Abstract
Objective
Transcriptomic biomarkers of psychiatric diseases obtained from a query of peripheral tissues that are clinically accessible (e.g., blood cells instead of post-mortem brain tissue) have substantial practical appeal to discern the molecular subtypes of common complex diseases such as major psychosis. To this end, spliceome-profiling is a new methodological approach that has considerable conceptual relevance for discovery and clinical translation of novel biomarkers for psychiatric illnesses. Advances in microarray technology now allow for improved sensitivity in measuring the transcriptome while simultaneously querying the “exome” (all exons) and “spliceome” (all alternatively spliced variants). The present study aimed to evaluate the feasibility of spliceome-profiling to discern transcriptomic biomarkers of psychosis.
Methods
We measured exome and spliceome expression in peripheral blood mononuclear cells from 13 schizophrenia patients, nine bipolar disorder patients, and eight healthy control subjects. Each diagnostic group was compared to each other, and the combined group of bipolar disorder and schizophrenia patients was also compared to the control group. Furthermore, we compared subjects with a history of psychosis to subjects without such history.
Results
After applying Bonferroni corrections for the 21,866 full-length gene transcripts analyzed, we found significant interactions between diagnostic group and exon identity, consistent with group differences in rates or types of alternative splicing. Relative to the control group, 18 genes in the bipolar disorder group, eight genes in the schizophrenia group, and 15 genes in the combined bipolar disorder and schizophrenia group appeared differentially spliced. Importantly, thirty-three genes showed differential splicing patterns between the bipolar disorder and schizophrenia groups. More frequent exon inclusion and/or over-expression was observed in psychosis. Finally, these observations are reconciled with an analysis of the ontologies, the pathways and the protein domains significantly over-represented among the alternatively spliced genes, several of which support prior discoveries.
Conclusions
To our knowledge, this is the first blood-based spliceome-profiling study of schizophrenia and bipolar disorder to be reported. The battery of alternatively spliced genes and exons identified in this discovery-oriented exploratory study, if replicated, may have potential utility to discern the molecular subtypes of psychosis. Spliceome-profiling, as a new methodological approach in transcriptomics, warrants further work to evaluate its utility in personalized medicine. Potentially, this approach could also permit the future development of tissue-sampling methodologies in a form that is more acceptable to patients and thereby allow monitoring of dynamic and time-dependent plasticity in disease severity and response to therapeutic interventions in clinical psychiatry.
Keywords: Gene expression, spliceome-profiling, blood-based biomarkers, psychosis, CNS disease diagnostics
INTRODUCTION
Schizophrenia and bipolar disorder are two of the most prevalent mental health disorders, collectively affecting approximately 1.5% of the population at some point in their lifetime [1, 2]. These are also among the most severe and debilitating disorders with substantial public health importance, typically developing in late adolescence or early adulthood and often imposing near-lifelong disability on affected individuals, as well as considerable strain on their caretakers and society at large. One way to combat schizophrenia and bipolar disorder would be to discover biological markers - or “biomarkers” - for these illnesses, which potentially could revolutionize their rational diagnosis and management. Conceivably, biomarkers could expedite and standardize the process of primary and differential diagnosis, which presently involves considerable time, effort, and uncertainty. Biomarkers also might allow for earlier identification of affected individuals, in turn hastening their receipt of effective (perhaps personalized) treatment and improving prognoses. Furthermore, biomarkers could form the basis for early intervention and prevention efforts targeting at- risk individuals, which might reduce the morbidity and prevalence of these crippling disorders. Collectively, these advances would translate into an enormous improvement in global public health.
Given such immense promise, biomarkers for schizophrenia and bipolar disorder have been pursued through many decades and approaches. As both disorders are thought to have relatively high heritability (~80%) [3, 4], biomarkers tied to specific candidate genes have often formed the basis for these pursuits. For example, both schizophrenia and bipolar disorder have been the subject of hundreds of candidate-gene association analyses [5–7] and, most recently, a handful of genome-wide association studies [8–14]. Through meta-analysis, our group and others have identified several genetic polymorphisms that influence the risk for one or the other disorder [6, 15–21]. Collectively, these polymorphisms may explain a small portion of the heritability of each disorder; however, the majority of heritable variance in susceptibility remains unexplained by such variants, as does the considerable amount of risk attributable to non-heritable factors. Furthermore, genetic polymorphisms, as (inherited or acquired) fixed factors, lack flexibility as biomarkers due to their inability to fluctuate over time within individuals and to reflect the course of illness, including periods of risk, prodrome, first episode, chronic illness, remission, and treatment-responsiveness or non-responsiveness. As such, comprehensive biomarker profiles of complex psychiatric disorders may require the integration of markers from multiple “omic” domains [22], including static genomic factors (i.e., DNA polymorphisms) as well as dynamic factors reflected in the transcriptome [i.e., the total messenger RNA (mRNA) expressed in a cell or tissue at a given time [23].
Over the past four years, we and others [24–28] have documented the potential utility of blood-based transcrip-tomic profiling of mRNA abundances by microarray as a source of biomarkers for schizophrenia and bipolar disorder. We first used mRNA expression patterns in circulating peripheral blood mononuclear cells (PBMCs) to identify a large number of genes whose expression levels distinguished patients with schizophrenia or bipolar disorder from each other and from unaffected control subjects based on liberal significance criteria (p<0.05) [29]. We later re-analyzed those data adopting a more conservative permutation-based approach toward the control of type-I errors, which reduced the number of genes identified as differentially expressed in the blood of schizophrenia patients (relative to control subjects) from 567 to 123. Six of these genes were differentially expressed in both PBMCs and postmortem brain tissue from a separate sample of schizophrenia patients, further supporting their candidacy as biomarkers [30]. Of these, SELENBP1 (which codes for selenium binding protein 1) subsequently emerged as the strongest putative biomarker based on the similar magnitude of its up-regulation in both PBMCs and brain, validation of this effect at the protein level, and replication of SELENBP1 dysregulation in an independent series of postmortem brain tissue samples [31]. Interestingly, the strongest result in our replication study was observed in the comparison of tissue from individuals with a history of psychosis (including all schizophrenia patients and a portion of bipolar disorder patients) and individuals without such a history (including the remaining non-psychotic bipolar disorder patients and all unaffected control subjects). This result in particular suggested that future pursuits of schizophrenia and bipolar disorder biomarkers in PBMCs might also profit from a focus on psychosis as a common feature of the disorders, which may be more strongly linked than either diagnosis to changes in the transcriptome. In fact, this conceptualization mirrors the emerging recognition of partially overlapping genetic contributions to the etiology of the two disorders [4, 32–35].
Toward facilitating the identification of mRNA bio- markers of disease, technology has been developed recently that exponentially increases the sensitivity and specificity of existing transcriptome-profiling systems (and eliminates 3′ bias) by yielding estimates of exon-level mRNA abundance. In addition to the ability to summate exon-level data into a more accurate measure of full-length gene expression, these microarrays allow for the identification and measurement of different splice variants of all expressed genes as well as the detection of novel splicing events. Given prior evidence of alternative splicing of select candidate genes in schizophrenia [36–40] and bipolar disorder [41], it was of keen interest to determine if a preliminary survey of the entire human “spliceome” in PBMCs could reveal useful biomarkers for these disorders, as well as their sometimes-shared clinical feature, psychosis. Additionally, the present study presents a critical overview on the feasibility of spliceome-profiling as a new methodological approach in transcriptomics, and in personalized medicine research more generally.
MATERIALS AND METHODS
Ascertainment
Subjects with schizophrenia (SCZ; n=13) or bipolar disorder (BPD; n=11) were recruited from the University of California, San Diego (UCSD) Psychopharmacology Research Initiatives Center for Excellence (PRICE) participant network. Healthy control subjects (CNT; n=10) were recruited from the same catchment area through the use of flyers and print advertisements. All participants underwent a brief initial phone screening to assess their appropriateness for possible inclusion in the study. During this screening, information was gathered related to the study’s inclusion and exclusion criteria, as well as basic demographics (e.g., age, sex, ancestry).
Inclusion criteria required participants to: 1) be between the age of 18 and 55 years; 2) have at least an eighth-grade education; 3) speak English as their first language; and 4) have no documented evidence of mental retardation. Subjects in the two patient groups (SCZ and BPD) were further required to have met criteria for their primary diagnosis (schizophrenia or bipolar disorder) for at least two years. Exclusion criteria were: 1) substance abuse or dependence in the past year; 2) neurologic problems (e.g., stroke, meningitis); 3) systemic medical illnesses (e.g., heart disease, diabetes); 4) history of head injury with documented loss of consciousness lasting longer than 10 minutes; 5) pregnancy; or 6) physical disabilities. Subjects in the CNT group were also excluded if they had a personal or family history of a psychotic disorder, bipolar disorder, major depressive disorder, or a cluster-A (schizotypal, schizoid, or paranoid) personality disorder. All participants gave written consent prior to enrollment in the study, and all study procedures were approved by the Institutional Review Board at UCSD.
Clinical Assessment and Data Analyses
Individuals satisfying inclusion and exclusion criteria during the initial phone screening were scheduled for a two-hour in-person clinical assessment using the Diagnostic Interview for Genetic Studies (DIGS) [42], which was administered by a trained Masters-level research assistant. The DIGS was used to verify information obtained from the phone screening, collect additional psychiatric data, and ultimately assist in accurate diagnosis and classification. To ensure accurate diagnosis and classification of participants, each DIGS was reviewed by two independent doctoral-level clinicians. When discrepancies in diagnoses occurred, an attempt was made to resolve them and come to a consensus diagnosis. If a consensus could not be reached, the participant was excluded from the study. Participants were also excluded if one or more of the inclusion or exclusion criteria were found not to be satisfied following review of the DIGS data.
The DIGS interview also yielded information on important covariates and potential confounding variables of relevance to the analyses of gene expression and alternative splicing. Age was measured continuously in years, whereas sex (male/female), ancestry (European, African, Hispanic, or Asian), current smoking status (yes/no), and history of psychosis (yes/no) were coded as categorical variables. Finally, each participant’s current medication regimen was reviewed and coded as a binary categorical variable (yes/no) for each of the predominant classes of medication used by subjects in the sample, including antipsychotic and mood-stabilizing drugs.
Continuously distributed demographic and clinical variables were compared between diagnostic groups by analyses of variance (ANOVAs), while categorical demographic and clinical variables were compared between the groups by χ2-tests. These analyses were conducted in Stata SE software, version 9.2 (StataCorp; College Station, TX)
mRNA Sample Acquisition, Stabilization, Isolation, and Storage
After completing the DIGS interview, subjects were scheduled to provide a 10-ml sample of blood on a subsequent visit. To control for potential environmental (e.g., diet) and biological (e.g., circadian) influences on gene expression, all blood draws were performed in the morning after subjects fasted overnight. Each blood sample was collected into an EDTA-coated collection tube and immediately transferred to an RNase-free laboratory, where all subsequent procedures took place. The blood sample was passed over a LeukoLOCK™ filter, which was flushed with PBS and then fully saturated with RNAlater® [43]. Each LeukoLOCK™ filter, containing bound, isolated, stabilized, and purified white blood cells, was sealed and stored in a sterile box at -20°C. Once mRNA samples were acquired from all 34 subjects, the entire batch of samples was processed to isolate mRNA. Eluted mRNA samples were stored at -20°C until transferred to the GeneChip™ Microarray Core (San Diego, CA) for quality assurance and microarray hybridization. LeukoLOCK™ filters, RNAlater®, and TRI reagent® were obtained from Applied Biosystems, Inc. (Foster City, CA), while all other reagents and supplies were obtained from VWR International, LLC (West Chester, PA) unless otherwise specified.
mRNA Quantitation and Quality Assurance
The concentration of mRNA in each DNA-free sample was quantified by the absorption of ultraviolet light at two wavelengths (260 and 280 nm), which was measured on a NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific; Wilmington, Delaware). The quantity of mRNA in each of the 34 samples far exceeded the minimally sufficient amount required for microarray hybridization. The purity of each mRNA sample was estimated by the 260:280 nm absorbance ratio, with an acceptable range designated a priori as 1.7–2.1. The sample from one BPD subject had a value below this range and thus was excluded from further analyses. The quality of each mRNA sample was quantified by the RNA Integrity Number (RIN) [44], which was determined on an RNA 6000 Labchip Kit on an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.; Santa Clara, CA). According to convention [44], a RIN of 6.0 or greater was deemed to be indicative of acceptable quality. The samples from one BPD subject and two CNT subjects had values below 6.0 and thus were excluded from further analyses. A total of 30 subjects (SCZ, n=13; BPD, n=9; CNT, n=8) provided samples with acceptable levels of mRNA quantity, purity, and quality, which were then assayed on GeneChip® Human Exon 1.0 ST Arrays (Affymetrix, Inc.; Santa Clara, CA) per the “Whole Transcript (WT) Sense Target Labeling Assay” protocol [45] using 1 μg of total RNA from each sample.
Microarray Data Import, Normalization, Transformation, and Summarization
Partek® Genomics Suite software, version 6.3 © 2008 (Partek Incorporated; St. Louis, MO), was utilized for all analytic procedures performed on microarray scan data. First, interrogating probes from the most conservative “Core” probe set were imported. According to the manufacturer, probes in this set “are supported with the most reliable evidence from RefSeq and full length mRNA GenBank records containing complete CDS (coding sequence) information” [46]. Corrections for background signal were applied using the robust multi-array average (RMA) method [47]. The set of 30 GeneChips was standardized using quan-tile normalization, and expression levels of each probe underwent log-2 transformation to yield distributions of data that more closely approximated normality. As each exon was typically measured by multiple probe sets, summari- zation of redundant probe sets was obtained by median polish. According to convention [48], probe sets with a maximum signal:noise ratio of less than 3.0 were excluded from subsequent analyses.
Microarray Data Analyses
Five sets of comparisons of diagnostic groups were performed, as follows: 1) BPD vs. CNT; 2) SCZ vs. CNT; 3) BPD vs. SCZ; 4) BPD+SCZ vs. CNT; and 5) subjects with a history of psychosis [PSYCH(+), which included all SCZ subjects and six of the nine BPD subjects] vs. subjects with no history of psychosis [PSYCH(−), which included all CNT subjects and the remaining three BPD subjects].
The principal analyses of these data were designed to detect expressed genes (i.e., full-length mRNA transcripts) that were significantly different in constitution between diagnostic groups, which might indicate different rates or types of alternative splicing events between them. These analyses compared groups of interest on the mean expression level of all exons in each gene on a gene-by-gene basis through analyses of covariance (ANCOVAs) and inspection of interaction terms. Four classes of factors were included in each ANCOVA model. First, we included diagnostic group as the primary factor of interest. Next, we included potentially important background biological characteristics of the subjects, including age, sex, and ancestry. Third, we included variables that accounted for expected (and ultimately observed) differences between the groups in their current rates of smoking and use of prescribed psychotropic medications. Comparisons involving the SCZ group included a variable indexing current use of an antipsychotic medication, while comparisons involving the BPD group included a variable to indicate whether or not each subject was currently on a mood-stabilizer; analyses that involved both patient groups included both medication indicator variables. Lastly, we included a series of factors that allowed for the detection of alternate splicing events. Since not all exons in a gene express at the same level, exon identity (ID) was added to the model to account for exon-to-exon differences. Since multiple measurements (on the multiple exons) come from the same subject, subject ID was added to the model to accommodate the assumption of independence that is fundamental to ANCOVA. The last, most critical term included was the interaction of exon ID with diagnostic group, which allowed for the detection of differences in the expression of any exons in the different diagnostic groups [49].
Once a gene was identified as being influenced by a significant diagnostic group-by-exon ID interaction (indicative of alternative splicing), post-hoc comparisons were made between the groups to identify dysregulated exons in these genes by simply comparing diagnostic groups on the mean expression level of each individual exon using ANCOVAs. These analyses included the same factors and interaction terms described above for whole-gene analyses, except for those factors that enabled the detection of alternate splicing (i.e., exon identity and its interaction with diagnostic group). These analyses were useful for identifying which exons contributed to or accounted for the significance of alternative splicing events identified in the principal omnibus analyses.
After all quality-control procedures were executed, 21,866 full-length gene transcripts and 232,448 exons were included in the analyses. Due to the large number of statistical tests to be performed, the probability of committing type-I errors (i.e., finding false-positive results) in this study was greatly inflated. We addressed the threat of inflation of the type-I-error rate in three ways. The first method was statistical, as we controlled family-wise error rates (FWERs) at 5% using Bonferroni corrections and estimated the false-discovery rate (FDR) for all comparisons (expressed as q-values, or the proportion of findings at a given significance level that are expected to be false discoveries). In general, this approach followed the guidelines provided by Mirnics et al. [50], with the exception that we did not filter genes based on any fold-change criterion, since it is not empirically known (nor would it be expected) that any fold-change criterion is either universally applicable to all genes or is biologically meaningful. Secondly, we required a significant interaction of diagnostic group and exon ID in omnibus analyses of full-length gene transcripts before comparing the expression levels of individual exons between groups. The third method was by data reduction through secondary analyses of groups of genes: after performing each type of analysis, the generated lists of significantly alternatively spliced genes were subjected to the DAVID algorithm [51] to determine if they were enriched for genes that aggregated in the same biological pathways (defined by BioCarta- or KEGG [52]), represented similar ontologies (defined by GOC [53]), or exhibited common protein domains (defined by InterPro [54] or PIR [55]).
RESULTS
Demographic, Substance Use, and Clinical Variables
The three ascertained diagnostic groups were comparable on demographic variables (Table 1). No significant differences were observed between SCZ, BPD, and CNT groups on age (p=0.788), sex (p=0.788), or ancestry (p=0.264). The number of current smokers differed significantly between the groups (p=0.003), with the CNT group having significantly fewer current smokers than either the SCZ group (χ2(1)=11.748, p=0.001) or the BPD group (χ2(1)= 6.296, p=0.012); however, notably, the rate of smoking did not differ between SCZ and BPD groups (χ2(1)=1.119, p=0.290).
Table 1.
Descriptive Statistics by Diagnostic Group
| Diagnostic Group | Omnibus | ||||
|---|---|---|---|---|---|
| Variable | BPD (n=9) | SCZ (n=13) | CNT (n=8) | Test Statistic | p |
| Age: mean years (s.d.) | 42 (8) | 44 (9) | 45 (7) | F(2,27)=0.240 | 0.788 |
| Sex: female n (%) | 2 (22) | 4 (31) | 3 (38) | χ2(2)=0.477 | 0.788 |
| Ancestry: n (%) | χ2(6)=7.654 | 0.264 | |||
| European | 7 (78) | 5 (39) | 5 (63) | ||
| African | 1 (11) | 6 (46) | 1 (12) | ||
| Hispanic | 0 (0) | 2 (15) | 1 (12) | ||
| Asian | 1 (11) | 0 (0) | 1 (12) | ||
| Current Smoker: n (%) | 5 (56) | 10 (77) | 0 (0) | χ2(2)=11.880 | 0.003 |
| Current Medication Use: n (%) | |||||
| Antipsychotic | 5 (56) | 12 (92) | 0 (0) | χ2(2)=17.191 | <0.001 |
| Mood Stabilizer | 7 (78) | 2 (15) | 0 (0) | χ2(2)=14.534 | 0.001 |
| History of Psychosis: n (%) | 6 (67) | 13 (100) | 0 (0) | χ2(2)=21.388 | <0.001 |
As expected, the groups also differed in their rates of use of antipsychotic (p<0.001) and mood-stabilizing medications (p=0.001). The rate of use of antipsychotic medication was higher in the SCZ group than in either the BPD group (χ2(1)=4.090, p=0.043) or, obviously, the CNT group (χ2(1)= 17.231, p<0.001), and the BPD group also used this class of medications at a higher rate than did the CNT group (χ2(1)=6.296, p=0.012). Mood stabilizers were used at a higher rate in the BPD group than in either the SCZ group (χ2(1)=8.564; p=0.003) or the CNT group (c2(1)=10.578, p=0.001), but the SCZ and CNT groups did not differ significantly in their rates of use of these medications (χ2(1)=1.360, p=0.243).
Lastly, the rate of lifetime psychotic symptoms differed significantly by diagnostic group (p<0.001). As a function of our ascertainment scheme and exclusion criteria, the CNT group contained no subjects with a history of psychosis, which was a significant difference from both the SCZ group (χ2(1)=21.000; p<0.001) and the BPD group (χ2(1)=8.242; p=0.004). All subjects in the SCZ group had a history of psychosis, whereas only two-thirds of the subjects in the BPD group reported such symptoms (χ2(1)=5.018; p=0.025).
Microarray Analyses: Alternatively Spliced Genes and Dysregulated Exons
Tables 2–6 provide information on genes that were found to be influenced by a Bonferroni-corrected significant interaction of diagnostic group and exon ID, indicative of differential patterns and/or rates of alternative splicing between the groups. Supplementary Tables 1–5 provide the individual significantly dysregulated exons of these genes which may have contributed to or accounted for the significant alternative splicing event(s) detected in each gene. In each table, the genes are sorted by p-values in ascending order, with the most significantly alternatively spliced gene at the top.
Table 2.
Alternatively Spliced Genes: Bipolar Disorder (BPD) vs. Control (CNT)
| Gene | Gene | Chromosomal | Accession | Transcript | Probe | Diagnostic Group-x-Exon ID Interaction | |||
|---|---|---|---|---|---|---|---|---|---|
| Symbol | Product | Locus | Number | ID | Sets (n) | F | p | Adjusted p | q |
| PTK2B | protein tyrosine kinase 2 beta | 8p21.1 | NM_173174 | 3091301 | 34 | 4.396 | 3.75e−12 | 8.19e−08 | 1.26e−11 |
| NCF4 | neutrophil cytosolic factor 4, 40kDa | 22q13.1 | NM_013416 | 3944543 | 11 | 10.114 | 9.07e−11 | 1.98e−06 | 3.05e−10 |
| UBR5 | ubiquitin protein ligase E3 component n-recognin 5 | 8q22 | NM_015902 | 3147321 | 64 | 2.822 | 1.50e−10 | 3.26e−06 | 5.03e−10 |
| COG4 | component of oligomeric golgi complex 4 | 16q22.1 | NM_015386 | 3697125 | 24 | 4.894 | 2.60e−10 | 5.67e−06 | 8.75e−10 |
| ROD1 | regulator of differentiation 1 (S. pombe) | 9q32 | NM_005156 | 3220977 | 18 | 5.428 | 4.18e−09 | 9.13e−05 | 1.41e−08 |
| C8orf41 | chromosome 8 open reading frame 41 | 8p12 | BC066935 | 3130823 | 12 | 7.534 | 5.58e−09 | 1.22e−04 | 1.88e−08 |
| ACTR10 | actin-related protein 10 homolog (S. cerevisiae) | 14q23 | NM_018477 | 3537813 | 17 | 5.429 | 1.11e−08 | 2.43e−04 | 3.74e−08 |
| C18orf10 | chromosome 18 open reading frame 10 | 18q12.2 | BC022199 | 3804358 | 10 | 7.891 | 5.26e−08 | 1.15e−03 | 1.77e−07 |
| FLOT2 | flotillin 2 | 17q11-q12 | NM_004475 | 3751121 | 18 | 4.594 | 1.56e−07 | 3.41e−03 | 5.25e−07 |
| SFRS2IP | splicing factor, arginine/serine-rich 2, interacting protein | 12q13.11 | NM_004719 | 3452145 | 26 | 3.620 | 1.77e−07 | 3.86e−03 | 5.95e−07 |
| FAM83H | family with sequence similarity 83, member H | 8q24.3 | NM_198488 | 3157722 | 11 | 6.667 | 1.93e−07 | 4.20e−03 | 6.47e−07 |
| SPATA3 | spermatogenesis associated 3 | 2q37.1 | BC047704 | 2531648 | 13 | 5.706 | 2.49e−07 | 5.44e−03 | 8.37e−07 |
| LIG1 | ligase I, DNA, ATP-dependent | 19q13.2-q13.3 | NM_000234 | 3866898 | 29 | 3.346 | 2.68e−07 | 5.86e−03 | 9.02e−07 |
| FCAR | Fc fragment of IgA, receptor for | 19q13.2-q13.4 | NM_002000 | 3841862 | 12 | 5.909 | 4.10e−07 | 8.95e−03 | 1.38e−06 |
| GK5 | glycerol kinase 5 (putative) | 3q23 | NM_001039547 | 2698693 | 19 | 4.160 | 5.53e−07 | 1.21e−02 | 1.86e−06 |
| MAMDC4 | MAM domain containing 4 | 9q34.3 | NM_206920 | 3194832 | 29 | 3.224 | 6.49e−07 | 1.42e−02 | 2.18e−06 |
| RIOK1 | RIO kinase 1 (yeast) | 6p24.3 | NM_031480 | 2893721 | 20 | 3.807 | 1.64e−06 | 3.58e−02 | 5.51e−06 |
| FBXO34 | F-box protein 34 | 14q22.3 | NM_017943 | 3536786 | 12 | 5.396 | 1.71e−06 | 3.73e−02 | 5.74e−06 |
Table 6.
Alternatively Spliced Genes: History of Psychosis [PSYCH(+)] vs. No History of Psychosis [PSYCH(−)]
| Gene | Gene | Chromosomal | Accession | Transcript | Probe | Diagnostic Group-x-Exon ID Interaction | |||
|---|---|---|---|---|---|---|---|---|---|
| Symbol | Product | Locus | Number | ID | Sets (n) | F | p | Adjusted p | q |
| SYNE1 | spectrin repeat containing, nuclear envelope 1 | 6q25 | NM_033071 | 2979871 | 167 | 2.788 | 3.41e−27 | 7.45e−23 | 8.21e−27 |
| KIAA0460 | KIAA0460 | 1q21.2 | BC045623 | 2358221 | 22 | 7.306 | 2.12e−18 | 4.63e−14 | 5.11e−18 |
| VPS13D | vacuolar protein sorting 13 homolog D (S. cerevisiae) | 1p36.22 | NM_015378 | 2320762 | 77 | 2.958 | 3.67e−15 | 8.02e−11 | 8.84e−15 |
| ZYX | zyxin | 7q32 | NM_003461 | 3029129 | 15 | 8.215 | 9.83e−15 | 2.15e−10 | 2.37e−14 |
| CLASP1 | cytoplasmic linker associated protein 1 | 2q14.2-q14.3 | NM_015282 | 2573641 | 53 | 3.503 | 9.92e−15 | 2.17e−10 | 2.39e−14 |
| JAK1 | Janus kinase 1 (a protein tyrosine kinase) | 1p32.3-p31.3 | NM_002227 | 2416522 | 25 | 5.425 | 2.41e−14 | 5.28e−10 | 5.81e−14 |
| UBR4 | ubiquitin protein ligase E3 component n-recognin 4 | 1p36.13 | NM_020765 | 2399409 | 115 | 2.370 | 1.31e−13 | 2.87e−09 | 3.16e−13 |
| DYNC1H1 | dynein, cytoplasmic 1, heavy chain 1 | 14q32.3-qter | NM_001376 | 3552847 | 84 | 2.657 | 2.97e−13 | 6.50e−09 | 7.16e−13 |
| SPG11 | spastic paraplegia 11 (autosomal recessive) | 15q14 | NM_025137 | 3621948 | 53 | 3.138 | 3.34e−12 | 7.31e−08 | 8.05e−12 |
| ESF1 | ESF1, nucleolar pre-rRNA processing protein, homolog (S. cerevisiae) | 20p12.1 | NM_016649 | 3898224 | 17 | 6.048 | 1.17e−11 | 2.57e−07 | 2.82e−11 |
| LRP1 | low density lipoprotein-related protein 1 (alpha-2-macroglobulin receptor) | 12q13-q14 | NM_002332 | 3417842 | 105 | 2.283 | 1.50e−11 | 3.28e−07 | 3.61e−11 |
| UBR2 | ubiquitin protein ligase E3 component n-recognin 2 | 6p21.1 | NM_015255 | 2907190 | 53 | 3.007 | 2.58e−11 | 5.65e−07 | 6.22e−11 |
| KIAA0082 | KIAA0082 | 6p21.2 | ENST00000373451 | 2905512 | 26 | 4.423 | 2.72e−11 | 5.94e−07 | 6.54e−11 |
| HD | huntingtin (Huntington disease) | 4p16.3 | NM_002111 | 2715820 | 89 | 2.364 | 7.20e−11 | 1.57e−06 | 1.73e−10 |
| ZZEF1 | zinc finger, ZZ-type with EF-hand domain 1 | 17p13.2 | NM_015113 | 3741875 | 70 | 2.600 | 7.47e−11 | 1.63e−06 | 1.80e−10 |
| UTRN | utrophin | 6q24 | NM_007124 | 2929168 | 74 | 2.521 | 1.10e−10 | 2.41e−06 | 2.65e−10 |
| EIF4EBP3 | eukaryotic translation initiation factor 4E binding protein 3 | 5q31.3 | NM_003732 | 2831719 | 64 | 2.675 | 1.18e−10 | 2.59e−06 | 2.84e−10 |
| DSC2 | desmocollin 2 | 18q12.1 | NM_004949 | 3802980 | 22 | 4.557 | 3.40e−10 | 7.43e−06 | 8.17e−10 |
| HIPK2 | homeodomain interacting protein kinase 2 | 7q32-q34 | NM_022740 | 3075778 | 19 | 4.993 | 3.87e−10 | 8.45e−06 | 9.29e−10 |
| HERC1 | hect d domain and RCC1 (CHC1)-like domain (RLD) 1 | 15q22 | NM_003922 | 3628650 | 87 | 2.303 | 4.59e−10 | 1.00e−05 | 1.10e−09 |
| MLL3 | myeloid/lymphoid or mixed-lineage leukemia 3 | 7q36.1 | NM_170606 | 3080033 | 69 | 2.514 | 5.09e−10 | 1.11e−05 | 1.22e−09 |
| MDN1 | MDN1, midasin homolog (yeast) | 6q15 | NM_014611 | 2964350 | 119 | 2.054 | 6.32e−10 | 1.38e−05 | 1.52e−09 |
| DIP2A | DIP2 disco-interacting protein 2 homolog A (Drosophila) | 21q22.3 | NM_015151 | 3924674 | 42 | 3.077 | 1.09e−09 | 2.39e−05 | 2.62e−09 |
| FAM62B | family with sequence similarity 62 (C2 domain containing) | 7q36.3 | NM_020728 | 3082248 | 31 | 3.581 | 1.25e−09 | 2.72e−05 | 2.99e−09 |
| C1orf198 | chromosome 1 open reading frame 198 | 1q42.13-q43 | AK096166 | 2460325 | 17 | 5.135 | 1.48e−09 | 3.24e−05 | 3.56e−09 |
| SMG5 | Smg-5 homolog, nonsense mediated mRNA decay factor (C. elegans) | 1q21.2 | NM_015327 | 2438042 | 31 | 3.562 | 1.50e−09 | 3.28e−05 | 3.60e−09 |
| EIF2C4 | eukaryotic translation initiation factor 2C, 4 | 1p34.3 | NM_017629 | 2330002 | 28 | 3.766 | 1.51e−09 | 3.30e−05 | 3.62e−09 |
| SPEN | spen homolog, transcriptional regulator (Drosophila) | 1p3 | NM_015001 | 2322103 | 35 | 3.321 | 1.85e−09 | 4.04e−05 | 4.43e−09 |
| ARHGAP25 | Rho GTPase activating protein 25 | 2p14 | NM_014882 | 2486927 | 13 | 6.088 | 2.65e−09 | 5.78e−05 | 6.34e−09 |
| ATXN7 | ataxin 7 | 3p21.1-p12 | NM_000333 | 2627390 | 23 | 4.099 | 3.71e−09 | 8.11e−05 | 8.89e−09 |
| VPS13B | vacuolar protein sorting 13 homolog B (yeast) | 8q22.2 | NM_017890 | 3108901 | 74 | 2.341 | 3.88e−09 | 8.48e−05 | 9.29e−09 |
| ZFX | zinc finger protein, X-linked | Xp21.3 | NM_003410 | 3971923 | 7 | 10.235 | 3.94e−09 | 8.61e−05 | 9.43e−09 |
| FCHSD1 | FCH and double SH3 domains 1 | 5q31.3 | NM_033449 | 2878778 | 24 | 3.972 | 4.48e−09 | 9.79e−05 | 1.07e−08 |
| MMP27 | matrix metallopeptidase 27 | 11q24 | NM_022122 | 3388730 | 10 | 7.326 | 4.55e−09 | 9.95e−05 | 1.09e−08 |
| ARHGAP27 | Rho GTPase activating protein 27 | 17q21.31 | NM_199282 | 3759778 | 18 | 4.727 | 4.95e−09 | 1.08e−04 | 1.19e−08 |
| DOCK5 | dedicator of cytokinesis 5 | 8p21.2 | NM_024940 | 3090512 | 67 | 2.401 | 6.79e−09 | 1.48e−04 | 1.62e−08 |
| PLEC1 | plectin 1, intermediate filament binding protein 500kDa | 8q24 | NM_201380 | 3157901 | 60 | 2.505 | 7.19e−09 | 1.57e−04 | 1.72e−08 |
| MYCBP2 | MYC binding protein 2 | 13q22 | NM_015057 | 3518496 | 97 | 2.096 | 7.76e−09 | 1.70e−04 | 1.86e−08 |
| ISCU | iron-sulfur cluster scaffold homolog (E. coli) | 12q24.1 | NM_014301 | 3430776 | 9 | 7.769 | 8.77e−09 | 1.92e−04 | 2.10e−08 |
| BIRC6 | baculoviral IAP repeat-containing 6 (apollon) | 2p22-p21 | NM_016252 | 2476219 | 85 | 2.187 | 8.83e−09 | 1.93e−04 | 2.11e−08 |
| MAP3K5 | mitogen-activated protein kinase kinase kinase 5 | 6q22.33 | NM_005923 | 2975867 | 34 | 3.215 | 9.09e−09 | 1.99e−04 | 2.17e−08 |
| TRERF1 | transcriptional regulating factor 1 | 6p21.1-p12.1 | NM_033502 | 2954025 | 32 | 3.281 | 1.28e−08 | 2.79e−04 | 3.05e−08 |
| NOTCH2 | Notch homolog 2 (Drosophila) | 1p13-p11 | NM_024408 | 2431112 | 37 | 3.043 | 1.43e−08 | 3.12e−04 | 3.41e−08 |
| FAM48A | family with sequence similarity 48, member A | 13q13.3 | NM_017569 | 3509910 | 31 | 3.322 | 1.44e−08 | 3.15e−04 | 3.44e−08 |
| SP100 | SP100 nuclear antigen | 2q37.1 | NM_001080391 | 2531377 | 28 | 3.503 | 1.45e−08 | 3.18e−04 | 3.47e−08 |
| C7orf26 | chromosome 7 open reading frame 26 | 7p22.1 | BC005121 | 2989141 | 10 | 6.904 | 1.60e−08 | 3.49e−04 | 3.81e−08 |
| PIP5K3 | phosphatidylinositol-3-phosphate/phosphatidylinositol5-kinase, type III | 2q33.3 | NM_015040 | 2525272 | 47 | 2.712 | 1.72e−08 | 3.77e−04 | 4.11e−08 |
| CTTN | cortactin | 11q13 | NM_005231 | 3338552 | 27 | 3.535 | 1.99e−08 | 4.34e−04 | 4.74e−08 |
| YEATS2 | YEATS domain containing 2 | 3q27.1 | NM_018023 | 2655168 | 39 | 2.938 | 1.99e−08 | 4.35e−04 | 4.75e−08 |
| PTK2 | PTK2 protein tyrosine kinase 2 | 8q24-qter | NM_005607 | 3156307 | 35 | 3.088 | 2.12e−08 | 4.63e−04 | 5.05e−08 |
| INTS3 | integrator complex subunit 3 | 1q21.3 | NM_023015 | 2359817 | 35 | 3.080 | 2.30e−08 | 5.03e−04 | 5.49e−08 |
| EIF4G2 | eukaryotic translation initiation factor 4 gamma, 2 | 11p15 | NM_001418 | 3362719 | 26 | 3.579 | 2.50e−08 | 5.47e−04 | 5.96e−08 |
| HMGCR | 3-hydroxy-3-methylglutaryl- Coenzyme A reductase | 5q13.3-q14 | NM_000859 | 2815965 | 27 | 3.500 | 2.64e−08 | 5.77e−04 | 6.29e−08 |
| LYRM1 | LYR motif containing 1 | 16p11.2 | NM_020424 | 3651588 | 9 | 7.344 | 2.68e−08 | 5.87e−04 | 6.40e−08 |
| NFKB2 | nuclear factor of kappa light polypeptide gene enhancer | 10q24 | NM_001077493 | 3261643 | 24 | 3.731 | 2.69e−08 | 5.88e−04 | 6.42e−08 |
| ASXL2 | additional sex combs like 2 (Drosophila) | 2p24.1 | NM_018263 | 2544925 | 23 | 3.819 | 2.75e−08 | 6.02e−04 | 6.56e−08 |
| CR1 | complement component (3b/4b) receptor 1 (Knops blood group) | 1q32 | NM_000651 | 2377332 | 19 | 4.272 | 2.81e−08 | 6.15e−04 | 6.70e−08 |
| SRRM2 | serine/arginine repetitive matrix 2 | 16p13.3 | NM_016333 | 3645253 | 40 | 2.873 | 2.84e−08 | 6.21e−04 | 6.77e−08 |
| PTPN1 | protein tyrosine phosphatase, non-receptor type 1 | 20q13 | NM_002827 | 3888721 | 20 | 4.120 | 3.21e−08 | 7.01e−04 | 7.64e−08 |
| TPR | translocated promoter region (to activated MET oncogene) | 1q25 | NM_003292 | 2448232 | 51 | 2.563 | 3.79e−08 | 8.29e−04 | 9.02e−08 |
| MACF1 | microtubule-actin crosslinking factor 1 | 1p32-p31 | NM_012090 | 2331213 | 115 | 1.923 | 3.93e−08 | 8.59e−04 | 9.35e−08 |
| JARID1C | jumonji, AT rich interactive domain 1C | Xp11.22-p11.21 | NM_004187 | 4009062 | 31 | 3.208 | 4.15e−08 | 9.08e−04 | 9.88e−08 |
| ITSN2 | intersectin 2 | 2pter-p25.1 | NM_006277 | 2544238 | 50 | 2.571 | 4.58e−08 | 1.00e−03 | 1.09e−07 |
| DFFB | DNA fragmentation factor, 40kDa, beta polypeptide | 1p36.3 | NM_004402 | 2317512 | 13 | 5.311 | 5.97e−08 | 1.31e−03 | 1.42e−07 |
| TTC15 | tetratricopeptide repeat domain 15 | 2p25.3 | NM_016030 | 2467691 | 20 | 4.020 | 5.98e−08 | 1.31e−03 | 1.42e−07 |
| NPAL3 | NIPA-like domain containing 3 | 1p36.12-p35.1 | NM_020448 | 2325410 | 18 | 4.259 | 6.90e−08 | 1.51e−03 | 1.64e−07 |
| ABCA1 | ATP-binding cassette, sub-family A (ABC1), member 1 | 9q3 | NM_005502 | 3218528 | 55 | 2.446 | 6.90e−08 | 1.51e−03 | 1.64e−07 |
| EIF4G3 | eukaryotic translation initiation factor 4 gamma, 3 | 1p36.12 | NM_003760 | 2400373 | 47 | 2.603 | 7.31e−08 | 1.60e−03 | 1.74e−07 |
| RBM26 | RNA binding motif protein 26 | 13q31.1 | NM_022118 | 3519119 | 23 | 3.680 | 7.36e−08 | 1.61e−03 | 1.75e−07 |
| PRKAR2A | protein kinase, cAMP-dependent, regulatory, type II, alpha | 3p21.3-p21.2 | NM_004157 | 2673730 | 16 | 4.559 | 7.91e−08 | 1.73e−03 | 1.88e−07 |
| ADRBK1 | adrenergic, beta, receptor kinase 1 | 11q13.1 | NM_001619 | 3336801 | 27 | 3.364 | 8.09e−08 | 1.77e−03 | 1.92e−07 |
| NUP214 | nucleoporin 214kDa | 9q34.1 | NM_005085 | 3191900 | 48 | 2.563 | 9.17e−08 | 2.00e−03 | 2.18e−07 |
| EPHB4 | EPH receptor B4 | 7q22 | NM_004444 | 3064293 | 23 | 3.649 | 9.17e−08 | 2.01e−03 | 2.18e−07 |
| SYNE2 | spectrin repeat containing, nuclear envelope 2 | 14q23.2 | NM_182914 | 3539724 | 134 | 1.809 | 9.59e−08 | 2.10e−03 | 2.28e−07 |
| DDX3X | DEAD (Asp-Glu-Ala-Asp) box polypeptide 3, X-linked | Xp11 | NM_001356 | 3974838 | 20 | 3.939 | 9.88e−08 | 2.16e−03 | 2.34e−07 |
| EP400 | E1A binding protein p400 | 12q24.33 | NM_015409 | 3438617 | 63 | 2.295 | 1.05e−07 | 2.31e−03 | 2.50e−07 |
| HERC2 | hect domain and RLD 2 | 15q13 | NM_004667 | 3614901 | 57 | 2.384 | 1.06e−07 | 2.31e−03 | 2.51e−07 |
| SMG7 | Smg-7 homolog, nonsense mediated mRNA decay factor (C. elegans) | 1q25 | NM_173156 | 2371255 | 31 | 3.106 | 1.07e−07 | 2.34e−03 | 2.54e−07 |
| PRKDC | protein kinase, DNA-activated, catalytic polypeptide | 8q11 | NM_001081640 | 3134034 | 99 | 1.968 | 1.13e−07 | 2.47e−03 | 2.68e−07 |
| CENTB2 | centaurin, beta 2 | 3q29 | NM_012287 | 2712040 | 26 | 3.387 | 1.14e−07 | 2.49e−03 | 2.71e−07 |
| HSPH1 | heat shock 105kDa/110kDa protein 1 | 13q12.3 | NM_006644 | 3508330 | 26 | 3.387 | 1.15e−07 | 2.51e−03 | 2.72e−07 |
| NFAT5 | Nuclear factor of activated T-cells 5, tonicity-responsive | 16q22.1 | NM_138714 | 3666779 | 39 | 2.781 | 1.17e−07 | 2.55e−03 | 2.77e−07 |
| CCNT1 | cyclin T1 | 12pter-qter | NM_001240 | 3453218 | 12 | 5.431 | 1.23e−07 | 2.68e−03 | 2.91e−07 |
| LRBA | LPS-responsive vesicle trafficking, beach and anchor containing | 4q31.3 | NM_006726 | 2789266 | 65 | 2.259 | 1.26e−07 | 2.75e−03 | 2.98e−07 |
| HERC3 | hect domain and RLD 3 | 4q21 | NM_014606 | 2735459 | 27 | 3.296 | 1.40e−07 | 3.07e−03 | 3.33e−07 |
| RAP1GDS1 | RAP1, GTP-GDP dissociation stimulator 1 | 4q23-q25 | NM_021159 | 2736853 | 16 | 4.440 | 1.44e−07 | 3.14e−03 | 3.40e−07 |
| NPL | N-acetylneuraminate pyruvate lyase (dihydrodipicolinate synthase) | 1q25 | NM_030769 | 2370926 | 18 | 4.126 | 1.45e−07 | 3.18e−03 | 3.44e−07 |
| SLC30A6 | solute carrier family 30 (zinc transporter), member 6 | 2p22.3 | NM_017964 | 2476116 | 17 | 4.229 | 1.83e−07 | 4.00e−03 | 4.33e−07 |
| AFF1 | AF4/FMR2 family, member 1 | 4q21 | NM_005935 | 2734784 | 33 | 2.947 | 2.03e−07 | 4.44e−03 | 4.80e−07 |
| NRD1 | nardilysin (N-arginine dibasic convertase) | 1p32.2-p32.1 | NM_002525 | 2412529 | 36 | 2.829 | 2.07e−07 | 4.52e−03 | 4.89e−07 |
| ST3GAL6 | ST3 beta-galactoside alpha-2,3-sialyltransferase 6 | 3q | NM_006100 | 2633256 | 15 | 4.549 | 2.08e−07 | 4.54e−03 | 4.91e−07 |
| NUP133 | nucleoporin 133kDa | 1q42.13 | NM_018230 | 2459866 | 32 | 2.984 | 2.15e−07 | 4.70e−03 | 5.08e−07 |
| ADAR | adenosine deaminase, RNA-specific | 1q21.1-q21.2 | NM_001111 | 2436754 | 27 | 3.234 | 2.33e−07 | 5.10e−03 | 5.51e−07 |
| GOLGA2 | golgi autoantigen, golgin subfamily a, 2 | 9q34.11 | NM_004486 | 3226431 | 16 | 4.337 | 2.40e−07 | 5.25e−03 | 5.67e−07 |
| DVL3 | dishevelled, dsh homolog 3 (Drosophila) | 3q27 | NM_004423 | 2655438 | 24 | 3.432 | 2.42e−07 | 5.30e−03 | 5.73e−07 |
| KIAA0319L | KIAA0319-like | 1p34.2 | NM_024874 | 2406139 | 31 | 3.016 | 2.44e−07 | 5.32e−03 | 5.75e−07 |
| COL13A1 | collagen, type XIII, alpha 1 | 10q22 | NM_005203 | 3250486 | 25 | 3.352 | 2.55e−07 | 5.57e−03 | 6.02e−07 |
| SAPS3 | SAPS domain family, member 3 | 11q13 | NM_018312 | 3337618 | 26 | 3.279 | 2.68e−07 | 5.86e−03 | 6.33e−07 |
| HOOK3 | hook homolog 3 (Drosophila) | 8p11.21 | NM_032410 | 3096368 | 20 | 3.778 | 2.68e−07 | 5.86e−03 | 6.33e−07 |
| GPBP1 | GC-rich promoter binding protein 1 | 5q11.2 | NM_022913 | 2810458 | 16 | 4.311 | 2.73e−07 | 5.98e−03 | 6.45e−07 |
| UBE3C | ubiquitin protein ligase E3C | 7q36.3 | NM_014671 | 3033924 | 34 | 2.871 | 2.87e−07 | 6.28e−03 | 6.78e−07 |
| BAZ2B | bromodomain adjacent to zinc finger domain, 2B | 2q23-q24 | NM_013450 | 2583014 | 46 | 2.506 | 3.37e−07 | 7.37e−03 | 7.95e−07 |
| PDDC1 | Parkinson disease 7 domain containing 1 | 11p15.5 | NM_182612 | 3358361 | 14 | 4.646 | 3.39e−07 | 7.41e−03 | 8.00e−07 |
| TSC2 | tuberous sclerosis 2 | 16p13.3 | NM_000548 | 3644375 | 42 | 2.601 | 3.41e−07 | 7.46e−03 | 8.05e−07 |
| ZDHHC17 | zinc finger, DHHC-type containing 17 | 12q21.2 | NM_015336 | 3423184 | 21 | 3.634 | 3.50e−07 | 7.65e−03 | 8.25e−07 |
| PXK | PX domain containing serine/threonine kinase | 3p14.3 | NM_017771 | 2626167 | 20 | 3.732 | 3.55e−07 | 7.75e−03 | 8.36e−07 |
| CHD4 | chromodomain helicase DNA binding protein 4 | 12p13 | NM_001273 | 3442054 | 43 | 2.568 | 3.73e−07 | 8.16e−03 | 8.79e−07 |
| COL19A1 | collagen, type XIX, alpha 1 | 6q12-q13 | NM_001858 | 2912649 | 32 | 2.914 | 4.15e−07 | 9.08e−03 | 9.78e−07 |
| ANKRD28 | ankyrin repeat domain 28 | 3p24.3 | NM_015199 | 2664452 | 34 | 2.831 | 4.26e−07 | 9.31e−03 | 1.00e−06 |
| PPT1 | palmitoyl-protein thioesterase 1 | 1p32 | NM_000310 | 2408189 | 10 | 5.797 | 4.50e−07 | 9.83e−03 | 1.06e−06 |
| USP34 | ubiquitin specific peptidase 34 | 2p15 | NM_014709 | 2555277 | 83 | 2.008 | 4.83e−07 | 1.06e−02 | 1.14e−06 |
| STXBP2 | syntaxin binding protein 2 | 19p13.3-p13.2 | NM_006949 | 3819016 | 17 | 4.041 | 4.93e−07 | 1.08e−02 | 1.16e−06 |
| MLL | myeloid/lymphoid or mixed-lineage leukemia | 11q23 | NM_005933 | 3351385 | 48 | 2.434 | 4.95e−07 | 1.08e−02 | 1.16e−06 |
| WDFY3 | WD repeat and FYVE domain containing 3 | 4q21.23 | NM_014991 | 2776372 | 87 | 1.973 | 5.42e−07 | 1.19e−02 | 1.28e−06 |
| MAPK14 | mitogen-activated protein kinase 14 | 6p21.3-p21.2 | NM_139012 | 2904877 | 17 | 4.021 | 5.49e−07 | 1.20e−02 | 1.29e−06 |
| TRIM33 | tripartite motif-containing 33 | 1p13.1 | NM_015906 | 2429069 | 28 | 3.072 | 5.53e−07 | 1.21e−02 | 1.30e−06 |
| ECOP | EGFR-coamplified and overexpressed protein | 7p11.2 | NM_030796 | 3051655 | 10 | 5.715 | 5.77e−07 | 1.26e−02 | 1.36e−06 |
| LMTK2 | lemur tyrosine kinase 2 | 7q21.3 | NM_014916 | 3014159 | 27 | 3.121 | 5.80e−07 | 1.27e−02 | 1.36e−06 |
| MGAM | maltase-glucoamylase (alpha-glucosidase) | 7q34 | NM_004668 | 3028011 | 36 | 2.729 | 5.85e−07 | 1.28e−02 | 1.37e−06 |
| DLL1 | delta-like 1 (Drosophila) | 6q27 | NM_005618 | 2986350 | 15 | 4.310 | 6.34e−07 | 1.39e−02 | 1.49e−06 |
| ARHGAP26 | Rho GTPase activating protein 26 | 5q31 | NM_015071 | 2833286 | 29 | 3.002 | 6.42e−07 | 1.40e−02 | 1.51e−06 |
| HCP5 | HLA complex P5 | 6p21.3 | NM_006674 | 2902326 | 9 | 6.127 | 6.96e−07 | 1.52e−02 | 1.64e−06 |
| VPS39 | vacuolar protein sorting 39 homolog (S. cerevisiae) | 15q | NM_015289 | 3620457 | 31 | 2.898 | 7.10e−07 | 1.55e−02 | 1.67e−06 |
| IVNS1ABP | influenza virus NS1A binding protein | 1q25.1-q31.1 | NM_006469 | 2448073 | 21 | 3.517 | 7.44e−07 | 1.63e−02 | 1.75e−06 |
| KIAA0746 | KIAA0746 | 4p15.2 | ENST00000264868 | 2764192 | 32 | 2.851 | 7.44e−07 | 1.63e−02 | 1.75e−06 |
| ITGA2B | integrin, alpha 2b | 17q21.32 | NM_000419 | 3759137 | 31 | 2.892 | 7.54e−07 | 1.65e−02 | 1.77e−06 |
| NPEPL1 | aminopeptidase-like 1 | 20q13.32 | NM_024663 | 3891048 | 18 | 3.820 | 8.00e−07 | 1.75e−02 | 1.88e−06 |
| FCER1G | Fc fragment of IgE, high affinity I, receptor for; gamma | 1q23 | NM_004106 | 2363562 | 6 | 8.586 | 8.52e−07 | 1.86e−02 | 2.00e−06 |
| ZFYVE26 | zinc finger, FYVE domain containing 26 | 14q24.1 | NM_015346 | 3569441 | 51 | 2.337 | 8.61e−07 | 1.88e−02 | 2.02e−06 |
| XYLT1 | xylosyltransferase I | 16p12.3 | NM_022166 | 3682445 | 15 | 4.243 | 8.67e−07 | 1.90e−02 | 2.03e−06 |
| TMEM87A | transmembrane protein 87A | 15q15 | ENST00000389834 | 3620515 | 18 | 3.778 | 1.01e−06 | 2.21e−02 | 2.37e−06 |
| CSNK1G2 | casein kinase 1, gamma 2 | 19p13.3 | NM_001319 | 3816153 | 12 | 4.857 | 1.03e−06 | 2.26e−02 | 2.42e−06 |
| KIAA0372 | KIAA0372 | 5q15 | ENST00000358746 | 2867693 | 46 | 2.416 | 1.05e−06 | 2.29e−02 | 2.45e−06 |
| DDX58 | DEAD (Asp-Glu-Ala-Asp) box polypeptide 58 | 9p12 | NM_014314 | 3203086 | 22 | 3.345 | 1.32e−06 | 2.88e−02 | 3.09e−06 |
| ERBB2IP | erbb2 interacting protein | 5q12.3 | NM_018695 | 2812435 | 30 | 2.872 | 1.32e−06 | 2.89e−02 | 3.09e−06 |
| ZNF638 | zinc finger protein 638 | 2p13.2-p13.1 | NM_014497 | 2488114 | 39 | 2.561 | 1.32e−06 | 2.89e−02 | 3.10e−06 |
| CPD | carboxypeptidase D | 17q11.2 | NM_001304 | 3716411 | 32 | 2.786 | 1.36e−06 | 2.96e−02 | 3.17e−06 |
| GGA2 | golgi associated, gamma adaptin ear containing, ARF binding protein 2 | 16p12 | NM_015044 | 3685183 | 28 | 2.958 | 1.41e−06 | 3.08e−02 | 3.30e−06 |
| DPP9 | dipeptidyl-peptidase 9 | 19p13.3 | NM_139159 | 3846926 | 24 | 3.185 | 1.45e−06 | 3.17e−02 | 3.40e−06 |
| CREBBP | CREB binding protein (Rubinstein-Taybi syndrome) | 16p13 | NM_004380 | 3677795 | 43 | 2.451 | 1.50e−06 | 3.28e−02 | 3.51e−06 |
| CDC2L5 | cell division cycle 2-like 5 | 7p13 | NM_003718 | 2998536 | 30 | 2.857 | 1.51e−06 | 3.30e−02 | 3.53e−06 |
| INPP5D | inositol polyphosphate-5-phosphatase, 145kDa | 2q37.1 | NM_005541 | 2532699 | 25 | 3.112 | 1.55e−06 | 3.39e−02 | 3.62e−06 |
| PIGT | phosphatidylinositol glycan anchor biosynthesis, class T | 20q12-q13.12 | NM_015937 | 3886889 | 14 | 4.287 | 1.62e−06 | 3.54e−02 | 3.79e−06 |
| JMJD2A | jumonji domain containing 2A | 1p34.1 | NM_014663 | 2333429 | 27 | 2.992 | 1.63e−06 | 3.56e−02 | 3.81e−06 |
| NCAPD3 | non-SMC condensin II complex, subunit D3 | 11q25 | NM_015261 | 3399545 | 39 | 2.538 | 1.69e−06 | 3.69e−02 | 3.95e−06 |
| VPS41 | vacuolar protein sorting 41 homolog (S. cerevisiae) | 7p1 | NM_014396 | 7385683 | 27 | 2.972 | 1.90e−06 | 4.16e−02 | 4.44e−06 |
| CRAT | carnitine acetyltransferase | 9q34.1 | NM_000755 | 3226844 | 19 | 3.548 | 1.99e−06 | 4.35e−02 | 4.65e−06 |
| PLXNC1 | plexin C1 | 12q23.3 | NM_005761 | 3426502 | 32 | 2.741 | 2.05e−06 | 4.47e−02 | 4.78e−06 |
| ACSL3 | acyl-CoA synthetase long-chain family member 3 | 2q34-q35 | NM_004457 | 2529546 | 20 | 3.442 | 2.10e−06 | 4.60e−02 | 4.91e−06 |
| PDS5A | PDS5, regulator of cohesion maintenance, homolog A (S. cerevisiae) | 4p14 | NM_015200 | 2766588 | 41 | 2.466 | 2.13e−06 | 4.65e−02 | 4.97e−06 |
| NNT | nicotinamide nucleotide transhydrogenase | 5p13.1-cen | NM_012343 | 2808438 | 27 | 2.956 | 2.16e−06 | 4.73e−02 | 5.05e−06 |
| TBC1D5 | TBC1 domain family, member 5 | 3p24.3 | NM_014744 | 2664891 | 23 | 3.193 | 2.22e−06 | 4.86e−02 | 5.19e−06 |
| EML4 | echinoderm microtubule associated protein like 4 | 2p22-p2 | NM_019063 | 2478748 | 30 | 2.811 | 2.25e−06 | 4.91e−02 | 5.25e−06 |
| MLL2 | myeloid/lymphoid or mixed-lineage leukemia 2 | 12q12-q14 | NM_003482 | 3453592 | 80 | 1.949 | 2.26e−06 | 4.93e−02 | 5.27e−06 |
| PTPRJ | protein tyrosine phosphatase, receptor type, J | 11p11.2 | NM_002843 | 3329983 | 29 | 2.854 | 2.26e−06 | 4.94e−02 | 5.27e−06 |
| TOP3A | topoisomerase (DNA) III alpha | 17p12-p11.2 | NM_004618 | 3748262 | 25 | 3.061 | 2.26e−06 | 4.95e−02 | 5.28e−06 |
BPD vs. CNT
When comparing the BPD and CNT groups, the expression levels of 18 different genes (out of 21,866 full-length transcripts surveyed) surpassed a stringent, Bonferroni-corrected threshold for significance of the interaction between diagnostic group and exon ID (Table 2). Supplementary Table 1 identifies the 33 individual exons of these 18 genes that were nominally significantly dysregulated (p<0.05) between the two groups; however, due to the large number of comparisons made at the exon level (232,448) and the consequent severity of the adjustment for multiple testing, the dysregulation of no individual exon nor any of those discussed in subsequent comparisons remained significant after Bonferroni correction. Relative to the CNT group, 19 of these exons were significantly down-regulated in the BPD group while the remaining 14 were up-regulated; this proportion of up- and down-regulated exons did not differ from the ratio that might be expected by chance (binomial p=0.095).
A prototypical example of differential splicing between the two groups is illustrated in Fig. (1), which shows each group’s pattern of expression of all exons of PTK2B, the gene with the smallest p-value for the interaction of diagnostic group (BPD vs. CNT) and exon ID. BPD and CNT groups exhibited highly comparable raw levels of expression of most exons of PTK2B, but intensities of some exons appeared to diverge, especially in the vicinity of two known splicing sites. When controlling for all covariates, this divergence was statistically significant at exons 2 (p=0.010), 31 (p=0.007), and 33 (p=0.027), while the apparent decrease in expression of exon 29 (a known alternatively spliced exon) in the BPD group was not significant (p=0.299).
Fig. (1). Alternative Splicing of PTK2B: BPD vs. CNT.
The figure illustrates a prototypical example of differential expression of alternatively spliced gene variants between two groups; in this case, BPD and CNT groups. The illustrated gene, PTK2B, produced the smallest p-value (8.19e−08) for the interaction of diagnostic group (BPD vs. CNT) and exon ID. The top panel shows the four known splice variants of the gene, while the bottom panel plots the raw microarray expression levels (signal intensities, unadjusted for covariates) of individual exons in the BPD and CNT groups (red triangles and blue squares, respectively). The lines representing signal intensity in each diagnostic group closely correspond to each other over most of the length of the gene, but intensities of some exons appeared to diverge, especially in the area of two known splicing sites. When controlling for all covariates, this divergence was statistically significant (*) at exons 2 (p=0.010), 31 (p=0.007), and 33(p=0.027), while the apparent decrease in expression of exon 29 (a known alternatively spliced exon) in the BPD group was not significant (p=0.299).
Results such as that observed for exons 2, 31, and 33 suggest several possibilities, including: 1) subjects in the BPD group, on average, express less of the known splice variants of PTK2B that include this exon (this is supported by the significantly lower mean expression level of this exon); 2) some subjects in the BPD group do not express much or any of the splice variants containing this exon, while other individuals in the BPD group express such splice variants at normal levels (this is supported by the larger-than-average standard error of expression of this exon); 3) the groups do not differ in their rates or levels of expression of different splice variants, but this exon somehow is selectively inhibited in its expression; 4) the BPD group expresses less of some novel, unrecognized splice variant(s) that include this exon; or, conceivably, 5) a type-I error has occurred. Against the last option, the q-values for PTK2B and the other 17 genes listed in Table 2 were found to range from 1.26e−11 to 5.74e−06, suggesting that these 18 results have a very low probability of representing false-discoveries. The genes on this list did not represent an enrichment of any particular ontology, pathway, or protein domain.
SCZ vs. CNT
Only eight genes surpassed the Bonferroni-corrected threshold for significance of the diagnostic group-by-exon ID interaction when comparing SCZ and CNT groups (Table 3). The expected FDRs for these genes ranged from 4.18e−10 to 7.57e−06. The 20 individual significantly dysregulated exons of these eight genes are presented in Supplementary Table 2. In contrast to the comparison of BPD and CNT groups, in which the slight majority (19/33) of significantly dysregulated exons were down-regulated in BPD, almost all (18/20) of the exons that were significantly dysregulated in the SCZ group were down-regulated. The chance of observing this ratio of down-regulated to up-regulated genes was extremely low (binomial p=1.81e−4).
Table 3.
Alternatively Spliced Genes: Schizophrenia (SCZ) vs. Control (CNT)
| Gene | Gene | Chromosomal | Accession | Transcript | Probe | Diagnostic Group-x-Exon ID Interaction | |||
|---|---|---|---|---|---|---|---|---|---|
| Symbol | Product | Locus | Number | ID | Sets (n) | F | p | Adjusted p | q |
| IRAK4 | interleukin-1 receptor- associated kinase 4 | 12q12 | NM_016123 | 3412296 | 16 | 6.333 | 1.10e−10 | 2.41e−06 | 4.18e−10 |
| CYC1 | cytochrome c-1 | 8q24.3 | NM_001916 | 3120051 | 8 | 11.177 | 6.44e−10 | 1.41e−05 | 2.44e−09 |
| CHI3L1 | chitinase 3-like 1 (cartilage glycoprotein-39) | 1q32.1 | NM_001276 | 2451593 | 13 | 6.702 | 1.77e−09 | 3.86e−05 | 6.71e−09 |
| FLJ46321 | FLJ46321 protein (FLJ46321), mRNA | 9q21.32 | ENST00000344803 | 3176711 | 9 | 7.076 | 2.57e−07 | 5.60e−03 | 9.73e−07 |
| ATXN3 | ataxin 3 | 14q24.3-q32.2 | NM_004993 | 3576889 | 12 | 5.492 | 3.54e−07 | 7.74e−03 | 1.34e−06 |
| DENND1A | DENN/MADD domain containing 1A | 9q33.2 | NM_020946 | 3224650 | 20 | 3.898 | 3.55e−07 | 7.75e−03 | 1.35e−06 |
| S100A12 | S100 calcium binding protein A12 | 1q21 | NM_005621 | 2435981 | 3 | 28.983 | 3.97e−07 | 8.67e−03 | 1.51e−06 |
| ARAF | v-raf murine sarcoma 3611 viral oncogene homolog | Xp11.4-p11.2 | NM_001654 | 3976299 | 17 | 3.946 | 2.00e−06 | 4.36e−02 | 7.57e−06 |
Notably, despite the small number of genes emerging as significant in this comparison, several ontologies were significantly over-represented by these eight genes, including the biological processes “phosphorylation” (p=0.025), “phosphorus metabolism” (p=0.036), “phosphate metabolism” (p=0.036), and “cellular metabolism” (p=0.044), as well as the molecular function “catalytic activity” (p=0.047).
BPD vs. SCZ
The comparison of BPD and SCZ groups revealed 33 genes that were influenced by a Bonferroni-adjusted significant interaction of diagnostic group and exon ID (Table 4). Expected FDRs for these 33 genes were between 9.28e−12 and 5.31e−06. Consistent with the observation of more frequent down-regulation of alternatively spliced exons in the SCZ group than in the BPD (each relative to the CNT group), 75/103 significantly dysregulated exons in this comparison (Supplementary Table 3) were down-regulated in the BPD group relative to the SCZ group; this ratio of down-regulated to up-regulated genes was significantly different from chance expectation (binomial p=4.00e−6). As with the comparison of BPD and CNT groups described above, no ontologies, pathways, or protein domains were significantly over-represented in this gene list.
Table 4.
Alternatively Spliced Genes: Bipolar Disorder (BPD) vs. Schizophrenia (SCZ)
| Gene | Gene | Chromosomal | Accession | Transcript | Probe | Diagnostic Group-x-Exon ID Inter−action | |||
|---|---|---|---|---|---|---|---|---|---|
| Symbol | Product | Locus | Number | ID | Sets (n) | F | p | Adjusted p | q |
| SLC44A2 | solute carrier family 44, member 2 | 19p13.1 | NM_020428 | 3820612 | 24 | 5.217 | 3.80e−12 | 8.30e−08 | 9.28e−12 |
| DDX24 | DEAD (Asp-Glu- Ala-Asp) box polypeptide 24 | 14q32 | NM_020414 | 3577513 | 15 | 7.001 | 3.02e−11 | 6.60e−07 | 7.38e−11 |
| FIG4 | FIG4 homolog (S. cerevisiae) | 6q21 | NM_014845 | 2920962 | 27 | 4.340 | 1.33e−10 | 2.90e−06 | 3.24e−10 |
| CTNNB1 | catenin (cadherin- associated protein), beta 1, 88kDa | 3p21 | NM_001904 | 2618940 | 18 | 5.072 | 3.62e−09 | 7.90e−05 | 8.83e−09 |
| ATG16L1 | ATG16 autophagy related 16-like 1 (S. cerevisiae) | 2q3 | NM_030803 | 2532793 | 30 | 3.477 | 1.86e−08 | 4.07e−04 | 4.55e−08 |
| EX-OC7 | exocyst complex component 7 | 17q25.1 | NM_001013839 | 3771336 | 28 | 3.566 | 2.68e−08 | 5.85e−04 | 6.53e−08 |
| MAPK14 | mitogen-activated protein kinase 14 | 6p21.3-p21.2 | NM_139012 | 2904877 | 17 | 4.832 | 3.00e−08 | 6.56e−04 | 7.32e−08 |
| LTBP1 | latent transforming growth factor beta binding protein 1 | 2p22-p21 | NM_206943 | 2476510 | 42 | 2.820 | 6.38e−08 | 1.39e−03 | 1.55e−07 |
| JMJD2A | jumonji domain containing 2A | 1p34.1 | NM_014663 | 2333429 | 27 | 3.513 | 6.91e−08 | 1.51e−03 | 1.68e−07 |
| LE−PRE1 | leucine proline-enriched proteoglycan (leprecan) 1 | 1p3 | NM_022356 | 2409004 | 29 | 3.374 | 7.22e−08 | 1.58e−03 | 1.76e−07 |
| MCM4 | minichromosome maintenance complex component 4 | 8q11.2 | NM_005914 | 3097152 | 23 | 3.800 | 9.47e−08 | 2.07e−03 | 2.31e−07 |
| EDEM1 | ER degradation enhancer, mannosidase alpha-like 1 | 3p26. | NM_014674 | 2608801 | 21 | 4.008 | 9.70e−08 | 2.12e−03 | 2.36e−07 |
| VPS13D | vacuolar protein sorting 13 homolog D (S. cerevisiae) | 1p36.22 | NM_015378 | 2320762 | 77 | 2.172 | 1.15e−07 | 2.50e−03 | 2.79e−07 |
| TMEM120−A | transmembrane protein 120A | 7q11.23 | NM_031925 | 3057520 | 13 | 5.490 | 1.17e−07 | 2.56e−03 | 2.85e−07 |
| ATG2B | ATG2 autophagy related 2 homolog B (S. cerevisiae) | 14q3 | NM_018036 | 3578278 | 30 | 3.233 | 1.40e−07 | 3.07e−03 | 3.42e−07 |
| LEPROT | leptin receptor overlapping transcript | 1p31.3 | NM_017526 | 2340433 | 28 | 3.277 | 2.51e−07 | 5.49e−03 | 6.12e−07 |
| PI4KA | phosphatidylinositol 4-kinase, catalytic, alpha | 22q11.2 | NM_058004 | 3953724 | 40 | 2.721 | 3.64e−07 | 7.95e−03 | 8.85e−07 |
| SASH1− | SAM and SH3 domain containing 1 | 6q24.3 | NM_015278 | 2930243 | 29 | 3.127 | 5.19e−07 | 1.13e−02 | 1.26e−06 |
| ENTPD6 | ectonucleoside triphosphate diphosphohydrolase 6 (putative function) | 20p11.2-p11.22 | NM_001247 | 3880706 | 22 | 3.614 | 5.52e−07 | 1.21e−02 | 1.34e−06 |
| BSDC1 | BSD domain containing 1 | 1p35.1 | NM_018045 | 2405036 | 24 | 3.405 | 7.17e−07 | 1.57e−02 | 1.74e−06 |
| TFR2 | transferrin receptor 2 | 7q22 | NM_003227 | 3064158 | 18 | 4.016 | 7.22e−07 | 1.58e−02 | 1.75e−06 |
| TTC15 | tetratricopeptide repeat domain 15 | 2p25.3 | NM_016030 | 2467691 | 20 | 3.767 | 7.35e−07 | 1.61e−02 | 1.79e−06 |
| CIRBP | cold inducible RNA binding protein | 19p13.3 | NM_001280 | 3815649 | 8 | 7.210 | 9.48e−07 | 2.07e−02 | 2.30e−06 |
| EI−F4EBP3 | eukaryotic translation initiation factor 4E binding protein 3 | 5q31.3 | NM_003732 | 2831719 | 64 | 2.178 | 1.08e−06 | 2.36e−02 | 2.63e−06 |
| FLJ13611 | hypothetical protein FLJ13611 | 5q12.3 | NM_001093756 | 2812315 | 14 | 4.607 | 1.15e−06 | 2.50e−02 | 2.78e−06 |
| COQ2 | coenzyme Q2 homolog,prenyltransferase (yeast) | 4q21.23 | NM_015697 | 2775965 | 13 | 4.846 | 1.15e−06 | 2.52e−02 | 2.80e−06 |
| C14orf118 | chromosome 14 open reading frame 118 | 14q22.1-q24.3 | NM_017926 | 3544905 | 15 | 4.378 | 1.26e−06 | 2.74e−02 | 3.05e−06 |
| UTRN | utrophin | 6q24 | NM_007124 | 2929168 | 74 | 2.063 | 1.27e−06 | 2.77e−02 | 3.08e−06 |
| JA-RID2 | jumonji, AT rich interactive domain 2 | 6p24-p23 | NM_004973 | 2896177 | 21 | 3.531 | 1.59e−06 | 3.48e−02 | 3.86e−06 |
| BRD1 | bromodomain containing 1 | 22q13.33 | NM_014577 | 3965314 | 17 | 3.980 | 1.70e−06 | 3.71e−02 | 4.12e−06 |
| C18orf10 | chromosome 18 open reading frame 10 | 18q12.2 | BC022199 | 3804358 | 10 | 5.706 | 1.88e−06 | 4.11e−02 | 4.57e−06 |
| TNKS | tankyrase, TRF1-interacting ankyrin-related ADP-ribose polymerase | 8p23.1 | NM_003747 | 3085270 | 29 | 2.960 | 1.94e−06 | 4.23e−02 | 4.70e−06 |
| SN−AP29 | synaptosomal-associated protein, 29kDa | 22q11.21 | NM_004782 | 3937755 | 11 | 5.255 | 2.19e−06 | 4.78e−02 | 5.31e−06 |
BPD+SCZ vs. CNT
When the BPD and SCZ groups were combined and jointly contrasted with the CNT group, 15 genes were found to exhibit significantly different patterns of alternative splicing between the groups after correcting for multiple comparisons using the Bonferroni method (Table 5). As with the prior comparisons of diagnostic groups, the expected FDRs among these 15 genes were exceedingly low, ranging from 9.07e−10 to 5.21e−06. Of the 24 exons that were significantly differentially expressed between groups in this comparison (Supplementary Table 4), 15 were up-regulated in the combined BPD+SCZ group relative to the CNT group while the remaining nine exons were significantly down-regulated; however, this ratio did not deviate from chance expectation (p=0.078). The gene list for this comparison was most significantly enriched with genes representing biological processes such as “phosphorus metabolism” (p=0.004) and “phosphate metabolism” (p=0.004), but numerous other broad ontologies were over-represented as well, including the biological processes “development” (p=0.007), “phosphorylation” (p=0.018), “protein modification” (p=0.029), and “metabolism” (p=0.021), and the molecular functions “protein binding” (p=0.020) and “nucleotide binding” (p=0.050).
Table 5.
Alternatively Spliced Genes: Bipolar Disorder (BPD) + Schizophrenia (SCZ) vs. Control (CNT)
| Gene | Gene | Chromosomal | Accession | Transcript | Probe | Diagnostic Group-x-Exon ID Interaction | |||
|---|---|---|---|---|---|---|---|---|---|
| Symbol | Product | Locus | Number | ID | Sets (n) | F | p | Adjusted p | q |
| VCL | vinculin | 10q22.1-q23 | NM_014000 | 3820612 | 37 | 3.392 | 2.99e−10 | 6.53e−06 | 9.07e−10 |
| PARP10 | poly (ADP-ribose)polymerase family, member 10 | 8q24.3 | NM_032789 | 3577513 | 20 | 4.021 | 5.95e−08 | 1.30e−03 | 1.81e−07 |
| IL3 | interleukin 3 (colony-stimulating factor, multiple) | 5q31.1 | NM_000588 | 2920962 | 7 | 8.607 | 8.71e−08 | 1.91e−03 | 2.64e−07 |
| PERQ1 | PERQ amino acid rich, with GYF domain 1 | 7q22 | NM_022574 | 2618940 | 30 | 3.107 | 1.67e−07 | 3.64e−03 | 5.05e−07 |
| FLJ46321 | FLJ46321 protein (FLJ46321), mRNA | 9q21.32 | ENST00000344803 | 2532793 | 9 | 6.632 | 1.78e−07 | 3.90e−03 | 5.41e−07 |
| PTPRC | protein tyrosine phosphatase, receptor type, C | 1q31-q32 | NM_002838 | 3771336 | 38 | 2.754 | 2.24e−07 | 4.90e−03 | 6.79e−07 |
| LIG1 | ligase I, DNA, ATP-dependent | 19q13.2-q13.3 | NM_000234 | 2904877 | 29 | 3.112 | 2.52e−07 | 5.50e−03 | 7.62e−07 |
| CYC1 | cytochrome c-1 | 8q24.3 | NM_001916 | 2476510 | 8 | 6.919 | 4.52e−07 | 9.89e−03 | 1.37e−06 |
| SMAD2 | SMAD family member 2 | 18q21.1 | NM_005901 | 2333429 | 16 | 4.200 | 4.75e−07 | 1.04e−02 | 1.44e−06 |
| IDH3A | isocitrate dehydrogenase 3 (NAD+) alpha | 15q25.1-q25.2 | NM_005530 | 2409004 | 15 | 4.202 | 1.05e−06 | 2.29e−02 | 3.17e−06 |
| ADRBK1 | adrenergic, beta, receptor kinase 1 | 11q13.1 | NM_001619 | 3097152 | 27 | 3.037 | 1.14e−06 | 2.49e−02 | 3.45e−06 |
| SRCAP | Snf2-related CBP activator protein | 16p11.2 | NM_006662 | 2608801 | 52 | 2.294 | 1.22e−06 | 2.67e−02 | 3.70e−06 |
| PLK4 | polo-like kinase 4 (Drosophila) | 4q28 | NM_014264 | 2320762 | 19 | 3.620 | 1.31e−06 | 2.87e−02 | 3.97e−06 |
| KIAA0460 | KIAA0460 | 1q21.2 | BC045623 | 3057520 | 22 | 3.312 | 1.63e−06 | 3.57e−02 | 4.94e−06 |
| FLOT2 | flotillin 2 | 17q11-q12 | NM_004475 | 3578278 | 18 | 3.682 | 1.72e−06 | 3.77e−02 | 5.21e−06 |
PSYCH(+) vs. PSYCH(−)
By far, the greatest disparity in patterns of expression of alternatively spliced genes was seen when comparing groups with a different history of psychosis. A total of 156 genes surpassed the Bonferroni-adjusted significance threshold for the interaction of diagnostic group with exon ID when comparing the PSYCH(+) and PSYCH(−) groups (Table 6). These 156 genes had associated FDRs ranging from 8.21e−27 to 5.28e−06. Of the 16,555 nominally significantly dysregulated exons of these 156 genes (Supplementary Table 5), only 64 were down-regulated in the PSYCH(+) group relative to the PSYCH(−) group, while the remaining 16,491 exons were up-regulated; this represented a highly significant departure from the ratio expected by chance (p<1.00e−10).
Many ontologies, pathways, and protein domains were significantly over-represented by alternatively spliced genes in the comparison of PSYCH(+) and PSYCH(−) groups (Table 7). Each of the five ontological terms significantly over-represented by alternatively spliced genes in the comparison of SCZ vs. CNT groups were also significantly over-represented in the comparison of PSYCH(+) vs. PSYCH(−) groups; these included “phosphorylation” (p=0.042), “phosphorus metabolism” (p=0.015), “phosphate metabolism” (p=0.015), “cellular metabolism” (p=0.021), and “catalytic activity” (p=0.009). Of note, genes in the NOTCH signaling pathway were also present in this list at a significantly higher rate than would be expected by chance (p=0.015), as were genes that bind to (p=0.005) or are expressed in (p=0.035) the cytoskeleton, genes involved in the ubiquitin cycle (p=0.047) or having ubiquitin-protein ligase activity (p=0.009), and genes that activate (p=7.50e−04) or regulate (p=5.20e−05) GTPase activity.
Table 7.
Ontologies, Pathways, and Protein Domains Significantly Over-Represented Among Alternatively Spliced Genes: History of Psychosis [PSYCH(+)] vs. No History of Psychosis [PSYCH(−)]
| Domain | Category | Term | n (%) of Genes on List | p |
|---|---|---|---|---|
| Ontology | Biological Process | biopolymer modification | 29 (5.2) | 1.20e−04 |
| macromolecule metabolism | 57 (10.0) | 1.40e−04 | ||
| protein metabolism | 46 (8.3) | 1.80e−04 | ||
| cellular macromolecule metabolism | 44 (7.9) | 1.80e−04 | ||
| cellular protein metabolism | 43 (7.7) | 2.70e−04 | ||
| biopolymer metabolism | 39 (7.0) | 3.90e−04 | ||
| protein modification | 27 (4.8) | 4.50e−04 | ||
| cellular physiological process | 98 (18.0) | 4.90e−03 | ||
| primary metabolism | 74 (13.0) | 7.50e−03 | ||
| protein amino acid phosphorylation | 12 (2.2) | 1.00e−02 | ||
| cell organization and biogenesis | 21 (3.8) | 1.30e−02 | ||
| phosphate metabolism | 15 (2.7) | 1.50e−02 | ||
| phosphorus metabolism | 15 (2.7) | 1.50e−02 | ||
| cellular metabolism | 74 (13.0) | 2.10e−02 | ||
| organ development | 10 (1.8) | 2.30e−02 | ||
| cell cycle | 12 (2.2) | 2.30e−02 | ||
| regulation of translational initiation | 3 (0.5) | 2.60e−02 | ||
| organ morphogenesis | 6 (1.1) | 2.70e−02 | ||
| endocytosis | 5 (0.9) | 2.90e−02 | ||
| Notch signaling pathway | 3 (0.5) | 3.00e−02 | ||
| nuclear organization and biogenesis | 2 (0.4) | 3.00e−02 | ||
| organelle organization and biogenesis | 13 (2.3) | 3.10e−02 | ||
| metabolism | 77 (14.0) | 3.80e−02 | ||
| phosphorylation | 12 (2.2) | 4.20e−02 | ||
| protein localization | 10 (1.8) | 4.30e−02 | ||
| ubiquitin cycle | 9 (1.6) | 4.70e−02 | ||
| Cellular Component | intracellular | 88 (16.0) | 6.10e−05 | |
| endomembrane system | 10 (1.8) | 1.80e−03 | ||
| eukaryotic translation initiation factor 4F complex | 3 (0.5) | 2.20e−03 | ||
| cytoplasm | 43 (7.7) | 6.20e−03 | ||
| nuclear envelope | 5 (0.9) | 7.30e−03 | ||
| intracellular organelle | 70 (13.0) | 9.60e−03 | ||
| organelle | 70 (13.0) | 9.70e−03 | ||
| nucleus | 44 (7.9) | 2.30e−02 | ||
| cytoskeleton | 14 (2.5) | 3.50e−02 | ||
| cell | 108 (19.0) | 4.00e−02 | ||
| Molecular Function | protein binding | 67 (12.0) | 8.20e−10 | |
| GTPase regulator activity | 11 (2.0) | 5.20e−05 | ||
| enzyme regulator activity | 15 (2.7) | 4.40e−04 | ||
| GTPase activator activity | 7 (1.3) | 7.50e−04 | ||
| binding | 103 (19.0) | 9.80e−04 | ||
| zinc ion binding | 29 (5.2) | 1.00e−03 | ||
| enzyme activator activity | 8 (1.4) | 1.40e−03 | ||
| microtubule binding | 4 (0.7) | 3.30e−03 | ||
| protein kinase activity | 13 (2.3) | 3.70e−03 | ||
| adenyl nucleotide binding | 21 (3.8) | 4.30e−03 | ||
| cytoskeletal protein binding | 9 (1.6) | 4.70e−03 | ||
| phosphotransferase activity, alcohol group as acceptor | 14 (2.5) | 5.50e−03 | ||
| tubulin binding | 4 (0.7) | 6.30e−03 | ||
| ATP binding | 20 (3.6) | 6.40e−03 | ||
| ubiquitin-protein ligase activity | 8 (1.4) | 8.80e−03 | ||
| catalytic activity | 57 (10.0) | 9.40e−03 | ||
| ligase activity | 10 (1.8) | 1.20e−02 | ||
| small GTPase regulator activity | 6 (1.1) | 1.30e−02 | ||
| transition metal ion binding | 29 (5.2) | 1.40e−02 | ||
| cation binding | 37 (6.6) | 1.60e−02 | ||
| kinase activity | 15 (2.7) | 1.60e−02 | ||
| ion binding | 39 (7.0) | 1.60e−02 | ||
| metal ion binding | 39 (7.0) | 1.60e−02 | ||
| protein serine/threonine kinase activity | 9 (1.6) | 1.70e−02 | ||
| ligase activity, forming carbon-nitrogen bonds | 8 (1.4) | 2.10e−02 | ||
| purine nucleotide binding | 22 (3.9) | 2.50e−02 | ||
| nucleotide binding | 24 (4.3) | 3.60e−02 | ||
| actin binding | 6 (1.1) | 3.70e−02 | ||
| transcription cofactor activity | 6 (1.1) | 4.20e−02 | ||
| transferase activity | 22 (3.9) | 4.60e−02 | ||
| transferase activity, transferring phosphorus-containing groups | 15 (2.7) | 4.80e−02 | ||
| Pathway | KEGG Pathway | notch signaling pathway | 4 (0.7) | 1.50e−02 |
| Protein Domain | InterPro Name | Zinc finger, PHD-type | 8 (1.4) | 1.10e−05 |
| Actin-binding, actinin-type | 5 (0.9) | 4.50e−05 | ||
| Bromodomain | 5 (0.9) | 2.70e−04 | ||
| Zinc finger, ZZ-type | 4 (0.7) | 3.20e−04 | ||
| Spectrin repeat | 7 (1.3) | 4.10e−04 | ||
| FY-rich, C-terminal | 3 (0.5) | 7.50e−04 | ||
| FY-rich, N-terminal | 3 (0.5) | 7.50e−04 | ||
| Involucrin repeat | 3 (0.5) | 7.50e−04 | ||
| Regulator of chromosome condensation, RCC1 | 4 (0.7) | 1.00e−03 | ||
| HECT | 4 (0.7) | 1.30e−03 | ||
| Calponin-like actin-binding | 5 (0.9) | 1.90e−03 | ||
| Protein kinase | 12 (2.2) | 2.20e−03 | ||
| Armadillo-like helical | 8 (1.4) | 2.90e−03 | ||
| Tyrosine protein kinase | 10 (1.8) | 8.10e−03 | ||
| SET-related region | 3 (0.5) | 8.90e−03 | ||
| Serine/threonine protein kinase | 10 (1.8) | 1.00e−02 | ||
| Cell surface receptor IPT/TIG | 3 (0.5) | 2.60e−02 | ||
| Src homology-3 | 6 (1.1) | 2.70e−02 | ||
| HEAT | 4 (0.7) | 3.00e−02 | ||
| Zinc finger, FYVE-type | 3 (0.5) | 3.50e−02 | ||
| Klarsicht/ANC-1/syne-1 homology | 2 (0.4) | 3.50e−02 | ||
| Initiation factor eIF-4 gamma, MA3 | 2 (0.4) | 4.20e−02 | ||
| DEAD/DEAH box helicase, N-terminal | 4 (0.7) | 4.30e−02 | ||
| Tyrosine protein kinase, active site | 4 (0.7) | 4.60e−02 | ||
| Helicase, C-terminal | 4 (0.7) | 4.80e−02 | ||
| Concanavalin A-like lectin/glucanase, subgroup | 4 (0.7) | 4.90e−02 | ||
| Plectin repeat | 2 (0.4) | 4.90e−02 | ||
| eIF4-gamma/eIF5/eIF2-epsilon | 2 (0.4) | 4.90e−02 | ||
| PIR Superfamily | SF010337:acute lymphoblastic leukemia protein, ALR type | 2 (0.4) | 1.00e−02 | |
| SF002662:plectin | 2 (0.4) | 2.00e−02 |
DISCUSSION
Transcriptomic biomarkers of psychiatric diseases obtained from a query of peripheral tissues that are clinically accessible (e.g., blood cells instead of post-mortem brain tissue) have substantial practical appeal to discern the molecular subtypes of common complex diseases such as major psychosis. To our knowledge, this is the first blood-based spliceome-profiling study of schizophrenia and bipolar disorder to be reported. The present study has relevance both as an original exploratory study as well as a new methodological approach in the study of genetic factors that may contribute to disease susceptibility. Beyond these qualities, the study suggests other potential applications of the methodology, such as the study of dynamic host responses to environmental factors/perturbations such as drug treatment or other exposures.
The chief result from this pilot study is the demonstration that exomic and spliceomic profiling can identify transcripts that reliably differentiate groups of psychiatric patients from each other and from non-mentally ill individuals. Second, this pilot study yielded information on numerous specific exons, alternatively spliced genes, and functionally or structurally related groups of those genes that are expressed in varying amounts in the peripheral blood of patients with schizophrenia, bipolar disorder, or psychosis. Third, from a methodological standpoint, a focus on spliceome-profiling may offer a renewed interest and deeper insights on peripheral tissue markers of central nervous system diseases, an area of scientific inquiry that has thus far lacked from a paucity of transcriptomics-based biomarkers in tissues that are clinically accessible in a form that is also acceptable to patients. Seen in this light, blood-based biomarkers deserve further exploration particularly by taking into account human variation in the spliceome. However, we underscore that the findings presented in this report are exploratory in nature and intended to serve as a baseline inquiry for spliceome-based biomarkers. These results will undoubtedly require replication and the test of triangulation by other independent biomarker technology platforms in different tissues before they can be considered solid leads in the pursuit of biomarkers for these disorders.
With these caveats in mind, several conclusions from this discovery-oriented exploratory study can be distilled. First, a relatively small number of genes display differential patterns of expression of alternative splice variants between diagnostic groups that are reliable after applying stringent corrections for multiple testing. The comparisons producing the fewest such differences were between the CNT group and either of the other ascertained diagnostic groups (SCZ and BPD), which is notable given the vast differences between the groups not just in diagnoses but in variables associated with the disorders (e.g., comorbid conditions, pharmacologic treatment) and potential confounding variables (e.g., smoking). In contrast, a fair number of genes showed differential patterns of expression of alternatively spliced variants between BPD and SCZ groups, which is considerable given their much higher degree of comparability on clinical and potential confounding variables. The largest number of significant differences in splicing patterns was observed when groups were contrasted on their history of psychosis without regard to diagnostic boundaries. Disentangling the source(s) of this enhancement is difficult. For some genes, it may be that the larger sample size of the PSYCH(+) and PSYCH(−) groups relative to any two individual ascertained diagnostic groups allowed some small effects to achieve statistical significance in the former contrast where they otherwise might not in the latter. Alternatively, the phenotype of psychosis simply may be more strongly linked than any particular diagnostic entity to the expression of alternatively spliced genes in PBMCs. Of course, both situations also may be operating concurrently.
A second conclusion can be drawn based on our analyses of biological pathways, ontologies, and protein domains represented by the alternatively spliced genes. Many of the structural or functional categories over-represented by the alternatively spliced genes were very broad in definition (e.g., cellular metabolism or catalytic activity), but some were quite specific and potentially informative. For example, the SCZ and combined BPD+SCZ groups (relative to the CNT group) showed differential splicing of genes linked to phosphorylation, phosphorus metabolism, and phosphate metabolism. Relative to the PSYCH(−) group, the PSYCH (+) group also showed an over-abundance of alternatively spliced genes representing these processes. In addition, the PSYCH(+) group expressed more differentially spliced variants of genes linked to very specific biological processes such as GTPase activity, as well as Notch and ubiquitin pathways, each of which independently has been linked to schizophrenia and/or psychosis previously [20, 56–59]. Thus, we conclude that the genes that show differential expression of alternatively spliced variants in schizophrenia or psychosis are not randomly distributed, but aggregate in mechanistically meaningful pathways, some of which are supported by prior work and some of which can foster new hypotheses.
A final (and perhaps most striking) general conclusion to be drawn from this study is that psychosis appears to be marked by a global up-regulation of exons in transcripts expressed in PBMCs. Of the 16,555 exons (in 156 genes) that were significantly differentially expressed in the PSYCH (+) group, 99.6% were up-regulated while just 0.4% were down-regulated. The strength of this result suggests that a systematic process (e.g., down-regulation of splicing factors) may be operating in psychosis, which leads to widespread over-expression of selected exons or more frequent inclusion of those exons in expressed transcripts. Aside from stimulating work to unravel the biological basis for the phenomenon, this finding may encourage the advancement of unidirectional hypotheses in subsequent analyses of splicing patterns in psychosis, which in turn would warrant the use of one-tailed statistical tests and the conservation of inferential power.
Beyond these conclusions, this exploratory study can serve as a useful comparator for future spliceome-wide or candidate-gene analyses of alternative splicing, as well as a replication study for prior reports of alternative splicing in these disorders. Significant changes in the expression of alternatively spliced transcripts in blood or brain tissue samples from schizophrenia patients have been reported for a number of genes, including CAMK2A [60], CTNNA2 [61], DRD3 [62], ERBB4 [36, 38], GRIN1 (NMDARI/NR1) [39], GRM3 [63], and QKI [64]. Further supporting the conceptualization of mRNA expression as perhaps the most basic endophenotype, dysregulated expression of exons in some of these alternatively spliced variants has been shown to be strongly governed by specific cis-acting polymorphisms in the same genes [36, 38, 41, 63]. Conversely, few significant findings have been reported in bipolar disorder; in fact, we could only find one study examining the issue, which reported a non-significant difference in the expression of ISYNA1 splice variants in blood from bipolar disorder patients relative to control subjects [65]. We too failed to find a significant interaction of diagnostic group (BPD vs. CNT) and exon ID for ISYNA1 (p=0.107), but we also failed to find evidence supporting the differential expression of splice variants of any of the genes previously found to be alternatively spliced in schizophrenia as well. Since none of these genes exhibited a Bonferroni-corrected (or even nominally) significant interaction of diagnostic group and exon ID, we did not interrogate specific exons of these genes in our primary analyses reported above. However, post-hoc inspection of the data did reveal nominally significant evidence for dysregulation of one exon in each of CTNNA2 (p=0.030), DRD3 (p=0.005), ERBB4 (p=0.040), and QKI (p=0.035).
We caution the reader that several caveats must be considered in comparing our results to those described above. First, in many instances different tissues are being compared (PBMCs and postmortem brain). Although expression of many full-length gene transcripts is known to be reasonably well correlated [Spearman’s rho(ρ)=0.73] between blood and brain tissue [25], it is not known if this correspondence extends to individual exons or all splice variants of each gene. In addition, while the overall correlation between blood and brain gene expression is reasonably high, the correlation varies widely for individual genes, with some showing near perfect correspondence and others showing little or no correlation. Although this has yet to be demonstrated across the whole human spliceome, similar results may be expected, and thus it is possible that the discrepancies between prior studies and our own are driven by tissue-specificity of splicing and expression of the evaluated genes. Second, our results for these genes were derived by microar-ray whereas the original discoveries were typically made by quantitative reverse-transcription PCR; thus, it is possible that our failure to provide independent replication of these discoveries was due to a lack of sensitivity of our chosen platform. Third, and of particular relevance to the results of Sartorius et al. [63], Silberberg et al. [38], Atz et al. [41] and Law et al. [36], is the fact that we did not genotype the DNA polymorphisms that these investigators showed to regulate gene expression and splicing; thus, it is possible that we may have replicated their findings if we controlled for the governing DNA polymorphisms these authors identified. Fourth, our sample size was relatively small, which may have prohibited us from detecting effects that may have attained statistical significance in a larger sample, and this problem was likely compounded by the stringency of the corrections for multiple testing that we applied. Fifth, we emphasize that many factors (and combinations of factors) would be expected to influence gene expression or alternate splicing levels in studies of chronic illnesses where patient samples often reflect cumulative exposures from drug treatment, diet, smoking, and co-morbid disease states in addition to those conferred by diagnostic categories. In particular, there was a significant disparity in smoking status between our various diagnostic groups, with no present smokers in the CNT group compared to greater than 50% of present smokers in both the BPD and SCZ groups. If smoking status significantly influences the rates and types of splicing events that occur in peripheral blood cells, then it is possible that some of the observed differences we ascribed to diagnoses may actually be type-I errors, and instead should be attributed solely to the diagnostic-group differences in smoking rates. The present study was limited in our ability to model all potentially influential covariates by constraints on our sample size. On the other hand, even with a large study sample, confounding variables such as differences in comedication use between diagnostic groups and healthy controls are difficult to avoid. Hence, future spliceome-profiling studies in drug-naive (and non-smoking) patients or individuals who present with a first episode of psychosis could conceivably provide invaluable additional insights to further extend the observations from the present exploratory investigation. Lastly, we note however that our findings in the present study were reconciled with an analysis of the ontologies, the biological pathways and the protein domains significantly over-represented among the alternatively spliced genes, several of which support prior discoveries.
Other alternatively spliced genes identified in our study do not necessarily replicate prior findings of altered expression of splice variants of candidate genes, but in another manner provide support for involvement of these genes in the disorder. For example, we found significantly differential expression of alternatively constituted CHI3L1 transcripts in the SCZ and CNT groups, which was accounted for by decreased expression of 11 different exons in the SCZ group. This result is highly consistent with results from Zhao et al., who first discovered that schizophrenia-associated risk haplotypes of CHI3L1 were associated with lower trans- criptional activity and lower expression of the gene [66]; however, the effect of the implicated polymorphisms of this gene on the expression of particular splice variants remains to be determined. Polymorphisms in BRD1 [67], IL3 [68, 69], JARID2 [70], MAPK14 [71], and SNAP29 [72, 73] have previously been associated with risk for schizophrenia and also appear to exhibit alternative splicing in the SCZ group of the present study relative to either the CNT or BPD groups. Again, whether these genetic associations and patterns of alternative splicing reflect a common mechanistic link to the disorder is an empirical question to be addressed subsequently.
In addition to pursuing such unanswered questions, future work should proceed along several other trajectories. First, based on sequence analysis of the 230 genes listed in Tables 2–6, we have determined that 216 of these genes harbor DNA polymorphisms in putative exonic splicing enhancers, 151 harbor polymorphisms in putative exonic splicing silencers, 36 harbor polymorphisms in canonical splice sites, and 1 gene harbors a polymorphism in a predicted novel splice site at an intron/exon boundary. These polymorphisms are prime candidates for genotyping and association analysis with the disorders under study as well as the expression levels of the alternatively spliced transcripts that these polymorphisms produce. Second, previously completed transcriptomic studies of schizophrenia and bipolar disorder may warrant re-examination in light of the patterns of differential alternate splicing we observed. Failure to replicate the specific (i.e., gene-level) results of transcriptomic studies of these disorders is not uncommon. Frequently cited causes of such discrepancies include tech- nical and methodological variables, or the potential that networks of genes (rather than specific transcripts) are more likely to generalize across subjects and studies [50]. These are indeed potentially valid explanations for the phenomenon; however, it is also possible that variable patterns of expression of splice variants (or differing prevalences of influential polymorphisms [74]) across samples have given the appearance of incomparable full-length gene expression, particularly when coupled with the 3′ bias inherent in previous generations of expression microarrays. Third, further control of potentially influential covariates (e.g., diet and exercise, dosage and duration of medication usage, length of illness, etc.) should be attempted; however, this will require much larger sample sizes so that the effects of these many factors (and their interactions) can be simultaneously modeled. Lastly, our results should be verified using more sensitive mRNA quantification methods, and verified independently in either spliceome-wide or targeted replication efforts, after which some of the identified candidate exons and alternatively spliced genes may be validated as useful biomarkers for these conditions.
An important lesson learned from prior efforts to develop biomarkers for psychiatric disorders is that no single tissue, molecule, or marker is likely to yield sufficient power for improving existing behavioral classification schemes; rather, integration of the best markers from multiple domains (e.g., DNA, brain mRNA, blood mRNA, protein, etc.) may lead to the most reliable profiles [22], which then can be used to begin appropriate, group-tailored interventions. Beyond this working model, it also may be necessary to introduce an iterative component to deal with the abundant phenotypic heterogeneity of the major psychoses, which potentially may map onto heterogeneity in etiologic and biomarker profiles. Thus, as broadly influential etiologic factors and their associated biomarkers are identified, segments of the larger schizophrenia and bipolar disorder phenotypes may be “carved out”, some of which [e.g., PSYCH(+)] are influenced by that factor and some of which [e.g., PSYCH(−)] are not. The subsequent detection of both environmental and biological influences on-and biomarkers of-such subgroups will be facilitated by their relative homogeneity; however, additional phenotypic “cleavage points” should be anticipated until groups with a highly similar etiologic and biomarker profile are obtained.
Supplementary Material
Acknowledgments
This work was supported in part by the UCSD Center for AIDS Research Genomics Core, and National Institutes of Health grants P30MH062512 (Igor Grant), R01MH079881, R25MH074508, R25MH081482, and R41MH079728 (I.P.E.), R01AG018386, R01AG022381, and R01AG022982 (W.S.K.), R01DA012846, R01DA018662, R01MH065562, R01MH 071912, and R21MH075027 (M.T.T.), P50MH081755 (Eric Courchesne/S.J.G.), and a NARSAD Young Investigator Award (S.J.G.).
LIST OF ABBREVIATIONS
- ANCOVA
Analysis of covariance
- ANOVA
Analysis of variance
- BPD
Bipolar disorder
- CDS
Coding sequence
- CNT
Control
- DIGS
Diagnostic Interview for Genetic Studies
- FDR
False-discovery rate
- FWER
Family-wise error rate
- ID
Identity
- mRNA
Messenger ribonucleic acid
- PBMC
Peripheral blood mononuclear cell
- PRICE
Psychopharmacology Research Initiatives Center for Excellence
- PSYCH
Psychosis
- QRT-PCR
Quantitative reverse-transcription PCR
- RIN
RNA Integrity Number
- RMA
Robust multi-array average
- SCZ
Schizophrenia
- UCSD
University of California, San Diego
- WT
Whole transcript
Footnotes
DUALITY/CONFLICT OF INTERESTS
None declared/applicable.
Supplementary material is available on the publishers Web site along with the published article.
References
- 1.Saha S, Chant D, McGrath J. Meta-analyses of the incidence and prevalence of schizophrenia: conceptual and methodological issues. Int J Methods Psychiatr Res. 2008;17(1):55–61. doi: 10.1002/mpr.240. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Merikangas KR, Akiskal HS, Angst J, et al. Lifetime and 12-month prevalence of bipolar spectrum disorder in the National Comorbid-ity Survey replication. Arch Gen Psychiatry. 2007;64(5):543–52. doi: 10.1001/archpsyc.64.5.543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Sullivan PF, Kendler KS, Neale MC. Schizophrenia as a complex trait: evidence from a meta-analysis of twin studies. Arch Gen Psychiatry. 2003;60(12):1187–92. doi: 10.1001/archpsyc.60.12.1187. [DOI] [PubMed] [Google Scholar]
- 4.Cardno AG, Marshall EJ, Coid B, et al. Heritability estimates for psychotic disorders: the Maudsley twin psychosis series. Arch Gen Psychiatry. 1999;56(2):162–8. doi: 10.1001/archpsyc.56.2.162. [DOI] [PubMed] [Google Scholar]
- 5.Kato T. Molecular genetics of bipolar disorder and depression. Psychiatry Clin Neurosci. 2007;61(1):3–19. doi: 10.1111/j.1440-1819.2007.01604.x. [DOI] [PubMed] [Google Scholar]
- 6.Allen NC, Bagade S, McQueen MB, et al. Systematic meta-analyses and field synopsis of genetic association studies in schizophrenia: the SzGene database. Nat Genet. 2008;40(7):827–34. doi: 10.1038/ng.171. [DOI] [PubMed] [Google Scholar]
- 7.Smoller JW, Gardner-Schuster E. Genetics of bipolar disorder. Curr Psychiatry Rep. 2007;9(6):504–11. doi: 10.1007/s11920-007-0069-8. [DOI] [PubMed] [Google Scholar]
- 8.Sklar P, Smoller JW, Fan J, et al. Whole-genome association study of bipolar disorder. Mol Psychiatry. 2008;13(6):558–69. doi: 10.1038/sj.mp.4002151. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Wellcome Trust Case Control Consortium. Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls. Nature. 2007;447(7145):661–78. doi: 10.1038/nature05911. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Baum AE, Akula N, Cabanero M, et al. A genome-wide association study implicates diacylglycerol kinase eta (DGKH) and several other genes in the etiology of bipolar disorder. Mol Psychiatry. 2008;13(2):197–207. doi: 10.1038/sj.mp.4002012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Lencz T, Morgan TV, Athanasiou M, et al. Converging evidence for a pseudoautosomal cytokine receptor gene locus in schizophrenia. Mol Psychiatry. 2007;12(6):572–80. doi: 10.1038/sj.mp.4001983. [DOI] [PubMed] [Google Scholar]
- 12.Sullivan PF, Lin D, Tzeng JY, et al. Genomewide association for schizophrenia in the CATIE study: results of stage 1. Mol Psychiatry. 2008;13(6):570–84. doi: 10.1038/mp.2008.25. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Shifman S, Johannesson M, Bronstein M, et al. Genome-wide association identifies a common variant in the reelin gene that increases the risk of schizophrenia only in women. PLoS Genet. 2008;4(2):e28. doi: 10.1371/journal.pgen.0040028. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Kirov G, Zaharieva I, Georgieva L, et al. A genome-wide association study in 574 schizophrenia trios using DNA pooling. Mol Psychiatry. 2008 doi: 10.1038/mp.2008.33. [Epub ahead of print] [DOI] [PubMed] [Google Scholar]
- 15.Chen C, Glatt SJ, Tsuang MT. The tryptophan hydroxylase gene influences risk for bipolar disorder but not major depressive disorder: results of meta-analyses. Bipolar Disord. doi: 10.1111/j.1399-5618.2008.00623.x. in press. [DOI] [PubMed] [Google Scholar]
- 16.Lasky-Su JA, Faraone SV, Glatt SJ, et al. Meta-analysis of the association between two polymorphisms in the serotonin transporter gene and affective disorders. Am J Med Genet B Neuropsy-chiatr Genet. 2005;133B(1):110–5. doi: 10.1002/ajmg.b.30104. [DOI] [PubMed] [Google Scholar]
- 17.Glatt SJ, Faraone SV, Tsuang MT. Association between a functional catechol O-methyltransferase gene polymorphism and schizophrenia: meta-analysis of case-control and family-based studies. Am J Psychiatry. 2003;160(3):469–76. doi: 10.1176/appi.ajp.160.3.469. [DOI] [PubMed] [Google Scholar]
- 18.Glatt SJ, Faraone SV, Tsuang MT. Meta-analysis identifies an association between the dopamine D2 receptor gene and schizophrenia. Mol Psychiatry. 2003;8(11):911–5. doi: 10.1038/sj.mp.4001321. [DOI] [PubMed] [Google Scholar]
- 19.Glatt SJ, Jonsson EG. The Cys allele of the DRD2 Ser311Cys polymorphism has a dominant effect on risk for schizophrenia: evidence from fixed- and random-effects meta-analyses. Am J Med Genet B Neuropsychiatr Genet. 2006;141(2):149–54. doi: 10.1002/ajmg.b.30273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Glatt SJ, Wang RS, Yeh YC, et al. Five NOTCH4 polymorphisms show weak evidence for association with schizophrenia: evidence from meta-analyses. Schizophr Res. 2005;73(2–3):281–90. doi: 10.1016/j.schres.2004.07.015. [DOI] [PubMed] [Google Scholar]
- 21.Baum AE, Hamshere M, Green E, et al. Meta-analysis of two genome-wide association studies of bipolar disorder reveals important points of agreement. Mol Psychiatry. 2008;13(5):466–7. doi: 10.1038/mp.2008.16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Ozdemir V, Williams-Jones B, Glatt SJ, et al. Shifting emphasis from pharmacogenomics to theragnostics. Nat Biotechnol. 2006;24(8):942–6. doi: 10.1038/nbt0806-942. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Duffus JH, Nordberg M, Templeton DM. IUPAC Glossary of Terms Used in Toxicology, 2nd Edition. Pure Applied Chemistry. 2007;79(7):1153–344. [Google Scholar]
- 24.Vawter MP, Ferran E, Galke B, et al. Microarray screening of lymphocyte gene expression differences in a multiplex schizophrenia pedigree. Schizophr Res. 2004;67(1):41–52. doi: 10.1016/s0920-9964(03)00151-8. [DOI] [PubMed] [Google Scholar]
- 25.Sullivan PF, Fan C, Perou CM. Evaluating the comparability of gene expression in blood and brain. Am J Med Genet B Neuropsy-chiatr Genet. 2006;141B(3):261–8. doi: 10.1002/ajmg.b.30272. [DOI] [PubMed] [Google Scholar]
- 26.Zvara A, Szekeres G, Janka Z, et al. Over-expression of dopamine D2 receptor and inwardly rectifying potassium channel genes in drug-naive schizophrenic peripheral blood lymphocytes as potential diagnostic markers. Dis Markers. 2005;21(2):61–9. doi: 10.1155/2005/275318. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Middleton FA, Pato CN, Gentile KL, et al. Gene expression analysis of peripheral blood leukocytes from discordant sib-pairs with schizophrenia and bipolar disorder reveals points of convergence between genetic and functional genomic approaches. Am J Med Genet B Neuropsychiatr Genet. 2005;136(1):12–25. doi: 10.1002/ajmg.b.30171. [DOI] [PubMed] [Google Scholar]
- 28.Bowden NA, Weidenhofer J, Scott RJ, et al. Preliminary investigation of gene expression profiles in peripheral blood lymphocytes in schizophrenia. Schizophr Res. 2006;82(2–3):175–83. doi: 10.1016/j.schres.2005.11.012. [DOI] [PubMed] [Google Scholar]
- 29.Tsuang MT, Nossova N, Yager T, et al. Assessing the validity of blood-based gene expression profiles for the classification of schizophrenia and bipolar disorder: a preliminary report. Am J Med Genet B Neuropsychiatr Genet. 2005;133(1):1–5. doi: 10.1002/ajmg.b.30161. [DOI] [PubMed] [Google Scholar]
- 30.Glatt SJ, Everall IP, Kremen WS, et al. Comparative gene expression analysis of blood and brain provides concurrent validation of SELENBP1 up-regulation in schizophrenia. Proc Natl Acad Sci USA. 2005;102(43):15533–8. doi: 10.1073/pnas.0507666102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Kanazawa T, Chana G, Glatt SJ, et al. The utility of SELENBP1 gene expression as a biomarker for major psychotic disorders: Replication in schizophrenia and extension to bipolar disorder with psychosis. Am J Med Genet B Neuropsychiatr Genet. doi: 10.1002/ajmg.b.30664. in press. [DOI] [PubMed] [Google Scholar]
- 32.Owen MJ, Craddock N, Jablensky A. The genetic deconstruction of psychosis. Schizophr Bull. 2007;33(4):905–11. doi: 10.1093/schbul/sbm053. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Potash JB. Carving chaos: genetics and the classification of mood and psychotic syndromes. Harv Rev Psychiatry. 2006;14(2):47–63. doi: 10.1080/10673220600655780. [DOI] [PubMed] [Google Scholar]
- 34.Badner JA, Gershon ES. Meta-analysis of whole-genome linkage scans of bipolar disorder and schizophrenia. Mol Psychiatry. 2002;7(4):405–11. doi: 10.1038/sj.mp.4001012. [DOI] [PubMed] [Google Scholar]
- 35.Berrettini W. Bipolar disorder and schizophrenia: convergent molecular data. Neuromolecular Med. 2004;5(1):109–17. doi: 10.1385/NMM:5:1:109. [DOI] [PubMed] [Google Scholar]
- 36.Law AJ, Kleinman JE, Weinberger DR, et al. Disease-associated intronic variants in the ErbB4 gene are related to altered ErbB4 splice-variant expression in the brain in schizophrenia. Hum Mol Genet. 2007;16(2):129–41. doi: 10.1093/hmg/ddl449. [DOI] [PubMed] [Google Scholar]
- 37.Zhao C, Xu Z, Chen J, et al. Two isoforms of GABA(A) receptor beta2 subunit with different electrophysiological properties: Differential expression and genotypical correlations in schizophrenia. Mol Psychiatry. 2006;11(12):1092–105. doi: 10.1038/sj.mp.4001899. [DOI] [PubMed] [Google Scholar]
- 38.Silberberg G, Darvasi A, Pinkas-Kramarski R, et al. The involvement of ErbB4 with schizophrenia: association and expression studies. Am J Med Genet B Neuropsychiatr Genet. 2006;141B(2):142–8. doi: 10.1002/ajmg.b.30275. [DOI] [PubMed] [Google Scholar]
- 39.Le Corre S, Harper CG, Lopez P, et al. Increased levels of expression of an NMDARI splice variant in the superior temporal gyrus in schizophrenia. Neuroreport. 2000;11(5):983–6. doi: 10.1097/00001756-200004070-00017. [DOI] [PubMed] [Google Scholar]
- 40.Huntsman MM, Tran BV, Potkin SG, et al. Altered ratios of alternatively spliced long and short gamma2 subunit mRNAs of the gamma-amino butyrate type A receptor in prefrontal cortex of schizophrenics. Proc Natl Acad Sci USA. 1998;95(25):15066–71. doi: 10.1073/pnas.95.25.15066. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Atz ME, Rollins B, Vawter MP. NCAM1 association study of bipolar disorder and schizophrenia: polymorphisms and alternatively spliced isoforms lead to similarities and differences. Psychiatr Genet. 2007;17(2):55–67. doi: 10.1097/YPG.0b013e328012d850. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 42.Nurnberger JI, Jr, Blehar MC, Kaufmann CA, et al. Diagnostic interview for genetic studies. Rationale, unique features, and training. Arch Gen Psychiatry. 1994;51:849–59. doi: 10.1001/archpsyc.1994.03950110009002. [DOI] [PubMed] [Google Scholar]
- 43.Gonzales J, Kemppainen J, Latham G, et al. Isolate RNA from white blood cells captured by a novel filter system. Ambion Tech-Notes. 2005;12(4):24–5. [Google Scholar]
- 44.Schroeder A, Mueller O, Stocker S, et al. The RIN: an RNA integrity number for assigning integrity values to RNA measurements. BMC Mol Biol. 2006;7:3. doi: 10.1186/1471-2199-7-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45.Affymetrix I. GeneChip® Whole Transcript (WT) Sense Target Labeling Assay Manual, Version 4. [Accessed April 14, 2009];P/N 701880. 2006 Rev. 5(2009) Available from : http://www.affymetrix.com/support/downloads/manuals/wt_sensetarget_label_manual.pdf.
- 46.Affymetrix I. Exon and Gene Array Glossary. [Accessed April 14, 2009];2008 Available from http://www.affymetrix.com/support/help/exon_glossary/index.affx.
- 47.Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics. 2003;4(2):249–64. doi: 10.1093/biostatistics/4.2.249. [DOI] [PubMed] [Google Scholar]
- 48.Handran S, Pickett S, Verdick D. Key Considerations for Accurate Microarray Scanning and Image Analysis. In: Kamberova G, Shah S, editors. DNA Array Image Analysis: Nuts & Bolts. DNA Press, LLC; Salem, MA: 2002. pp. 83–98. [Google Scholar]
- 49.Partek® software On-line Help, version 6.3. Copyright © 2008 Partek Inc., St. Louis, MO, USA. [Accessed April 14, 2009].
- 50.Mirnics K, Levitt P, Lewis DA. Critical appraisal of DNA microar-rays in psychiatric genomics. Biol Psychiatry. 2006;60(2):163–76. doi: 10.1016/j.biopsych.2006.02.003. [DOI] [PubMed] [Google Scholar]
- 51.Dennis G, Jr, Sherman BT, Hosack DA, et al. DAVID: Database for Annotation, Visualization, and Integrated Discovery. Genome Biol. 2003;4(5):p3. [PubMed] [Google Scholar]
- 52.Kanehisa M, Goto S. KEGG: kyoto encyclopedia of genes and genomes. Nucleic Acids Res. 2000;28(1):27–30. doi: 10.1093/nar/28.1.27. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 53.Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet. 2000;25(1):25–9. doi: 10.1038/75556. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 54.Mulder NJ, Apweiler R, Attwood TK, et al. InterPro, progress and status in 2005. Nucleic Acids Res. 2005;33(Database issue):D201–5. doi: 10.1093/nar/gki106. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 55.Wu CH, Yeh LS, Huang H, et al. The Protein Information Resource. Nucleic Acids Res. 2003;31(1):345–7. doi: 10.1093/nar/gkg040. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 56.Wang Z, Wei J, Zhang X, et al. A review and re-evaluation of an association between the NOTCH4 locus and schizophrenia. Am J Med Genet B Neuropsychiatr Genet. 2006;141B(8):902–6. doi: 10.1002/ajmg.b.30383. [DOI] [PubMed] [Google Scholar]
- 57.Thomas EA. Molecular profiling of antipsychotic drug function: convergent mechanisms in the pathology and treatment of psychiatric disorders. Mol Neurobiol. 2006;34(2):109–28. doi: 10.1385/MN:34:2:109. [DOI] [PubMed] [Google Scholar]
- 58.Hashimoto R, Yoshida M, Kunugi H, et al. A missense polymorphism (H204R) of a Rho GTPase-activating protein, the chimerin 2 gene, is associated with schizophrenia in men. Schizophr Res. 2005;73(2–3):383–5. doi: 10.1016/j.schres.2004.01.017. [DOI] [PubMed] [Google Scholar]
- 59.Pickard BS, Malloy MP, Clark L, et al. Candidate psychiatric illness genes identified in patients with pericentric inversions of chromosome 18. Psychiatr Genet. 2005;15(1):37–44. doi: 10.1097/00041444-200503000-00007. [DOI] [PubMed] [Google Scholar]
- 60.Novak G, Seeman P, Tallerico T. Increased expression of calcium/calmodulin-dependent protein kinase IIbeta in frontal cortex in schizophrenia and depression. Synapse. 2006;59(1):61–8. doi: 10.1002/syn.20211. [DOI] [PubMed] [Google Scholar]
- 61.Mexal S, Berger R, Pearce L, et al. Regulation of a novel alphaN-catenin splice variant in schizophrenic smokers. Am J Med Genet B Neuropsychiatr Genet. 2007 doi: 10.1002/ajmg.b.30679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 62.Schmauss C. Enhanced cleavage of an atypical intron of dopamine D3-receptor pre-mRNA in chronic schizophrenia. J Neurosci. 1996;16(24):7902–9. doi: 10.1523/JNEUROSCI.16-24-07902.1996. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 63.Sartorius LJ, Weinberger DR, Hyde TM, et al. Expression of a GRM3 splice variant is increased in the dorsolateral prefrontal cortex of individuals carrying a schizophrenia risk SNP. Neuropsychopharmacology. 2008 doi: 10.1038/sj.npp.1301669. [DOI] [PubMed] [Google Scholar]
- 64.Aberg K, Saetre P, Jareborg N, et al. Human QKI, a potential regulator of mRNA expression of human oligodendrocyte-related genes involved in schizophrenia. Proc Natl Acad Sci USA. 2006;103(19):7482–7. doi: 10.1073/pnas.0601213103. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 65.Shamir A, Shaltiel G, Mark S, et al. Human MIP synthase splice variants in bipolar disorder. Bipolar Disord. 2007;9(7):766–71. doi: 10.1111/j.1399-5618.2007.00440.x. [DOI] [PubMed] [Google Scholar]
- 66.Zhao X, Tang R, Gao B, et al. Functional variants in the promoter region of Chitinase 3-like 1 (CHI3L1) and susceptibility to schizophrenia. Am J Hum Genet. 2007;80(1):12–8. doi: 10.1086/510438. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 67.Severinsen JE, Bjarkam CR, Kiaer-Larsen S, et al. Evidence implicating BRD1 with brain development and susceptibility to both schizophrenia and bipolar affective disorder. Mol Psychiatry. 2006;11(12):1126–38. doi: 10.1038/sj.mp.4001885. [DOI] [PubMed] [Google Scholar]
- 68.Sun S, Wang F, Wei J, et al. Association between interleukin-3 receptor alpha polymorphism and schizophrenia in the Chinese population. Neurosci Lett. 2008;440(1):35–7. doi: 10.1016/j.neulet.2008.05.029. [DOI] [PubMed] [Google Scholar]
- 69.Chen X, Wang X, Hossain S, et al. Interleukin 3 and schizophrenia: the impact of sex and family history. Mol Psychiatry. 2007;12(3):273–82. doi: 10.1038/sj.mp.4001932. [DOI] [PubMed] [Google Scholar]
- 70.Pedrosa E, Ye K, Nolan KA, et al. Positive association of schizophrenia to JARID2 gene. Am J Med Genet B Neuropsychiatr Genet. 2007;144B(1):45–51. doi: 10.1002/ajmg.b.30386. [DOI] [PubMed] [Google Scholar]
- 71.Olsen L, Hansen T, Jakobsen KD, et al. The estrogen hypothesis of schizophrenia implicates glucose metabolism: association study in three independent samples. BMC Med Genet. 2008;9:39. doi: 10.1186/1471-2350-9-39. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 72.Saito T, Guan F, Papolos DF, et al. Polymorphism in SNAP29 gene promoter region associated with schizophrenia. Mol Psychiatry. 2001;6(2):193–201. doi: 10.1038/sj.mp.4000825. [DOI] [PubMed] [Google Scholar]
- 73.Wonodi I, Hong LE, Avila MT, et al. Association between polymorphism of the SNAP29 gene promoter region and schizophrenia. Schizophr Res. 2005;78(2–3):339–41. doi: 10.1016/j.schres.2005.03.023. [DOI] [PubMed] [Google Scholar]
- 74.Benovoy D, Kwan T, Majewski J. Effect of polymorphisms within probe-target sequences on olignonucleotide microarray experiments. Nucleic Acids Res. 2008;36(13):4417–23. doi: 10.1093/nar/gkn409. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.