Supporting the generalist genes hypothesis for intellectual ability/disability: the case of SNAP25

Authors


Corresponding author: T. S. Rizzi, Functional Genomics, Centre for Neurogenomics and Cognitive Research, VU Medical Center and VU University, De Boelelaan 1087, 1081 HV, Amsterdam, The Netherlands. E-mail: t.rizzi@vumc.nl

Abstract

Intellectual disability (ID) is an unresolved health care problem with a worldwide prevalence rate of 2–3%. For many years, research into the genetic causes of ID and related disorders has mainly focused on chromosomal abnormalities or X-linked genetic deficits. Only a handful of autosomal genes are known to cause ID. At the same time it has been suggested that at least some cases of ID represent an extreme form of normal intellectual ability and therefore that genes important for intellectual ability in the normal range may also play a role in ID. In this study, we tested whether the autosomal SNAP25 gene, which was previously associated with variation in intellectual ability in the normal range, is also associated with ID. The gene product of SNAP25 is an important presynaptic plasma membrane protein, is known to be involved in regulating neurotransmitter release, and has been linked to memory and learning by its effect on long term potentiation in the hippocampus. Allele frequencies of two genetic variants in SNAP25 previously associated with intellectual ability were compared between a group of 636 ID cases (IQ < 70) and a control group of 361 persons of higher than average intellectual ability. We observed a higher frequency of the putative risk allele of rs363050 (P = 0.02; OR = 1.24) in cases as compared to controls. These results are consistent with a role of SNAP25 in ID, and also support the notion that ID reflects the lower extreme of the quantitative distribution of intellectual ability.

Intellectual disability (ID) affects 2–3% of the (child) population, (Bhasin et al. 2006; Kraijer & Plas 2006). ID is defined as ‘an IQ < 70 with deficits in two or more adaptive skills starting at a childhood age’ (American Psychiatric Association, DSM-IV 1994, Widiger et al. 1994). ID is a complex disorder in which a large number of intellectual skills (e.g. language, motor, social, emotional, visuospatial) are suboptimally developed in the patient. The severity of disability ranges from profound to mild disability, although the majority of cases (60–85%) are classified as ‘mild disability’ (Ropers & Hamel 2005).

The causes of ID are enormously heterogeneous, and in a significant proportion of patients with ID the cause remains unexplained (Curry et al. 1997; Schaefer & Bodensteiner 1992). Some of the known causes of ID include environmental factors like cerebrovascular incidents associated with premature birth, and genetic factors like chromosomal abnormalities, or rare mutations with major gene effects (Dimauro et al. 1989; Piecuch et al. 1997).

Genetic causes of ID are thought to be present in 25–50% of cases (Mclaren & Bryson 1987). The Online Mendelian Inheritance in Man (OMIM) database contains 488 identified ID genes (March 2012). Most likely numerous additional ID genes remain to be identified (Inlow & Restifo 2004). Current genetic research typically focuses on chromosomal abnormalities and X-linked genetic effects which tend to be related to the more severe forms of ID (Chelly et al. 2006; Rauch et al. 2006). From all reported cases of individuals with moderate to severe ID, only about 50% can be traced back to a known cause (Curry et al. 1997; Schaefer & Bodensteiner 1992). In mild ID (IQ between 50 and 70) about 30% has a known cause (Chelly et al. 2006; Inlow & Restifo 2004; Ropers & Hamel 2005). It has been proposed that the lack of major gene findings for mild ID suggests that mild ID is influenced by multiple genes, each of relatively small effect (Ropers 2008). It is also estimated that most remaining ID genes are autosomal (Chelly et al. 2006).

According to the two related ‘Common Disorders are Quantitative Traits’ and ‘Generalist genes’ hypotheses (Plomin & Kovas 2005; Plomin et al. 2009), common disorders, such as ID, are the quantitative extreme of a normally distributed trait (intellectual ability), and therefore the same genetic factors are largely responsible for both upper and lower extremes. Genetic polymorphisms associated with variation in normal intellectual ability may thus also be involved in some cases of ID. In this study, we set out to test whether the SNAP-25 gene, which has been associated with variation in intellectual ability in a non-clinical, population-based sample (Gosso et al. 2006, 2008), is also important in ID. If that is the case, that would support the notion that at least some cases of ID are indeed at the lower extreme of the normally distributed trait ‘intellectual ability’.

Materials and methods

ID cohort (cases)

The ID cohort consisted of 636 Dutch children (518 males, 118 females; mean age of 7.7; SD 3.2) with intellectual delay (IQ < 70), without chromosomal rearrangements and without fragile X syndrome, recruited through the Clinical Genetics department of the VU Medical Centre. Anonymous blood-samples of the ID cohort were collected and genomic DNA was isolated from all samples using Flexigene AGF3000 technology (QIAGEN, Valencia, CA, USA) on an automated AutoGeneFlex 3000 isolator (AutoGene, Holliston, MA, USA) according to the protocols supplied by the provider. DNA was collected for diagnostic purposes and parents consented to anonymous use of the DNA samples for scientific purposes.

Higher than average IQ cohort (controls)

Controls were derived from a ‘higher than average IQ’ (HTA-IQ) cohort consisting of 361 (170 males and 191 females) Dutch individuals aged between 13 and 14 years old at the time of inclusion in the study (mean age = 13.5, SD = 0.5). The HTA-IQ cohort is part of the Amsterdam Growth and Health Study (AGAHLS) (Kemper & van Mechelen 2004; Kemper et al. 1997). This is a longitudinal study that started in 1976 and recruited children who followed the highest level of secondary education in two Dutch secondary schools. Subjects who reported to have been premature at birth were excluded from the study (Kemper & van Mechelen 2004). DNA was isolated from 361 of these participants. A subsample (N = 260) had also performed a Dutch standardized IQ test (‘Groninger IQ Test, GIT’) (Luteijn & Ploeg, 1983), which correlates well to the widely used Wechsler Adult Intelligence Scale (WAIS) (Wechsler 1997), with correlations of standardized GIT and WAIS-IQ scores ranging from 0.72 to 0.91 (Luteijn & Ploeg 1983). In this study, IQ scores in the subsample were only used to confirm the ‘higher than average IQ’ level of this group, which was used as ‘controls’ for the ID cases. The mean IQ of this sample was 107.3 (SD = 13.8), which is conform expectation as these individuals were sent to the highest level of secondary school. Both the fact that ID is generally detected at an early age (Fidler et al. 2010) and the fact that all controls entered the highest level of secondary education in the Netherlands, render it highly unlikely that individuals marked as controls would later develop into cases. The study was approved by the medical ethical committee of the VU University Medical Center, and all subjects gave their written informed consent (provided by the parents as the subjects were aged 13–16 years).

Genotyping

We selected two SNPs (rs363050 and rs363039) in the SNAP-25 gene that were previously reported to be associated with intellectual ability in two independent Dutch samples (Gosso et al. 2006, 2008). The selected genetic variants are located in the first intron of SNAP25 and are not in strong linkage disequilibrium (LD) (r2 < 0.40).

A TaqMan assay with specific fluorogenic probes in the high throughput 5′ nuclease assay (TaqMan, PE Applied Biosystems, Foster City, CA, USA) was used for genotyping the two SNPs in the SNAP25 gene (rs363039 and rs363050). Deviation from Hardy–Weinberg equilibrium for all genotyped markers was tested using PLINK (Purcell et al. 2007). Alleles previously reported to be associated with increased intellectual ability by Gosso et al. (2006, 2008) are G for rs363039 (minor allele: A, major allele: G) and A for rs363050 (minor allele: G, major allele: A). Thus, for both SNPs the minor allele is defined as the ‘putative risk allele’ for ID.

Statistical analysis

A logistic regression of case–control status on genotype implemented in Plink (Purcell et al. 2007) was performed, adjusting for the effect of sex. One-sided tests were conducted to test the hypothesis that the putative risk alleles for ID were more frequent in the ID cases as compared to the HTA-IQ controls. The Bonferroni corrected significance level was set at 0.05/2 = 0.025, i.e. correcting for testing two SNPs.

Brain expression analysis

As the selected SNPs (rs363050 and rs363039) are intronic and do not have a known functional role, we investigated whether they were associated with expression of the SNAP25 gene in the brain. To this end, we used the publicly available brain expression dataset (Myers et al. 2007) that includes genotypes and brain expression data. Brain cortex samples were available from 193 individuals of European descent with age at death greater than or equal to 65 years with no clinical history of stroke, cerebrovascular disease, Lewy bodies, or co-morbidity (Myers et al. 2007). All 193 samples were genotyped using Affymetrix GeneChip Human Mapping 500 K and the expression analysis was done using Illumina HumanRefseq-8 Expression BeadChip (Myers et al. 2007). For all 193 individuals identified from the Myers' database, genomic coverage in the SNAP25 genomic area (±1.5 Mb) was increased by using genomic imputation (MACH; Li et al. 2009). The reference panel used was the HapMap II phased data (NCBI build 36, UCSC hg18). For the brain expression phenotype we made use of the available SNAP25 mRNA intensity information [transcript variant 2 (NM_130811) isoform SNAP25B]. Genetic association of imputed genotypes for all 193 individuals from the Myers' database was carried out using a weighted linear regression analysis implemented in MACH2QTL (Li et al. 2009). This sample size (193) is sufficient to detect SNPs explaining 4% of the variance in expression of SNAP25, given a Bonferroni corrected significance level of 0.025.

In silico binding site analysis

We further investigated if the SNAP25 genetic variants might have functional effects by affecting transcription factor binding sites (TFB). For this we used the UCSC browser, which includes experimental outcomes from published studies (Fields 2007; Fujita et al. 2011; Valouev et al. 2008). In addition, we investigated whether the two variants might alter binding of the transcription factors using the JASPAR binding site prediction program (Sandelin et al. 2004). We selected a 500 bp region surrounding the two SNPs and used the web interface for an online sequence analysis of regulatory regions present in the region. The TFB site models for each sequence were selected if the scoring matrices were above 90%. The analysis was done for each allele separately. We ran similar analyses for SNPs in high LD (r2 > 0.9) with the two target SNPs.

Results

Quality control

In total, 997 DNA samples were available for genotyping, 636 in the ID cohort and 361 in the HTA-IQ cohort. The genotyping success rate was 96% average. Missingness occurred due to ambiguous genotype calling in small proportion of the TaqMan assays. Both SNP's where in Hardy–Weinberg equilibrium in both cohorts separately [ID cohort rs363050: P = 0.47, rs363039: P = 0.10; HTA-IQ cohort rs363050: P = 0.73, rs363039: P = 0.69 as well as for the both cohorts combined rs363050: P = 0.36, rs363039: P = 0.25)].

Descriptives

Frequencies of the putative ID-risk alleles (G in rs363050 and A in rs363039) were 0.47 and 0.35, respectively in the ID cohort and 0.41 and 0.30 in the HTA-IQ groups, see Table 1.

Table 1. Descriptives of cases of intellectual disability (ID) and controls of higher than average IQ (HTA-IQ)
 ID cohortHTA-IQ cohort
Age in years (SD)7.7 (3.2)13.5 (0.5)
Males/females (%)81/1947/53
Total N636361
rs363050 Genotypic frequencies (GG/AG/AA)0.227/0.484/0.2890.174/0.473/0.353
Risk allele (G) frequency rs3630500.470.41
rs363039 Genotypic frequencies (AA/AG/GG)0.139/0.427/0.4340.084/0.433/0.483
Risk allele (A) frequency rs3630390.350.30

Figure 1 shows the allele frequencies of the putative risk alleles for cases and controls and for males and females separately.

Figure 1.

Putative risk allele frequencies of the two selected SNAP25 SNPs across intellectual disability (ID) cases and higher than average IQ (HTA-IQ) controls, and across males and females. Error bars denote 95% CI.

Case–control association analysis

As the cases and controls were not matched for sex (i.e. the cases included relatively more males, which is likely related to the higher prevalence of ID in boys than in girls) all analyses included sex as a covariate, to correct for possible confounding of sex with genotype. The putative risk allele of rs363039 (A) was not significantly associated with ID (OR = 1.11, P = 0.182), although a relative large difference in allele frequency in females in the hypothesized direction was observed. The putative risk allele of rs363050 (G) was associated with ID (OR = 1.24, P = 0.020), and showed a higher frequency in cases compared to controls. Analyses for rs363050 were also conducted separately for males and females and showed a significant association in females (OR = 1.44, P = 0.02) but not in males (OR = 1.14, P = 0.17), although the direction of the effect was similar. A separate analysis was run to test for a sex-by-genotype interaction, which showed no significant interaction (P = 0.12). Together with the sex-adjusted analyses, this suggests that although the effect was strongest in females, both males and females showed an association in the same direction.

For rs363039 the test in females and males only did not reach significance, although females showed a trend (OR = 1.34, P = 0.06)

Brain expression and in silico analysis

There was no evidence for an association of either variant with SNAP25 brain expression, using the Myers database (rs363039: P = 0.89; rs363050: P = 0.60). In addition, neither of the genetic variant was inside reported binding site regions nor in the predicted sites (scoring matrices were below 90%). We also checked SNPs that were in high LD (r2 > 0.9) with rs363050 or rs363039 using the CEU hapmap LD structure. There were three SNPs in high LD with rs363050 (rs6104571, rs363016 and rs12626080), which are located in the second intron of SNAP25. None of them were associated with expression of SNAP25, however, two of them were located inside predicted TFB sites: rs6104571 (MEF2A binding site) and rs12626080 (FOXL1 binding site).

Recently Söderqvist et al. (2010) conducted an in silico analysis using the TESS (transcription element search system) program to search for a functional role of the rs363039 SNP. They reported the predicted presence of a glucocorticoid receptor (NR3C1) binding site in carriers of the A allele (i.e. the putative risk allele) of rs363039, which was lacking in carriers of the G allele. Instead, G allele carriers were predicted to have a zinc finger protein (ZNF589, alias SZF1) binding site. However, these predicted binding sites for rs363039 could not be confirmed by us using data from UCSC or JASPAR.

Discussion

In this study, we tested whether two genetic variants in the SNAP25 gene, which were previously associated with variation in intellectual ability in the normal range, are associated with ID risk. For one SNP (rs363050) we observed a significantly higher frequency of the putative risk allele in ID cases as compared to the controls of higher than average IQ. For the second SNP (rs363039) the allele frequency difference between cases and controls was not statistically significant although there was a trend in the expected direction.

In silico analysis showed that neither SNP was related to differential brain expression of SNAP25. This suggests that if there is a regulatory role, it must be very small (explaining <4% of the variance in SNAP25 brain expression), and will therefore go undetected with the current sample size of brain expression analysis. Both SNPs were also not predicted to alter TFB sites. However, there were three SNPs in high LD with rs363050 (rs6104571, rs363016 and rs12626080). Two of these were located inside predicted TFB sites: rs6104571 (MEF2A binding site) and rs12626080 (FOXL1 binding site). The MEF2A binding site is of particular interest as the MEF2A factor was recently reported to be involved short-term synaptic plasticity in mice (Akhtar et al. 2012). Further research is needed to elucidate the functional role of these SNPs or other SNPs in LD with them.

The SNAP25 gene codes for a presynaptic plasma membrane protein that is an integral component of the vesicle docking and fusion machinery that regulates neurotransmitter release (Horikawa et al. 1993; Oyler et al. 1989; Seagar & Takahashi 1998). It is also implicated in axonal growth and synaptic plasticity (Osen-Sand et al. 1993) and is shown to be involved in hippocampal long-term potentiation (LTP), which is thought to be a form of synaptic plasticity that underlies memory and learning (Bliss & Collingridge 1993, Hou et al. 2004; Martin et al. 2000; Morris et al. 1986; Roberts et al. 1998). Pavlowsky et al. (2011) recently reviewed all known ID-related genes and concluded that gene-products of these genes are enriched in synapses, thereby ‘supporting the unifying synapse-based theory for cognitive deficits’ (Pavlowsky et al. 2011). Although there have also been failures to replicate (e.g. see the recent study by Chabris et al. 2011 who failed to replicate the rs363050 effect on ‘g’ in a sample of 6464 individuals), SNAP25 has been suggested to be associated not only with ID and intellectual ability but also with related traits such as autism (Guerini et al. 2011), ADHD (Forero et al. 2009) and working memory capacity (Söderqvist et al. 2010).

Although our results warrant replication in other ID cases and controls to further confirm the role of SNAP25 in ID, we set out to put the ‘Common disorders are quantitative traits’ and ‘Generalist Genes’ hypotheses to a test (Plomin & Kovas, 2005; Plomin et al. 2009). We reasoned that if ID is indeed the extreme of the quantitative trait intelligence, than it is likely that genes associated with intelligence are also of importance to ID. Our results support this view by showing that SNAP25 also plays a role in the lower extreme of the quantitative trait ‘intellectual ability’. This also suggests that genes that have already been identified for ID may be of importance in normal intellectual functioning. It should be noted however, that we do not intend to suggest that all cases of ID represent the lower extreme of the normally distributed trait of intellectual ability. ID is a heterogeneous disorder with multiple causes, some of which are environmental, some are monogenic, X-linked, and in addition some may indeed be at the lower extreme of intellectual ability and share common (genetic) causes.

Intellectual ability is considered to be influenced by hundreds of genetic variants each of small effect (Davies et al. 2011). Gene finding for intellectual ability therefore necessitates large sample sizes. If, however some of the genetic variants underlying normal intellectual ability also influence the lower (and upper) extremes of the distribution, then statistical power of genome-wide association analyses for normal intellectual ability can be greatly enhanced by selecting extreme cases of the distribution (i.e. ID vs. high IQ) (Plomin et al. 2009). Gene-finding efforts for both intellectual ability and disability may thus benefit from taking a generalist genes view.

Acknowledgements

We are extremely grateful to all the families who took part in this study (high and low IQ cohort). The AGAHLS was financially supported by the Netherlands Heart Foundation grant 76051-79051, Dutch Prevention Fund grants 28-189a, 28-1106, and 28-1106-1, the Dutch Ministry of Well Being and Public Health grant 90-170, the Dairy Foundation on Nutrition and Health, the Dutch Olympic Committee/Netherlands Sports Federation, Heineken Inc, and the Scientific Board of Smoking and Health. Statistical analyses were carried out on the Genetic Cluster Computer (http://www.geneticcluster.org) which is financially supported by the Netherlands Scientific Organization (NWO 480-05-003) along with a supplement from the Dutch Brain Foundation and the VU University Amsterdam. We further wish to acknowledge the financial support of NWO-VI-016-065-318, and the Center for Neurogenomics and Cognitive Research (CNCR) at the VU University. None of the authors have conflict of interest to declare.

Ancillary