ANK3 is among the most frequently reported risk genes in bipolar disorder, 1 with further evidence for association with schizophrenia, Reference Ripke, Sanders, Kendler, Levinson, Sklar and Holmans2 suggesting a common genetic mechanism in these two clinically related disorders. ANK3 sequence variants have been associated with clinical characteristics of bipolar disorder and schizophrenia, including impaired cognitive function Reference Roussos, Katsel, Davis, Bitsios, Giakoumaki and Jogia3 and mood dysregulation. Reference Roussos, Giakoumaki, Georgakopoulos, Robakis and Bitsios4 Gene variants are also associated with disturbed brain connectivity, Reference Linke, Witt, King, Nieratschker, Poupon and Gass5 and ankyrin-3 is expressed throughout the brain, Reference Rueckert, Barker, Ruderfer, Bergen, O'Dushlaine and Luce6 with functions related to clustering of sodium and potassium channels, axodendritic polarity and action potential generation. Despite indications of altered gene expression, Reference Roussos, Katsel, Davis, Bitsios, Giakoumaki and Jogia3,Reference Kato, Hayashi-Takagi, Toyota, Yoshikawa and Iwamoto7 it is still unclear whether ANK3 mRNA levels are altered in bipolar disorder and schizophrenia. Here we determined ANK3 mRNA levels in peripheral blood in individuals with bipolar disorder, with schizophrenia and in healthy controls. We also investigated potential cis-acting expression quantitative trait loci (eQTLs) in the ANK3 gene.
Method
Our sample (total n = 779) comprised 227 patients with bipolar disorder (type 1 n = 147, type 2 n = 65 and bipolar disorder not otherwise specified n = 15) (BD group), 273 patients with schizophrenia (schizophrenia n = 209, schizophreniform disorder n = 19 and schizoaffective disorder n = 45) (SZ group) according to the Structured Clinical Interview for DSM-III-R (SCID), Reference Spitzer, Williams, Gibbon and First8 as well as 279 healthy controls. All participants were of White Northern European ethnicity. See online data supplement and Table DS1 for further details.
Quantitative reverse transcription polymerase chain reaction was performed using TaqMan Gene Expression Assays (ANK3: Hs00241738_m1; Life Technologies Corporation, Carlsbad, California, USA) and the gene GUSB (glucuronidase beta) as an endogenous control. Genotyping was performed in a subset of 685 individuals with Affymetrix Genome-Wide Human SNP array 6.0 (Affymetrix Inc., Santa Clara, California, USA) and additional imputation conducted with MACH using the European samples available in the Phase I release of the 1000 Genomes project (www.sph.umich.edu/csg/abecasis/MACH/download/1000G-PhaseI-Interim.html). Further information on RNA measurement, genotyping procedures and imputation is presented in the online supplement.
Analysis of covariance (ANCOVA) was selected to investigate ANK3 mRNA level differences between participants with bipolar disorder, those with schizophrenia and healthy controls in the IBM SPSS software package for Windows, version 20. ANK3 mRNA was entered as a dependent variable and diagnostic spectrum (bipolar disorder, schizophrenia and healthy controls) as an independent variable, with age and gender as covariates. To further elucidate the differences in mRNA level between participants with bipolar disorder, those with schizophrenia and healthy controls, logistic regression was performed with mRNA, age and gender as predictors. Additionally, ANCOVA analyses were performed in males and females separately for main diagnostic spectra, as well as in the total sample for the diagnostic subcategories. Association analyses between 2398 imputed ANK3 single nucleotide polymorphisms (SNPs) and mRNA levels in a subset of 685 participants were performed with a linear regression model, with gender and age as covariates, using PLINK (version 1.07; http://pngu.mgh.harvard.edu/purcell/plink/). Additional information is presented in the online data supplement.
Results
The mRNA levels differed significantly between the BD, SZ and healthy control groups (P = 2.8 × 10-4; Fig. 1 and online Fig. DS1). The effect size (partial eta squared) was 0.021 for diagnostic spectrum (online Table DS2). The post hoc logistic regression tests showed significantly higher mRNA levels in the BD group (P = 9.0 × 10-5, odds ratio (OR) = 2.29) and the SZ group (P = 0.010, OR = 1.73) compared with the healthy control group, but there was no significant difference between the BD and SZ group (P = 0.13, OR = 1.37). Results from the post hoc logistic regression analysis are presented in online Table DS3. The results remained largely the same when analysed in males and females separately (online Table DS4, Table DS5 and Fig. DS2). With respect to the diagnostic subcategories, there were significantly higher mRNA levels in bipolar type 1 disorder, bipolar type 2 disorder and schizophrenia compared with healthy controls, and significantly higher mRNA levels in bipolar type 2 disorder than in schizophrenia and schizophreniform disorder (online Table DS6, Table DS7 and Fig. DS1). ANK3 SNPs close to one of the transcription start sites were significantly associated with ANK3 mRNA levels in the total subsample as well as in the BD, SZ and healthy control groups separately (online Table DS8 and Fig. DS3).
ANK3 mRNA expression levels (mean (s.d. = 1) value) according to diagnostic spectrum.
ANK3 mRNA levels are significantly higher in bipolar spectrum disorder (BD) (P = 9.0 × 10–5, odds ratio (OR) = 2.29) and schizophrenia spectrum disorder (SZ) (P = 0.010, OR = 1.73) compared with healthy controls (CTR).
Discussion
We found ANK3 gene expression levels to be significantly increased in the BD and SZ groups compared with healthy controls. This effect was more pronounced for bipolar disorder than schizophrenia, with higher ANK3 expression in the affective end (bipolar type 2 disorder) than in the schizophrenia end of the psychosis spectrum (schizophrenia and schizophreniform disorder) (Table DS6, Table DS7 and Fig. DS1). Additionally, potential cis-acting eQTLs were identified. These results are in line with sequence variant findings, as ANK3 SNPs have been associated with both bipolar disorder and schizophrenia, albeit with stronger evidence in the former. Thus, increased ANK3 mRNA might be involved in a common pathophysiological mechanism underlying these disorders, supporting a continuum model of psychosis. This is in accordance with a finding of elevated ANK3 mRNA levels in peripheral blood in bipolar disorder, Reference Kato, Hayashi-Takagi, Toyota, Yoshikawa and Iwamoto7 although post-mortem brain findings of down-regulation of ANK3 mRNA in superior temporal gyrus in individuals with schizophrenia (n = 46) has also been reported. Reference Roussos, Katsel, Davis, Bitsios, Giakoumaki and Jogia3
Measurement of ANK3 mRNA in peripheral blood may not correlate with central nervous system (CNS) expression, but there is evidence for correlation between levels of transcripts expressed in blood and multiple CNS tissues. Reference Sullivan, Fan and Perou9
The effect sizes for SNP associations with psychiatric disorders are very low (OR<1.2), providing no clinical relevance. The current effect sizes for ANK3 mRNA level in bipolar disorder (OR = 2.29) and schizophrenia (OR = 1.73) were larger, indicating that mRNA level is a more robust marker than sequence variants for further assessment of underlying mechanisms in psychotic disorders. ANK3 mRNA level might be useful for investigations of the relevance of the ANK3 gene in brain networks, pharmacological interventions and clinical phenotypes in bipolar disorder and schizophrenia.
Acknowledgements
We thank the patients and controls for their participation in the study, and the health professionals who facilitated our work. We also thank Thomas D. Bjella for assistance with the database and Lars J. Hansson for performing the RNA measurement procedures.
eLetters
No eLetters have been published for this article.