Molecular and Genetic Evidence for Abnormalities in the Nodes of Ranvier in Schizophrenia FREE

During the past decade, multiple lines of evidence (gene and protein expression, genetic association, neuroimaging) have suggested that the disconnectivity syndrome thought to underlie the neurobiology of schizophrenia (SZ) is in part due to abnormalities in myelination and oligodendroglial function.^1- 3 Efficient and fast neurotransmission is dependent on saltatory conduction, which is in turn dependent on the integrity of myelin sheaths, oligodendrocytes, and the nodes of Ranvier (NOR).⁴ The NOR span the entire length of myelinated axons and comprise myelin sheath–free segments where ion (sodium and potassium) exchange can take place, propagating action potentials from the axon initial segment down the length of the axon to synaptic terminals. In the central nervous system, NOR are formed and maintained through reciprocal interactions between neurons and oligodendrocytes.⁵ Multiple and specific neuronal and oligodendroglial proteins such as ankyrin G (AnkG) interact to ensure the high concentration (>1200 voltage-gated sodium ion channels/μm²) and anchoring of voltage-gated sodium ion channels (Nav; Nav1.6 in adults) to the NOR and maintain tight junctions between the axolemma and paranodal myelin loops.⁵ The AnkG protein belongs to a family of scaffolding proteins that interact with several proteins, including the cytoskeletal proteinβIV spectrin and Nav channels.^5- 6 The AnkG and Nav subunits also interact in cis with neurofascin 186 and paranodal contactin.^5- 6 Contactin and Caspr interact in trans with cell adhesion molecules of oligodendroglial paranodal loops that include the neurofascin 155 isoform and Tag1 to ensure the integrity of the myelin-axolemma tight junction.^5- 6 Abnormalities in the expression of oligodendroglial and neuronal genes and proteins in SZ could hamper saltatory conduction by affecting the formation, maintenance, and integrity of the NOR and contribute to the hypothesized disconnectivity syndrome.⁴

The current studies were undertaken to advance the observation of myelin and oligodendrocyte gene and protein expression abnormalities in SZ to the functional domain and to the study of genetic association based on direct neurobiological observations in the brains of persons with SZ with specific gene and protein expression abnormalities. Exploratory analysis of data from a large postmortem microarray study of multiple brain regions of persons with SZ^7- 8 suggested the abnormal expression of many neuron- and oligodendrocyte-enriched genes that encode proteins associated with paranodal junctions and NOR.⁹ We therefore sought to identify the involvement of axoglial interacting genes and proteins in SZ more directly by using postmortem brain tissue from an independent cohort of persons with SZ and unaffected comparison controls.

ANK3 was among the genes that were abnormally expressed in independent gene and protein expression studies. Because recent genome-wide association studies have implicated ANK3 in bipolar disorder (BD)^10- 13 and SZ¹⁴ and because several studies have suggested commonalities in the neurobiology,^15- 17 etiology, and genetic associations^{12- 13,18- 19} between SZ and BD, we also sought to determine whether ANK3 genetic variants affected the expression of ANK3 in the same SZ and control brain specimens and in a larger postmortem collection of racially homogeneous cases and controls (n = 272). The influence of the SZ risk–associated ANK3 polymorphism on cognitive domains of significance to SZ and on brain activity during performance of a working memory task was also examined in healthy controls (n = 513 and n = 52, respectively).

Brain tissue specimens were derived from the Brain Bank of the Department of Psychiatry, Mount Sinai School of Medicine, New York, New York/James J. Peters Veterans Affairs Medical Center, Bronx, New York. The precise tissue-handling procedures have been described in detail.^20- 23 The cause of death, type of antipsychotic medications used, mean age, postmortem interval, tissue pH, and sex distributions of the subjects are shown in eTable 1.

Gray matter from 14 cortical-frontal (Brodmann area [BA] 8, 10, 44, 46, 4), cingulate (BA 23/31, 24/32), parietal (BA 7), temporal (BA 20, 21, 22, 36/28, 38), and occipital (BA 17) brain regions and 3 noncortical brain regions (caudate, hippocampus, and putamen) were used for the Affymetrix HG-U133AB GeneChip gene expression analysis (n = 17-39) (Affymetrix, Santa Clara, California) as described in detail previously.^7- 8,24 Similarly prepared aliquots from the superior temporal gyrus (STG) and primary visual cortex were used in quantitative polymerase chain reaction (qPCR) and quantitative Western blotting analysis in a larger and independent set of brain tissue specimens (eTable 1).^25- 26

Additional studies for genetic association of ANK3 rs9804190 were carried out using case-control samples from the Brain Bank. For analytical purposes, only white American individuals with Western European ancestry (SZ, n = 208; comparison control, n = 64) were used for the clinical genetic investigation of the ANK3 rs9804190 polymorphism.

Learning on Genetics of Schizophrenia Spectrum Cohort and Cognitive Task Performance. We examined the effect of the ANK3 rs9804190 polymorphism using intermediates phenotypes in 530 individuals from the Learning On Genetics Of Schizophrenia Spectrum (LOGOS) cohort. This cohort has been described in detail previously.²⁷

British Cohort and Neuroimaging Data. We further explored the effect of the ANK3 rs9804190 polymorphism on brain activity (on functional magnetic resonance imaging [fMRI]) during a working memory task. Fifty-two healthy volunteers of white British descent were assessed with the Structured Clinical Interview for DSM-IV-TR Axis I Disorders²⁸ and the Family Interview for Genetic Studies²⁹ to exclude any personal psychiatric history or family history (up to second-degree relatives) of SZ or affective disorders. Participants were further screened to exclude past, current, and hereditary medical disorders, DSM-IV lifetime drug or alcohol dependence and drug or alcohol abuse in the preceding 6 months, and contraindications to MRI. For all participants, an estimate of the current full-scale IQ was obtained on the day of scanning using the Wechsler Adult Intelligence Scale–Revised,³⁰ while handedness was based on self-report. Written informed consent was obtained from all subjects before study participation. The study was approved by the Joint Ethics Committee of the Institute of Psychiatry and the South London and Maudsley NHS Foundation Trust.

To assess the effects of neuroleptic exposure on the expression of selected NOR genes of interest, groups of 8 male Sprague-Dawley rats (6-8 months of age) received daily subcutaneous injections of haloperidol (2 mg/kg) or the saline vehicle for 21 days. Ank3, Nfasc, Nrcam, and Scn8a expressions were assessed by the procedures described earlier using rat-specific primers and probes (eTable 2).

Microarray analysis was performed using the Affymetrix HG-U133AB GeneChip set as described previously.^7- 8 Statistical comparisons were made using the GeneSpring GX 7.3.1 cross-gene error model (Agilent Technologies/Silicon Genetics, Santa Clara), based on the deviation from 1.0 algorithm. Filtering conditions were a combination of confidence (P ≤ .05) and fold change (≥1.6) with Benjamini and Hochberg³¹ multiple testing corrections.

The messenger RNA (mRNA) levels of selective NOR genes were measured by qPCR in a larger independent (from the microarray cases and controls) cohort (eTable 1) using TaqMan probes and primer sets (Applied Biosystems, Foster City, California). TaqMan probe identification numbers for selected nodal genes and normalization controls are listed in eTable 2. For relative quantification of mRNA expression, geometric means of the expression of 3 housekeeping genes were calculated using the standard curve method. The housekeeping genes PGK1, PPIA, and RPLPO were selected for their stability after comparison of several different endogenous controls using geNorm (http://medgen.ugent.be/~jvdesomp/genorm/). Two housekeeping genes (Gadph and Ppia) were used as the endogenous references in the rat studies.

Protein abundance was measured in the STG from patients with SZ and comparison control subjects (n = 8/group) previously selected for qPCR analysis using Western blotting. Blots were probed with mouse monoclonal anti-AnkG antibody (IgG₁, 1:200 vol/vol dilution; NeuroMab, Davis, California), which binds specifically to 2 different isoforms (approximately 270 kDa and approximately 250 kDa) and rabbit anti–glyceraldehyde-3-phosphate dehydrogenase antibody (1:5000 vol/vol dilution; LifeSpan Biosciences, Seattle, Washington). Visualization and quantification of bands were performed with the Odyssey 2.1 software (Li-COR Biosciences, Lincoln, Nebraska). To account for gel-to-gel variability, the relative expression value of AnkG and glyceraldehyde-3-phosphate dehydrogenase in each sample was calculated as a ratio between the averaged intensities of the band in the experimental sample and in the standard calibrator (a mix of small aliquots of tissue from all samples).

Samples of DNA from the postmortem cohort were extracted from the STG using the Genomic DNA-Tissue MiniPrep kit (Zymo Research, Irvine, California). We initially genotyped the qPCR cohort (n = 48) for ANK3 rs9804190, rs10994336, and rs10761482 polymorphisms that contribute to SZ and BD risk^10- 11,14 (Polymorphic DNA Technologies, Alameda, California). Genotyping of the LOGOS cohort was performed blind to phenotype measures with a competitive allele-specific PCR system (KBioscience, Herts, England). Brain Bank cohort genotypes were analyzed by Children's Hospital of Philadelphia, Philadelphia, Pennsylvania, using the Affymetrix single-nucleotide polymorphism 6.0 array. Genotyping of the British cohort was determined by a TaqMan allelic discrimination assay (Applied Biosystems, assay C_2584015_10). Genotyping quality control for ANK3 rs9804190 in both the Brain Bank and LOGOS cohorts was performed in more than 10% of the samples by duplicate checking (rate of concordance in duplicates >99%). Call rate was approximately 97% for all polymorphisms. The minor allele frequencies of the ANK3 rs9804190 variant were identical across the different cohorts (eTable 3).

The 3-back task was administered in a block design. Participants were instructed to respond by button press to the target letter. In the baseline (0-back) condition, the designated target letter was X. In the experimental condition (3-back), the target letter was defined as any letter that was identical to the one presented 3 trials back. In each condition, a series of 14 letters were visually presented for 2 seconds each. Responses were monitored via a magnetic resonance–compatible button box held in the subject's dominant hand. There were 12 epochs in all, each lasting 30 seconds, with a total experiment duration of 6 minutes. The ratio of target to nontarget letters presented per block ranged from 2:12 to 3:11. Reaction time to target letters and number of correct responses (accuracy) were recorded.

Both anatomical and gradient-echo echoplanar imaging data were acquired during the same session using a 1.5-T neuro-optimized Signa MRI system (General Electric, Milwaukee, Wisconsin). A total of 150 T2*-weighted echoplanar imaging brain volumes depicting blood oxygenation level–dependent contrast were acquired at each of the 16 axial planes (repetition time, 2000 milliseconds; echo time, 40 milliseconds; flip angle, 70°; slice thickness, 7 mm; slice skip, 0.7 mm; matrix size, 64 × 64; voxel dimensions, 3.75 × 3.75 × 7.7 mm). For the purposes of anatomical localization, a T1-weighted 3-dimensional inversion recovery prepared spoiled gradient recalled sequence was used to acquire structural images in 124 axial planes (repetition time, 1800 milliseconds; echo time, 5.1 milliseconds; inversion time, 450 milliseconds; flip angle, 20°; slice thickness, 1.5 mm; matrix size, 256 × 192; field of view, 240 × 180 mm; voxel dimensions, 0.9375 × 0.9375 × 1.5 mm; number of excitations, 1).

Multiple statistical procedures were used for different aspects of the study. Comparison of the demographic variables was performed using separate 1-way analyses of variance or the nonparametric Mann-Whitney U test as appropriate, based on deviation from normality. A 2-tailed t test was used to compare the relative abundance of AnkG protein and the relative ANK3 mRNA expression of NOR genes in qPCR and genotyping experiments using SPSS version 17 statistical software (SPSS Inc, Chicago, Illinois).

We used QTPHASE software (http://www.mrc-bsu.cam.ac.uk/personal/frank/software/unphased/) from the UNPHASED package³² version 3.1.4 for the analysis of genotype associations in the LOGOS cohort and ANK3 expression levels in the qPCR cohort. The ANK3 rs9804190 polymorphism was examined for allelic association with SZ in the Brain Bank cohort with PLINK software³³ (http://pngu.mgh.harvard.edu/~purcell/plink/). Quality control for the Brain Bank cohort was done as described previously¹⁰; correction for population stratification was achieved using the eigenstrat method.³⁴ Logistic regression analysis was carried out to evaluate the association of rs9804190 with SZ. P values from the UNPHASED and PLINK analyses were corrected for multiple testing by running 100 000 permutations of the data. Additionally, to lower the possibility of type I statistical error in the LOGOS analysis, we set the level of significance to .01.

For the fMRI data analysis, we used the Statistical Parametric Mapping (SPM5) software (http://www.fil.ion.ucl.ac.uk/spm/software/spm5) implemented in MATLAB 7.1 (MathWorks Inc, Natick, Massachusetts) for data processing and analysis. Preprocessing of the images involved realignment to correct for movement, normalization into Montreal Neurological Institute space using the participant's structural MRI images, and smoothing with an 8-mm full-width half-maximum isotropic gaussian kernel. A first-level fixed-effects model was computed for each participant. The fMRI responses were convolved with a canonical hemodynamic response function with the vectors of interest and a high-pass filter (128 seconds) was applied to remove low-frequency noise. A hemodynamic response function–shaped low-pass filter was used to filter low-frequency noise artifacts. Contrast images representing the 3-back vs 0-back conditions were produced for each subject and entered in the second-level, random-effects, 2-sample t test analysis (group 1: CC; group 2: CT and TT). Results were considered significant at P < .05. False discovery rate corrected for multiple comparisons. Stereotactic coordinates of the peak maxima of the suprathreshold clusters were converted from the Montreal Neurological Institute spatial array to that of Talairach and Tournoux (http://imaging.mrc-cbu.cam.ac.uk/imaging/MniTalairach), and corresponding BAs were identified with the Talairach Daemon Client (http://www.talairach.org). Measures of brain activation (weighted parameter estimates) from the suprathreshold cluster were extracted with the Marsbar toolbox (http://marsbar.sourceforge.net) to examine the effect size of the genotype on the level of activation, expressed as Cohen d.

We initially conducted an exploratory analysis using microarray data (contrast analysis t scores⁷) from 17 different brain regions representing all 4 cortical lobes, the hippocampus, and basal ganglia in persons with SZ (n = 9-21) vs unaffected comparison controls (n = 8-18) (eTable 1). We found significant dysregulation in the expression of genes known to participate in the development, organization, and maintenance of the NOR across all brain regions (Table 1). The majority of the genes, classified by gene ontology analysis as associated with cell adhesion, were downregulated in the assayed brain regions of persons with SZ.

Eight genes known to be involved in the organization and maintenance of NOR were selected from those shown in Table 1 for qPCR analysis in an independent and larger sample of well-matched cases (n = 22) and controls (n = 24). The expression of these genes was assessed in the STG and primary visual cortex (eTable 1 and eTable 2), brain regions where we found significant and minimal alterations of genes expression, respectively, in our microarray analysis (Table 1). Other relevant genes (eg, Nav1.6) that were expressed at too low a level for detection by microarray were also included in this analysis. As shown in Table 2, the STG of this independent cohort of persons with SZ evidenced significantly reduced mRNA expression of ANK3 (P = .01; fold change [FC] = −1.25) (Figure 1), neurofascin (NFASC) (P = .02; FC = −1.33), neuronal cell adhesion molecule (NRCAM) (P = .01; FC = −1.3), and Nav1.6 sodium channel,α subunit (SCN8A) (P = .04; FC = −1.34) and a trend for contactin 2 (CNTN2 or TAG1) (P = .08; FC = −1.3). These changes were gene specific given that the mRNA levels of other nodal proteins, including contactin 1 (CNTN1), contactin-associated protein-like 2 (CNTNAP2), or Nav1.2 sodium channel,α subunit (SCN2A), did not show significant differences among persons with SZ and comparison controls (all P > .19). Brain region specificity was evident from the lack of changes in any of these genes in the primary visual cortex (all P > .12). Sample pH, postmortem interval, or age of the donors did not alter the outcome of the analysis when entered into statistical analyses as covariates.

All of the persons with SZ included in these studies had received antipsychotic medications for many years. To test whether the expression of axoglial junction genes is affected by the classic and most frequently prescribed antipsychotic medication in this cohort, the expression levels of Ank3, Nrcam, Nfasc, and Scn8a were measured in the cerebral cortex of rats receiving long-term treatment with haloperidol and compared with saline vehicle–treated controls. Three weeks of relatively high-dose (2 mg/kg/d) treatment with haloperidol did not influence the cerebral cortical expression of the examined nodal genes (all P > .50.).

Based on the multiple lines of evidence described earlier supporting decreased ANK3 mRNA expression levels, the protein levels of AnkG in STG were measured in a subset of individuals with SZ (n = 8) vs comparison subjects (n = 8) well matched for age, sex, pH, and postmortem interval from the larger set of subjects used in the qPCR studies described earlier (eTable 1). As shown in Figure 1, the levels of AnkG protein were significantly decreased (P = .03) in the STG of persons with SZ relative to controls.

Given the association of ANK3 with BD^10- 13 and SZ¹⁴ in genetic studies and the evidence of ANK3 gene and AnkG protein expression deficits in the brain of persons with SZ, we determined whether ANK3 genetic variants are associated with ANK3 gene expression in the STG. Because of the low frequency of minor (T) allele homozygotes, carriers of the minor allele (C/T and T/T) were grouped together. A significant main effect of genotype was observed for rs9804190 in the STG with ANK3 gene expression in individuals with SZ (P = .03). Among persons with SZ, those who were homozygous for the C allele had approximately 27% lower levels of ANK3 expression compared with individuals with at least 1 minor T allele (Figure 1). The UNPHASED analysis revealed that the rs9804190 C allele was associated significantly (P = .006) with lower ANK3 expression in the STG of individuals with SZ; the estimated additive genetic value between the C and T alleles was−4.17. A significant effect of the C/C allele in controls was not found (P = .82). No significant effect was observed for other ANK3 -associated single-nucleotide polymorphisms,^10- 13 rs10994336 or rs10761482, in the STG (all P > .18) or any of the ANK3 genotypes in the primary visual cortex (all P > .13).

Based on the association of the ANK3 rs9804190 C allele with lower ANK3 expression levels and the finding of decreased ANK3 expression in SZ, the association of the rs9804190 C allele, which increases the risk for SZ, was examined in a case-control study. Using a larger postmortem cohort of white persons with SZ (n = 208) relative to controls (n = 64), the study revealed a significant association of the ANK3 rs9804190 C allele with SZ (permuted P = .03; odds ratio = 1.77; 95% confidence interval, 1.08-2.88).

We further examined whether the ANK3 rs9804190 C allele is associated with poorer performance in SZ-related behavioral assays. Therefore, the association of the rs9804190 C allele with performance on a test of cognition sensitive to SZ was investigated in 513 demographically homogeneous, healthy, young, white individuals without psychopathology (LOGOS cohort²⁷). There were no differences in demographic characteristics based on ANK3 rs9804190 genotype (eTable 4). Table 3 shows the association of the ANK3 rs9804190 polymorphism with a broad array of phenotypic cognitive function measures as revealed by QTPHASE analysis. The rs9804190 C allele was associated with statistically significant (P < .01) lower correct responses on the 2-back and 3-back conditions of the N-back sequential letter task, fewer completed categories and an increased number of total and perseverative errors during the Wisconsin Card Sorting Test, and lower self-directedness and higher self-transcendence scores on the Temperament and Character Inventory. Results were not altered when age, education, smoking status (cigarettes per day), and IQ were entered as covariates. There were no significant associations with baseline acoustic startle, prepulse inhibition of the startle response, and other cognitive tasks or personality dimensions, with or without covarying for age, IQ, and smoking status (cognitive tests and prepulse inhibition: all P > .05).

The final experiment used blood oxygenation level–dependent fMRI to examine the effect of the ANK3 rs9804190 polymorphism in an independent cohort of healthy volunteers (n = 52). Because of the small number of TT homozygotes, all carriers of the T variant were grouped together for further analysis. Homozygotes of the ANK3 rs9804190 risk C allele compared with the T allele carriers showed increased activation (as measured by fMRI) in 2 prefrontal clusters centered in the left inferior frontal gyrus (BA 11/47; x = −36, y = 40, z = −2; voxels = 218; z score = 3.34) and the left middle frontal gyrus (BA 45/46; x = −46, y = 30, z = 20; voxels = 114; z score = 3.15) during performance of the 3-back version of the N-back working memory task (Figure 2). There was no effect of genotype on demographic or task performance variables or IQ (all P > .53) (eTable 5).

These results provide insights into the mechanistic processes and genetic substrates associated with the frequently observed myelin- and oligodendrocyte-related deficits in SZ.^1- 2 They point to an abnormality in the expression of genes and proteins associated with the integrity of the NOR and by extension with the efficiency of neurotransmission. The gene expression results are strengthened by replication in multiple cohorts and diverse methods and by the observation of a novel genetic association of at least 1 of the key proteins of the NOR, AnkG, with SZ and with cognitive functions known to be impaired in SZ. Although the ANK3 single-nucleotide polymorphism is within a noncoding region of the ANK3 gene, the gene expression findings indicate an association not only with SZ but also with reduced gene expression in the STG. The genetic association findings are strengthened by their consistency across multiple study groups and by the fact that the ANK3 genetic association with SZ is also known to be associated with increased risk for BD.^10- 13 There is evidence suggestive of etiological convergence between SZ and BD^12,18- 19 consistent with these results.

The molecular mechanism(s) of the ANK3 genetic susceptibility for SZ contributes to reduced gene and protein expression of ANK3 /AnkG, with disease-relevant anatomic specificity. The expression of ANK3 and other nodal genes is reduced in the STG, a brain region that has been repeatedly shown to be vulnerable in SZ, but not in the visual cortex where SZ-dependent abnormalities are rarely observed.^8,36 The current observations derived from a specific evidence-based hypothesis testing approach: preliminary findings from multiregional microarray expression studies of oligodendrocyte- and myelin-associated deficits in SZ led to hypothesized abnormalities of the NOR. This hypothesis was tested and supported by independent gene and protein expression experiments. The observation of abnormal expression of known NOR-enriched genes and proteins led directly to genetic association studies and linkage of genetic association with cognitive, behavioral, and neuroimaging intermediate phenotypes repeatedly linked to SZ.

The mRNA expressions of the NOR-enriched proteins ANK3, NFASC, NRCAM, SCN8A, and CNTN2 were all significantly downregulated in the STG of persons with SZ. Increasing evidence points to a pivotal role of AnkG in the initial assembly and concentration of Nav channels and cell adhesion molecules to central nervous system NOR.^5- 6 In addition, a significant role for cell adhesion molecules in Nav channels clustering has been demonstrated in in vitro experiments where treatments with antibodies against the cell adhesion molecules lead to a failure of both Nav channels and AnkG to accumulate at the NOR.^5- 6 That many of these proteins demonstrated to be crucial to the normal functioning of the NOR and to saltatory conduction are abnormally expressed in SZ and the genetic linkage of ANK3 with SZ attest to their importance to the neurobiology of the disease. However, despite significant recent advances, the mechanisms of formation and maintenance of the NOR remain poorly understood and the full complement of known genes and proteins that contribute to their assembly, maintenance, and function is still incomplete.^5- 6 One speculative mechanism is that NOR abnormalities will affect the conduction velocity, leading to attenuated, slow, abnormally variable or absent action potentials.⁴ Indeed, in an ANK3 knockout mouse model, the generation and persistence of action potentials were dramatically inhibited.³⁷ Such abnormalities will limit the distance over which cortical neurons can fire in synchrony, affecting the normal long-rangeγ-oscillations between cortical subfields that could lead to attention deficits.⁴

AnkG and many of the proteins whose mRNA expressions were studied here are not exclusively localized to the NOR. AnkG is also essential for the localization of many membrane proteins to the axon initial segment,³⁸ including the Nav channels that are required for action potential generation.³⁷ Cruz et al³⁹ showed that the density of the AnkG-immunoreactive axon initial segment in subjects with SZ was significantly reduced in superficial, but not in deep, cortical layers compared with both healthy comparison subjects and subjects with major depressive disorder. This might provide an additional mechanism in which AnkG and Nav clustering abnormalities impair action potential generation, disrupting neurotransmission and the capacity of some neurons to fire in the repetitive and synchronous fashion required for cortical network oscillations.

While the ANK3 single-nucleotide polymorphisms identified are within noncoding regions of the gene, they nevertheless support a role for ANK3 in susceptibility to SZ and performance in cognitive measures. Although our strategy of a deductive, direct, postmortem neurobiology study minimized serendipitous findings, it did involve multiple statistical tests. Thus, the potential for spurious associations because of multiple testing is an important consideration. However, a consistent pattern of ANK3 rs9804190 association was found in healthy subjects, involving cognition and functional imaging, and in ANK3 gene and protein expression levels in brain tissue. The likelihood that the same ANK3 risk allele was consistently associated with illness or with illness-associated abnormalities across multiple geographically diverse samples by chance seems small.

ANK3 has recently been shown to be a susceptibility gene for BD^10- 11 and SZ,¹⁴ further supporting the notion of shared genetic vulnerabilities between BD and SZ.^12,18- 19 However, to our knowledge, there has been no previous report of an association of the ANK3 rs9804190 variant with SZ. This may reflect the fact that P values reported in this a priori hypothesis–driven postmortem brain-based genetic association study would not be significant after genome-wide association study correction for hypothesis-free multiple testing.

The absence of an ANK3 genetic variant effect on ANK3 mRNA expression levels in healthy individuals might be due to a lack of power. In patients with SZ, additive effects such as epistasis or environmental factors might augment the effect size of ANK3 genetic variants. We identified an effect of the rs9804190 C allele in a larger population of healthy subjects (LOGOS cohort) and using a sensitive approach (neuroimaging cohort), further supporting the notion of inadequate power.

Antipsychotic medication exposure is a confounding factor in postmortem studies of SZ and therefore can contribute to the detected gene expression abnormalities of the axoglial genes. We did not find an altered expression level in the cerebral cortex of rats treated with haloperidol. However, the possibility of antipsychotic medication exposure confounds cannot be entirely ruled out, as haloperidol exposure in rats might not exactly mimic the neuroleptic-induced changes in the brain of patients with SZ. Nevertheless, never-medicated, healthy, and asymptomatic ANK3 risk allele carriers present with poorer working memory and executive function, indices of reduced functionality (self-directedness), and increased prefrontal activity during a working memory task. These findings indicate a penetrant mechanism of the ANK3 gene in behavior that might increase the risk for SZ through poor cognition. Collectively, these findings identify SZ-associated abnormalities in NOR genes and proteins and provide significant evidence for neurobiological mechanisms and functions that might be key to the pathophysiology of SZ and to the development of targeted therapeutic interventions.