Background

Autism (MIM 209850) is a kind of complex neurodevelopmental disorder clinically characterized by three core deficits: impairment of social interaction, deficits in verbal communication, as well as stereotypic and repetitive behaviors [1, 2]. The onset of autism is usually before the age of 3 years old. Autism shows a remarkable sex bias, with a male-to-female ratio of over 4:1 [3]. The global prevalence of autism has risen to 6.2 ‰ [4]. Autism patients in China are estimated in the range of 0.6–1.8 millions according to the WHO’s report.

It is commonly accepted that there is a strong genetic basis associated with environmental factors in the etiology of autism [5]. The importance of genetic factors has been demonstrated by family and twin-based studies. The recurrence risk of autism in sib-ships is 3–8 %, approximately 50–100 times greater than in the general population. Twin studies have documented a higher concordance rate in monozygotic (60–90 %) than in dizygotic twins (3–10 %) [6–8]. However, the genetic factors for autism are still largely elusive.

Recently, several studies have implicated the involvement of neuroligin family in the incidence of autism [9–13]. Neuroligins are postsynaptic cell adhesion molecules and are involved in the NRXN-NLGN-SHANK pathway, which are most likely associated with synaptogenesis and balance between synapse excitatory and inhibitory. Neuroligin family includes five members of type I transmembrane proteins (NLGN1, NLGN2, NLGN3, NLGN4X and NLGN4Y). The NLGN3, NLGN4X and NLGN4Y genes are located on sex chromosomes. Neuroligins are evolutionarily highly conserved, sharing 52 % identity averagely. The neuroligin proteins are composed of several domains, including a cleavable signal peptide, an extracellular noncatalytic acetylcholinesterase homology domain, a carbohydrate attachment region, a single transmembrane domain and a short C-terminal tail containing a type I PDZ-binding motif [14, 15].

In 2003, a mutation in NLGN3 (p.R451C) and a mutation in NLGN4X (p.D396 fs) were reported in two unrelated Swedish autism families, respectively [10]. In the following years, several other mutations in the NLGN3, NLGN4X and NLGN4Y genes have been reported to be related with autism [11, 12, 16–20]. In vitro and in vivo experiments have indicated that the autism-related neuroligin mutations may affect synapse maturation and function [19, 21, 22]. Comprehensive analyzing all of the mutations found in neuroligin family, the mutations found in autism patients from various ethnic backgrounds present conflicting results and failed to be validated by other research groups, regarding the etiology of ASD, because of genetic heterogeneity and/or differences in their ethnic background. Furthermore, mutations in these genes have been reported only in less than 1 % autism patients hining that neuroligin family mutations are rare mutation with low frequencies and may account for only a small proportion of autism patients.

Here, we sequenced the entire coding regions of NLGN3 and NLGN4X gene from 318 unrelated sporadic autism cases, and identified a NLGN3 variation and three NLGN4X variations.

Methods

Subjects

The patients were comprised of 318 unrelated Chinese autism patients (270 males and 48 females, 366 X chromosomes in total, mean age is 6 years within a range of 3–18 years) and were recruited from psychiatric department of the Second Xiangya Hospital at Central South University or Elim Autism Training Department of the Qingdao Municipal Autism Research Institute. Subjects were diagnosed with the diagnostic and statistical manual of mental disorders-IV criteria autistic disorder and the childhood autism rating scale. Routine chromosome analysis revealed a normal karyotype of 46, XY or 46, XX for all subjects. All subjects were tested negatively for Fragile-X syndrome using PCR procedure followed by denaturing polyacrylamide gel electrophoresis analysis. Meanwhile, the samples were excluded for other neurologic/medical conditions. The controls were consisted of 453 unrelated Chinese healthy volunteers (339 males and 114 females, 567 X chromosomes in total, mean age is 20 years within a range of 3–46 years) without a history of autism. All the patients and their direct families as well as the controls were Chinese Han descent. Part of the patients and controls were enrolled in other studies described previously [23, 24].The study was approved by the local ethics committee, and all the patients and controls have given written informed consent that complies with all the declaration of Helsinki Principles.

PCR amplification and sequencing

Genomic DNA was extracted from the peripheral blood using the standard phenol–chloroform method. The entire coding regions of the NLGN3(NM_181303) and NLGN4X(NM_020742) gene both in patients and controls were analyzed by PCR and direct sequencing using 22 pairs of primers(primer sequences were shown in the supplementary file 1). The PCR products were verified by 6 % polyacrylamide gel electrophoresis. DNA sequencing was performed using ABI 3100/3130 DNA analyzer and the data were analyzed employing DNASTAR software. Amplification of abnormalities was confirmed in a second independent PCR amplification and DNA sequencing. The standard DNA sequences for the neuroligin3 and neuroligin4X were collected from UCSC database (http://genome.ucsc.edu/).

Statistical analysis

In this study, all statistical analysis was performed using SPSS 13.0 for Windows. The allelic(A, a) and genotypic(AA, Aa, aa, A and a) distributions of the cases and controls as well as Odds ratio (OR) were determined by Fisher’s exact test, respectively. All statistical tests were two-sided. p values lower than 0.05 were considered to be significant. Data for the Hardy–Weinberg equilibrium was evaluated at http://analysis.bio-x.cn/myAnalysis.php.

Evolutionary conservation analysis and functional prediction

Protein sequences for the neuroligin3 and neuroligin4X of different species were collected from UCSC database (http://genome.ucsc.edu/). Analysis of evolutionary conservation was performed using the softwares ClustalX, Phylop and GERP, which were used to analyze multiple sequence alignments. Function prediction of the substitutions was performed using programs MutationTaster, Poly-Phen-2 and SIFT. These three softwares were used to evaluate disease-causing potential of sequence alterations.

Results

To study the contribution of neuroligin family in the etiology of autism, we analyzed the whole exons (excluding the 3′ UTR) and the exon flanking intron sequences of NLGN3 gene and NLGN4X gene in 318 unrelated sporadic autism cases and 453 controls. As shown in Table 1 and Fig. 1, several variations and single nucleotide polymorphisms (SNPs) were identified. In the patient cohort we identified ten reported SNPs and four novel missense variations. The four missense variations were absent in the 453 controls and have not been reported in dbSNP database, 1000 Genomes Project and NHLBI ESP. In the control cohort, except the ten reported SNPs found in patient cohort, we identified another three reported SNPs.

Table 1 The variations found in NLGN3 and NLGN4X gene
Full size table
Fig. 1
figure 1

Identification of NLGN3and NLGN4X missense mutations in sporadic autism patients. a Schematic representation of the mutations p.G426S(NLGN3), p.G84R-(NLGN4X), p.Q162A(NLGN4X) and p.A283T(NLGN4X). Except the mutation p.G84R(NLGN4X), the other three missense mutations were analyzed by genotyping the patient, parents and control. Because of absence of the sample from the father, the mutation p.G84R(NLGN4X) was only analyzed in patient, mother and control. The black triangles show the location of the mutation. b Secondary structure elements of NLGN3 and NLGN4X and the location of the mutations identified in our study. Neuroligin3 and neuroligin4X contain an extracellular acetylcholinesterase (AChE)-homologous domain that contains one site of alternative splicing (SSA), a highly glycosylated region (CH), a transmembrane domain (TM) and terminate in PDZ-domain-binding sites. The black arrows show the location of the mutation identified in NLGN4X gene and the turquoise arrow shows the location of the mutation identified in NLGN3 gene

Full size image

The four novel missense variations found in patients (p.G426S in NLGN3, and p.G84R, p.Q162 K, p.A283T in NLGN4X) were identified in unrelated single patient and no patient carried more than one variation. The two variations p.G426S and p.Q162 K were observed in two different female patients, whereas the other two variations p.G84R and p.A283T were observed in two different male patients. We further investigated the inherited status of the four variations by analyzing the parents’ DNA samples. As shown in Table 1, the variations p.G426S and p.Q162 K were de novo mutations, while the variations p.G84R and p.A283T were inherited from their asymptomatic mothers, respectively.

Analysis of evolutionary conservation showed that the four variations are located in the conserved extracellular noncatalytic acetylcholinesterase homology domain(Fig. 2). Functional prediction of these substitutions using programs MutationTaster, Poly-Phen-2 and SIFT revealed that p.G84R and p.A283T were “Probably damaging”; however, p.G426Sand p.Q162 K were likely to be “benign” (Table 2). By analysis, the mutation p.G84R and p.A283T were predicted to probably alter protein function.

Fig. 2
figure 2

Multiple sequence alignment for mutations of NLGN3/4X. a Alignment of NLGN3 mutation in different species. b Alignment of NLGN4X mutations in different species. Protein sequences for NLGN3/4 were collected from UCSC database and NCBI database. We used protein sequence from human, chimp, gorilla, mouse, horse, dog, opossum, chicken, x.tropicalis and zebrafish for alignment. Multiple sequence alignment of NLGN3/4 was made by the software ClustalX. The mutations are highlighted by black arrows

Full size image
Table 2 Evolutionary conservation analysis and function prediction of missense variations in NLGN3 and NLGN4X gene
Full size table

In order to analyze the association of NLGN3 and NLGN4X with autism in Chinese population, we analyzed the allelic and genotypic distribution frequencies of the common SNPs (rs115732184, rs116209949, rs180681274, rs7051693, rs190142919, rs139034847, rs7049300, rs3747333 and rs3747334) presented both in patients and controls (Table 3). The p values of allelic frequencies for rs3747333 and rs3747334 in 318 autism patients and 453 control subjects were both 5.09E–005. The p values of genotypic frequencies for rs3747333 and rs3747334 in 318 autism patients and 453 control subjects were both 0.003 in male patients and 0.036 in female patients. The odds ratio was 4.685 (95 % CI 2.073–10.592). All the genotypes were in Hardy–Weinberg equilibrium. The statistical analysis implied an association of the gene NLGN4X with autism in Chinese population.

Table 3 Case–control analysis for nine common SNPs in Chinese autism patients
Full size table

Discussion

In this study, we found four missense variations (p.G426S in NLGN3, and p.G84R, p.Q162 K, p.A283T in NLGN4X) in our autism cohort. These four missense variations were not identified in 453 control individuals and have not been reported in dbSNP database, 1000 Genomes Project and NHLBI ESP, implying they were probably patient-specific. Interestingly, as shown in Table 4, the four patients who carry with these missense mutations identified in NLGN3 and NLGN4X exhibited some deficit in cognitive and verbal development. We thus speculated that the variations mentioned above may be involved in the development of the disease in the four patients.

Table 4 Clinical features of Autism patients with missense mutations in NLGN3 and NLGN4X gene
Full size table

All four missense mutations reported here are located in the conserved extracellular noncatalytic acetylcholinesterase homology domain (Fig. 2). The acetylcholinesterase homology domain of neuroligins essential for binding to neurexin and triggering synapse activity [25]. As shown in Fig. 1b, p.Q162 K is located at splice site A (SSA) that product multiple isoforms and alternative splicing, thus its mutation may interfere with alternative splicing and the formation of neuroligin–neurexin complex [26]. The p.G426S mutation is located in a predicted EF-hand domain [27]. The EF-hand domains are known to concentrate Ca2+ and confer protein’s structural integrity. As the interaction between neuroligin and neurexin depends on Ca2+ concentration [28]. p.G426S mutation may affect the binding of neuroligin and neurexin by abnormal concentrate of Ca2+. p.G84R, p.Q162 K and p.A283T change the polarity of the amino acid and maybe accordingly change the protein conformation.

Given that the two asymptomatic mothers were carriers of p.G84R and p.A283T mutation respectively, X-chromosome inactivation (XCI) may be involved in the pathogenessis of autism. The abnormal X chromosomes of the two mothers may have been inactivated selectively by XCI, thus they did not suffer autism. This may lead to genetic counseling implications that carrier mothers may be at risk to have affected offspring.

In conclusion, we identified the common SNP (rs3747333 and rs3747334) in the NLGN4X gene significantly associated with risk for autism. We have identified four rare missense variations in the NLGN3 and NLGN4X genes, reinforced the importance of neuroligin family in the etiology of autism. These variations may be related to abnormal synaptic homeostasis and therefore make it more predisposed to autism. However, further investigations regarding the functional consequences of these mutations are to be resolved to elucidate the specific mechanism affecting synapse signaling activity and the pathogenesis of autism in future studies.