Molecular Psychiatry (2012) 17, 267–279 & 2012 Macmillan Publishers Limited All rights reserved 1359-4184/12 www.nature.com/mp ORIGINAL ARTICLE Transcriptional regulation of neurodevelopmental and metabolic pathways by NPAS3 L Sha1,7, L MacIntyre2,7, JA Machell1, MP Kelly3, DJ Porteous1, NJ Brandon3, WJ Muir4,{, DH Blackwood4, DG Watson2, SJ Clapcote5 and BS Pickard2,6 1Department of Medical Genetics, Institute for Genetics and Molecular Medicine, University of Edinburgh, Molecular Medicine Centre, Western General Hospital, Edinburgh, UK; 2Strathclyde Institute of Pharmacy and Biomedical Sciences, University of Strathclyde, Glasgow, UK; 3Pfizer, Pfizer Global Research and Development, Neuroscience Research Unit, Groton, CT, USA; 4Division of Psychiatry, University of Edinburgh, Royal Edinburgh Hospital, Edinburgh, UK; 5Institute of Membrane and Systems Biology, University of Leeds, Leeds, UK and 6CeNsUS—Centre for Neuroscience, University of Strathclyde, Glasgow, UK

The basic helix-loop-helix PAS (Per, Arnt, Sim) domain NPAS3 is a replicated genetic risk factor for psychiatric disorders. A knockout (KO) mouse model exhibits behavioral and adult neurogenesis deficits consistent with human illness. To define the location and mechanism of NPAS3 etiopathology, we combined immunofluorescent, transcriptomic and metabonomic approaches. Intense Npas3 immunoreactivity was observed in the hippocampal subgranular zone—the site of adult neurogenesis—but was restricted to maturing, rather than proliferating, neuronal precursor cells. Microarray analysis of a HEK293 cell line over-expressing NPAS3 showed that transcriptional targets varied according to context and C-terminal deletion. The most highly up-regulated NPAS3 target gene, VGF, encodes secretory peptides with established roles in neurogenesis, depression and . VGF was just one of many NPAS3 target also regulated by the SOX family of transcription factors, suggesting an overlap in neurodevelopmental function. The parallel repression of multiple glycolysis genes by NPAS3 reveals a second role in the regulation of glucose . Comparison of wild-type and Npas3 KO metabolite composition using high-resolution mass spectrometry confirmed these transcriptional findings. KO brain tissue contained significantly altered levels of NAD þ , glycolysis metabolites (such as dihydroxyacetone phosphate and fructose-1,6-bisphosphate), pentose phosphate pathway components and Kreb’s cycle intermediates (succinate and a-ketogluta- rate). The dual neurodevelopmental and metabolic aspects of NPAS3 activity described here increase our understanding of mental illness etiology, and may provide a mechanism for innate and medication-induced susceptibility to diabetes commonly reported in psychiatric patients. Molecular Psychiatry (2012) 17, 267–279; doi:10.1038/mp.2011.73; published online 28 June 2011 Keywords: schizophrenia; ; transcriptomics; metabolomics; glycolysis; SOX transcription factor

Introduction A role for the NPAS3 gene in psychiatric illness risk was suggested through the study of a mother and Schizophrenia and bipolar disorder are common, daughter diagnosed with schizophrenia and mild lifelong psychiatric illnesses affecting mood, percep- learning disability who carried a chromosomal tion and cognition. A strong genetic contribution to abnormality disrupting the gene.1,2 Subsequent gene- these conditions is indicated by family and epide- specific and genome-wide case–control association miological studies. studies have linked single nucleotide polymorphisms at the NPAS3 with increased risk of schizo- phrenia,3,4 bipolar disorder3–5 and major depression.4 Correspondence: Dr BS Pickard, Strathclyde Institute of Phar- Three common NPAS3 exonic variants have also been macy and Biomedical Science, University of Strathclyde, Andrew recently associated with increased risk of schizo- Hamnett Wing, 161 Cathedral Street, Glasgow G4 0RE, UK. phrenia.6 Genetic variation at the NPAS3 locus has E-mail: [email protected] 7 also been linked with response to treatment with the Equal authorship status. 7 { antipsychotic drug, iloperidone. Deceased. Received 24 October 2010; revised 6 May 2011; accepted 10 May NPAS3 encodes a member of the bHLH-PAS 2011; published online 28 June 2011 (basic helix-loop-helix—Per, Arnt, Sim) domain NPAS3 transcription factor targets L Sha et al 268 transcription factor family that form functional Prolong antifade reagent with DAPI (40,6-diamidino- heterodimers.8,9 Npas3 knockout (KO) mice display 2-phenylindole) nuclear stain (Invitrogen). neuroanatomical, memory and behavioral phenotypes typical of a model of human psychiatric disorders.10,11 NPAS3 and SOX expression constructs and standard/ They also exhibit a deficit in adult hippocampal circadian cell culture neurogenesis12—a potential cellular pathology of The full-length NPAS3 open reading frame (acc. schizophrenia and depression.13,14 Importantly, a novel NM_001164749) was cloned into a TET-inducible compound acting on mitochondria, P7C3, reverses the expression plasmid (pT-REx-DEST30; Invitrogen) neurogenesis, neuroanatomical, electrophysiological using a restriction digest, Gateway (Invitrogen) clon- and behavioral phenotypes of the Npas3 KO.15 ing linker ligation and the BP/LR reactions (Invitro- Here, we report results from parallel experimental gen). The truncated form, DNPAS3, was generated by approaches designed to identify the function of cleavage and removal of sequence between internal NPAS3 in health and mental illness. The hippocam- and multiple cloning site XhoI sites thus deleting pal neurogenesis deficit associated with Npas3 dele- the second PAS domain and the putative transactiva- tion suggested that NPAS3 might directly regulate tion domain. Plasmids were stably integrated into the developmental pathway associated with new HEK293 [T-REx-293] cells (Invitrogen), using Genet- neuron production and maturation and we describe icin and Blasticidin. HEK293 cells were maintained an immunofluorescence study to determine NPAS3’s in Dulbecco’s Modified Eagle Medium (DMEM) spatio-temporal contribution to this process. supplemented with 10% fetal bovine serum (‘stan- As a transcription factor, NPAS3 is particularly dard culture conditions’). amenable to a transcriptomics approach because SOX expression constructs were generously gifted by observed expression changes are likely to represent Chuanju Liu (SOX5/SOX6/SOX9) and Fabien Muri- direct activity rather than secondary or homeostatic sier/Friedrich Beerman (SOX10). SOX11 was cloned reactions. We employ an in vitro system; over- from human cDNA as described in Li et al. (submitted expression of full-length (FLNPAS3) and artificially for publication, Acta Neuropsychiatr). All SOX micro- truncated (DNPAS3) forms of NPAS3 in the human array experiments were carried out as transient embryonic kidney cell line, HEK293, followed by transfections of HEK293 cells using Optimem/Lipofec- microarray analysis not to model psychiatric illness tamine 2000/plasmid DNA (Invitrogen) incubation for but rather to efficiently identify target genes. We 6 h followed by 24 h in standard culture conditions. compare NPAS3 targets with those of a known For circadian induction, cells were maintained in family of neurodevelopmental regulators, the SOX DMEM alone for 36 h. At the zero hour time point, the family of transcription factors.16–18 We also describe cell medium was replaced with DMEM supplemented an immunofluorescence study to determine NPAS3’s with 50% horse serum plus tetracycline in order to spatio-temporal contribution to hippocampal neuro- induce circadian cycling/NPAS3 over-expression.21,22 genesis. The observation that the related NPAS2 At þ 2 h, cells were washed with DMEM and then acts as a functional equivalent of the CLOCK incubated with the same plus tetracycline for the circadian regulator in certain brain regions19 remaining period of the experiment. At either þ 12 or prompted us to examine NPAS3 targets in cells þ 24 h, cells were removed, washed and frozen. stimulated to commence synchronous circadian cy- cling. Finally, we assess the biological relevance of Illumina microarray analysis and data normalization the in vitro microarray data to the in vivo pathologies For baseline and circadian NPAS3 over-expression in the Npas3 KO mouse. This was achieved by high- experiments, parental HEK293 [T-REx-293] cells were resolution mass spectrometry-based metabonomic used as negative controls. Samples (n = 2) were comparison of wild-type and mutant brain tissue. assessed for FLNPAS3 over-expression, DNPAS3 Our findings indicate that NPAS3 contributes to over-expression and parental negative controls. In both neurodevelopmental transcription factor net- the SOX microarray studies, DNA-free Lipofecta- works and the regulation of brain glucose metabolism. mine2000 transfections (n = 4) were used as shared negative controls compared with transfections with each of the SOX expression constructs (n = 3). RNA Materials and methods extraction (RNeasy kit, Qiagen, Crawley, UK) and synthesized microarray probes (Illuminas TotalPrep Immunofluorescence RNA Amplification Kit, Ambion, Austin, TX, USA) Immunofluorescence of frozen brain sections was were quantified and quality checked using an Agilent carried out as previously described.20 Antibodies (Santa Clara, CA, USA) Bioanalyzer. An Illumina (San against Npas3, Gfap, Dcx and Nestin were all obtained Diego, CA, USA) Beadstation platform was used in from Santa Cruz Biotechnology (Santa Cruz, CA, conjunction with Sentrixs HumanRef-8 v2 chips USA). 1:400 donkey secondary antibodies against (24 500 gene transcripts) to detect goat or rabbit IgG, conjugated to Alexa Fluor 594 were profiles. This work was carried out in the Wellcome applied for red fluorescence, or to Alexa Fluor 488 or Trust Clinical Research Facility at the Western General FITC for green fluorescence (Invitrogen Life Technol- Hospital, Edinburgh. Microarray data analysis was ogies, Paisley, UK). Sections were mounted with carried out using BRB-ArrayTools 3.8.0 freeware

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 269 (Biometric Research Branch, National Cancer Institute, 13 mm syringe filters with 0.2 mm nylon membrane NIH, Bethesda, MA, USA). Regulated genes were (Sigma Aldrich, Gillingham, UK) before LC–MS analysis. categorized using IPA (Ingenuity Pathway Analysis) and GeneCodis2 for over-representation of parti- LC–MS analysis of polar metabolites. Analysis was cular gene ontologies or biological processes. ProSim carried out using a Finnigan LTQ-Orbitrap fitted with Ternary Diagram freeware (http://www.prosim.net/en/ a Surveyor HPLC pump (Thermo Fisher, Hemel resources/download.html) was used to visualize nor- Hempstead, UK) using 30 000 resolution. The malized/de-transformed microarray expression data. software program XCalibur (version 2.0, Thermo Fisher, Hemel Hempstead, UK) was used to acquire Confirmation by QPCR the LC–MS data. Analyses were carried out in positive Oligo-dT-primed cDNA synthesis of microarray and negative modes over mass range of 60–1000 m/z. sample RNA was performed (Roche, Welwyn Garden The capillary temperature was set at 250 1C and in City, UK) followed by triplicate QPCR with SYBR positive ionization mode the ion spray voltage was green QPCR Master Mix (Invitrogen) and a real-time 4.5 kV, the capillary voltage 30 V and the tube lens QPCR machine (Bio-Rad, Hemel Hempstead, UK). voltage was 105 V. In negative ionization mode, the For relative quantification of mRNA expression, ion spray voltage was À4.5 kV, the capillary voltage geometric means were calculated using the compara- was À25 V, the tube lens voltage was À95.0 V. The tive double delta method. Primers used in QPCR are sheath and auxiliary gas flow rates were 45 and 15, included in Supplementary Table 9. The housekeep- respectively (manufacturer units). A ZIC-HILIC ing gene (18sRNA) was selected as the endogenous column (5 mm, 150 Â 4.6 mm2; HiChrom, Reading, references in this QPCR studies. UK) was used with a binary gradient. Solvent A was 0.1% v/v formic acid in HPLC water and B was 0.1% Dual luciferase assay formic acid in acetonitrile. A flow rate of 0.3 ml per 50 human VGF sequence (2029 kb), containing promo- minute was used and injection volume was 10 ml. ter, exon1, intron1 and part of exon2, was amplified by The gradient program used was 80% B at 0 min to 50% PCR and cloned into digested pGL3 reporter vectors. B at 12 min to 20% B at 28 min to 80% B at 37 min HEK293 and SHSY-5Y cell samples were collected at with total run time of 45 min. The instrument was 24 h after transfection. The activities of Firefly externally calibrated before analysis and internally luciferase (expressed from all pGL3 reporter vectors) calibrated using lock masses at m/z 83.06037 and m/z and Renilla luciferase (expressed from co-transfected 195.08625. Samples were analyzed sequentially and pRL-TK vector) were measured sequentially from each the vial tray temperature was set at a constant sample by using the Dual luciferase reporter (DLR TM) temperature of 4 1C. Raw XCalibur data files from Assay System with the Plate-reader (Synergy HT, wild-type and Npas3 KO samples were processed BioTek, Potton, UK). For each sample, the Firefly using Sievet version 1.2 (Thermo Fisher). luciferase activity was normalized by the Renilla luciferase activity and then compared with the Results mean value of control samples. Dual luciferase assay results were statistically analyzed using SigmaPlotr Npas3 expression in the dentate gyrus supports its role (Bruxton, Seattle, WA, USA). One-way analysis of in neurogenesis variance test was used to compare the VGF promoter We observed the strongest NPAS3 immunofluores- activity of different cells among treatment groups. cence originating in the subgranular zone of the dentate gyrus with immunopositive processes radiat- Metabonomics ing into the granule cell layer proper (Figure 1j). Extraction of polar metabolites from brain tissue Additional expression included nuclear distribution homogenates for liquid chromatography–mass in ependymal cells and axo-dendritic staining adja- spectrometry (LC–MS) analysis. The deleted mouse cent to the cell soma in many cortical neurons. To Npas3 locus (on a 129S6/SvEvTac background) was define the stage of Npas3 action in newly forming backcrossed with mice from the C57BL/6J strain neurons, we co-stained with antibodies against such that, at the stage of metabonomic analysis, the Nestin, Gfap and Dcx (Figures 1a–i). Only Dcx mice were on average 62.5% C57BL/6J. Sagitally showed an overlapping expression pattern, indicating dissected half-brains from homozygote Npas3 KO that Npas3 acts during neuronal maturation/differ- (n = 4) and wild-type (n = 4) littermates were snap entiation. frozen in isopentane and stored at À80 1C. A two-step metabolite extraction method using methanol, water Identifying NPAS3 target genes in standard culture and chloroform as extraction solvents produced a conditions biphasic solution comprising polar and non-polar Unsupervised hierarchical clustering of gene expres- fractions.23 Only the polar fractions were analyzed in sion profiles using centered correlation and average this study. In all, 200 ml of collected polar extract was linkage showed reproducible transcriptional changes added to 600 ml of 1:1 acetonitrile:water. Samples were after FLNPAS3 or DNPAS3 over-expression (Supple- diluted to combat salt interference and to prevent ion mentary Figure 1a). Using the SAM algorithm suppression. Samples were filtered using Acrodisc (Significance Analysis of Microarrays) with a 0.01

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 270

Figure 1 Npas3 expression in the mouse dentate gyrus. Npas3, Gfap, Nestin and Dcx protein expression was compared using immunofluorescence on frozen mouse brain sections. Npas3 (green; a, d, g) expression was mainly non-nuclear and localized to the inner face/subgranular zone of the dentate gyrus. Anti-Gfap, -Nestin and -Dcx/Doublecortin antibodies (red; b, e, h) were used in co-immunofluorescence studies (c, f, i). Only Dcx co-localized with Npas3 in the subgranular zone (yellow; i). A composite image (j) shows Npas3 immunoreactivity throughout the hippocampus.

target proportion of false discoveries and 100 permu- SOX11 confirmed the microarray findings (Supple- tations, 3476 genes were found to discriminate mentary Figure 2a). between FLNPAS3 and control. In all, 282 genes showed X1.5-fold up-regulation by FLNPAS3, and NPAS3 and SOX transcription factors share target 359 genes were similarly down-regulated (Tables 1a genes, including VGF and 1b; Figure 2a). NPAS3, as a consequence of The up-regulated set of genes (Table 1a) was analyzed experimental over-expression, was the most up- for over-representation of particular biological pro- regulated gene (62-fold) but was only detected in cesses using online tool GeneCodis2.25 Thirty-six of FLNPAS3 arrays as the chip’s oligonucleotide probe is the 282 up-regulated genes (corrected hypergeometric complementary to 30 sequence absent in DNPAS3. P-value = 2.07 Â 10À8) were categorized by the gene There was substantial correlation of target term ‘transcription’ (GO:0006350), being fold-change between FLNPAS3 and DNPAS3 cells in either transcription factors or DNA-binding standard culture conditions, with linear regression r2 with regulatory function (Supplementary Table 3). values of 0.59 (up-regulated genes) and 0.78 (down- The most highly up-regulated NPAS3 target gene was regulated genes) (Supplementary Figures 1b and c). A VGF (2.92-fold), which encodes a group of processed similar C-terminal deletion of a related bHLH tran- and secreted neuropeptides including TLQP-21. The scription factor, ARNT (HIF1 b), also largely pre- human VGF promoter containing an ‘E box’ motif served regulatory function.24 QPCR directed against (consensus bHLH transcription factor binding site), four selected regulated genes, VGF, HK2, ENO2 and was cloned into the pGL3 luciferase reporter plasmid.

Molecular Psychiatry Table 1 The 51 genes most up-regulated (a) and 50 most down-regulated (b) by FLNPAS3 in standard cell culture conditions

Illumina ID Accession Symbol Description Fold up- Glycolysis Hypoxia Circadian regulation

(a) ILMN_1752550 NM_022123.1 NPAS3 Neuronal PAS domain protein 3 62.49 ILMN_1757497 NM_003378.2 VGF VGF nerve growth factor inducible 2.92 ILMN_1705153 NM_021076.2 NEFH Neurofilament, heavy polypeptide 2.58 200 kDa ILMN_1716264 NM_014391.2 ANKRD1 Ankyrin repeat domain 1 (cardiac 2.40 muscle) ILMN_1660841 NM_001001870.1 MGC14376 Hypothetical protein MGC14376 2.36 ILMN_1730818 NM_174947.2 C19orf30 19 open reading frame 30 2.33 ILMN_1655741 NM_181617.1 KRTAP21-2 Keratin-associated protein 21-2 2.31 ILMN_1789096 NM_152672.4 OSTalpha Organic solute transporter a 2.23 ILMN_1734011 NM_001010866.1 RP13-15M17.2 Hypothetical protein LOC199953 2.20 ILMN_1725524 NM_003727.1 DNAH17 Dynein, axonemal, heavy chain 17 2.15 ILMN_1680085 NM_004881.2 TP53I3 Tumor protein inducible protein 3 2.12 ILMN_1657148 NM_152480.1 C19orf23 open reading frame 23 2.08 ILMN_1684886 NM_013452.2 VCX Variable charge, X-linked 2.07 ILMN_1669410 NM_001275.3 CHGA Chromogranin A (parathyroid secretory 2.07 protein 1) ILMN_1695199 NM_003543.3 HIST1H4H Histone cluster 1, H4h 2.05 ILMN_1667295 NM_138440.2 VASN Vasorin 2.01 ILMN_1810191 NM_003706.1 PLA2G4C Phospholipase A2, group IVC (cytosolic, 1.98 calcium-independent)

ILMN_1747506 NM_014681.4 DHX34 DEAH (Asp-Glu-Ala-His) box 1.92 Sha targets L factor transcription NPAS3 polypeptide 34 ILMN_1754894 NM_174896.2 C1orf162 Chromosome 1 open reading frame 162 1.92 ILMN_1810514 NM_014655.1 SLC25A44 Solute carrier family 25, member 44 1.92 al et ILMN_1800425 NM_003047.2 SLC9A1 Solute carrier family 9 (sodium/ 1.92 hydrogen exchanger), member 1 ILMN_1754061 NM_022917.4 NOL6 Nucleolar protein family 6 (RNA- 1.92 associated) ILMN_1667194 NM_144684.1 ZNF480 Zinc-finger protein 480 1.89 ILMN_1658800 NM_015695.2 BRPF3 Bromodomain and PHD finger 1.88 containing, 3 ILMN_1666690 NM_052957.3 ACRC Acidic repeat containing 1.83 ILMN_1669113 NM_012068.3 ATF5 Activating transcription factor 5 1.83 ILMN_1769931 NM_005066.1 SFPQ Splicing factor proline/glutamine-rich 1.82 ILMN_1710544 NM_002589.2 PCDH7 Protocadherin 7 1.80 ILMN_1786024 NM_001018052.1 POLR3H Polymerase (RNA) III (DNA directed) 1.80 polypeptide H (22.9 kDa) ILMN_1718977 NM_015675.2 GADD45B Growth arrest and DNA-damage- 1.80 Yes Yes inducible, b ILMN_1694731 NM_001287.3 CLCN7 Chloride channel 7 1.79 ILMN_1697268 NM_032048.2 EMILIN2 Elastin microfibril interfacer 2 1.79 ILMN_1757406 NM_005319.3 HIST1H1C Histone cluster 1, H1c 1.77 Yes ILMN_1725510 NM_014762.3 DHCR24 24-Dehydrocholesterol reductase 1.75 oeua Psychiatry Molecular ILMN_1807719 NM_004937.2 CTNS Cystinosis, nephropathic 1.74 ILMN_1784037 NM_001083621.1 ZBTB40 Zinc-finger and BTB domain 1.74 containing 40 271 272 oeua Psychiatry Molecular

Table 1 Continued

Illumina ID Accession Symbol Description Fold up- Glycolysis Hypoxia Circadian regulation

ILMN_1761411 NM_024834.2 C10orf119 Chromosome 10 open reading frame 119 1.73 ILMN_1795671 NM_018269.1 ADI1 Acireductone dioxygenase 1 1.72 ILMN_1787826 NM_078483.2 SLC36A1 Solute carrier family 36 (proton/amino- 1.72 acid symporter), member 1 ILMN_1760650 NM_001384.4 DPH2 DPH2 homolog (S. cerevisiae) 1.72 ILMN_1722059 NM_002967.2 SAFB Scaffold attachment factor B 1.71 ILMN_1689959 NM_020170.3 NCLN Nicalin homolog (zebrafish) 1.70 ILMN_1733847 NM_003857.2 GALR2 Galanin 2 1.70 ILMN_1762037 NM_144582.2 TEX261 Testis expressed 261 1.70 Yes ILMN_1665212 NM_014329.3 EDC4 Enhancer of mRNA decapping 4 1.69

ILMN_1754643 NM_022719.2 DGCR14 DiGeorge syndrome critical region gene 1.69 targets factor transcription NPAS3 14 ILMN_1797903 NM_014480.1 ZNF544 Zinc-finger protein 544 1.69 ILMN_1743397 NM_178517.3 PIGW Phosphatidylinositol glycan anchor 1.69 biosynthesis, class W ILMN_1737298 NM_005911.4 MAT2A Methionine adenosyltransferase II, a 1.67 ILMN_1668012 NM_014251.1 SLC25A13 Solute carrier family 25, member 13 1.67 (citrin) ILMN_1694837 NM_015590.2 GPATCH4 G patch domain containing 4 1.67 Sha L tal et Illumina ID Accession Symbol Description Fold down- Metabolic Hypoxia Circadian regulation

(b) ILMN_1756417 NM_181726.2 ANKRD37 Ankyrin repeat domain 37 À4.49 Yes ILMN_1758164 NM_003155.2 STC1 Stanniocalcin 1 À4.05 Yes ILMN_1661599 NM_019058.2 DDIT4 DNA-damage-inducible transcript 4 À3.75 Yes ILMN_1695880 NM_002317.3 LOX Lysyl oxidase À3.53 Yes ILMN_1659990 NM_013332.3 HIG2 Hypoxia-inducible protein 2 À3.27 Yes ILMN_1755974 NM_005165.2 ALDOC Aldolase C, fructose-bisphosphate À3.21 Glycolysis Yes ILMN_1728057 NM_145267.2 C6orf57 Chromosome 6 open reading frame 57 À2.85 ILMN_1748124 NM_198057.2 TSC22D3 TSC22 domain family, member 3 À2.77 Yes ILMN_1785703 NM_198271.2 LMOD3 Leiomodin 3 (fetal) À2.64 ILMN_1809931 NM_006096.2 NDRG1 N- downstream regulated gene 1 À2.63 Yes ILMN_1759092 NM_178124.3 CXorf40A Chromosome X open reading frame 40A À2.60 ILMN_1697448 NM_006472.2 TXNIP Thioredoxin interacting protein À2.57 Yes ILMN_1658289 NM_032118.2 WDR54 WD repeat domain 54 À2.55 ILMN_1715324 NM_014234.3 HSD17B8 Hydroxysteroid (17-b) dehydrogenase 8 À2.48 ILMN_1724658 NM_004052.2 BNIP3 BCL2/adenovirus E1B 19 kDa interacting À2.44 Yes protein 3 ILMN_1776602 NM_194431.1 RNASE4 Ribonuclease, RNase A family, 4 À2.40 Yes ILMN_1671791 NM_004563.2 PCK2 Phosphoenolpyruvate carboxykinase 2 À2.39 Glyc/TCA ILMN_1719476 XR_015982.1 LOC731786 Similar to 60S ribosomal protein L32 À2.38 ILMN_1693789 NM_001632.3 ALPP Alkaline phosphatase, placental (Regan À2.38 isozyme) ILMN_1723486 NM_000189.4 HK2 Hexokinase 2 À2.36 Glycolysis Yes ILMN_1666750 NM_145111.2 C7orf38 Chromosome 7 open reading frame 38 À2.36 Table 1 Continued

Illumina ID Accession Symbol Description Fold down- Metabolic Hypoxia Circadian regulation

ILMN_1809292 NM_032549.2 IMMP2L IMP2 inner mitochondrial membrane À2.34 peptidase-like ILMN_1800796 NM_000997.3 RPL37 Ribosomal protein L37 À2.33 ILMN_1756338 NM_030971.3 SFXN3 Sideroflexin 3 À2.32 ILMN_1696183 NM_005331.3 HBQ1 Hemoglobin, theta 1 À2.31 ILMN_1651610 XM_001126202.1 LOC730525 Hypothetical protein LOC730525 À2.30 ILMN_1780825 NM_006270.3 RRAS Related RAS viral (r-ras) oncogene À2.30 homolog ILMN_1693334 NM_000917.2 P4HA1 Proline 4-hydroxylase, a polypeptide I À2.29 Yes Yes ILMN_1798581 NM_032485.4 MCM8 Minichromosome maintenance complex À2.29 component 8 ILMN_1671661 NM_016371.2 HSD17B7 Hydroxysteroid (17-b) dehydrogenase 7 À2.28 ILMN_1753342 NM_002970.1 SAT1 Spermidine/spermine N1- À2.27 acetyltransferase 1 ILMN_1708778 NM_000050.4 ASS1 Argininosuccinate synthetase 1 À2.25 Urea ILMN_1679093 NM_016535.3 ZNF581 Zinc-finger protein 581 À2.24 ILMN_1726460 NM_001034996.1 RPL14 Ribosomal protein L14 À2.23 ILMN_1692145 NM_021030.2 ZNF14 Zinc-finger protein 14 À2.23 ILMN_1797668 NM_002196.2 INSM1 Insulinoma-associated 1 À2.22 ILMN_1787885 NM_024815.3 NUDT18 Nudix (nucleoside diphosphate linked À2.22 moiety X)-type motif 18 ILMN_1772492 NM_033412.1 MCART1 Mitochondrial carrier triple repeat 1 À2.21

ILMN_1809750 NM_198795.1 TDRD1 Tudor domain containing 1 À2.20 Sha targets L factor transcription NPAS3 ILMN_1801584 NM_003467.2 CXCR4 Chemokine (C-X-C motif) receptor 4 À2.17 Yes

ILMN_1781388 NM_021965.3 PGM5 Phosphoglucomutase 5 À2.16 Glycolysis al et ILMN_1760727 NM_001097577.1 ANG Angiogenin, ribonuclease, RNase A À2.16 family, 5 ILMN_1765796 NM_001975.2 ENO2 Enolase 2 (g, neuronal) À2.15 Glycolysis Yes ILMN_1671766 NM_000505.3 F12 Coagulation factor XII (Hageman factor) À2.12 ILMN_1809291 NM_004615.2 TSPAN7 Tetraspanin 7 À2.12 ILMN_1755075 NM_004508.2 IDI1 Isopentenyl-diphosphate delta À2.12 Yes isomerase 1 ILMN_1680279 NM_018561.3 USP49 Ubiquitin-specific peptidase 49 À2.10 ILMN_1770922 NM_018004.1 TMEM45A Transmembrane protein 45A À2.10 Yes ILMN_1733559 NR_003287.1 LOC100008589 28S Ribosomal RNA À2.10 ILMN_1782743 NM_015416.3 LETMD1 LETM1 domain containing 1 À2.10

Abbreviation: PAS, Per, Arnt, Sim. Enrichment for glycolysis and hypoxia gene targets is evident in the latter group. oeua Psychiatry Molecular 273 NPAS3 transcription factor targets L Sha et al 274 ab15 15 14 13 13 12 11 11 10 FLNPAS3 FLNPAS3 9 9 8

7 7 6 5 5 5791113 15 5 7 9 11 13 15 CONT CONT

cd15 15 14 14 13 13 12 12 11 11 10 10 FLNPAS3 FLNPAS3 9 9 8 8 7 7 6 6 5 5 5 7 9 11 13 15 5 7 9 11 13 15 CONT Delta NPAS3 Figure 2 Scatter plots of microarray data. Diagonals represent equal expression in compared experimental groups. Genes in red show at least 1.5-fold expression difference between the two conditions. Control and FLNPAS3-expressing HEK293 cell lines are shown in (a). FLNPAS3 displayed predominantly inhibitory activity at þ 12 h after circadian induction (b) but not at þ 24 h (c). Only the þ 12 h time point exposed functional differences between FLNPAS3 and DNPAS3 (d). The arrowed gene indicates exogenous FLNPAS3.

In HEK293 cells, luciferase activity linked to the VGF interaction of NPAS3 with SOX transcriptional net- promoter was significantly increased upon FLNPAS3 works (Supplementary Figure 2b). or DNPAS3 co-expression compared with control vector transfection (Supplementary Figure 4a). In The set of NPAS3 down-regulated genes is enriched for SH-SY5Y neuroblastoma cells, luciferase activity metabolic function was only significantly increased after FLNPAS3 NPAS3-repressed genes were queried using IPA over-expression (Supplementary Figure 4b). (Supplementary Table 6), GeneCodis2 (Supplemen- We compared microarray data from NPAS3 over- tary Table 3) and a search for genomic clustering of expression with those from SOX transcription factor targets (Supplementary Figure 7). A cluster of chro- over-expression (Supplementary Tables 11a–d) in mosome 1 histone genes was down-regulated by order to identify shared neurodevelopmental path- NPAS3, suggesting global alterations in chromatin ways. SOX genes within the same functional group state—potentially relating to the neuronal differentia- (SOX5/6; SOX D group, SOX9/10; SOX E group) tion process. IPA identified significant enrichment of displayed closely related target profiles, confirming the glycolysis/gluconeogenesis pathway in the gene their established functional conservation and redun- set (16 genes, P = 6.14 Â 10À6). The glycolysis link dancy. NPAS3 target profiles closely matched those of (eight genes) was confirmed using GeneCodis2 (cor- SOX5/6 with the VGF gene being most highly rected hypergeometric P-value = 5.21 Â 10À7). Addi- up-regulated by all three transcription factors. tionally, 21 genes participating in the ‘oxidation: NPAS3–SOX target overlaps are summarized in reduction’ biological process (GO:0055114) were also Supplementary Figure 5. Three members of the SOX significantly down-regulated (P = 6.30 Â 10À7). Visual family of transcription factors, SOX3, SOX11 and inspection and comparison with published data27–30 SOX12 involved in neuronal proliferation and differ- suggested that many genes were targets for the entiation17,26 were confirmed as directly down-regu- hypoxia response pathway (Table 1b). Glycolysis lated by NPAS3 over-expression, further supporting regulation can also be considered part of the hypoxia

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 275 response. Together, these findings strongly suggest that, cells (100 increasing and 100 decreasing in expression in addition to a neurodevelopmental role, NPAS3 between þ 12 and þ 24 h), 14 NPAS3-regulated glyco- regulates the anaerobic catabolism of glucose into pyru- lysis genes and histone gene clusters located on vate and, more generally, enzymatic redox processes. 1 and 6. The randomly selected genes showed no deviation, indicating no systemic bias An interaction between NPAS3 activity and the between the three experimental conditions (Figure circadian rhythm 3a). The circadian rhythm impacted on FLNPAS3 and A circadian rhythm was successfully induced in DNPAS3 activities in three ways. First, global NPAS3 HEK293 cells (Supplementary Figure 8) in order to activity was almost completely inhibitory at þ 12 h monitor NPAS3 activity in the context of a daily unlike þ 24 h (Figures 2b and c). Second, significant rhythm. A comparison of þ 12 and þ 24 h microarray differences between FLNPAS3 and DNPAS3 activities data in control cells (no NPAS3) identified a set of genes on circadian (and glycolysis, to a lesser extent) target (Supplementary Table 11e), showing putative circadian genes were exposed at þ 12 h (Figures 3b and c; regulation including FOS, FOSB, EGR1, EGR2, EGR3 Supplementary Table 11f) but not at þ 24 h. This was and PER2, all with established circadian rhythm roles. evident as reduced DNPAS3 potency at þ 12 h as Circadian microarray data were visualized using highlighted in Figure 2d. Third, the sets of genes ternary plots to highlight the influence of FLNPAS3, regulated by FL/DNPAS3 at þ 12 h (Supplementary DNPAS3 and control conditions on individual gene Table S11g) and þ 24 h (Supplementary Table 11h) expression levels. Expression data assessed comprised: showed considerable differences; for example, substan- 109 randomly selected genes, 200 genes showing tial down-regulation of the chromosome 1 histone normal circadian regulation in the parental control cluster was observed at þ 12 but not þ 24 h (Figure 3d).

ab 0.7 0.3 0.7 0.3 Control 0.6 0.4 0.6 0.4

0.5 0.5 0.5 0.5

0.4 0.6 0.4 0.6

FLNPAS3 S3

NPA 

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.8 0.7 0.6 0.5 0.4 0.3 0.2

cd0.7 0.3 0.7 0.3

0.6 0.4 0.6 0.4

0.5 0.5 0.5 0.5

0.4 0.6 0.4 0.6

0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.8 0.7 0.6 0.5 0.4 0.3 0.2 Figure 3 Ternary plots illustrate influences on target gene expression. In all, 109 randomly selected genes are plotted for þ 12 h (green) and þ 24 h (red) time points (a). The absence of substantial deviation from the center indicates the no systemic bias between the three conditions—arrows indicate how inhibitory action of an experimental condition would directionally skew the distribution of points. In all, 100 up-regulated genes (circles) and 100 down-regulated genes (squares) between þ 12 and þ 24 h (b). The substantially deviating group (green squares) reflects strong and specific inhibition by DNPAS3 of the group of circadian genes down-regulated between þ 12 h and þ 24 h (Supplementary Tables 11e and f). FLNPAS3 and DNPAS3 inhibit the expression of 15 glycolysis genes at þ 24 h (c), although FLNPAS3 activity is much stronger than DNPAS3 at þ 12 h (see Figure 2d). Chromosome 1 histones (squares), but not chromosome 6 histones (circles), are substantially inhibited by both FLNPAS3 and DNPAS3 at þ 12 h but not þ 24 h (d).

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 276

Figure 4 Glycolysis, tricarboxylic acid cycle and urea cycle showing coincident findings from transcriptomic and metabonomic analyses. The enzyme or regulatory protein is displayed in italics with transcript fold down-regulation upon NPAS3 over-expression (standard culture conditions) shown in brackets. Large black arrows indicate metabolite abundance changes in the knockout mouse brain.

Altered metabolite levels in brain tissue from Npas3 KO pathological activity of NPAS3. We concede that mice NPAS3 over-expression in the HEK293 cell line will To determine the in vivo actions of Npas3, we applied not fully replicate the in vivo neuronal state. How- high-resolution mass spectrometry to homozygote KO ever, it may avoid confounding in vivo factors such as and wild-type littermate brain tissue. In both positive developmental/homeostatic compensation, cell-type ionization mode (Supplementary Tables 10a and b) complexity and variable physiological states. An and negative ionization mode (Supplementary Tables additional reason for the approach taken was the 10c and d), statistically significant changes in multi- failure to identify, using the Gene Expression Atlas at ple polar metabolites were observed. Of particular ArrayExpress (http://www.ebi.ac.uk/arrayexpress/), a note were the decreased levels of dihydroxyacetone laboratory neuronal cell line expressing endogenous phosphate/glycerone phosphate and octulose-1,8-bi- NPAS3 at levels high enough to warrant a ‘knock- sphosphate in the KO (6 and 5% of wild-type levels, down’ strategy. We believe that certain observations respectively) and increased levels of NAD þ (nicoti- suggest the in vitro data are biologically relevant. namide adenine dinucleotide) (Supplementary Figure First, overlaps in SOX target gene sets reflect evolu- 9), cystathione and dTDP-glucose/galactose (12.4-, tionary relatedness. Second, many up-regulated 3.7- and 3.4-fold increases, respectively). Many NPAS3/SOX targets are neuronal specific. The perturbations relate to glycolysis, the tricarboxylic laboratory that originally derived the cell line in acid cycle and the urea cycle (Figure 4). 1973 has speculated that HEK293 cells might actually be of neuronal origin (http://www.fda.gov/ohrms/ Discussion dockets/ac/01/transcripts/3750t1_01.pdf). Finally, exogenous transcription factors (including SOX Our experimental approaches produced novel genes) can show dominant effects on cell gene and complementary insights into the normal and transcription.31

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 277 The target similarity between NPAS3 and SOX5/6 Kreb’s cycle intermediates 2-oxoglutarate/a-ketoglu- may reflect a shared role in neuronal differentia- taric acid and Succinate and, additionally, may tion.16,32 The co-localization of Npas3 and Doublecortin inhibit the mitochondrial G3P (glycerol-3-phosphate) indicates participation in a later, post-prolifer- shuttle (Figure 4) contributing to the altered ative, stage of neuronal differentiation33 matching the dihydroxyacetone phosphate, G3P and NAD þ con- observations of Pieper et al.15 This may refine our centrations. understanding of the link between adult neurogenesis These findings may have bearing on the metabolic pathologies and psychiatric illness.13 syndrome that can occur in psychiatric patients. This The top NPAS3, SOX5 and SOX6 up-regulated is most frequently encountered as a medication side- gene, VGF, encodes a precursor for a number of effect.48,49 However, schizophrenia has been repeat- processed and secreted signaling peptides such as edly associated with an increased predisposition to TLQP-21. VGF is a highly relevant target gene as it metabolic dysfunction and type II diabetes in medica- shows; regulation by circadian rhythm,34 involvement tion-naive individuals.50 Studies suggest that glyco- in metabolic control,35,36 contribution to activity- lytic dysfunction,51,52 mitochondrial failure and related adult neurogenesis,37 and association with oxidative stress53 may be innate metabolic pathologies neurological diseases38–41 including depression37,42 in schizophrenia and bipolar disorder. and psychosis.43 Importantly, VGF and the related We have revealed unsuspected global metabolic SCG2 protein (up-regulated here by SOX9/10/11) deficits in the Npas3 KO brain. Other rodent have been identified as discriminating cerebrospinal models of psychiatric illness merit similar biochem- fluid biomarkers for schizophrenia and depres- ical profiling as a means to uncover novel patho- sion,43,44 suggesting they may be reporting neurogenic logies and opportunities for rational therapeutic pathologies. development. NPAS3 regulates hypoxia/glycolysis pathways in a circadian-dependent context. NPAS2 has also been implicated in the redox-sensitive transcription of a Conflict of interest 45,46 metabolic gene, lactate dehydrogenase. NPAS3 The authors declare no conflict of interest. target gene sets show substantial differences between þ 12 and þ 24 h. Binding partner combinations (via the bHLH domains that mediate DNA binding and Acknowledgments dimerization) may underlie the ‘cross-talk’ between NPAS3, circadian and hypoxia pathways. The ob- We acknowledge Royal Society of Physicians in served DNPAS3-specific regulation of circadian targets Edinburgh Sim Fellowship to BSP and a China at þ 12 h suggests that truncation permits abnormal Council Studentship to LS. SJC was supported in interaction with circadian transcription factor part- part by a NARSAD Young Investigator Award. ners present at þ 12 h but absent at þ 24 h. Altered circadian biology, manifest as sleep or metabolic References dysfunction, has been described in individuals diag- 47 nosed with depression and bipolar disorder. 1 Kamnasaran D, Muir WJ, Ferguson-Smith MA, Cox DW. Disrup- Despite differing experimental paradigms and tion of the neuronal PAS3 gene in a family affected with contrasting detection technologies, the in vitro and schizophrenia. J Med Genet 2003; 40: 325–332. 2 Pickard BS, Malloy MP, Porteous DJ, Blackwood DH, Muir WJ. in vivo actions of NPAS3 converged at the glycolysis Disruption of a brain transcription factor, NPAS3, is associated pathway. Perhaps most striking was the 12-fold with schizophrenia and learning disability. Am J Med Genet B increase in abundance of oxidized coenzyme NAD þ Neuropsychiatr Genet 2005; 136: 26–32. (nicotinamide adenine dinucleotide). This may 3 Pickard BS, Christoforou A, Thomson PA, Fawkes A, Evans KL, reflect a slow-down in glycolysis (which generates Morris SW et al. Interacting at the NPAS3 locus alter risk of schizophrenia and bipolar disorder. Mol Psychiatry 2009; the reduced form, NADH) or abnormally oxida- 14: 874–884. tive conditions in the cell. The decrease in 4 Huang J, Perlis RH, Lee PH, Rush AJ, Fava M, Sachs GS et al. fructose-1,6-bisphosphate and dihydroxyacetone Cross-disorder genomewide analysis of schizophrenia, bipolar phosphate concentrations provides direct evidence disorder, and depression. Am J Psychiatry 2010; 167: 1254–1263. 5 Ferreira MA, O’Donovan MC, Meng YA, Jones IR, Ruderfer DM, for glycolysis abnormalities in the KO. Both glucose Jones L et al. Collaborative genome-wide association analysis and the glycogen synthesis intermediate, Uridine supports a role for ANK3 and CACNA1C in bipolar disorder. Nat diphosphate-glucose/-galactose, were increased in Genet 2008; 40: 1056–1058. KO brain, potentially indicating glycogen storage in 6 Macintyre G, Alford T, Xiong L, Rouleau GA, Tibbo PG, Cox DW. response to ineffective glycolysis (Thymidine diphos- Association of NPAS3 exonic variation with schizophrenia. Schizophr Res 2010; 120: 143–149. phate-glucose/-galactose level changes might be 7 Lavedan C, Licamele L, Volpi S, Hamilton J, Heaton C, similarly explained). Altered concentrations of sedo- Mack K et al. Association of the NPAS3 gene and five other heptulose and octulose-1,8-bisphosphate may also loci with response to the antipsychotic iloperidone identified indicate a compensatory impact on the pentose in a whole genome association study. Mol Psychiatry 2009; 14: 804–819. phosphate pathway initiating from glucose-6-phos- 8 Brunskill EW, Witte DP, Shreiner AB, Potter SS. Characterization phate. We speculate that the proposed mitochondrial of , a novel basic helix-loop-helix PAS gene expressed in the deficit in Npas3 KO mice15 may cause the increases in developing mouse nervous system. Mech Dev 1999; 88: 237–241.

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 278 9 Gilles-Gonzalez MA, Gonzalez G. Signal transduction by 29 Greijer AE, van der Groep P, Kemming D, Shvarts A, Semenza GL, heme-containing PAS-domain proteins. J Appl Physiol 2004; 96: Meijer GA et al. Up-regulation of gene expression by hypoxia is 774–783. mediated predominantly by hypoxia-inducible factor 1 (HIF-1). 10 Brunskill EW, Ehrman LA, Williams MT, Klanke J, Hammer D, J Pathol 2005; 206: 291–304. Schaefer TL et al. Abnormal neurodevelopment, neurosignaling 30 Mense SM, Sengupta A, Zhou M, Lan C, Bentsman G, Volsky DJ and behaviour in Npas3-deficient mice. Eur J Neurosci 2005; 22: et al. Gene expression profiling reveals the profound upregulation 1265–1276. of hypoxia-responsive genes in primary human astrocytes. Physiol 11 Erbel-Sieler C, Dudley C, Zhou Y, Wu X, Estill SJ, Han T et al. Genomics 2006; 25: 435–449. Behavioral and regulatory abnormalities in mice deficient in the 31 Takahashi K, Yamanaka S. Induction of pluripotent stem cells NPAS1 and NPAS3 transcription factors. Proc Natl Acad Sci USA from mouse embryonic and adult fibroblast cultures by defined 2004; 101: 13648–13653. factors. Cell 2006; 126: 663–676. 12 Pieper AA, Wu X, Han TW, Estill SJ, Dang Q, Wu LC et al. The 32 Lefebvre V. The SoxD transcription factors—Sox5, Sox6, and neuronal PAS domain protein 3 transcription factor controls FGF- Sox13—are key cell fate modulators. Int J Biochem Cell Biol 2010; mediated adult hippocampal neurogenesis in mice. Proc Natl 42: 429–432. Acad Sci USA 2005; 102: 14052–14057. 33 Brown JP, Couillard-Despres S, Cooper-Kuhn CM, Winkler J, 13 Kempermann G, Krebs J, Fabel K. The contribution of failing adult Aigner L, Kuhn HG. Transient expression of doublecortin during hippocampal neurogenesis to psychiatric disorders. Curr Opin adult neurogenesis. J Comp Neurol 2003; 467: 1–10. Psychiatry 2008; 21: 290–295. 34 Cirelli C, Tononi G. Gene expression in the brain across the sleep- 14 Reif A, Schmitt A, Fritzen S, Lesch KP. Neurogenesis and waking cycle. Brain Res 2000; 885: 303–321. schizophrenia: dividing neurons in a divided mind? Eur Arch 35 Bartolomucci A, Possenti R, Levi A, Pavone F, Moles A. The role of Psychiatry Clin Neurosci 2007; 257: 290–299. the vgf gene and VGF-derived peptides in nutrition and metabo- 15 Pieper AA, Xie S, Capota E, Estill SJ, Zhong J, Long JM et al. lism. Genes Nutr 2007; 2: 169–180. Discovery of a proneurogenic, neuroprotective chemical. Cell 36 Jethwa PH, Ebling FJ. Role of VGF-derived peptides in the control 2010; 142: 39–51. of food intake, body weight and reproduction. Neuroendocrinol- 16 Azim E, Jabaudon D, Fame RM, Macklis JD. SOX6 controls dorsal ogy 2008; 88: 80–87. progenitor identity and interneuron diversity during neocortical 37 Thakker-Varia S, Krol JJ, Nettleton J, Bilimoria PM, Bangasser DA, development. Nat Neurosci 2009; 12: 1238–1247. Shors TJ et al. The neuropeptide VGF produces antidepressant- 17 Haslinger A, Schwarz TJ, Covic M, Chichung Lie D. like behavioral effects and enhances proliferation in the hippo- Expression of Sox11 in adult neurogenic niches suggests a campus. J Neurosci 2007; 27: 12156–12167. stage-specific role in adult neurogenesis. Eur J Neurosci 2009; 38 Pasinetti GM, Ungar LH, Lange DJ, Yemul S, Deng H, Yuan X et al. Identification of potential CSF biomarkers in ALS. Neurology 29: 2103–2114. 2006; 66: 1218–1222. 18 Lefebvre V. The SoxD transcription factors – Sox5, Sox6, and 39 Selle H, Lamerz J, Buerger K, Dessauer A, Hager K, Hampel H et al. Sox13 – are key cell fate modulators. Int J Biochem Cell Biol 2010; Identification of novel biomarker candidates by differential 42: 429–432. peptidomics analysis of cerebrospinal fluid in 19 Reick M, Garcia JA, Dudley C, McKnight SL. NPAS2: an analog of Alzheimer’s disease. Comb Chem High Throughput Screen 2005; operative in the mammalian forebrain. Science 2001; 293: 8: 801–806. 506–509. 40 Ruetschi U, Zetterberg H, Podust VN, Gottfries J, Li S, 20 Knight HM, Pickard BS, Maclean A, Malloy MP, Soares DC, Hviid Simonsen A et al. Identification of CSF biomarkers for McRae AF et al. A cytogenetic abnormality and rare coding frontotemporal dementia using SELDI-TOF. Exp Neurol 2005; 196: variants identify ABCA13 as a candidate gene in schizophrenia, 273–281. bipolar disorder, and depression. Am J Hum Genet 2009; 85: 41 Altar CA, Vawter MP, Ginsberg SD. Target identification for CNS 833–846. diseases by transcriptional profiling. Neuropsychopharmacology 21 Balsalobre A, Damiola F, Schibler U. A serum shock induces 2009; 34: 18–54. circadian gene expression in mammalian tissue culture cells. Cell 42 Malberg JE, Monteggia LM. VGF, a new player in antidepressant 1998; 93: 929–937. action? Sci Signal 2008; 1: pe19. 22 Huang TS, Grodeland G, Sleire L, Wang MY, Kvalheim G, Laerum 43 Huang JT, Leweke FM, Oxley D, Wang L, Harris N, Koethe D et al. OD. Induction of circadian rhythm in cultured human mesench- Disease biomarkers in cerebrospinal fluid of patients with first- ymal stem cells by serum shock and cAMP analogs in vitro. onset psychosis. PLoS Med 2006; 3: e428. Chronobiol Int 2009; 26: 242–257. 44 Huang JT, Leweke FM, Tsang TM, Koethe D, Kranaster L, Gerth 23 Wu H, Southam AD, Hines A, Viant MR. High-throughput tissue CW et al. CSF metabolic and proteomic profiles in patients extraction protocol for NMR- and MS-based metabolomics. Anal prodromal for psychosis. PLoS One 2007; 2: e756. Biochem 2008; 372: 204–212. 45 Rutter J, Reick M, McKnight SL. Metabolism and the 24 Reisz-Porszasz S, Probst MR, Fukunaga BN, Hankinson O. control of circadian rhythms. Annu Rev Biochem 2002; 71: Identification of functional domains of the aryl hydrocarbon 307–331. receptor nuclear translocator protein (ARNT). Mol Cell Biol 1994; 46 Rutter J, Reick M, Wu LC, McKnight SL. Regulation of clock and 14: 6075–6086. NPAS2 DNA binding by the redox state of NAD cofactors. Science 25 Nogales-Cadenas R, Carmona-Saez P, Vazquez M, Vicente C, 2001; 293: 510–514. Yang X, Tirado F et al. GeneCodis: interpreting gene lists 47 Lamont EW, Legault-Coutu D, Cermakian N, Boivin DB. The role of through enrichment analysis and integration of diverse biological circadian clock genes in mental disorders. Dialogues Clin Neurosci information. Nucleic Acids Res 2009; 37(Web Server issue): 2007; 9: 333–342. W317–W322. 48 Ma D, Chan MK, Lockstone HE, Pietsch SR, Jones DN, Cilia J et al. 26 Axell MZ, Zlateva S, Curtis M. A method for rapid derivation and Antipsychotic treatment alters protein expression associated with propagation of neural progenitors from human embryonic stem presynaptic function and nervous system development in rat cells. J Neurosci Methods 2009; 184: 275–284. frontal cortex. J Proteome Res 2009; 8: 3284–3297. 27 Benita Y, Kikuchi H, Smith AD, Zhang MQ, Chung DC, 49 McQuillin A, Rizig M, Gurling HM. A microarray gene expression Xavier RJ. An integrative genomics approach identifies study of the molecular pharmacology of lithium carbonate on hypoxia inducible factor-1 (HIF-1)-target genes that form mouse brain mRNA to understand the neurobiology of mood the core response to hypoxia. Nucleic Acids Res 2009; 37: stabilization and treatment of bipolar affective disorder. Pharma- 4587–4602. cogenet Genomics 2007; 17: 605–617. 28 Chi JT, Wang Z, Nuyten DS, Rodriguez EH, Schaner ME, Salim A 50 Fernandez-Egea E, Bernardo M, Donner T, Conget I, Parellada E, et al. Gene expression programs in response to hypoxia: cell type Justicia A et al. Metabolic profile of antipsychotic-naive specificity and prognostic significance in human cancers. PLoS individuals with non-affective psychosis. Br J Psychiatry 2009; Med 2006; 3:e47. 194: 434–438.

Molecular Psychiatry NPAS3 transcription factor targets L Sha et al 279 51 Herberth M, Koethe D, Cheng TM, Krzyszton ND, Schoeffmann S, dorsolateral prefrontal cortex indicates the involvement of cytoske- Guest PC et al. Impaired glycolytic response in peripheral blood leton, oligodendrocyte, energy metabolism and new potential mononuclear cells of first-onset antipsychotic-naive schizophre- markers in schizophrenia. JPsychiatrRes2009; 43: 978–986. nia patients. Mol Psychiatry 2010 (e-pub ahead of print). 53 Do KQ, Cabungcal JH, Frank A, Steullet P, Cuenod M. Redox 52 Martins-de-Souza D, Gattaz WF, Schmitt A, Maccarrone G, dysregulation, neurodevelopment, and schizophrenia. Curr Opin Hunyadi-Gulyas E, Eberlin MN et al. Proteomic analysis of Neurobiol 2009; 19: 220–230.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)

Molecular Psychiatry