Abstract
Background. Recent studies have shown that Ras-like without CAAX2 (RIT2) polymorphism is a susceptible factor for Parkinson’s disease (PD) and autism spectrum disorder (ASD). SNP rs12456492 and rs16976358 show the emerging evidence of increased risk of PD and ASD, respectively. A meta-analysis examining the relationship between rs12456492 and PD was reported, but the association between rs16976358 and ASD has not been investigated. Methods. We searched literature from the databases PubMed, Embase, Google Scholar, ScienceDirect, EBSCOhost, OVID, Web of Science, and Wiley up to February 2021. Three studies including 1160 ASD cases and 1367 controls were eventually enrolled in the meta-analysis based on strict inclusion and exclusion criteria. Results. All genetics models indicate a significant association between rs16976358 polymorphism and ASD susceptibility (C vs. T: ; CC vs. TT: ; CT vs. TT: ; CC+CT vs. TT: ; CC vs. CT+TT: ; TT+CC vs. CT: ). The results of sensitivity analysis and publication bias of Begg’s and Egger’s tests were stable in the models of allele (C vs. T), codominant (CC vs. TT), dominant (CC+CT vs. TT), and recessive (CC vs. CT+TT). Conclusions. Our meta-analysis exhibits that the allele C, CC, and CT genotyping of rs16976358 suggest the risk for ASD, but additional studies using a large sample size and ethnically diverse populations need to be included in the future.
1. Introduction
Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by global developmental delay, speech and language impairment, intellectual disability, restricted interests, and social communication deficits. It affects approximately 1-2% of the population worldwide with a ratio of 3 : 1 in males compared with females [1]. The etiology of ASD is diverse and complex, affected by both genetic and environmental factors. Since the 2000s, genetic studies have revealed that the genetic traits leading to ASD were multigenic and strongly heterogeneous. Many susceptible genomic regions were identified by a large-scale genome sequencing following single nucleotide polymorphism (SNP) studies. More SNPs in variant genes were revealed to be associated with ASD, such as microtubule affinity-regulating kinase 1 (MARK1), oxytocin receptor (OXTR), SH3 and multiple ankyrin repeat domains 3 (SHANK3), and gamma-aminobutyric acid type A receptor subunit beta3 (GABRB3) [2–5]. However, these mutations only partially account for ASD pathomechanism. It is still of importance to investigate other genes, which may contribute to the risk of ASD.
Recently, Ras-like without CAAX2 (RIT2) has been introduced as a genetic risk gene for neurological and psychiatric disorders such as Parkinson’s disease (PD), schizophrenia, and ASD. RIT2, as a member of the Ras superfamily, is highly expressed in the brain, in particular the retinal ganglion cells and dopaminergic neurons [6, 7]. It acts as GTPases and regulates the intracellular signaling cascades in neuronal differentiation, neurogenesis, neurite growth, and branching [8]. In 2012, a novel locus, RIT2 rs12456492, was identified as a risk factor for PD patients in the white population by conducting a meta-analysis of genome-wide association study (GWAS) [9]. This rs12456492 SNP is also confirmed based on Asian populations with Parkinson’s disease [10]. In 2016, Liu et al. found that, in a Japanese and Han Chinese populations, a RIT2 rs16976358 polymorphism had the strongest association with ASD [11], suggesting that RIT2 polymorphism also indicates the risk of ASD. To drive a precise estimation of the association between rs16976358 and ASD, we perform a meta-analysis to investigate the role of rs16976358 in the pathogenesis of ASD, which provides broad possibilities for further studies.
2. Methods
2.1. Literature Search
The databases PubMed, Embase, Google Scholar, ScienceDirect, EBSCOhost, OVID, Web of Science, and Wiley were used for the literature search. The searching strategy was applied by the following terms: “Autism spectrum disorder or autism or ASD” and “Ras-like without CAAX 2 or RIT2 or Rin or rs16976358” and “polymorphism or SNP”, with the last report up to February 26th, 2021. In addition, we also examined the references in the retrieved papers to identify extra studies.
2.2. Inclusion and Exclusion Criteria
The inclusion criteria are listed as follows: (1) association between rs16976358 and ASD risk; (2) case-control studies; (3) those providing sufficient genotyping data to calculate the odds ratio (OR) and 95% confidence interval (CI); and (4) patients with ASD were diagnosed according to the Diagnostic and Statistical Manual of Mental Disorders (5th edition) or the Autism Diagnostic Inventory-Revised (ADI-R) criteria. Studies were excluded for the following reasons: (1) republished or duplicated studies; (2) animal tests or reviews or nonrelated; and (3) insufficient genotyping data.
2.3. Data Extraction
Data were collected independently by two investigators according to the inclusion and exclusion criteria described above. In each study, the following information was extracted from the eligible studies: first author, year of publication, study type, region, ethnicity of the sample population, genotyping method, sample size of case, and control groups. Genotyping distributions and minor allele frequency (MAF) were separately listed to facilitate analysis. For the data which are not provided in published papers or the supplementary materials, the relevant information was obtained by direct communications with the corresponding authors.
2.4. Statistical Analysis
Statistical analyses were performed by STATA software, version 15.0 (STATA Corp., College Station, TX, USA). The association between rs16976358 and risk of ASD was evaluated by pooled OR and 95% CI. Six genetic models were established for analysis, including allele (C vs. T), codominant (CC vs. TT), codominant (CT vs. TT), dominant (CC+CT vs. TT), recessive (CC vs. CT+TT), and over dominant (TT+CC vs. CT). The significance of OR was determined by the -test, in which was considered as statistically significant. We examined the heterogeneity with the -test and statistics using the fixed- and random-effect model. If , the fixed-effect model was adopted to calculate the pooled ORs; if , the random-effect model was applied instead. Sensitivity analysis was carried out by sequentially omitting one study at a time to estimate the stability of the results. Publication bias among studies was determined using funnel plot and Begg’s and Egger’s tests. A meta-regression analysis was used to assess the impact on moderator covariables on the heterogeneity of results.
3. Results
3.1. Study Characteristics
Sixty-four initial studies were retrieved from the databases (PubMed: 4, Embase: 6, Google Scholar: 20, ScienceDirect: 13, EBSCOhost: 3, OVID: 10, Web of Science: 6, and Wiley: 2). Thirty-four studies were remained by removing duplicates. After browsing through the titles and abstracts, we excluded 5 reviews, 3 animal tests, 22 irrelevant studies, and 1 research article because of sufficient genotyping information. Finally, three studies published between 2016 and 2017 were selected with a collection of 1160 ASD cases and 1367 controls (Figure 1). All of these three studies were carried out in Asian populations. The characteristics and genotyping distribution are separately listed in Tables 1 and 2.
3.2. Meta-analysis Results
The association between rs16976358 and ASD risk was assessed by 6 genetic models, including allele (C vs. T), codominant (CC vs. TT), codominant (CT vs. TT), dominant (CC+CT vs. TT), recessive (CC vs. CT+TT), and over dominant (TT+CC vs. CT). Odds ratio (OR), 95% confidence interval (CI), and values from these models in the studies were, respectively, shown in Table 3. Meta-analysis was performed on the combined population by the fixed-effect or random-effect model based on the heterogeneity. The heterogeneity was significant in models of codominant (CT vs. TT) and dominant (CC+TT vs. TT) (). statistics showed that all the genetic models had high heterogeneity () (Table 4). In overall population, we found that rs16976358 was significantly associated with ASD risk in all the genetic models (C vs. T: , 95% CI: 1.517-2.057, ; CC vs. TT: , 95% CI: 2.630-7.369, ; CT vs. TT: , 95% CI: 1.127-2.311, ; CC+CT vs. TT: , 95% CI: 1.270-2.498, ; CC vs. CT+TT: , 95% CI: 2.217-6.176, ; TT+CC vs. CT: , 95% CI: 0.479-0.916, ) (Figure 2). Thus, the C allele and CC and CT genotyping of rs16976358 polymorphism indicate ASD susceptibility.
3.3. Metaregression and Sensitive Analysis
Metaregression was performed to explore the effect of covariables on the heterogeneity caused in our study. Different publication years and regions had no impacts on the heterogeneity found in any genetic models (Table 5). Unfortunately, we do not have sufficient data to analyze other factors contributing to the heterogeneity. To estimate the stability of the results, sensitivity analysis was carried out in the 6 genetic models. In Figure 3, pooled OR and 95% CI values were not influenced by sequentially omitting one study at a time, may suggesting the high stability of this meta-analysis. To analyze it in detail, we found that the results in the codominant (CT vs. TT) were unstable after omitting Liu et al. (estimated , 95% -2.629, ) and Hamedani et al.’ s studies (estimated , 95% -2.036, ). In the model of over dominant (TT+CC vs. CT), sensitivity was also unstable after deleting Liu et al. (estimated , 95% -1.099, ) and Hamedani et al.’s studies (estimated , 95% -1.090, ). But there was high sensitivity in the remained models of allele (C vs. T), codominant (CC vs. TT), dominant (CC+CT vs. TT), and recessive (CC vs. CT+TT) () (Table 6).
3.4. Publication Bias
Funnel plot and Begg’s and Egger’s tests were carried out to evaluate the publication bias in our study. A publication bias was found in codominant (CT vs. TT) and dominant (CC+CT vs. TT) models in funnel plot. One paper (Emamalizadeh et al.) was slightly apart from the symmetrical shape in these 2 models (Figure 4). However, no publication bias was shown in Begg’s and Egger’s tests (Table 7). Based on these results, the number of studies analyzed in our current meta-analysis may be not sufficient.
4. Discussion
In two distinct GWAS reports, RIT2 gene was identified as a new locus for both Parkinson’s disease (PD) and autism spectrum disorder (ASD) [9, 11, 12]. SNP analysis results in each study showed that rs12456492 and rs16976358 were associated with the risk of PD and ASD, respectively. Multiple meta-analysis of the relationship between rs12456492 and PD revealed that the G allele and GG and GA genotyping of rs12456492 (A/G) polymorphism may increase PD susceptibility [13, 14]. Recently, the number of the investigations on the association between rs16976358 and ASD is also increased [15–17]. The studies exhibit that the C allele of rs12456492 (T/C) indicates a possible genetic risk for ASD, but not all the genotypes show a significant difference in ASD patients compared to control populations.
For this reason, we carried out a meta-analysis on six genetic models (the allele (C vs. T), codominant (CC vs. TT), codominant (CT vs. TT), dominant (CC+CT vs. TT), recessive (CC vs. CT+TT), and over dominant (TT+CC vs. CT)). We found that CC and CT genotyping of rs16976358 might also underlie the ASD susceptibility in Asian populations. A high heterogeneity was found in all the genetic models in tests, which could be caused by the heterogeneity of clinical information, methods, and statistics. Moreover, an increasing number of publications could be a possibility to reduce the heterogeneity. To uncover the reason of heterogeneity, the sensitive analyses were further performed in all the 6 models. Results showed the estimated OR in the models of allele (C vs. T), codominant (CC vs. TT), dominant (CC+CT vs. TT), and recessive (CC vs. CT+TT) stayed steadily between the lower and upper CI limits after omitting any of one study, suggesting that the data in most models show stable and reliable results. Collecting numerous of literature may increase the sensitivity in models of codominant (CT vs. TT) and over dominant (TT+CC vs. CT).
Publication bias was not detected in Begg’s and Egger’s tests but slightly found in the funnel plot when analyzing the genetic models of codominant (CT vs. TT) and dominant (CC+CT vs. TT). This bias could be due to the fact that researchers in Emamalizadeh et al. did not find significant results in the dominant model (CC+CT vs. TT). In addition, it may suggest that the number of studies included in the meta-analysis was not sufficient. Our meta-analysis suggests that the C allele and CC and CT genotyping of rs16976358 polymorphism might be associated with ASD risk, but this results should be verified in the future with a large number of studies and an increasing sample size.
The mechanism of the correlation between RIT2 rs16976358 and susceptibility to ASD is not well understood. Since SNP rs16976358 is located in the far downstream region of RIT2 gene, one possibility is that SNP rs16976358 may affect gene and protein expression by altering the distant elements [15]. RIT2, as a functional GDP/GTP regulator switch, plays an important role in various signaling pathways within the nervous system. It is involved in EGF receptor- and mSos-mediated signaling pathways and in calcium/calmodulin-mediated cellular processes [18]. Calmodulin regulates numerous Ca2+-dependent enzymes and modulates neuronal functions including synaptic development and plasticity, learning, and memory formation [19, 20]. RIT2 signaling is involved in mediating neurogenesis, neuronal differentiation, and neurite growth and branching, showing an important role in normal neuronal development. Furthermore, RIT2 interacts with dopamine active transporters (DATs) in the dopamine (DA) signaling which is mediated by protein kinase C (PKC) activation [21]. In this regard, dysregulated expression of DAergic RIT2 would impact striatal function and DA-dependent behaviors such as movement, motivation, and addictive behaviors [22]. Thus, RIT2 rs16976358 may disrupt the central dopaminergic system, which is closely related to the ASD pathogenesis [23]. In addition, the investigation of causative linked variants between rs16976358 and ASD susceptibility revealed that the allelic changes may affect the regulatory motif of SOX transcription factor, suggesting another possible mechanism in the emerging etiology of ASD [17]. However, further studies need to be carried out to elucidate this mechanism.
Our meta-analysis has some limitations. Firstly, the publications in English were included in our studies. Secondly, the number of literature and sample size was not large. Thirdly, only Asian populations were analyzed in the current meta-analysis, as the RIT2 rs16976358 polymorphism has not been reported in Caucasian populations. Lastly, despite publication of year and region of patients, we could not analyze the other potential influencing factors due to the insufficiency of data. All these limitations could cause bias in the results, though we try to eliminate the bias.
In conclusion, our meta-analysis indicates that RIT2 rs16976358 polymorphism is significantly associated with ASD in Asian populations. The allele C, CC, and CT genotyping of rs16976358 suggest the risk of ASD. However, more additional studies from different countries and ethnic populations may be helpful to supply a more reliable estimation of the association between rs16976358 and ASD susceptibility.
Data Availability
The data used to support the findings of this study are included within the article.
Conflicts of Interest
The authors declare no competing interest.
Authors’ Contributions
J.W. and H.W. conceived the study, designed the experiments, and interpreted the results. J.W. and S.W performed data collection and analysis. J.Z. inspected the methods and results. J.W. wrote the draft. S.W, J.Z., and H.W. helped in the discussion of the results and revised the study with comments.
Acknowledgments
We thank Prof. Susan Shur-Fen Gau (Department of Psychiatry, National Taiwan University Hospital) and Dr. Xiaoxi Liu (Department of Human Genetics, University of Tokyo) for providing their original data for meta-analysis.