p = 0.009), or opposing the T-T haplotype (associating with lower impulsivity scores) to the others (p = 0.003). In other words, the 39 UTR haplotypes of the SNAP-25 gene had a more pronounced effect on the measured phenotype than the contributing SNPs, separately.

Functional analysis of SNAP-25 39 UTR haplotypes
Based on the results of the association study, the 39 UTR haplotypes showing an association with impulsivity were subjected to in vitro functional analysis. According to sequence alignment analysis both SNPs (rs3746544 and rs1051312) in the 39 UTR are supposed to alter duplex formation of miR-641 and SNAP-25 mRNA, as the two polymorphisms are localized in the target site of the seed region of the miRNA-641 (Figure 2A). To study the molecular effect of the target sequence variants, pMIR-REPORT luciferase constructs containing the complete 39 UTR region with the different haplotype variants were analyzed. This included the G-C haplotype as well, although it could not be detected in our investigated population. The lowest relative luciferase activity could be observed in case of the T-T haplotype, which generates a perfect target sequence for the seed region of miR-641. A single nucleotide change of the target sequence in constructs with the G-T and T-C haplotypes resulted in a 1.8 and a 2.1-fold elevation of relative luciferase activity, respectively. Although the single nucleotide change elevated significantly the reporter activity, there was no difference between G-T and T-C haplotypes, suggesting that a single base change alters significantly the binding of miR-641, but the position of the SNP within the seed region does not play a crucial role in this system. The G-C haplotype resulted in two mismatches in the target region of miR-641, and caused a 4.6-fold higher luciferase activity compared to the T-T form ( Figure 2B), demonstrating that two mismatches in the target sequence have a more pronounced effect on miRNA function than one mismatch. The control construct had the same length, but an independent sequence without any binding site of miR-641. As expected, the corresponding luciferase activity was significantly higher than that of any of the constructs containing the 39 UTR of the SNAP-25 gene.

Discussion
Recent studies have demonstrated that regulation of protein synthesis is much more complex than was earlier predicted. One major component of this network is the system of micro-RNAs (miRNAs) demonstrated by the fact that as high as 5% of the genes are predicted to encode miRNAs. It is also noteworthy that at least 30% of protein coding genes are suggested to be regulated by miRNAs [29], while this number could reach even 60% according to other estimations [30]. Alterations in miRNA profiles have been shown in the background of several illnesses, such as cancers, autoimmune diseases or cardiovascular disorders [31]. MiRNA profiling can be a useful tool in tumor diagnostics [32], and several miRNA-based therapeutic protocols have been elaborated. For example a miR-122 inhibitor was shown to decrease the amount of hepatitis C virus RNA, whereas a modulator of miR-208 level seemed to be effective against cardiac hypertrophy [33].
As a single miRNA might regulate a set of target genes, or the 39 UTR of a specific mRNA might bind a number of miRNAs, the translational regulation by microRNAs is a complex system. Interestingly, as few as 20 SNPs have been identified in genomic regions coding for miRNA seed sequences. On the other hand, more than a hundred thousand SNPs are predicted to change miRNA target sites, but less than 1% of these functional variants have been validated experimentally [30].
The rs3746544 and rs1051312 SNPs in the 39 UTR region of the SNAP-25 gene have been thoroughly investigated as possible risk factors of ADHD [34,35,36] or bipolar disorder [12], however the molecular function of these variants has not yet been studied. Thus the question has been raised if these SNPs are genetic markers or causal polymorphisms [14], and according to our knowledge, our study is the first to analyze the effect of these SNPs on miRNA binding efficiency. Both studied SNPs are localized in the binding region of miR-641 in a close proximity of each other.
Here we demonstrated that a single mismatch in the 39 UTR of the SNAP-25 gene caused by any of the two SNPs resulted in about 2-fold reporter activity, while the rs3746544 G-rs1051312 C haplotype possessing two mismatches in the miR-641 target site led to a more than 4-fold elevation in the luciferase level. These results confirm the molecular functional role of rs3746544 and rs1051312 haplotypes in the SNAP-25 gene and are in agreement with the results of previous studies demonstrating the importance of optimal SNAP-25 level. The coloboma mouse, which is the animal model of ADHD has SNAP-25 deficiency and changes in SNAP-25 expression were shown to play a role in altered neuronal function [37]. It was also demonstrated, that atomoxetine which is an orally administered medicine used for the treatment of ADHD resulted in the significant up-regulation of SNAP-25 both on mRNA and protein level [38].
Here we aimed to investigate the impulsive behavior of a healthy population of European descent, and found a nominally significant association (p = 0.042) between rs1051312 and impulsivity which did not survive the Bonferroni correction for multiple testing (p,0.0125 for 4 SNPs). If, however, the 39 UTR haplotypes were applied in the association study instead of the SNPs, lower impulsivity scores were observed in the presence of the T-T haplotype of the rs3746544 and rs1051312 SNPs as compared to the other haplotypes (p = 0.003). These results are in good agreement with our data obtained by the molecular analysis where the lowest reporter activity was measured in the presence of the 39 UTR with T-T haplotype. Our findings are also in agreement with family studies, which showed association between the rs3746544G [39] and the rs1051312C alleles [40], although these results were not significant.
On the other hand the T allele of both SNPs was also shown to be the risk factor of ADHD. It is also notable, however, that the association had a modest statistical significance [10] and odds ratio was 1.15 and 1.06 for the rs3746544 and rs1051312 SNPs, respectively [14]. This shows that SNAP-25 is only one of the numerous genetic components of this phenotype. Taking into consideration that our study investigated impulsivity instead of ADHD in a healthy population, all data support, that SNAP-25 is one of the several genetic factors of impulsivity and related psychiatric disorders. Interestingly the haplotype with double mismatch (G-C) in the miRNA target site of SNAP-25 gene was completely undetectable in our healthy population of European descent (Hungarian) according to our previous [19] and current studies. It is important to note that haplotypes were not estimated but individually measured by a direct molecular haplotype analysis method developed earlier [19,41]. A theoretically possible explanation of the absence of the G-C haplotype in our healthy volunteer population could be the more severe effect of this haplotype on SNAP-25 gene expression leading to pathophysiological consequences. Alternatively, the missing haplotype might be explained by the evolutionally young age of one of the SNPs, therefore recombinant haplotypes of these 39 UTR SNPs, could not spread out in the population yet.
Although the putative biological effect of SNPs in regulatory regions both in 59 or 39 region is the modulation of protein level by different mechanisms, interestingly, much less data are available about the role of 59 polymorphisms of the SNAP-25 gene. The rs6039769 SNP was shown to be in association with early-onset bipolar disorder [15], however no association could be found between ADHD and this locus [13]. On the other hand, the other polymorphism, located approximately 2 kb 59 from the transcriptional start site (rs6077690) was demonstrated to be in association with ADHD [13]. Analysis of the two promoter SNPs and their haplotypes did not reveal any association between the polymorphisms and impulsivity. In conclusion, our results confirmed the findings of previous studies investigating SNAP-25 and ADHD.
Our study focused on a healthy population, but even in this setting an association could be observed between the haplotypes of the 39 UTR SNPs (rs3746544 and rs1051312) and normal individual variability of impulsivity. In vitro functional analyses suggested that these loci are miR-SNPs altering the binding efficiency of miR641 in a luciferase reporter system.