Convergent functional genomics of genome-wide association data for bipolar disorder: Comprehensive identification of candidate genes, pathways and mechanisms

Authors

  • H. Le-Niculescu,

    1. Laboratory of Neurophenomics, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    2. INBRAIN, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    3. Institute of Psychiatric Research, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • S.D. Patel,

    1. Laboratory of Neurophenomics, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    2. INBRAIN, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    3. Institute of Psychiatric Research, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • M. Bhat,

    1. Laboratory of Neurophenomics, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    2. Institute of Psychiatric Research, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • R. Kuczenski,

    1. Department of Psychiatry, UC San Diego, La Jolla, California
    Search for more papers by this author
  • S.V. Faraone,

    1. Department of Psychiatry, SUNY Upstate Medical University, Syracuse, New York
    Search for more papers by this author
  • M.T. Tsuang,

    1. Department of Psychiatry, UC San Diego, La Jolla, California
    Search for more papers by this author
  • F.J. McMahon,

    1. Mood and Anxiety Disorders Branch, NIMH, Bethesda, Maryland
    Search for more papers by this author
  • N.J. Schork,

    1. Scripps Genomic Medicine, The Scripps Research Institute, La Jolla, California
    Search for more papers by this author
  • J.I. Nurnberger Jr.,

    1. Institute of Psychiatric Research, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    Search for more papers by this author
  • A.B. Niculescu III

    Corresponding author
    1. Laboratory of Neurophenomics, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    2. INBRAIN, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    3. Institute of Psychiatric Research, Department of Psychiatry, Indiana University School of Medicine, Indianapolis, Indiana
    • Assistant Professor of Psychiatry and Medical Neuroscience, Indiana University School of Medicine; Staff Psychiatrist, Indianapolis VA Medical Center, Director, INBRAIN and Laboratory of Neurophenomics, Institute of Psychiatric Research, 791 Union Drive, Indianapolis, IN 46202-4887.
    Search for more papers by this author

  • Please cite this article as follows: Le-Niculescu H, Patel SD, Bhat M, Kuczenski R, Faraone SV, Tsuang MT, McMahon FJ, Schork NJ, Nurnberger Jr JI, Niculescu AB. 2009. Convergent Functional Genomics of Genome-Wide Association Data for Bipolar Disorder: Comprehensive Identification of Candidate Genes, Pathways and Mechanisms. Am J Med Genet Part B 150B:155–181.

Abstract

Given the mounting convergent evidence implicating many more genes in complex disorders such as bipolar disorder than the small number identified unambiguously by the first-generation Genome-Wide Association studies (GWAS) to date, there is a strong need for improvements in methodology. One strategy is to include in the next generation GWAS larger numbers of subjects, and/or to pool independent studies into meta-analyses. We propose and provide proof of principle for the use of a complementary approach, convergent functional genomics (CFG), as a way of mining the existing GWAS datasets for signals that are there already, but did not reach significance using a genetics-only approach. With the CFG approach, the integration of genetics with genomics, of human and animal model data, and of multiple independent lines of evidence converging on the same genes offers a way of extracting signal from noise and prioritizing candidates. In essence our analysis is the most comprehensive integration of genetics and functional genomics to date in the field of bipolar disorder, yielding a series of novel (such as Klf12, Aldh1a1, A2bp1, Ak3l1, Rorb, Rora) and previously known (such as Bdnf, Arntl, Gsk3b, Disc1, Nrg1, Htr2a) candidate genes, blood biomarkers, as well as a comprehensive identification of pathways and mechanisms. These become prime targets for hypothesis driven follow-up studies, new drug development and personalized medicine approaches. © 2008 Wiley-Liss, Inc.

INTRODUCTION

The recent availability of massively parallel genotyping technologies has made genome wide association studies (GWAS) feasible, with initial interesting results reported in a variety of complex disorders [GWAS, 2007; McPherson et al., 2007; Kingsmore et al., 2008; Willer et al., 2008]. However, the number of SNPs identified unambiguously, after correction for multiple comparisons, is relatively small, and the number of known genes unambiguously implicated by them is even smaller [Zeggini et al., 2007]. At least part of the problem facing genetic-only approaches in complex disorders may be related to extreme genetic heterogeneity [Walsh et al., 2008]. Given the mounting convergent evidence implicating many more genes in complex disorders [Walsh et al., 2008; Sun et al., 2008a] than the small number identified by the first-generation GWAS to date, there is a strong need for improvements in methodology. One strategy is to include in the next generation of GWAS larger number of subjects, and/or pool independent studies into meta-analyses [Zeggini et al., 2008]. We propose the use of a complementary approach, convergent functional genomics (CFG) [Niculescu et al., 2000a; Ogden et al., 2004; Le-Niculescu et al., 2007a,b; Le-Niculescu et al., 2008a,b], as a way of mining the existing GWAS datasets for signals that are there already, but did not reach significance using a genetics-only approach. With the CFG approach, the integration of genetics with genomics, of human and animal model data, and of multiple independent lines of evidence converging on the same genes offers a way of extracting signal from noise, and prioritizing candidates for future focused validatory studies-individual candidate gene association studies with more SNPs tested per gene, deep re-sequencing, and/or biological validation such as transgenic animal work [Le-Niculescu et al., 2008b].

As part of a CFG strategy, we have used data from three published GWAS datasets for bipolar disorder [GWAS, 2007; Baum et al., 2008]. We integrated those data with human postmortem brain gene expression data and human blood gene expression data, as well as with relevant animal model brain and blood gene expression data generated by our group [Niculescu et al., 2000a; Ogden et al., 2004; Le-Niculescu et al., 2007b, 2008a,b]. In addition, we have integrated as part of this comprehensive approach other published human genetic (linkage or association) data for bipolar and related disorders to date, and relevant mouse genetic (QTL or transgenic) data. Genes were prioritized based on a scoring of multiple independent lines of evidence, followed by pathway analyses of the top candidate genes. Finally, we have looked at whether the top candidate genes identified by our analysis are represented in a recently published independent GWAS [Sklar et al., 2008].

METHODS

Genome-Wide Association Data for Bipolar Disorder

The GWA data for the bipolar study from the Wellcome Trust is available at http://www.wtccc.org.uk/info/access_to_data_samples.shtml [2007]. The GWA data from NIMH and German studies is available at http://mapgenetics.nimh.nih.gov/bp_pooling [Baum et al., 2008]. We have used the genotypic test P-value (standard analysis). We used two nominal P-value thresholds for SNP selection-a lower stringency threshold (P < 0.05), and a higher stringency threshold (P < 0.001). The GWA data from the STEP-BD study, used as a replication cohort to test our top findings, is available at http://pngu.mgh.harvard.edu/∼purcell/bpwgas. No Bonferroni correction for number of SNPs tested was performed.

Gene Identification

To identify the genes that correspond to the selected SNPs, the lists of SNPs from the GWAS was uploaded to the SNPPER website (http://snpper.chip.org). In the cases where a SNP mapped to a region close to multiple genes, we selected all the genes that were provided by SNPper. SNPs for which no gene was identified were not included in our subsequent analysis.

Human Postmortem Brain Gene Expression

Information about our candidate genes was obtained using GeneCards (http://www.genecards.org), the Online Mendelian Inheritance of Man database (http://ncbi.nlm.nih.gov/entrez/query.fcgi?db=OMIM), as well as database searches using PubMed (http://ncbi.nlm.nih.gov/PubMed) and various combinations of keywords (gene name, bipolar, depression, human, postmortem, brain).

Human Genetic (Linkage, Association) Convergence

To designate convergence for a particular gene, the gene had to map within 10 cM [see Niculescu et al., 2000b for detailed discussion] of a microsatellite marker for which at least one published study showed evidence for linkage for bipolar disorder or depression, or a positive association study for the gene itself was reported in the literature. The University of Southampton's sequence-based integrated map of the human genome (The Genetic Epidemiological Group, Human Genetics Division, University of Southampton: http://cedar.genetics.soton.ac.uk/public_html/) was used to obtain cM locations for both genes and markers. The sex-averaged cM value was calculated and used to determine convergence to a particular marker. For markers that were not present in the Southampton database, the Marshfield database (Center for Medical Genetics, Marshfield, WI: http://research.marshfieldclinic.org/genetics) was used with the NCBI Map Viewer web-site to evaluate linkage convergence.

We have established in the lab manually curated databases of all the published human postmortem and human genetic literature to date on bipolar and related disorders. These large databases have been used in our CFG cross-validation analyses.

Animal Model Brain and Blood Gene Expression Data

For animal model brain and blood gene expression evidence, we have used previously generated data from two different animal models for bipolar disorder developed by our group, one pharmacogenomic and one transgenic [Ogden et al., 2004; Le-Niculescu et al., 2007a,b, 2008a,b].

Mouse Genetic (QTL, Transgenic) Convergence

To search for mouse genetic evidence—quantitative trait loci (QTL) or transgenic—for our candidate genes, we utilized the MGI_3.54—Mouse Genome Informatics (Jackson Laboratory, Bar Harbor, ME) and used the search menu for mouse phenotypes and mouse models of human disease/abnormal behaviors, using the following sub-categories: abnormal emotion/affect behavior and abnormal sleep pattern/circadian rhythm. To designate convergence for a particular gene, the gene had to map within 10 cM of a QTL marker for the abnormal behavior, or a transgenic mouse of the gene itself displayed that behavior.

Convergent Functional Genomics (CFG) Analysis Scoring

Genes from GWAS data that had SNPs with nominal P-values of <0.05 received 1 point; those that had SNPs with nominal P-values of <0.001 received 2 points (see Fig. 1). All other cross-validating lines of evidence (other human data, animal model data) received a maximum of 1 point each (for human genetic data, 0.5 points if it is linkage, 1 point if it is association; for mouse genetic data, 0.5 points if it is QTL, 1 point if it is transgenic). Thus the maximum possible CFG score for each gene is 12 (6 = 2 × 3 points from the three GWAS, and 6 points from the other lines of evidence). As we are interested in discovering signal in GWAS, we weighted data from GWAS more heavily, bringing the data from this one methodological approach on par with the data from all the other methodological approaches combined. It has not escaped our attention that other ways of weighing the scores of line of evidence may give slightly different results in terms of prioritization, if not in terms of the list of genes per se. Nevertheless, we feel this simple scoring system provides a good separation of genes based on our focus on identifying signal in the GWAS.

Figure 1.

Convergent functional genomics. Multiple independent lines of evidence for Bayesian cross-validation of GWAS data.

Pathway Analysis

Ingenuity 6.0 (Ingenuity Systems, Redwood City, CA) was employed to analyze the molecular networks, biological functions and canonical pathways of the top candidate genes resulting from our CFG analysis (Fig. 3), as well as to identify genes in our datasets that are the target of existing drugs (Table IIS).

We have also used another independent pathway analysis package, MetaCore (GeneGo, Encinitas, CA) to analyze genes functions in diseases (Fig. 5).

RESULTS

Top Candidate Genes

In order to minimize false negatives, we initially cast a wide net, using as a filter a minimal requirement for a gene to have both some genetic and some functional genomic evidence (Table IS). We thus generated an initial list of 1,529 unique genes with P < 0.05 in at least one of the three primary GWAS analyzed, that also had some functional (gene expression) evidence (human or animal model data), implicating them in bipolar disorder or depression. Of interest, a similar analysis for a recent independent GWAS (STEP-BD) [Sklar et al., 2008] yielded just 96 additional new genes (see Supplementary Information—Table IS) over the 1,529 we originally identified, suggesting that: (1) with our genetic-genomic filtering of the three GWAS in the primary analysis we are already capturing most of the genes that may be involved in bipolar disorder, with additional studies providing an asymptotic contribution beyond this point; and (2) that the number of genes potentially involved, directly or indirectly, in bipolar disorder may be indeed quite large, up to 10% of the genome.

In order to minimize false positive, we then used a CFG analysis integrating multiple lines of evidence to prioritize this initial list of 1,529 genes, and focused our subsequent analyses on only the top CFG scoring candidate genes. Forty-one genes had a CFG score of 6 and above (≥50% of maximum possible score) (Fig. 2). One hundred thirteen genes had a CFG score of 5 and above (≥2 + 2 + 1 = maximum score for gene expression data in human brain and blood + maximum score for gene expression data in animal models brain and blood + at least one nominal P-value signal in a GWAS) (Table I).

Figure 2.

Top candidate genes for bipolar disorder identified by CFG of GWAS data. CFG score depicted on the right side of the pyramid. Bold font—the gene has human postmortem evidence. Underlined—the gene has additional human genetic evidence beyond the GWAS data. Red—the gene has blood evidence making it a possible biomarker.

Table I. Top Candidate Genes for Bipolar Disorder Identified by Convergent Functional Genomics (CFG) of Genome-Wide Association Studies (GWAS) Data
Gene symbol/nameGWAS WTC P-valueGWAS NIMH P-valueGWAS German P-valueMouse genetic evidence (QTL, TG)Mouse models brain evidence [Ogden et al., 2004; Le-Niculescu et al., 2008b]Mouse models blood evidence [Le-Niculescu et al.,

2008a,b]

Additional human genetic evidence (linkage or association)Human postmortem brain evidenceHuman blood evidence [Le-Niculescu et al., 2008a]CFG score
  1. I, increased; D, decreased in expression. For human blood data: I, increased in high mood (mania); D, decreased in high mood (mania)/increased in low mood (depression). [For human blood data, where references other than Le-Niculescu et al., 2008a are cited, the studies are in lymphoblastoid cell lines without correlation with mood state, I, increased; D, decreased]. In METH, methamphetamine, VPA, valproate; PFC, prefrontal cortex; AMY, amygdala; CP, caudate putamen; NAC, nucleus accumbens; VT, ventral tegmentum; DBP, DBP knock-out mice; NST, nonstressed; ST, stressed; BP, bipolar disorder; MDD, major depressive disorder; TG, transgenic. For additional human genetic evidence, (Assoc)—genetic association evidence; where that is not mentioned, the evidence is only linkage.

  2. Gene symbols underlined are blood biomarker candidate genes.

  3. Bold values signify P < 0.001.

Klf122.76E-036.77E-041.68E-04Abnormal emotion/affect behaviorDBP ST AMY (I) 13q22.1 BP [Potash et al., 2003] (D) [Le-Niculescu et al., 2008a]8.0
Kruppel-like factor 12   Abnormal sleep pattern/circadian rhythmDBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Arntl7.71E-043.84E-023.72E-02(TG) Abnormal Sleep Pattern/ Circadian RhythmPFC Meth (D) [Ogden et al., 2004] 11p15.2 (Assoc) BP [Mansour et al., 2006; Nievergelt et al., 2006](I) BP [Nakatani et al., 2006] 8.0
aryl hydrocarbon receptor nuclear translocator-like          
Bdnf1.05E-023.76E-021.91E-03(TG) Abnormal emotion/affect behaviorPFC Meth (D) [Ogden et al., 2004] 11p14.1 (Assoc) BP [Sklar et al., 2002; Liu et al., 2008, in press](D) MDD [Duman and Monteggia, 2006](D) BP [Karege et al., 2004]8.0
brain-derived neurotrophic factor      (Assoc) MDD [Schumacher et al., 2005](D) BP [Knable et al., 2004; Torrey et al., 2005]  
       BP [McInnes et al., 1996; Detera-Wadleigh et al., 1999; Neves-Pereira et al., 2002]   
Aldh1a11.29E-021.58E-043.34E-02Abnormal sleep pattern/circadian rhythmDBP NST PFC (D)Meth (D)9q21.13 BP [Macgregor et al., 2004](I) BP [Pennington et al., 2007] 8.0
aldehyde dehydrogenase family 1, subfamily A1    DBP ST AMY (I) [Le-Niculescu et al., 2008b]     
A2bp13.42E-054.23E-041.59E-04 VT VPA (D) [Ogden et al., 2004] 16p13.2 BP [Ewald et al., 2002]  7.5
ataxin-2-binding protein 1          
Mbp 8.30E-038.19E-04 DBP NST PFC (D)Meth (I)18q23 BP [Freimer et al., 1996; Ewald et al., 1999; Baron, 2001; Schulze et al., 2003](D) BP [Tkachev et al., 2003; Sun et al., 2006](I) [Le-Niculescu et al., 2008a]7.5
myelin basic protein    DBP ST PFC (D) MDD [Coon et al., 1996](D) Female BP, (I) Male BP [Chambers and Perrone-Bizzozero, 2004]  
     DBP ST AMY (I) [Le-Niculescu et al., 2008b]     
Ak3l19.80E-051.79E-022.57E-02Abnormal emotion/affect behaviorDBP ST AMY (D) 1p31.3 BP [Rice et al., 1997; Ewald et al., 2002](D) MDD [Klempan et al., 2007] 7.0
adenylate kinase 3 alpha-like 1   Abnormal sleep pattern/circadian rhythmDBP ST PFC (I) MDD [Nurnberger et al., 2001]   
     DBP ST AMY (D) [Le-Niculescu et al., 2008b]     
Gsk3b9.82E-031.62E-026.72E-03(TG) Abnormal emotion/affect behaviorCP VPA (D) [Ogden et al., 2004] 3q13.33 (Assoc) BP [Szczepankiewicz et al., 2006; Lachman et al., 2007](D) BP [Nakatani et al., 2006; Vawter et al., 2006] 7.0
glycogen synthase kinase 3 beta    PFC Meth (D) [Ogden et al., 2004] BP [Bailer et al., 2002; Benedetti et al., 2004; Maziade et al., 2005; Nishiguchi et al., 2006](I) MDD [Vawter et al., 2006]  
     DBP NST PFC (D)     
     DBP NST AMY (I) [Le-Niculescu et al., 2008b]     
Nrcam1.63E-035.94E-048.60E-04Abnormal sleep pattern/circadian rhythmDBP NST AMY (I) [Le-Niculescu et al., 2008b] 7q31.1 BP [Detera-Wadleigh et al., 1999; Evans et al., 2007]  7.0
neuronal cell adhesion molecule          
Pcdh99.77E-031.19E-034.80E-04Abnormal emotion/affect behaviorDBP NST AMY (I) [Le-Niculescu et al., 2008b] 13q21.32 BP [Potash et al., 2003](D) MDD [Klempan et al., 2007] 7.0
Protocadherin 9   Abnormal sleep pattern/circadian rhythm      
Cd443.48E-023.94E-031.06E-02 CP Meth (I) [Ogden et al., 2004]Meth (D)11p13 BP [McInnes et al., 1996] (I) BP [Middleton et al., 2005]6.5
CD44 antigen (homing function and Indian blood group system)          
Kcnk11.89E-027.60E-033.47E-04   1q42.2 BP [Curtis et al., 2003; Macgregor et al., 2004](D) BP [Jurata et al., 2004](I) BP [Matigian et al., 2007]6.5
potassium channel, subfamily K, member 1          
Mbnl22.94E-034.64E-024.02E-04 AMY VPA (D) [Ogden et al., 2004]DBP NST (D) [Le-Niculescu et al., 2008b]13q32.1 BP [Liu et al., 2003; Maziade et al., 2005; Goes et al., 2007]  6.5
muscleblind-like 2 (Drosophila)          
Nav24.16E-035.77E-042.04E-03(TG) Abnormal emotion/affect behavior  11p15.1 BP [Detera-Wadleigh et al., 1999](D) BP [Kim et al., 2007] 6.5
neuron navigator 2          
Nos11.72E-023.73E-024.56E-02Abnormal emotion/affect behaviorDBP NST AMY (D) [Le-Niculescu et al., 2008b] 12q24.22 (Assoc) BP [Fallin et al., 2005](I) BP [Benes et al., 2006] 6.5
Nitric oxide synthase 1, neuronal (Nos1), mRNA   Abnormal sleep pattern/circadian rhythm  BP [Morissette et al., 1999; Chagnon et al., 2004; Fallin et al., 2005]   
Oprm17.82E-047.31E-031.90E-03(TG)  6q25.2 BP [Cheng et al., 2006](I) BP [Ryan et al., 2006] 6.5
Opioid receptor, mu 1   Abnormal emotion/affect behavior      
Pcdh74.08E-041.71E-028.05E-04 AMY VPA (D) [Ogden et al., 2004] 4p15.1 BP [Detera-Wadleigh et al., 1999; Lambert et al., 2005]  6.5
Protocadherin 7          
Prkce4.59E-032.37E-041.20E-02(TG) Abnormal emotion/affect behavior  2p21 BP [Etain et al., 2006](D) BP [Torrey et al., 2005] 6.5
protein kinase C, epsilon          
Ptprm1.74E-021.10E-022.41E-04   18p11.23 BP [Segurado et al., 2003](I) BP [Nakatani et al., 2006](D) [Le-Niculescu et al., 2008a]6.5
protein tyrosine phosphatase, receptor type, M          
Qki3.06E-02 7.74E-05 CP VPA (I) [Ogden et al., 2004] 6q26 BP [Cheng et al., 2006](D) MDD [Klempan et al., 2007](D) BP [Matigian et al., 2007]6.5
quaking homolog, KH domain RNA binding (mouse)    DBP AMY (I) [Le-Niculescu et al., 2008b]     
Rora1.90E-043.55E-046.36E-03 DBP NST AMY (I) 15q21-q22 MDD [Zubenko et al., 2002]  6.5
RAR-related orphan receptor alpha    DBP ST AMY (I)     
     DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Rorb1.29E-025.88E-041.95E-02(TG) Abnormal emotion/affect behaviorDBP ST AMY(I) 9q21.13 BP [Macgregor et al., 2004]  6.5
RAR-related orphan receptor beta    DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Ryr31.21E-032.89E-046.09E-03(TG) Abnormal emotion/affect behaviorCP VPA (I) [Ogden et al., 2004] 15q13.3 MDD [Levinson et al., 2007]  6.5
ryanodine receptor 3          
Cacna1a2.99E-022.12E-027.04E-04Abnormal emotion/affect behavior  19p13.13 MDD [Zubenko et al., 2003](D) BP [Iwamoto et al., 2004] 6.0
calcium channel, voltage-dependent, P/Q type, alpha 1A subunit          
Cdh135.89E-032.50E-039.08E-04Abnormal emotion/affect behaviorDBP NST AMY (D) [Le-Niculescu et al., 2008b] 16q23.3 BP [Etain et al., 2006]  6.0
cadherin 13          
Dapk14.02E-025.97E-054.04E-02Abnormal emotion/affect behaviorAMY VPA (D) [Ogden et al., 2004] 9q21.33 BP [Segurado et al., 2003]  6.0
death-associated protein kinase 1          
Disc11.31E-022.99E-036.08E-03(TG) Abnormal emotion/affect behavior  1q42.2 (Assoc) BP [Hodgkinson et al., 2004; Maeda et al., 2006; Millar et al., 2007; Hennah et al., 2008] (D) BP [Maeda et al., 2006]6.0
disrupted in schizophrenia 1      BP [Curtis et al., 2003; Macgregor et al., 2004]   
Gria11.47E-026.55E-039.19E-03Abnormal emotion/affect behaviorVT Meth (D) [Ogden et al., 2004] 5q33.2 BP [Morissette et al., 1999; Sklar et al., 2004; Etain et al., 2006](I) BP, MDD [Choudary et al., 2005] 6.0
glutamate receptor, ionotropic, AMPA1 (alpha 1)          
Grik15.39E-042.79E-033.36E-02Abnormal emotion/affect behavior  21q21.3 BP [Detera-Wadleigh et al., 1999; Morissette et al., 1999](D) BP [Iwamoto et al., 2004; Nakatani et al., 2006] 6.0
glutamate receptor, ionotropic, kainate 1       (I) DLPFC -MDD [Choudary et al., 2005]  
        (D) AnCg –BP [Choudary et al., 2005]  
Htr2a1.86E-024.52E-021.65E-03(TG) Abnormal emotion/affect behavior  13q14.2 (Assoc) BP [Ranade et al., 2003]0, [Ranade et al., 2003](D) BP [Knable et al., 2004; Torrey et al., 2005] 6.0
Seratonin receptor 2A   Abnormal sleep pattern/circadian rhythm  BP [Arranz et al., 1997; Badenhop et al., 2002](D) MDD [Klempan et al., 2007]  
Kcnd25.78E-034.08E-035.24E-05Abnormal emotion/affect behaviorDBP ST PFC (D) [Le-Niculescu et al., 2008b] 7q31.31 BP [Detera-Wadleigh et al., 1999; Evans et al., 2007]  6.0
Potassium voltage-gated channel, Shal-related family, member 2 (Kcnd2), mRNA          
Lmo76.62E-051.11E-028.17E-03Abnormal emotion/affect behavior  13q22.2 BP [Potash et al., 2003] (D) Anti-depressant treatment [Kalman et al., 2005]6.0
LIM domain only 7   Abnormal sleep pattern/circadian rhythm      
Mycbp25.66E-042.92E-022.39E-02Abnormal emotion/affect behavior  13q22.3 BP [Potash et al., 2003]; MDD [Zubenko et al., 2003](I) BP [Pennington et al., 2007] 6.0
MYC binding protein 2          
Myt1l2.25E-041.31E-021.25E-02Abnormal sleep pattern/circadian rhythmDBP ST PFC (D) [Le-Niculescu et al., 2008b] 2p25.3 BP [Detera-Wadleigh et al., 1999]  6.0
myelin transcription factor 1-like          
Nrg11.07E-052.19E-034.51E-03   8p12 (Assoc) BP [Green et al., 2005; Walss-Bass et al., 2006; Thomson et al., 2007](I) BP [Tkachev et al., 2003] 6.0
neuregulin 1      BP [Cichon et al., 2001; Zubenko et al., 2003; Park et al., 2004](D) MDD [Bertram et al., 2007]  
Scamp11.71E-021.31E-022.46E-03 DBP ST PFC (D) [Le-Niculescu et al., 2008b]DBP NST (D) [Le-Niculescu et al., 2008b]5q14.1 (D) [Le-Niculescu et al., 2008a]6.0
secretory carrier membrane protein 1          
Slc8a14.57E-032.77E-042.28E-02Abnormal emotion/affect behaviorDBP ST AMY (I) 2p22.1 BP [Etain et al., 2006]  6.0
solute carrier family 8 (sodium/calcium exchanger), member 1    DBP ST AMY (D) [Le-Niculescu et al., 2008b]     
Syn31.67E-044.94E-034.17E-03   22q12.3 (Assoc) BP [Lachman et al., 2006](D) BP [Vawter et al., 2002] 6.0
synapsin IIIa      BP [Kelsoe et al., 2001; Potash et al., 2003; Lachman et al., 2006]   
Tiam17.39E-051.82E-032.65E-03Abnormal emotion/affect behavior  21q22.11 BP [Detera-Wadleigh et al., 1999; Morissette et al., 1999](D) MDD [Aston et al., 2005] 6.0
T-cell lymphoma invasion and metastasis 1          
Tshz21.98E-028.22E-033.58E-04Abnormal emotion/affect behavior  20q13.2 BP [Radhakrishna et al., 2001] (D) [Le-Niculescu et al., 2008a]6.0
teashirt family zinc finger 2          
Zhx22.47E-032.86E-021.69E-03Abnormal emotion/affect behavior  8q24.13 BP [Cichon et al., 2001; Badenhop et al., 2002; Park et al., 2004](D) BP [Kim et al., 2007](I) BP [Middleton et al., 2005]6.0
Zinc fingers and homeoboxes 2          
Acacb2.94E-027.84E-041.42E-03   12q24.11 BP [Chagnon et al., 2004; Maziade et al., 2005] (D) [Le-Niculescu et al., 2008a]5.5
acetyl-Coenzyme A carboxylase beta          
App3.37E-029.86E-037.81E-03(TG) Abnormal emotion/affect behavior  21q21.3 BP [Morissette et al., 1999](I) BP [Jurata et al., 2004] 5.5
amyloid beta (A4) precursor protein   (TG) Abnormal sleep pattern/circadian rhythm      
Atxn11.11E-035.55E-036.58E-03 CP METH (D) [Ogden et al., 2004] 6p22.3 BP [Turecki et al., 2001] (I) [Le-Niculescu et al., 2008a]5.5
Ataxin 1    DBP PFC (D) [Le-Niculescu et al., 2008b]     
C14orf1452.27E-041.89E-021.03E-03   14q31.1 BP [Segurado et al., 2003] (D) [Le-Niculescu et al., 2008a]5.5
chromosome 14 open reading frame 145          
C18orf11.16E-044.21E-033.04E-03   18p11.21 BP [Detera-Wadleigh et al., 1999; Baron, 2001] (D) [Le-Niculescu et al., 2008a]5.5
Chromosome 18 open reading frame 1          
Cacnb22.40E-096.57E-034.23E-02 AMY VPA (D) [Ogden et al., 2004] 10p12.33 BP [Rice et al., 1997; Faraone et al., 1998; Foroud et al., 2000; Baron, 2001; McInnis et al., 2003; Lambert et al., 2005; Etain et al., 2006]  5.5
calcium channel, voltage-dependent, beta 2 subunit    CP VPA (I) [Ogden et al., 2004]     
     DBP NST AMY (D) [Le-Niculescu et al., 2008b]     
Camk2a 1.76E-023.62E-02(TG) Abnormal emotion/affect behaviorDBP NST AMY (I) 5q32 BP [Sklar et al., 2004; Etain et al., 2006](D) BP [Xing et al., 2002] 5.5
calcium/calmodulin-dependent protein kinase II alpha   (TG) Abnormal sleep pattern/circadian rhythmDBP ST PFC (I) [Le-Niculescu et al., 2008b]  (I) MDD [Novak et al., 2006]  
        (I) MDD [Tochigi et al., 2008]  
Camk2d1.69E-021.20E-032.90E-03 DBP ST PFC (I) [Le-Niculescu et al., 2008b] 4q26 BP [Lambert et al., 2005] (I) [Le-Niculescu et al., 2008a]5.5
calcium/calmodulin-dependent protein kinase II, delta          
Celsr11.85E-038.84E-044.85E-02Abnormal emotion/affect behavior  22q13.31(I) BP [Ryan et al., 2006] 5.5
Cadherin, EGF LAG seven-pass G-type receptor 1 (flamingo homolog, Drosophila)   Abnormal sleep pattern/circadian rhythm      
Clstn27.57E-034.25E-041.33E-02 DBP NST AMY (I) [Le-Niculescu et al., 2008b] 3q23 BP [Dick et al., 2003]  5.5
calsyntenin 2          
Crebbp5.02E-031.39E-033.64E-03(TG) Abnormal emotion/affect behaviorDBP ST PFC (D) [Le-Niculescu et al., 2008b] 16p13.3 BP [Ewald et al., 2002]  5.5
CREB binding protein          
Cugbp22.84E-053.38E-032.66E-02   10p14 MDD [Zubenko et al., 2003] BP [Etain et al., 2006] (I) BP [Matigian et al., 2007]5.5
CUG triplet repeat, RNA binding protein 2          
Dcamkl18.55E-032.36E-035.27E-03 AMY VPA (D) [Ogden et al., 2004]DBP NST (D) [Le-Niculescu et al., 2008b]13q13.3 BP [Maziade et al., 2005]  5.5
doublecortin and CaM kinase-like 1          
Diaph12.62E-024.70E-023.38E-03 CP VPA (I) [Ogden et al., 2004] 5q31.3 MDD [Zubenko et al., 2003] (I) BP [Matigian et al., 2007]5.5
diaphanous homolog 1 (Drosophila)          
Dpp101.31E-051.67E-032.70E-03 DBP NST AMY (I) [Le-Niculescu et al., 2008b] 2q14.1 BP [Maziade et al., 2005; Etain et al., 2006]  5.5
dipeptidylpeptidase 10          
Eif2c21.81E-02 2.48E-04 DBP ST AMY (I) 8q24.3 BP [Segurado et al., 2003]  5.5
eukaryotic translation initiation factor 2C, 2    DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Fam13a13.37E-034.77E-053.94E-02   4q22.1 BP [Curtis et al., 2003] (D) [Le-Niculescu et al., 2008a]5.5
family with sequence similarity 13, member A1          
Fgf126.14E-042.50E-039.57E-03   3q28 BP [Bailer et al., 2002; Liu et al., 2003; Schosser et al., 2004; Maziade et al., 2005](D) MDD [Evans et al., 2004] 5.5
fibroblast growth factor 12          
FLJ109869.77E-032.09E-032.29E-04   1p32.1 BP [Cichon et al., 2001] (I) [Le-Niculescu et al., 2008a]5.5
hypothetical protein FLJ10986          
Foxp14.80E-039.66E-045.33E-03 DBP NST AMY (D) 3p13 BP [McInnes et al., 1996; Etain et al., 2006; Evans et al., 2007]  5.5
Forkhead box P1 (Foxp1), mRNA    DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Fut94.03E-036.07E-045.34E-03 CP Meth (I) [Ogden et al., 2004] 6q16.1 BP [Schulze et al., 2004; Lambert et al., 2005; Goes et al., 2007] MDD [Camp et al., 2005]  5.5
fucosyltransferase 9 (alpha (1,3) fucosyltransferas)    DBP NST PFC (I)     
     DBP ST PFC (D)     
     DBP ST AMY (I) [Le-Niculescu et al., 2008b]     
Gnai14.98E-037.55E-031.55E-02 DBP ST PFC (D) [Le-Niculescu et al., 2008b] 7q21.11 BP [Lambert et al., 2005](D) BP [Jurata et al., 2004] 5.5
guanine nucleotide binding protein, alpha inhibiting 1          
Grm11.28E-033.67E-035.74E-03(TG) Abnormal emotion/affect behavior  6q24.3 BP [Rice et al., 1997; Ewald et al., 2002](D) BP [Iwamoto et al., 2004] 5.5
glutamate receptor, metabotropic 1          
Grm33.43E-023.18E-037.36E-03 PFC VPA (D) [Ogden et al., 2004] 7q21.12 BP [Lambert et al., 2005; Etain et al., 2006](I) BP [Choudary et al., 2005] 5.5
glutamate receptor, metabotropic 3    DBP ST AMY (I) [Le-Niculescu et al., 2008b]  (D) MDD [Klempan et al., 2007]  
Gsta21.14E-031.93E-031.89E-03  VPA (D)6p12.2 BP [Lambert et al., 2005](I) BP [Benes et al., 2006] 5.5
glutathione S-transferase, alpha 2 (Yc2)          
Iqgap28.17E-035.83E-036.65E-04Abnormal emotion/affect behaviorDBP ST AMY (D) [Le-Niculescu et al., 2008b] 5q13.3  5.5
IQ motif and Sec7 domain 1          
Itgav4.68E-021.09E-021.56E-02 DBP NST AMY (I) [Le-Niculescu et al., 2008b] 2q32.1 BP [Cichon et al., 2001] (I) BP [Middleton et al., 2005]5.5
integrin beta 1 (fibronectin receptor beta)          
Kif1A5.31E-046.77E-031.00E-02 CP VPA (I) [Ogden et al., 2004] 2q37.3 BP [Lambert et al., 2005]  5.5
kinesin family member 1A          
Ndufs24.27E-021.08E-024.67E-02 AMY VPA (I) [Ogden et al., 2004] 1q23 BP [Fallin et al., 2004] (D) BP [Middleton et al., 2005]5.5
NADH dehydrogenase (ubiqui) Fe-S protein 2, 49kDa (NADH-coenzyme Q reductase)          
Nfib 3.47E-031.44E-04 DBP ST AMY (I) [Le-Niculescu et al., 2008b] 9p24.1 BP [Segurado et al., 2003] (D) [Le-Niculescu et al., 2008a]5.5
nuclear factor I/B          
Nr3c14.03E-033.71E-022.96E-02(TG) Abnormal emotion/affect behavior  5q31.3 BP [Etain et al., 2006] MDD [van West et al., 2006](D) BP,MDD [Torrey et al., 2005] 5.5
nuclear receptor subfamily 3, group C, member 1 (glucocorticoid receptor)       (D) BP [Knable et al., 2004]  
        (I) MDD [Sequeira et al., 2007]  
Pde10a1.50E-029.64E-031.50E-03(TG) Abnormal emotion/affect behaviorDBP NST AMY (D) 6q27 BP [Cheng et al., 2006]  5.5
phosphodiesterase 10A    DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Pftk16.54E-041.55E-032.26E-03 DBP ST AMY (D) [Le-Niculescu et al., 2008b] 7q21.13 BP [Lambert et al., 2005; Etain et al., 2006]  5.5
PFTAIRE protein kinase 1          
Pik3r1 6.99E-049.97E-03Abnormal emotion/affect behaviorDBP ST PFC (D) [Le-Niculescu et al., 2008b] 5q13.1(D) MDD [Aston et al., 2005] 5.5
phosphatidylinositol 3-kinase, regulatory subunit, polypeptide 1 (p85 alpha)          
Plxna22.98E-024.71E-043.18E-02   1q32.2 BP [Segurado et al., 2003] (D) [Le-Niculescu et al., 2008a]5.5
Plexin A2          
Ptn2.85E-021.90E-024.56E-03 CP Meth (I) [Ogden et al., 2004] 7q33 BP [Segurado et al., 2003](I) MDD [Tochigi et al., 2008] 5.5
pleiotrophin (heparin binding growth factor 8, neurite growth-promoting factor 1)          
Ptprt6.27E-033.45E-031.12E-02 DBP ST AMY (I) [Le-Niculescu et al., 2008b] 20q12 BP [Radhakrishna et al., 2001](D) MDD [Aston et al., 2005] 5.5
Protein tyrosine phosphatase, receptor type, T       (I) MDD-suicide [Sequeira et al., 2007]  
Rasgrf21.27E-022.35E-029.06E-04Abnormal emotion/affect behavior  5q14.1 (D) [Le-Niculescu et al., 2008a]5.5
Ras protein-specific guanine nucleotide-releasing factor 2          
Sod1  1.58E-02(TG) Abnormal emotion/affect behaviorAMY VPA (I) [Ogden et al., 2004]DBP NST (D) [Le-Niculescu et al., 2008b]21q22.11 BP [Detera-Wadleigh et al., 1999; Morissette et al., 1999](D) BP [Benes et al., 2006] 5.5
superoxide dismutase 1, soluble          
Spast9.86E-035.03E-033.76E-02  Meth (D)2p22.3 BP [Etain et al., 2006](I) BP [Nakatani et al., 2006] 5.5
spastin          
Syne11.92E-051.31E-033.29E-03   6q25.1 BP [Cheng et al., 2006] (D) [Le-Niculescu et al., 2008a]5.5
synaptic nuclear envelope 1          
Tnik1.67E-027.43E-047.05E-03   3q26.2 BP [Cichon et al., 2001] (I) BP [Matigian et al., 2007]5.5
TRAF2 and NCK interacting kinase          
Trpm36.42E-033.49E-042.61E-03  VPA (D)9q21.13 BP [Macgregor et al., 2004]  5.5
transient receptor potential cation channel, subfamily M, member 3          
Zdhhc144.09E-034.59E-033.56E-02 DBP ST AMY (I) [Le-Niculescu et al., 2008b] 6q25.3 BP [Cheng et al., 2006] (D) [Le-Niculescu et al., 2008a]5.5
zinc finger, DHHC domain containing 14          
Adcy11.88E-021.18E-033.58E-02(TG) Abnormal emotion/affect behavior  7p13(I) BP [Bezchlibnyk et al., 2001] 5.0
adenylate cyclase 1          
Adcyap12.38E-021.32E-02 (TG) Abnormal emotion/affect behaviorVT VPA (D) [Ogden et al., 2004] 18p11.32 (Assoc) BP [Ishiguro et al., 2001]  5.0
adenylate cyclase activating polypeptide 1   (TG) Abnormal sleep pattern/circadian rhythm      
Ank24.77E-041.34E-028.90E-03Abnormal emotion/affect behaviorDBP ST PFC (I) [Le-Niculescu et al., 2008b] 4q25 BP [Lambert et al., 2005]  5.0
ankyrin 2, brain          
Chrna7 2.03E-031.33E-02(TG) Abnormal emotion/affect behavior  15q13.3 (Assoc) BP [Hong et al., 2004]BP [De Luca et al., 2006] 5.0
cholinergic receptor, nicotinic, alpha 7      MDD [Lai et al., 2001; Levinson et al., 2007]   
Drd2 1.20E-025.78E-03(TG) Abnormal emotion/affect behavior  11q23.2 (Assoc) BP [Li et al., 1999; Massat et al., 2002](I) BP [Ryan et al., 2006] 5.0
dopamine receptor 2      BP [Craddock et al., 1995; Peroutka et al., 1998; Serretti et al., 2000](D) MDD [Torrey et al., 2005]  
Dst2.56E-023.29E-024.12E-03(TG) Abnormal emotion/affect behavior  6p12.1 (D) [Le-Niculescu et al., 2008a]5.0
dystonin          
Elavl22.26E-024.47E-034.53E-02Abnormal emotion/affect behaviorAMY VPA (D) [Ogden et al., 2004] 9p21.3 BP [Lambert et al., 2005; McQueen et al., 2005]  5.0
ELAV (embryonic lethal, abnormal vision, Drosophila)-like 2 (Hu antigen B)   Abnormal sleep pattern/circadian rhythm      
Epha53.28E-021.61E-021.88E-02Abnormal emotion/affect behavior  4q13.1 BP [Zubenko et al., 2003; Lambert et al., 2005](D) MDD [Aston et al., 2005] 5.0
EPH receptor A5      BP [Etain et al., 2006]   
Gaa1.48E-022.91E-021.01E-02Abnormal emotion/affect behaviorDBP NST AMY (I) [Le-Niculescu et al., 2008b] 17q25.3 MDD [Curtis et al., 2003; Camp et al., 2005] BP [Dick et al., 2003; Schulze et al., 2004]  5.0
glucosidase, alpha, acid          
Gna126.67E-031.57E-023.18E-03Abnormal emotion/affect behavior  7p22.2 MDD [Camp et al., 2005] (I) BP [Middleton et al., 2005]5.0
guanine nucleotide binding protein, alpha 12   Abnormal sleep pattern/circadian rhythm      
Hmox1 2.87E-021.89E-05Abnormal emotion/affect behavior  22q12.3 BP [Detera-Wadleigh et al., 1999; Baron, 2001; Kelsoe et al., 2001; Potash et al., 2003](I) BP [Benes et al., 2006] 5.0
heme oxygenase (decycling) 1          
Impa23.93E-023.18E-021.44E-02   18p11.21 (Assoc) BP [Sjoholt et al., 2004; Ohnishi et al., 2007](D) BP [Yoon et al., 2001] 5.0
inositol monophosphatase (IMPase)          
Kcnab11.65E-026.37E-032.39E-02Abnormal emotion/affect behaviorVT VPA (I) [Ogden et al., 2004] 3q25.31 BP [Badenhop et al., 2002; Curtis et al., 2003]  5.0
potassium voltage-gated channel, shaker-related subfamily, beta member 1    DBP NST AMY (D) DBP ST PFC (D) [Le-Niculescu et al., 2008b]     
Kcnb11.61E-031.90E-022.25E-03Abnormal emotion/affect behaviorDBP NST PFC (I) 20q13.13 BP [Radhakrishna et al., 2001]  5.0
potassium voltage gated channel, Shab-related subfamily, member 1    DBP NST AMY (I)     
     DBP ST PFC (D)     
     DBP ST AMY (I) [Le-Niculescu et al., 2008b]     
Large4.32E-033.50E-032.75E-03Abnormal emotion/affect behavior  22q12.3 BP [Detera-Wadleigh et al., 1999; Baron, 2001; Kelsoe et al., 2001; Potash et al., 2003] (D) Anti-depressant treatment [Kalman et al., 2005]5.0
like-glycosyltransferase          
Lef1 3.84E-042.23E-02Abnormal emotion/affect behavior  4q25 BP [Lambert et al., 2005](D) BP [Benes, 2007] 5.0
lymphoid enhancer-binding factor 1          
Mdh1 8.45E-04 Abnormal emotion/affect behavior VPA (D)2p15 BP [Liu et al., 2003; Maziade et al., 2005](D) BP [Jurata et al., 2004] 5.0
malate dehydrogenase 1, NAD (soluble)       (I) MDD [Beasley et al., 2006]  
Ncam12.77E-022.61E-028.62E-03   11q23.1 (Assoc) BP [Arai et al., 2004; Atz et al., 2007](D) BP [Atz et al., 2007] 5.0
Neural cell adhesion molecule 1       (D) MDD [Tochigi et al., 2008]  
Nfia3.96E-028.70E-031.09E-02Abnormal emotion/affect behaviorDBP NST AMY (I) [Le-Niculescu et al., 2008b] 1p31.3 BP [Cichon et al., 2001]  5.0
nuclear factor I/A   Abnormal sleep pattern/circadian rhythm      
Olig21.49E-028.96E-038.47E-03Abnormal emotion/affect behavior  21q22.11 BP [Detera-Wadleigh et al., 1999; Morissette et al., 1999](D) BP [Tkachev et al., 2003] 5.0
oligodendrocyte lineage transcription factor 2       (D) MDD [Aston et al., 2005]  
Pard31.58E-023.48E-021.38E-02Abnormal emotion/affect behavior  10p11.22 BP [Rice et al., 1997](I) BP [Ryan et al., 2006] 5.0
Par-3 partitioning defective 3 homolog (C. elegans)          
Pdlim51.39E-031.73E-031.50E-03   4q22.3 (Assoc) BP [Kato et al., 2005](D) MDD [Iga et al., 2006] 5.0
PDZ and LIM domain 5          
Ppm1b7.73E-034.62E-021.31E-02Abnormal emotion/affect behaviorCP VPA (I) [Ogden et al., 2004] 2p21 BP [Etain et al., 2006]  5.0
protein phosphatase 1B, magnesium dependent, beta isoform          
Ptprk2.50E-021.37E-031.54E-03Abnormal emotion/affect behaviorDBP ST AMY (D) [Le-Niculescu et al., 2008b] 6q22.33 BP [Park et al., 2004]  5.0
protein tyrosine phosphatase, receptor type, K          
Rxrg1.43E-031.83E-023.04E-02Abnormal emotion/affect behaviorDBP ST PFC (D) [Le-Niculescu et al., 2008b] 1q23.3 BP [Fallin et al., 2004]  5.0
retinoid X receptor gamma   Abnormal sleep pattern/circadian rhythm      
Sparc 1.11E-024.55E-02Abnormal emotion/affect behaviorNAC Meth (I) [Ogden et al., 2004] 5q33.1 BP [Morissette et al., 1999; Etain et al., 2006](I) BP [Iwamoto et al., 2004] 5.0
secreted protein, acidic, cysteine-rich (osteonectin)    DBP NST AMY (D) [Le-Niculescu et al., 2008b]     
Stk247.83E-031.70E-027.95E-03Abnormal emotion/affect behavior Meth (D)13q32.2 BP [Detera-Wadleigh et al., 1999; Kelsoe et al., 2001; Liu et al., 2003; Maziade et al., 2005]  5.0
serine/threonine kinase 24 (STE20 homolog, yeast)          
Tpst24.36E-036.59E-034.67E-02Abnormal emotion/affect behavior  22q12.1 BP [Kelsoe et al., 2001](I) BP [Nakatani et al., 2006] 5.0
Tyrosylprotein sulfotransferase 3   Abnormal sleep pattern/circadian rhythm      

As a way of testing the validity of our approach, we have examined if our top findings were over-represented in an independent GWAS of bipolar disorder [Sklar et al., 2008]. Thirty of the top 41 genes identified by our approach had a P-value of <0.05 in that independent study, an estimated fourfold enrichment over what would be expected by chance alone in that study (see Table II).

Table II. Replication of Findings
Gene symbol/nameCFG scoreP

-value <0.05 in an independent GWAS [Sklar et al., 2008]

  1. Examination of our top candidate genes from Figure 2 in an independent bipolar GWAS [Sklar et al., 2008]. Thirty of our top 41 genes had a P < 0.05 in the Sklar et al. study. As there were 3,654 genes at P < 0.05 in that study, and the number of genes in the human genome is estimated at 20,500 [Clamp et al., 2007], the enrichment factor provided by our approach is (30/41)/(3654/20500) = 4.1-fold.

Klf12/Kruppel-like factor 128.0 
Arntl/aryl hydrocarbon receptor nuclear translocator-like8.00.0255
Bdnf/brain-derived neurotrophic factor8.0 
Aldh1a1/aldehyde dehydrogenase family 1, subfamily A18.0 
A2bp1/ataxin-2-binding protein 17.50.004176
Mbp/myelin basic protein7.50.001165
Ak3l1/adenylate kinase 3 alpha-like 17.0 
Gsk3b/glycogen synthase kinase 3 beta7.0 
Nrcam/neuronal cell adhesion molecule7.00.04352
Pcdh9/Protocadherin 97.0 
Cd44/CD44 antigen6.5 
Kcnk1/potassium channel, subfamily K, member 16.50.04384
Mbnl2/muscleblind-like 26.50.01614
Nav2/neuron navigator 26.50.001869
Nos1/Nitric oxide synthase 16.50.02122
Oprm1/Opioid receptor, mu 16.50.02105
Pcdh7/Protocadherin 76.5 
Prkce/protein kinase C, epsilon6.50.02484
Ptprm/protein tyrosine phosphatase, receptor type, M6.50.0101
Qki/quaking homolog, KH domain RNA binding6.5 
Rora/RAR-related orphan receptor alpha6.50.01628
Rorb/RAR-related orphan receptor beta6.50.0008992
Ryr3/ryanodine receptor 36.50.008071
Cacna1a/calcium channel, voltage-dependent, P/Q type, alpha 1A subunit6.00.002702
Cdh13/cadherin 136.00.00801
Dapk1/death-associated protein kinase 16.00.001561
Disc1/disrupted in schizophrenia 16.00.008606
Gria1/glutamate receptor, ionotropic, AMPA1 (alpha 1)6.00.006843
Grik1/glutamate receptor, ionotropic, kainate 16.00.04468
Htr2a/Seratonin receptor 2A6.00.005598
Kcnd2/Potassium voltage-gated channel, Shal-related family, member 2 (Kcnd2), mRNA6.00.03855
Lmo7/LIM domain only 76.00.006589
Mycbp2/MYC binding protein 26.0 
Myt1l/myelin transcription factor 1-like6.00.01648
Nrg1/neuregulin 16.00.0008814
Scamp1/secretory carrier membrane protein 16.00.02253
Slc8a1/solute carrier family 8 (sodium/calcium exchanger), member 16.00.007436
Syn3/synapsin IIIa6.00.02029
Tiam1/T-cell lymphoma invasion and metastasis 16.00.002492
Tshz2/teashirt family zinc finger 26.00.01729
Zhx2/Zinc fingers and homeoboxes 26.0 

Candidate Blood Biomarkers

Of the top candidate genes from Table I (see also Fig. 2), 32 out of 113 have prior blood gene expression evidence implicating them as potential blood biomarkers. The additional evidence provided by GWAS data indicates a genetic rather than purely environmental (medications, stress) basis for their alteration in disease, and their potential utility as trait rather than purely state markers.

Pathways and Mechanisms

We classified our top candidate genes from Table I into biological groups of interest previously reported to have relevance to the pathophysiology of bipolar and related disorders (see Table III). Ingenuity pathway analysis was carried out on the top 41 genes (Fig. 3A), as well as on the more extensive list of 113 top genes (Fig. 3B). Ingenuity was employed to analyze the molecular networks, biological functions and canonical pathways of the top candidate genes resulting from our CFG analysis (Fig. 3A,B), as well as to identify genes in our datasets that are the target of existing drugs (Table IIS). We have also used another independent pathway analysis package, MetaCore (GeneGo, Encinitas, CA) to analyze genes functions in diseases (Fig. 5). Finally, a model summarizing the data is depicted in Figure 4.

Table III. Top Candidate Genes and Biological RolesThumbnail image of
  • Top candidate genes (CFG score 5 and above—Table I) were classified into biological groups of interest previously reported to have relevance to the pathophysiology of bipolar and related disorders. Blue dots indicate there is also data showing alterations in expression of that gene in brains from subjects with bipolar and related disorders. Red dots indicate there is also data showing alterations in expression of that gene in bloods from subjects with bipolar and related disorders.

  • Figure 3.

    Ingenuity Pathway Analysis of top candidate genes. A: Analysis of top 41 candidate genes (CFG score of 6 and above). B: Analysis of top 113 candidate genes (CFG score of 5 and above).

    Figure 4.

    A comprehensive model for bipolar disorder pathophysiology.

    DISCUSSION

    Our CFG approach helped prioritize, as expected, genes for which there was consistent evidence among the three GWAS datasets, or stronger evidence in one or another of the datasets. However, it also prioritized genes with weaker evidence in the GWAS data, but with strong independent evidence in terms of gene expression studies and other prior human or animal genetic work.

    At the top of our list of candidate genes we have four genes: Arntl, Bdnf, Aldh1a1, and Klf12. Notably, of the four top candidate genes for bipolar disorder identified by our combined approach (Klf12, Arntl, Bdnf, Aldh1a1) (Fig. 2), one of them—Klf12 (Kruppel-like factor 12), had not been previously suspected to be involved in bipolar disorder, or indeed in neuropsychiatric disorders. It shows modest but consistent signal (P < 10−3, 10−4) across all three primary GWAS datasets. Klf12 maps to a mouse QTL for abnormal emotion/affect behavior, and to a linkage locus on chromosome 13q22.1 previously implicated in bipolar disorder [Potash et al., 2003]. Klf12 is a transcription factor, more specifically a zinc finger transcriptional repressor. Other transcription factor top candidate genes identified by our analysis include Mytl1, Tshz2, and Zhx2 (Fig. 2, and Tables I and III). Transcription factors are particularly interesting as effectors of broad phenotypic changes, due to the large number of genes they regulate. It is thus possible that by themselves, or in oligogenic combinations, they can account for complex disorders such as bipolar disorder. In our own animal model work, Klf12 was inversely changed in the pre-frontal cortex (decreased) and the amygdala (increased) of Dbp KO ST manic-like mice [Le-Niculescu et al., 2008b]. We have also identified Klf12 as a candidate blood biomarker in recent human studies, increased in expression in low mood (depression) [Le-Niculescu et al., 2008a]. The model that emerges, then, is that Klf12 may be involved in suppressing genes involved in elevated mood. Gain of function mutations or promoter mutations that lead to overexpression are likely to manifest as depressive phenotypes, and loss of function mutations or promoter mutations that lead to decreased expression, as manic phenotypes.

    Arntl (aryl hydrocarbon receptor nuclear translocator-like), also a transcription factor, is a circadian clock gene. Other circadian top candidate genes identified by our analysis include Rorb, Rora, and Rxrg (Fig. 2, and Tables I and III). Circadian rhythm and sleep abnormalities have long been described in bipolar disorder—excessive sleep in the depressive phase, reduced need for sleep in the manic phase [Bauer et al., 2006]. Sleep deprivation is one of the more powerful and rapid acting treatment modalities for severe depression, and can lead to precipitation of manic episodes in bipolar patients [Wirz-Justice et al., 2004]. Clock genes expression levels (Dbp, Per1, and Per2) have been reported to be changed by sleep deprivation in rodents [Wisor et al., 2002]. Seasonal affective disorder (SAD), a variant of bipolar disorder [Magnusson and Partonen, 2005], is tied to the amount of daylight, which is a primary regulator of circadian rhythms and clock gene expression; associations between polymorphisms in the clock genes Arntl, Per2, and Npas2 and SAD have previously been reported [Johansson et al., 2003; Partonen et al., 2007]. We had previously described the identification of clock gene D-box binding protein (Dbp) as a potential candidate gene for bipolar disorder [Niculescu et al., 2000b], using a CFG approach. Dbp was changed in expression by acute methamphetamine treatment in rat pre-frontal cortex (PFC) [Niculescu et al., 2000b], and mapped near a human genetic linkage locus for bipolar disorder [Morissette et al., 1999] and for depression [Zubenko et al., 2002] on chromosome 19q13. Subsequently, Dbp was also reported changed in expression by acute and chronic amphetamine treatments in mice [Sokolov et al., 2003]. Moreover, Dbp knock-out mice have abnormal circadian and homeostatic aspects of sleep regulation [Franken et al., 2000]. More recently, we have conducted extensive behavioral and gene expression studies in Dbp KO mice. These mice display a bipolar-like phenotype [Le-Niculescu et al., 2008b], which is modulated by stress. Decreases in Dbp expression have also been recently reported in fibroblasts from bipolar subjects [Yang et al., 2008]. In parallel, work carried out by us using an expanded CFG approach in a mouse pharmacogenomic model for bipolar disorder identified Arntl and a series of other clock genes (Cry2, Csnk1d, and Ccr4/nocturnin), as potential bipolar candidate genes [Ogden et al., 2004]. Following that, three independent reports have shown some suggestive association for Arntl in human bipolar samples [Mansour et al., 2006; Nievergelt et al., 2006; Shi et al., 2008]. Arntl is upstream of Dbp in the circadian clock intracellular molecular machinery, driving the transcription of Dbp [Ripperger and Schibler, 2006; van der Veen et al., 2006]. An increase in Arntl gene expression was reported in postmortem brains from bipolar subjects [Nakatani et al., 2006]. Overall, Arntl and related circadian clock genes are compelling candidates for involvement in bipolar disorders, especially the core clinical phenomenology of cycling and switching from depression to mania [Bunney and Bunney, 2000; Wager-Smith and Kay, 2000; Niculescu et al., 2000b; Niculescu and Kelsoe, 2001; Kelsoe and Niculescu, 2002; Lenox et al., 2002; Hasler et al., 2006; Wirz-Justice, 2006; McClung, 2007; Le-Niculescu et al., 2008b].

    Bdnf is a growth factor involved in neurotrophicity and synaptic transmission. Other growth factor top candidate genes identified by our analysis include Nrg1, Fgf12, and Ptn (Fig. 2, and Tables I and III). Bdnf has been previously implicated in a variety of neuropsychiatric disorders, by both animal model and human studies: depression [Pezawas et al., 2008; Sen et al., 2008], bipolar disorder [Ogden et al., 2004], anxiety, alcoholism [Rodd et al., 2007], and schizophrenia [Le-Niculescu et al., 2007a; Chao et al., 2008]. Notably, there are several candidate gene association studies to date implicating Bdnf in bipolar disorder [Fan and Sklar, 2008; Liu et al., in press].

    Aldh1a1 has been previously implicated in brain development [Denisenko-Nehrbass et al., 2000], schizophrenia [Galter et al., 2003], and alcoholism [Moore et al., 2007]. An intriguing finding is that of Oprm1 (opioid receptor mu 1) as a top candidate gene for bipolar. Oprm1 has been implicated in pain regulation [Oertel and Lotsch, 2008], substance abuse disorders [Luo et al., 2008], attachment behaviors [Barr et al., 2008], and suicide [Hishimoto et al., 2008]. Earlier work by us using animal models and a CFG approach had identified an overlap between candidate genes involved in mood regulation and pain regulation, such as Penk (preproenkephalin) [Ogden et al., 2004; Le-Niculescu et al., 2008b].

    A surprising finding is that of amyloid beta precursor protein (App), an Alzheimer disease (AD) candidate gene, among the top candidate gene for bipolar disorder (Table I), as well as the overall amyloid pathway being among the top canonical pathways identified (Fig. 3A). Another key gene involved in AD, Gsk3b, is also present on our list of top candidate genes. There is an interesting epidemiological literature showing increased AD in bipolar patients, and the prophylactic effect of the mood stabilizer lithium on the incidence of AD in bipolar patients [Nunes et al., 2007]. Notably, Gsk3b is a target of lithium treatment [Beaulieu et al., 2008a], as well as of serotonergic anti-depressants [Beaulieu et al., 2008b]. App has recently been shown to have a neurotrophic role [Oh et al., 2008], similar in some ways to growth factors such as Bdnf. App has also been reported to be increased in expression in bipolar postmortem brains compared to normal controls [Jurata et al., 2004]. It remains unclear if App's role in AD is pathogenic or is in fact a defense/compensatory mechanisms to try to maintain neuronal survival [Rohn et al., 2008]. If the later scenario is true, new compounds being developed for AD that target App might not stop the illness. Regardless if that turns out to be the case or not, drugs that regulate App levels may have an impact on mood (i.e., downregulation of App may be depressogenic), a particular concern given the prevalence of depression in the elderly in general [Alexopoulos et al., 2005], and in AD patients in particular [Sun et al., 2008b].

    Limitations and Confounds

    No correction of best P-values for number of SNPs tested/gene size effect was performed. While this is arguably a valid statistical issue for genetic studies by themselves, some of the multiple SNPs tested per gene could be in linkage disequilibrium, and the Bonferroni correction might be too conservative [Rice et al., 2008]. Moreover, it could introduce a bias against large-size genes, which generally have more SNPs tested than smaller genes. Of course, the converse is true if we do not correct for number of SNPs tested and one would expect some noise due to gene size effects. However, we did not observe a significant correlation between gene size and our top candidate genes (Supplementary Information—Fig. 1S and Table IIIS). That may be due to the fact that we are using this evidence for integration across platforms and modalities, along with a series of other lines of evidence that have their own attendant noise, as part of a Bayesian-like approach to pull signal from noise and prioritize findings. The convergence of lines of evidence arguably factors out the noise of the different individual approaches, and makes our network-like CFG approach relatively resilient to error even when one or another of the nodes (lines of evidence) is weak (Fig. 1).

    Our approach relies on a list of genes from the GWAS datasets generated by SNPPER identifying SNPs in genes. We may thus be missing genes where the assignment is not made by the software, and discarding SNPs that fall into regulatory regions, such as promoter or enhancer regions. Moreover, genes where the illnesses associated SNPs do not lead to a change in expression levels are not included in our CFG-GWA cross-validation. Similarly, genes that have changes in expression levels but no intragenic SNP in the GWAS datasets are not included. Interestingly, some of these later genes may be changed in expression as a consequence of distal regulatory SNPs or other genes in a network, an exciting area for future system biology studies awaiting better bioinformatic tools and data analysis now on the horizon [Stumpf et al., 2008].

    Other animal models data could potentially be used for CFG cross-validation, in addition to the data from the pharmacogenomic (methamphetamine/valproate) [Ogden et al., 2004] and the genetic (DBP knock-out mouse) [Le-Niculescu et al., 2008b] models that we generated and used. However, these are some of the best animal models with corresponding comprehensive brain and blood gene expression datasets published to date. Moreover, we relied, as an additional line of evidence, on an extensive public mouse QTL/transgenic database.

    As new human blood, postmortem brain, and human genetic studies are published, new evidence will be available for some of the genes we have identified. However, any new evidence will not remove genes from our results, but rather move them up higher in the prioritization list/pyramid (Fig. 2).

    Different ways of weighing the lines of evidence included in the CFG analysis rather than the equal weight approach we have used may become available in the future, based on more empirical and quantitative methods. Other ways of weighing the scores of line of evidence may give slightly different results in terms of prioritization, if not in terms of the list of top genes per se.

    Pathways identified by Ingenuity and GeneGo may be based on some of the same body of knowledge and published literature used in our direct CFG scoring. However, it is reassuring to see that different independent systematization and curation efforts lead to a consistent picture of genes involved in behavior, neurological disease, psychological disorders, and nervous system development coming up at the top of the over-represented pathways from our top candidate genes for bipolar disorder identified by our genetic–genomic combined approach.

    Conclusions and Future Directions

    In spite of these notable limitations, our analysis is arguably the most comprehensive integration of genetics and functional genomics to date in the field of bipolar disorder, yielding a series of candidate genes, blood biomarkers, pathways and mechanisms, that are prime targets for follow-up hypothesis driven studies. Such studies may include individual candidate gene association studies with more SNPs tested per gene, deep re-sequencing, and/or biological validation such as cell culture [Pletnikov et al., 2007] and transgenic animal work [Hikida et al., 2007; Le-Niculescu et al., 2008b].

    First, the model that emerges from this work (Fig. 4) is consistent with mood being a function of trophicity [Niculescu, 2005], through energy metabolism [Quiroz et al., 2008] as well as cellular growth and proliferation [Le-Niculescu et al., 2008a]. Speculatively, from an evolutionary standpoint, it may make sense for the organism to react to a favorable environment by activity and expansion, and to an unfavorable environment by inactivity and retraction-the “mood as a muscle” model [Niculescu, 2005]. In this view, high resources translate into high mood and high libido, as the environment is favorable and can support growth, expansion and progeny. The threshold to pain may be elevated [Ogden et al., 2004], so activity can occur even in the face of actual injuries. Conversely, low resources translate into a low mood and low libido, as the environment is unfavorable and cannot support more growth, expansion and progeny. The threshold to pain is reduced, so one can react and retract in the face of potential injuries [Niculescu and Akiskal, 2001a,b]. In clinical illness (bipolar disorder, depression), this congruence between mood and environment is arguably lost and/or the mood reaction to environmental cues is disproportionate.

    Second, despite the fact that our analysis uses only data from human and animal studies focused on bipolar and related disorders, it is likely that some of the genes and pathways identified in this report are not implicated only in bipolar disorder and depression, but also in other psychiatric disorders, such as schizophrenia [Le-Niculescu et al., 2007a]. Indeed, we provide some evidence for that (Fig. 5). While some of this overlap might be due to limitations in precision of diagnostic ascertainments in human studies and limitations in specificity to a disorder in animal studies, an alternative and more compelling explanation is that the genetic and neurobiological structure of psychiatric disorders is modular in a Lego-like fashion [Niculescu et al., 2006], with building blocks in different permutations leading to different clinical disorders.

    Figure 5.

    Genetic co-morbidity. MetaCore analysis (GeneGo, Encinitas, CA) of top candidate genes involvement in diseases. A: Analysis of top 41 candidate genes (CFG score of 6 and above). B: Analysis of top 113 candidate genes (CFG score of 5 and above). P-value indicates over-representation of these genes in different disease categories, based on bioinformatic analyses of published literature—derived connections.

    Third, our work provides additional integrated evidence focusing attention on and prioritizing a number of genes as candidate blood biomarkers for bipolar disorder, with an inherited genetic basis (Table I). While prior evidence existed as to alterations in gene expression levels of those genes in whole-blood samples or lymphoblastoid cell lines (LCLs) from mood disorders patients, it was unclear prior to our analysis whether those alterations were truly related to the disorder or were instead related to medication effects and environmental factors, or indeed were frankly artifactual.

    Last but not least, our work provides a proof of principle for how such a combined approach, integrating functional and genotypic data, can be used for other complex disorders-psychiatric and non-psychiatric. What we are beginning to see across GWAS of complex disorders are not necessarily the same genes showing the strongest signal, but rather consistency at the level of gene families or biological pathways. The distance from genotype to phenotype may be a bridge too far for genetic-only approaches, given the intervening complex layers of epigenetics, gene expression regulation and endophenotypes [Tan et al., 2008]. Using GWAS data in conjunction with gene expression data as part of CFG or integrative genomics [Degnan et al., 2008] approaches, followed by pathway—level analysis of the prioritized candidate genes, can serve as the necessary Rosetta Stone for unraveling the genetic code of complex disorders such as bipolar disorder. A whole body of work will then need to follow in terms of personalizing diagnosis and treatment based on particular combinations of genes and gene expression patterns, leading to major re-evaluations of current clinical nosology.

    Acknowledgements

    This work was supported by a grant (1 R01 MH 071912) from the U.S. National Institute of Mental Health to M.T.T. and A.B.N., and a NARSAD Mogens Schou Young Investigator award to ABN. We are grateful to Nick Craddock, Mick O'Donovan, Mike Owen and Pamela Sklar for sharing of data and very helpful discussions. We would also like to thank Griffin Fitzgerald, Bhavana Pandya and Jesse Townes for their help with database construction and bioinformatic analyses. Supplementary information for this paper is available from the journal website. Additional information is available at www.neurophenomics.info.

    Ancillary