Bipolar Disorder (BD) is one of the most prevalent mental illnesses characterized by mood swings between episodes of mania or hypomania and depression (McIntyre et al., 2020). Genetic factors play a critical role in BD's etiology (Craddock and Sklar, 2013), supported by the estimated heritability as high as ~80% (McGuffin et al., 2003). Family, twin, and adoption studies show that the lifetime prevalence of BD is ~5–10% among first-degree relatives (Lichtenstein et al., 2009) and ~40–70% among monozygotic twins of patients (Bertelsen et al., 1977; Kieseppä et al., 2004). Executive functions are higher-order cognitive processes that enable an individual to make complex decisions and process goal-oriented thoughts and behavior (Jurado and Rosselli, 2007; Miyake and Friedman, 2012). Executive function deficits have been well-reported in BD (Torres et al., 2007; Hellvin et al., 2012; Samamé et al., 2014; King et al., 2019) and their non-affected first-degree relatives (Bora et al., 2008; Kosger et al., 2015), suggesting that executive function deficits are inheritable features in BD (Frantom et al., 2008; Miskowiak et al., 2017). Considering that executive function is genetically less complicated than the disorder as a whole, it may help to understand the mechanism underlying BD by studying the effect of BD-related genes on executive function.

The T allele of the intronic single nucleotide polymorphism (SNP) rs10994336 within the ANK3 gene are considered a risk allele for BD in genome-wide association studies (GWAS) (Ferreira et al., 2008; Scott et al., 2009; Psychiatric GWAS and Consortium Bipolar Disorder Working Group, 2011; Mühleisen et al., 2014; Stahl et al., 2019). Ankyrin 3 (ANK3) encodes Ankyrin-G, a cytoplasmic adaptor protein primarily located at the axonal initial segments and the nodes of Ranvier (Kordeli et al., 1995). Ankyrin-G plays many roles in neurotransmission, ion channel stabilization, cortical development (Durak et al., 2015), myelination (Ching et al., 1999), and neurogenesis in the adult brain (Paez-Gonzalez et al., 2011). AnkG hemizygous mice generated by the gene trapping approach demonstrated obvious cognitive impairment in an animal model study (Liu et al., 2017) and knockdown of ANK3 in drosophila is associated with memory deficits (Iqbal et al., 2013). Several studies have examined the association between rs10994336 polymorphism and executive functions in BD, a study of 49 patients with BD observed no effect of the rs10994336 genotype on executive function, memory, or general intelligence (Hori et al., 2014) and another study on 47 patients with BD either detected no association between the rs10994336 genotype and executive function (Ruberto et al., 2011). Since BD is a complex disorder and is influenced by a number of genetic variants and environmental exposures, the effect of a single risk allele on phenotype is small, thus, the larger the sample size is, the more power to detect the effect. Considering that the sample size included in those studies were relatively small, further larger sample studies are required to examine whether rs10994336 polymorphism could influence executive functions in BD.

Epigenetic modification including DNA methylation may play a critical role in BD's pathogenesis (Rakyan et al., 2011; Fries et al., 2016; Greenberg and Bourc'his, 2019). Both candidate studies and genome-wide methylation approaches using brain or blood samples have demonstrated promising methylation alterations associated with BD (Gürel et al., 2020). Many risk loci associated with BD are located in non-coding regions of the genome, suggesting that gene regulation plays a role in the disease pathology. Furthermore, different genotypes at a certain locus can alter DNA methylation pattern through allele-specific methylation (Schalkwyk et al., 2010; Bell et al., 2011). These sites are named methylation quantitative trait loci (meQTLs) and have been found throughout the genome in multiple tissues (Gibbs et al., 2010; Schalkwyk et al., 2010; Bell et al., 2011). Interestingly, it has been observed that meQTLs overlap with risk SNPs associated with diseases like BD, schizophrenia (SCZ), and Alzheimer's disease (Gamazon et al., 2013; Gaunt et al., 2016; Hannon et al., 2016). A previous study observed a significant overlap of meQTL between ancestral groups, developmental stages, and tissue types (Smith et al., 2014). Another study also suggests that DNA methylation contains a significant heritable component that is highly stable across the lifespan (Gaunt et al., 2016). These converging results shed light on the interpretation of genetic variants in non-coding regions that the influence of genetic variation on phenotype is potentially mediated through the allele-specific epigenetic processes.

In this study, we sought to study the associations between the rs10994336 variant of ANK3 and executive functions in patients with BD and healthy controls (HCs), and examine whether rs10994336 polymorphism functions as a meQTL and changes the site methylation to mediate the influences of rs10994336 on executive functions.

In the present study, we found that the risk T-allele carriers of rs10994336 showed a worse WCST performance compared to those with homozygous CC genotype in the BD group. Moreover, rs10994336 affects the methylation level of specific CpG site within ANK3 as a meQTL in the BD and HC groups, which mediates the effect of rs10994336 on executive functions. To the best of our knowledge, it is the first study to document the mediation effect of ANK3 methylation between BD-related genetic variant rs10994336 within ANK3 and executive functions in BD.

Our results suggest that the risk allele of rs10994336 of ANK3 is associated with a poor executive performance in patients with BD after adjusting for age, gender, and education. This is consistent with the result of a previous study conducted among 173 patients with first-episode psychosis showing that the polymorphism rs1938526 of ANK3 gene, which was in high linkage disequilibrium with rs10994336, was associated with cognitive domains of reasoning and problem solving, though at a marginal significance (Cassidy et al., 2014). Moreover, the loss of association between rs10994336 and executive functions in the HC group was consistent with two previous studies which detected no effect of this SNP on measures of executive function either in HC individuals (Hori et al., 2014) or in healthy males (Roussos et al., 2011). However, two studies on BD failed to examine the association between rs10994336 and executive function, memory, or general intelligence (Ruberto et al., 2011; Hori et al., 2014). It should be noted that the sample size included in those studies were relatively small which reduced the statistical power to test the association. Considering the complex interplay between genetic and environmental factors in the etiology contributing to BD, we speculate that the association between rs10994336 and executive function is predominant in patients with BD, who have a predisposition to severe cognitive deficits due to the complex pathophysiological processes induced by genetic factors. Studies with larger samples are needed to test whether these findings can be replicated.

The polymorphism rs10994336 is located in the intron of ANK3 gene, and its function was unclear until now. We tested whether this SNP affected the methylation level of ANK3. Considering variables such as age, gender, and medication use were believed to influence DNA methylation (Burghardt et al., 2020; Gardea-Resendez et al., 2020; Webb et al., 2020), we add these confounding factors in our linear regression model as covariates to explore the allele-associated methylation. After adjusting these confounders, the effect of rs10994336 polymorphism on methylation of the CpG site cg02172182 within the gene body of ANK3 was observed in HC and replicated in patients with BD in the same direction where the risk T-allele carriers presented a reduced methylation relative to homozygous CC participants, suggesting that the effect of genotype on methylation is not driven by the disease and rs10994336 could be referred to as a reliable meQTL. Cg02172182 locates at the gene body region of ANK3 and ~1.5 kb upstream of rs10994336, which functions by cis-acting meQTL that can act over a range of distances, from a few base pairs to distances of over 500 kb. Our mediation analysis model in BD supports the theory that methylation acts as a mediator between genotype and phenotype, as the effect of rs10994336 polymorphism on PE was significantly mediated by the methylation of cg02172182. Although the methylation patterns are highly heterogeneous between different tissue types and at different life stages, genetic effects on methylation are found to be rather stable across life course and different tissue types (Smith et al., 2014; Gaunt et al., 2016). A previous study identified separately and compared genome-wide meQTLs from brain prefrontal cortex, whole blood, and saliva and observed significant overlap of meQTLs among tissues (Smith et al., 2014). Moreover, another study also implicated that the DNA methylation status of many CpG sites in the brain were mirrored in the blood and that common meQTLs were also detected between these two tissues (Gaunt et al., 2016). These converging findings support the theory that blood tissue could be used as a surrogate tissue to explore disease relevant processes within the brain. In addition, multiple strands of evidence indicate that DNA methylation plays an important role in several neurobiological processes including neurogenesis (Costello, 2003), neural plasticity (Dulac, 2010), and the formation and maintenance of memories (Halder et al., 2016), which in turn impacts on executive function. Prompted by these findings, it is valid to speculate that the function of rs10994336 as a meQTL in whole blood tissue may mirror the function in brain tissues which influences the methylation of ANK3 gene and further results in executive function impairment in patients with BD.

There are several limitations to the present study. First, the sample size of our study did not have sufficient power to detect the difference of frequencies of rs10994336, or the difference of methylation of ANK3 between BD and HC. Since rs10994336 is a well-established BD risk variant, the difference of genotype or allele frequencies between BD and control will lead to the difference of methylation of this gene, which will have an effect on brain function. This may explain why we only detected the association of rs10994336 with executive functions in BD. Second, the actual effect of rs10994336 polymorphism on ANK3 methylation in brain tissues in which the pathological processes for psychiatric disorder are likely involved could not be directly observed in our study, since brain tissues are not readily accessible in living patients. However, research findings indicate a common overlap of meQTLs between different tissue types including blood and brain tissues. Third, we did not stratify our samples according to those confounding factors such as age, medication use, and disease status (manic, depressive, and remission) which could affect DNA methylation considering that our sample size was not large enough. However, age and medication use were adjusted in our linear regression models and the effect of rs10994336 polymorphism on cg121722182 methylation remained significant. Future studies with larger samples could further exclude those confounders. Finally, a potential limitation is that the precise mechanism by which methylation of cg121722182 influences executive functions is unclear since it locates at the gene body of ANK3 while the role of methylation in gene bodies is not well-established, but some evidence indicates that methylation in gene bodies is associated with the activation of genes in contrast to the methylation observed in promoter regions which are generally considered to be involved in the content of silence of genes (Greenberg and Bourc'his, 2019). Future studies are needed to address this question.

In summary, using executive functions as a highly heritable endophenotype and combining genetic and epigenetic data, we found that allelic variation of rs10994336 in ANK3 impacts executive functions by moderating methylation, supporting the theory that methylation acts as a mediator between genotype and phenotype. This provides insights into the mechanism of how intron risk variants associated with BD contribute to disease susceptibility.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Ethics Committee of First Affiliated Hospital of China Medical University. Written informed consent to participate in this study was provided by the participants' legal guardian/next of kin.

YT and FW contributed to the conception and design of the study and provided the technical support. LT performed the data analyses and wrote the manuscript. LT, JL, YW, YZ, JD, and YC contributed to acquisition of data and manuscript preparation. XG helped perform the analysis with constructive discussions. All authors contributed to the article and approved the submitted version.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.