Molecular Psychiatry (2006) 11, 158–171 & 2006 Nature Publishing Group All rights reserved 1359-4184/06 $30.00 www.nature.com/mp ORIGINAL ARTICLE Mapping of hippocampal clusters regulated by the amygdala to nonlinkage sites for schizophrenia RE Burke1, J Walsh1, D Matzilevich1,3 and FM Benes1,2,3 1Program in Structural and Molecular Neuroscience, McLean Hospital, Belmont, MA, USA; 2Program in Neuroscience, Harvard Medical School, Boston, MA, USA and 3Department of Psychiatry, Harvard Medical School, Boston, MA, USA

A recent study using a ‘partial’ rodent model of schizophrenia has employed amygdalar activation to induce reported changes in the expression of hippocampal associated with metabolic and signaling pathways in response to amygdalar activation. The amygdalo– hippocampal pathway plays a central role in the regulation of the stress response and emotional learning. In the current study, we have performed a mapping analysis to determine whether genes showing changes in response to environmental stress may form clusters and, if so, whether they might show a topographical association with linkage sites for schizophrenia. When the hippocampal genes showing changes in expression were topographically mapped on specific rat , significant clustering was observed on chromosomes 1, 4 and 8, although chromosome 1 showed the largest amount of clustering. When these same rodent genes were mapped to human chromosomes, most of the genes found on chromosome 1 in rat mapped to chromosome 11 in human. The vast majority of the genes showing changes in regulation were excluded from known linkage sites for schizophrenia. Based on these findings, we postulate that environmental factors may contribute to the endophenotype for schizophrenia through the activation and/or deactivation of specific genetic clusters, ones that do not appear to be directly associated with susceptibility genes for this disorder. Molecular Psychiatry (2006) 11, 158–171. doi:10.1038/sj.mp.4001769; published online 22 November 2005 Keywords: rodent model; picrotoxin; chromosomes 1; chromosome 11; environmental regula- tion; homology

Introduction hippocampal pathway plays a central role in mediat- ing environmental stress responses and also contri- After many years of intensive family studies and butes to the encoding of context-dependent explicit linkage analysis,1 it is now generally believed that memory by the hippocampus.7 Evidence for GABAer- schizophrenia is best explained by a two factor model gic dysfunction has been reported in both the in which both genetic and environmental factors are amygdala8 and hippocampus9–15). Based on these needed to express the clinical phenotype.2,3 To date, studies, particularly those showing changes in the susceptibility genes for this disorder have eluded regulation of the GABA receptor, and the preferential identification and this has made it difficult to under- A distribution of abnormalities in sectors CA3 and CA2 stand how their expression may be influenced by (for reviews, see, Benes and Berretta6), the amygdalo– environmental factors. Indeed, there may be sets of hippocampal pathway is believed to play a pivotal genes independently regulated by environmental role in the pathophysiology of psychotic disor- influences that exist in parallel with susceptibility ders.16,17 Using the ‘partial’ model in which picro- genes and possibly influence their expression pat- toxin, a noncompetitive antagonist of the GABA-A terns.4,5 receptor, is infused into the basolateral nucleus of the To explore this hypothesis, we have used a ‘partial’ amygdala,18,19 we reported complex changes in the rodent model6 to investigate how hippocampal genes genetic regulation of hippocampal cells, particularly showing changes in regulation in response to amyg- in the monoamine and peptide G- coupled dalar activation are topographically mapped in rela- pathways in response to amygdalar activation.20 tion to the rat and . The amygdalo– These data support a model in which amygdalar activation of the hippocampus results in regulatory Correspondence: Dr FM Benes, Program in Structural and changes of long-term potential (LTP) and/or apopto- Molecular Neuroscience, McLean Hospital, 115 Mill Street, sis.21,22 Accordingly, these findings have raised Belmont, MA 02478, USA. E-mail: [email protected] important questions as to whether the various genes Received 9 June 2004; revised 28 June 2005; accepted 4 October showing changes in expression in response 2005; published online 22 November 2005 to amygdalar activation of the hippocampus are Mapping hippocampal gene clusters RE Burke et al 159 randomly distributed throughout the genome or from the Online Mendelian Inheritance in Man whether perhaps they may be clustered together on (OMIM) database32 and compared to the locations of particular chromosomes, perhaps in arrangements the human homologs of our rat genes-of-interest to that would facilitate epigenetic regulation.23 It is also detect the amount of overlap (Figure 2). The OMIM important to know whether there is a positional database is useful as a curated repository of informa- relationship between genes associated with the heri- tion about genetic disease, including current informa- tability of schizophrenia24 versus those more likely to tion about linkage areas with significant logarithm-of- be associated with the response to environmental odds (LOD) scores. MIM #181500 was used for the stress. Such information can potentially help define sites of genetic linkage as it provides an overview of whether the interaction of inherited genes and the published literature. Published support for these environmental factors in schizophrenia might occur linkage sites can be found in each MIM listed in the at the genomic level or perhaps at the level of Works Cited (181500, 603342, 181510, 600511, functional neural circuits. 603013, 603176, 605419, 603206, 600850, 192430, To begin to address these issues, we describe below 188400 and 604906, more than 200 linkage studies the results of a topographical analysis of the distribu- in all). tion of hippocampal genes that show regulatory In order to determine whether or not the 78 genes- changes following acute amygdalar activation. To of-interest were randomly distributed across the our knowledge, this is the first study to evaluate the rat genome (in other words, if the distribution of potential association between environmentally regu- the genes-of-interest was independent of the chromo- lated genes and established linkage sites for schizo- some number or if a disproportionate number of phrenia. genes-of-interest localized to one or more chromo- somes), an R  C (21 row  2 column) G-test 33 Materials and methods of independence was used. An a priori assumption of a random distribution of the genes-of-interest The ‘partial’ model study was conducted as des- seems appropriate given the relatively large number cribed.20 Briefly, a subepileptic dose of picrotoxin of genes from each chromosome on the Affymetrix (PICRO; 50 ng/0.8 ml; RBI, Natick, MA, USA) or RG-U34A chip and the relatively small number vehicle (VEH) was continuously infused over a period of genes-of-interest. The human genome contains of 96 h via a cannula stereotaxically placed in the nonrandom clusters of both genes with high expres- basolateral nucleus of the amygdala of 20 Sprague– sion levels expressed in many different tissues Dawley rats. The animals were killed 96 h later (ridges) and genes with low expression levels and and the hippocampus was removed, flash frozen, more tissue-specific expression (antiridges).34 How- and its RNA was processed for interrogation ever, it seems most appropriate to determine if there is with Affymetrix RG-U34A rat genome arrays. Statis- any clustering first by assuming a random distribu- tical analyses were performed on the arrays using tion, then assessing whether the clustering fits into the Affymetrix MicroArray Suite (Affymetrix, Santa either of the two preceding categories. The G-test is an Clara, CA, USA), dChip (www.dchip.org),25,26 and appropriate statistical test for determining whether or GenMAPP (www.genmapp.org)27 to identify only not an independent relationship exists between two biologically relevant genes-of-interest that showed categories or attributes.33 In this case, the number of significantly altered expression levels in the PICRO genes changed and the chromosome number (e.g. rats when compared to VEH rats (determined ) were the two categories. The absence by t-test (Pp0.25))20). The data used for the current of an association between genes-of-interest and a analysis can be accessed through the following particular chromosome (12, for example) would URL: www.mclean.harvard.edu/research/mrc/bene- suggest that the changes in gene expression induced sville.php. by the ‘partial’ model are randomly distributed across The NetAffx database (www.netaffx.com)28 was the rat genome when considered on a chromosome- used to obtain the genomic locations of the genes-of- by-chromosome basis. Information concerning how interest in this study. NetAffx utilizes the University many genes from each rat chromosome are on the RG- of California at Santa Cruz (UCSC) Genome Bioinfor- U34A chip was derived from the UCSC Genome matics database for this information.29 Of the 85 Bioinformatics29 and Affymetrix NetAffx28 web pages. genes-of-interest, 78 had documented > 99% se- The expression levels and breadth of expression quence homology with a known region of the rat of the genes-of-interest were also visualized using genome and were included in the analysis. NetAffx HTMseq (http://bioinfo.amc.uva.nl/HTMseq)34 to was also used to locate the appropriate human determine if they fit criteria for ridge- or antiridge- homolog (if one has been identified) for each gene- like clusters. of-interest. The rat and human genes were appro- The gene distribution of the genes-of-interest on priately placed on ideograms obtained from the NCBI individual chromosomes was investigated next using Genome Map Viewer and the University of Washing- dChip. The dChip statistical platform25,26 uses a ton, Department of Pathology, respectively.30,31 ‘Genome Information File’ constructed from resources Human chromosomal locations with significant in the UCSC Genome Bioinformatics Database and the linkage scores for schizophrenia (SZ) were derived Affymetrix NetAffx site to measure the relative

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 160 CHROMOSOME 1 CHROMOSOME 2 GNRH receptor 1p13 1p11 2q11 1p12 Opioid receptor MOR-1 1p11 1p11 2q12 Complex I, 13kDa subunit 1q11 2q13 2q12 G gamma 8 2q14 1q12 1q12 2q15 Lactate dehydrogenase C, testis 2q16 cAMP phosphodiesterase PDE 4D 1q21 2q16 1q21 2q21 Cluster p value=0.027251 Uncoupling protein 2 2q22 1q32 Uncoupling protein 3 2q23 1q22 1q32 Cholecystokinin B receptor 2q24 1q32 Cluster p value=0.001184 1q31 2q25 Dopamine D4 receptor Small proline-rich protein 1q32 1q37 2q26 2 Tyrosine hydroxylase 1q33 2q31 1q34 1q41 1q35 2q32 Glutamate receptor GluR-B Cluster p value=0.000008 1q36 7-dehydrocholesterol reductase 2q33 1q37 2 1q41 1q41 1q42 Fos-related antigen 1 2q34 1q43 1q41 1q51 Carnitine palmitoyltransferase I, liver 2q41 1q52 1q43 1q53 Acetylcholine M1 receptor 2q42 1q54 1q43-q51 2q43 1q55 Fatty acid coA ligase 5 2q44 1q55 2q45

CHROMOSOME 3 CHROMOSOME 4

3p13 4q11 3p12 3p11 4q12 3q11 4q13 3q12 4q21

Acetylcholine M2 receptor 3q21 Glycerol-3-phosphate dehydrogenase 4q22 4q22 3q22 3q21-q23 3q23 4q23 Caspase-2 4q23 3q24 26S subunit, TBP-1 4q24 3q24 3q31 4q31 3q32 4q32 3q33 4q33 3q34 3q35 Adrenergic beta 2 receptor 3q35 4q34 3q36 Cluster p value=0.017237 4q41 G1/S specific cyclin D2 3q41 Bcl-X 4q44 4q42 3q41.2 Lactate dehydrogenase B, heart 4q43 4q44 3q42 kRAS B 4q44 3q43

CHROMOSOME 5 CHROMOSOME 6

6q11 5q11 Opioid receptor KOR-1 5q11 5q12 6q12 Trifunctional enzyme beta 6q12 5q13 6q13 6q14 2,4 Dienoyl-coA reductase 5q21 6q15 5q21 I-kappa B 6q16 6 5q22 5q23 Aldolase B, liver 6q21 5q24 5 6q22 6q23 5q31 5-HT-6 receptor 5 5q32 6q24 20S subunit, component c8 5q33 6q24 5q34 5q35 6q31 TSH receptor 6q25 5q36 6q32 Phospholipase A2 5q36 6q33

Figure 1 Locations of genes-of-interest on the rat genome with clusters identified. Clusters are identified in blue on chromosomes 1, 4 and 8. The distribution of genes across the genome followed a random distribution; however, these clusters on specific coding regions of these particular chromosomes indicate the possibility of another level of regulation occurring in the ‘partial’ model as a response to PICRO infusion.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 161 CHROMOSOME 7 CHROMOSOME 8

Complex V, ATP synthase delta 7q11 7q11 8q11 7q12 8q12

7q13 8q13 Leukotriene A4 hydrolase 7 8q21 7q21 8q22 8q23 7q22 Complex IV, COX 5a TRH receptor Cluster p value= 0.045401 8q24 7q22 8q24 Smad3 8q24 7q31 Squalene monooxygenase 7q31 20S subunit, component c9 7q32 8q24 7q33 8q31

7q34 Carnitine palmitoyltransferase I,muscle 7q34 Chemokine receptor CCR-2 8q32 8q32 7q35 Gi alpha I 2 7q36 8q32

CHROMOSOME 9 CHROMOSOME 10

9q11 10q11 9q12 UDP-galactose transporter 10 9q13 10q12

Adrenergic alpha 1b receptor 9q21 10q21 10q21

9q22 10q22 10q23 9q31 14-3-3 protein for MSF L1 subunit 10q24 10q24 9q32 Arachidonate 12-lipoxygenase 10q25 10q24 9q33 10q26 9q34 9q35 LSH receptor 10q31 9q36 9q36 Protein kinase C epsilon subunit 10q32+1 Somatostatin type 2 receptor 10q32.1 9q37 9q36 10q32+2 9q38 10q32+3

CHROMOSOME 11 CHROMOSOME 12

11p12 12p12 11p11

12p11 11q11

12q11 5-lipoxygenase activating protein 12q11 11q12 12q12

11q21 12q13 Complex IV, COX 6a, liver 12 Complex V, OSCP 12q14 11q21 12q15 11q22

11q23 12q16

Figure 1 Continued.

distance of selected genes-of-interest from one an- distributed on that chromosome. If Xi is the rank of other on a single chromosome and assigns a P-value gene i on the chromosome (in other words, the rank of using normalized rank distances. The gene i divided by the number of all the genes on the number where transcription begins for a certain gene chromosome), the distribution of these ranks is (starting with 0 at the telomeric end of the p arm) is normalized to a distribution between 0 and 1. Under used as the location of that gene on the chromosome. these conditions, y=X(n)ÀX(i) represents the normal- The dChip algorithm assumes a null hypothesis that n ized rank distance of a particular stretch of DNA selected genes on a chromosome will be randomly between two genes. The probability (P) that y is less

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 162 CHROMOSOME 13 CHROMOSOME 14

13p13 14p22 Phosphoglycerate mutase, brain 14p22 13p12 13p11 14p21 Betaglycan 14p21 13q11 Chemokine receptor CXCR-4 13q11 13q12 14p11

13q13 Cholecystokinin A receptor 14 14q11 5-HT-5B receptor 13q21 13

13q22 FSH receptor 14q21 Phosphoglycerate mutase, muscle 13q23 13 13q24 14q21-q22 13q25

13q26 14q22 13q27

CHROMOSOME 15 CHROMOSOME 16

15p16 16p16 Protein kinase C delta subunit 16p16 16p15 15p15 Bone morphogenic protein 16p14 Lipoprotein lipase 15p14 15p14 16p14 15p13 Adenylate cyclase AC4 16p13 15p13 16p12 15p12 Squalene synthetase 15p12 16p11 15p11 15q11 16q11

15q12 16q12+1 Endothelin receptor B 15 16q12+2 15q21 16q12+3 16q12+4 15q22 15q23 16q12+5 15q24 15q25

CHROMOSOME 17 CHROMOSOME 18

18p13 17p14

17p13 18p12 Smad7 18p12 17p12 6-phosphofructokinase C, platelet 17 18p11 Melanocortin receptor MC4 17p11 18p11 17q11 18q11 Angiotensin receptor AT1A-R 17q12 17q12+1 Acetylcholine M3 receptor 17q12.1 18q12+1 17q12+2 Isopentenyl diphosphate isomerase 17q12.2 18q12+2 17q12+3 18q12+3

18q13

Figure 1 Continued. than or equal to the observed value can then be P½Xði þ 2ÞÀXðiÞoyŠ calculated using the following equation: ¼ 1 Àð1 À yÞnÃyÃð1 À yÞðnÀ1Þ P½Xði þ 1ÞÀXðiÞoŠ n ¼ 1 À Pðnone of X yÞ ¼ 1 Àð1 À yÞ io À Pðexact 1 of XioyÞ ¼ 1 À Pðnone of XioyÞ ¼ P½Xð1ÞoyŠ ¼ P½ðXð2ÞoyŠ ...etc:

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 163 CHROMOSOME 19 CHROMOSOME 20 Table 1 G test of independence

19p14 Chromosome Number of Number of Total 19p13 genes genes 20p12 19p12 Lmp7 changed unchanged 20p12 19p11 Gi alpha o 1 19p11 1 14 1062 1076 20p11 2 4 414 418 3 4 285 289 19q11 Smad1 4 5 848 853 19q11 5 5 305 310 20q11 6 4 196 200 7 5 242 247

20q12 8 5 160 165 19q12 9 2 214 216 10 5 322 327 20q13 11 1 70 71 12 2 31 33 13 3 142 145 X CHROMOSOME Y CHROMOSOME 14 4 81 85 Xq11 15 4 76 80 Xq12 16 2 67 69

Xq13 17 4 78 82 Xq14 18 2 42 44

Yq11 19 2 19 21 Xq21 20 1 74 75 X 0 23 23 Xq22 Total 78 4752 4830 Xq31 A G-test of independence was used to determine if, given Xq32 Yq12 Xq33 the number of genes-of-interest on each chromosome Xq34 (second column) and the total number of genes from each Xq35 chromosome represented on the Affymetrix RG-U34A chip Xq36 (third column), the distribution of the genes-of-interest was

Xq37 random across the entire genome. The G-test assumes a null hypothesis of independence between two categories or Figure 1 Continued. groups; in this case, the groups are the chromosomes and the genes-of-interest. In other words, the test assumes that there is no dependent relationship between a certain In this way, the algorithm measures the relative chromosome and a disproportionate number of genes-of- ‘tightness’ of a gene cluster of n genes on a chromo- interest, and this hypothesis was validated (P = 0.92). some against n genes randomly placed on the chromosome. The random distribution of these genes microarray analysis (see Benes et al.20) for both saline is ranked and normalized so that order statistics can controls (n = 4) and 96 h picrotoxin-treated animals then be used to calculate a P-value.25,26 Only clusters (n = 4) was used. Complementary DNA was synthe- containing more than two members were considered. sized from 1 mg of total RNA using Superscript II The Genome Information File used in this algorithm reverse transcriptase and oligo-dT primers (Invitro- was constructed in the ChipInfo program35 using gen). The 20 ml reaction volume was diluted five-fold information from the Rat Genome Sequencing Con- for use in the PCR. Based on the expression profiles sortium (RGSC) v3.1 (http://www.hgsc.bcm.tmc.edu). obtained in the original microarray analysis and to The Genome Information File contains the location of their locations on a specific chromosome (clustered each rat gene in terms of its corresponding chromoso- versus nonclustered), five genes were selected for mal number and the base pair number where its qRT-PCR: carnitine palmitoyl transferase I liver (Cpt- transcription starts and ends on that chromosome 1a), angiotensin II receptor, type I (ATIA), MI (numbered starting from the telomeric end of the p muscarinic acetylcholine receptor (MIach-R), D2 arm). This information is maintained in the UCSC dopamine receptor (D2R) and protein kinase C delta Genome Bioinformatics database and was obtained (PKC delta). A primer set for each gene was designed and parsed with the appropriate Affymetrix ID for each with Primer3 software (http://www.genome.wi.mit.e- gene obtained from NetAffx through the use of dChip. du/cgi-bin/primer/primer3.cgi). Amplicons were de- signed to be between 100 and 265 base pairs in length Quantitative RT-PCR (Table 3). The PCR amplifications were performed on For real-time quantitative polymerase chain reaction a Smartcycler (Cepheid, Sunnyvale, CA, USA) using (qRT-PCR), total RNA obtained from RNA used in the Dynamo HS SYBR Green qPCR Kit (Finnzymes,

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 164 Espoo, Finland) with 300 nM primer concentrations. Results For the signal detection, the Smartcycler was pro- grammed to an initial step of 15 min at 951C, followed Figure 1 displays rat ideograms with our genes-of- by 40 thermal cycles of 20 s at 951C, and 30 s for both interest mapped to their respective loci. As shown in annealing 601C and extension 301C in a volume of Table 1, our genes-of-interest were randomly distrib- 25 ml. Data were collected between 72 and 791C, uted across the genome, showing no preference for depending on amplicon melting temperature. A melt one chromosome over another (P = 0.92). However, curve analysis was performed at the end of each real- three clusters of genes located significantly closer time quantitative PCR experiment to confirm the together on specific chromosomes than expected by a specificity of the PCR amplicons from each primer random distribution were detected on chromosomes: pair. Relative standard curve methods were used to 1, 4 and 8. Table 2 presents the name, location, t-test determine the transcript levels for all genes targets P-value, direction of regulation and GenMAPP path- and constitutively expressed ‘housekeeping gene’ way for each gene in the clusters. GAPDH were constructed using serial dilutions of Chromosome 1 had by far the largest cluster, with cDNA from a control sample. Dilution curves, con- 12 members: the cholecystokinin B receptor, the trols and experimental samples were all run in acetylcholine M1 receptor, the dopamine D4 receptor, duplicate. Reported values were normalized to the tyrosine hydroxylase, 7-dehydrocholesterol reduc- internal control rat Glyceraldehyde 3 phosphate tase, g gamma 8, lactate hydrogenase C, uncoupling dehydrogenase (accession number M17701), which 2 and 3, a Fos-related antigen, carnitine was not regulated in the microarray or quantitative palmitoyltransferase I and a fatty acid CoA ligase. As PCR analysis using a relative expression software tool suggested in Figure 1, the dChip analysis indicated (REST) for group-wise comparison of the resulting that this group of 12 genes shows significant cluster- expression ratios.36 ing (P = 0.027). Of these, nine showed particularly

Table 2 Genes in clusters

Gene name Location P-value Direction GenMAPP pathway

Chromosome 1 G gamma 8 1q12 0.21 1.05m G protein signaling Lactate dehydrogenase C, testis 1q21 0.23 1.11m Glycolysis and gluconeo Uncoupling protein 2 1q32 0.16 1.12m Electron transport chain Uncoupling Protein 3 1q32 0.20 1.05m Electron transport chain Cholecystokinin B receptor 1q32 0.07 1.08m Peptide GPCRs Dopamine D4 receptor 1q37 0.24 1.07m Monoamine GPCRs Tyrosine hydroxylase 1q41 0.09 1.07m Catecholamine biosynthesis Fos-related antigen 1 1q41 0.24 1.06m Wnt signaling 7-Dehydrocholesterol reductase 1q41 0.15 1.07k Cholesterol biosynthesis Carnitine palmitoyltrans. I, liver 1q43 0.18 1.05m Fatty acid b-oxidation Acetylcholine M1 receptor 1q43 0.23 1.08m Monoamine GPCRs Fatty acid CoA ligase 5 1q55 0.23 1.09k Fatty acid degradation

P-value of cluster 0.03 Of the tightest nine members 1.2 Â 10À3 Of the tightest six members 8.0 Â 10À6

Chromosome 4 G1/S specific cyclin D2 4 0.15 1.11k Wnt signaling Lactate dehydrogenase B, heart 4q44 0.22 1.07k Glycolysis and gluconeogen kRAS B 4q44 0.16 1.10k G protein signaling

P-value of cluster 0.02

Chromosome 8 20S subunit, component c9 8 0.09 1.25k Proteasome degradation Complex IV, COX 5a 8q24 0.14 1.10k Electron transport chain Smad 3 8q24 0.22 1.05m TGF-b signaling pathway

P-value of cluster 0.05

Genes-of-interest located in significant clusters identified by the dChip algorithm are listed. While P-values and even direction of regulation varied, the unlikely chromosomal locations of the genes in these clusters adjacent to one another raise the possibility of another level of regulation occurring, possibly epigenetic in nature.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 165 tight clustering (P = 0.001), while a subgroup of six regulated genes: G1/S specific cyclin D2, lactate genes showed the greatest degree of clustering dehydrogenase B and kRAS B (P = 0.017). Chromo- (P =8Â 10À6). Of these twelve genes, 10 were upregu- some 8 also showed a small cluster of three genes: a lated in PICRO rats when compared to the VEH group. 20S subunit of the proteasome, cytochrome c oxidase Chromosome 4 has a small cluster of three down- 5a and Smad 3 (P = 0.045). In each of these cases, the

CHROMOSOME 2 CHROMOSOME 1 Trifunctional enzyme beta 2p23 25 36.3 5-HT-6 receptor 24 Protein kinase C, epsilon subunit 36.1 36.2 1p35-36 p 23 Phospholipase A2 2p21 34.3 35 22 34.2 1p35 p 34.1 Betaglycan 21 LSH receptor 33 2p21 32 1p33-p32 16 15 NRD convertase* 14 FSH receptor 31 = locus implicated in schizophrenia 2p21-p16 1p32.2-p32.1 13 = locus implicated in schizophrenia 12 22 11.2 11.1 11.1 21 11.2 12 Cluster p value= 0.007199 13 13 12 14.1 11 14.2 11 14.3 Chemokine receptor CXCR-4 12 21 2q21 1q21-q22 22 21 23 22 Glycerol-3 phosphate dehydrogenase 23 24 2q24.1 24 25 31 32.1 31 32.2 32.3 Chaperonins 60 and 10 q 33 2q33.1 32 34 q 35 41 Serine:pyruvate aminotransferase 1q42.1 (2) 36 2q36-q37 42 (mitochondrial)* 43 Acetylcholine M3 receptor 37 44 1q43 *Not a GenMAPP gene *Not a GenMAPP gene

CHROMOSOME 3 CHROMOSOME 4

26 25 16 p 24 Protein kinase C delta subunit p 15.3 23 15.2 Cholecystokinin A receptor 22 3p21.31 15.1 4p15.1 – p15.2 14 13 21 Gi alpha I 2 3p21 12 11 = locus implicated in schizophrenia = locus implicated in schizophrenia 11 12 14 Chemokine receptor CCR-2 3p21 13 13 GNRH receptor 4q21.2 12 21 11.1 11.1 22 11.2 12 23 13.1 24 13.2 25 13.3 26 21 27 22 Smad 1 23 28 4q28 Angiotensin receptor AT1A-R 24 3q24 31.1 25 31.2 Neuropeptide Y-1 receptor 31.3 26.1 4q31.3-q32 q 26.2 Transferrin receptor* q 32 26.3 3q26.2 Glutamate receptor (GluR-B) 27 33 4q32-q33 28 34 Somatostatin –14* 29 3q28 35 *Not a GenMAPP gene *Not a GenMAPP gene

CHROMOSOME 5 CHROMOSOME 6

Lmp7 p 25 15.3 Complex I, 13kDa subunit 6p21.3 5p15.33 24 15.2 23 p 15.1 6p23 I-kappa B 14 6p22.3 22 6p21.33 13 21.3 20S subunit, component c9 6p21.32 12 = locus implicated in schizophrenia 21.2 11 11.1 21.1 11.2 Allograft inflammatory factor 1* cAMP phosphodiesterase PDE 4D 12 6p21.3 5q11.2 – q13.3 12 5q12 11.2 11.1 11.1 Olfactory receptor protein* 13 Cluster p value= 0.000793 12 6p21.3 13 14 G1/S specific cyclin D3 14 6p21 15 15 16 Glutathione transferase subunit 8* 21 6p12.1 22 21 23 Adrenergic alpha 1b receptor 6q13-q26 5q23-q32 22

31 23 Adrenergic beta 2 receptor 32 5q31-q32 24 q 33 Opioid receptor MOR-1 25 6q24-q25 = locus implicated in schizophrenia 34 26 35 q 27 *Not a GenMAPP gene *Not a GenMAPP gene Figure 2 Human ideograms with susceptibility sites and human homologs identified. All 24 human chromosomes with homologs of our 78 genes-of-interest are appropriately placed above and genetic linkage susceptibility sites derived from OMIM are labeled in red.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 166 cluster members were generally regulated in the same appropriately noted. Red boxes on the ideograms direction in the PICRO group when compared to the connote locations where published studies have VEH group. With the exception of cytochrome c found significant linkage to schizophrenic illness; oxidase Va, HTMseq detected very low expression only five of the 78 genes are located on four levels for these genes and generally tissue-specific susceptibility loci for schizophrenia. The opioid expression.34 receptor MOR-1, cAMP phosphodiesterase 4D, carni- Figure 2 presents human ideograms with the tine-o-palmitoyltransferase I (muscle), cytochrome c human homologous genes to the genes-of-interest oxidase 6a (liver) and mevalonate kinase overlap

CHROMOSOME 8 CHROMOSOME 7 23 p Squalene synthetase 22 8p23.1-p22 p 22 21 Lipoprotein lipase 21 8p22 8p21 15 12 14 = locus implicated in schizophrenia 13 11.2 Fibroblast growth factor receptor 1 Phosphoglycerate mutase, muscle 12 11.1 beta-isoform* 8p11.2-p11.1 7p13-p12 11.1 11.2 11.1 11.2 11.1 = locus implicated in schizophrenia Opioid receptor KOR-1 12 8q11.2 11.2 *Not a GenMAPP gene 13

21 21.1 21.2 2,4-Dienoyl-coA reductase 22 21.3 8q21.3 Cluster p value=0.047233 Neuronal potassium channel alpha 8q22.3-q24.1 31 22 subunit* Acetylcholine M2 receptor TRH receptor 32 8q23 7q31-q35 23 33 14-3-3 protein for MSF S1 subunit q 34 Caspase - 2 35 8q23.1 7q34-q35 24.1 Squalene monooxygenase 36 24.2 8q24.1 q Growth factor (Arc) *Not a GenMAPP gene 24.3 8q24.3

CHROMOSOME 9 CHROMOSOME10 Isopentenyl diphosphate isomerase 10p15.3 p 24 15 23 p 22 14 6-phosphofructokinase C, platelet 10p15.3-p15.2 21 13 = locus implicated in schizophrenia 13 12 = locus implicated in schizophrenia 12 11.2 11.1 11.1 11.1 11.1 12 11.2 13 21 21 Aldolase B, liver CaMKII gamma subunit* 10q22 9q21.3-q22.2 22 22 Cluster p value= 0.001882 Complex I, chain I 31 23 10q23.2-q23.3 32 Gamma-adducin* 33 24 10q24.2-q24.3 q Fatty acid coA ligase 5 25 34 q 10q25.1-q25.2 Phosphoglycerate mutase, brain 10q25.3 26 Adrenergic alpha 2a receptor 10q24-q26 *Not a GenMAPP gene *Not a GenMAPP gene

Dopamine D4 receptor CHROMOSOME 11 11p15.5 CHROMOSOME12 CCK-B receptor 11p15.4 p 15 G1/S specific cyclin D2 Tyrosine hydroxylase 13 12p13 11p15.5 p Lactate dehydrogenase C, testis Lactate dehydrogenase B, heart 12p12.2-p12.1 14 11p15.5-p15.3 12 kRAS B Brain-derived neurotrophic factor* 11.2 12p12.1 = locus implicated in schizophrenia 13 11p13 11.1 12 26S subunit, TBP-1 11 11p12-p13 12 11.2 = locus implicated in schizophrenia 11.1 Acetylcholine M1 receptor 11 11q13 13 12 Fos-related antigen 1 Cluster p value= 0.007409 11q13 14 13 Uncoupling proteins 2 and 3 15 11q13 Lin-7-Ba* Carnitine palmitoyltransferase I, 12q21 11q14-q21 14 21 liver 11q13.1-q13.2 21 Leukotriene A4 Hydrolase 22 7-Dehydrocholesterol reductase 11q13.2-q13.5 22 12q22 Initiation factor associated 67kDa Protein kinase MUK2* 23 23 11q13-q14 protein* 12q23.1 24.1 Mevalonate kinase 12q24 24.2 12q24 24 Complex IV, COX 6a, liver q 25 24.3 12q24.2 *Not a GenMAPP gene q

*Not a GenMAPP gene Figure 2 Continued.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 167 susceptibility loci at 6q24, 5q12, 22q13 and 12q24 (2), change = þ 3.9; P = 0.009) showed a significant upre- respectively. In nonlinkage areas, chromosome 11 gulation that was predicted by the gene expression shows the largest aggregation of genes showing profiling data previously reported. For the unclus- changes in expression in the picrotoxin experiment. tered genes, the angiotensin A1 receptor (fold The majority of these genes were localized to change = þ 4.37; P = 0.04) and protein kinase C delta chromosome 1 in the rat, suggesting that significant (fold change = À3.73; P = 0.001) showed an up- and homology exists between these two sites (Figure 2). downregulation, respectively. Two additional non- QRT-PCR studies were performed on two clustered clustered genes, the D2R (fold change = 1.06; P = 0.92) and four nonclustered different genes to further and G3PDH (fold change = 1.01; P = 0.97), showed no validate the gene expression profiling results (Table change in regulation. In both the clustered and 3). As with the D4 and CCK-B receptors that unclustered genes, the direction of the changes was were previously reported, the M1 receptor (fold identical to that seen with gene expression profiling. change = þ 2.77; P = 0.02) and CPT (fold However, the magnitude of the changes in expression

CHROMOSOME 13 CHROMOSOME 14 p p 13 13 12 12 11.2 11.2 = locus implicated in schizophrenia 11.1 = locus implicated in schizophrenia 11.1 11 11.1 Adenylate cyclase AC4 12 11.2 5-lipoxygenase activating protein 14q11.2 13 13q12 12 14 13 21 Bone morphogenic protein 21 14q22-q23

Endothelin receptor ETB-R 23 20S subunit, component c8 22 14q23 13q22 31 24 Proteasome subunit RC8* 14q23 13q32 32 q 31 33 TSH receptor 14q31 13q34 34 32 q *Not a GenMAPP gene *Not a GenMAPP gene

CHROMOSOME 15 CHROMOSOME 16

p 13 p 12 Ubiquitin ligase (Nedd4)* 13.3 11.2 15q 13.2 11.1 13.1 11.1 11.2 12 12 = locus implicated in schizophrenia 13 14 15q15 15 11.1 11.1 Smad 3 21 15q21-q22 11.2 12.1 12.2 = locus implicated in schizophrenia Slow-twitch alpha tropomyosin* 13 Gi alpha o 1 22 15q22.1 21 16q13 23 22 24 q 23 25 Complex IV, COX 5a 15q25 24 q 26

*Not a GenMAPP gene *Not a GenMAPP gene

CHROMOSOME 17 CHROMOSOME 18 p

Arachidonate 12-lipoxygenase p 13 17p13.3 14-3-3 protein, mitochondrial MSF 11.3 L subunit* 17p13.3 = locus implicated in schizophrenia 12 18p 11.2 11.2 11.1 11.1 11.1 11.1 11.2 11.2 12 Proteasome subunit R-IOTA* 12 17q13 = locus implicated in schizophrenia 21 Plakoglobin* 17q21 Smad7 Frizzled 2 receptor 21 18q21.1 22 17q21.1 23 Melanocortin receptor MC4 22 18q22 24 Somatostatin type 2 receptor q 17q24 q 23 25

*Not a GenMAPP gene *Not a GenMAPP gene Figure 2 Continued.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 168 CHROMOSOME 19p CHROMOSOME 20 p Complex V, ATP synthase delta 13.3 19p13.3 13 = locus implicated in schizophrenia = locus implicated in schizophrenia 13.2 12 13.1 11.2 12 11.1 11 11.1 Bcl-X 11 20q11.1 12 11.2

13.1 12 13.1 13.2 G gamma 8 13.2 Skeletal muscle myosin light 19q13.2-q13.3 13.3 chain kinase* 20q13.31 q q 13.3 13.4

*Not a GenMAPP gene *Not a GenMAPP gene

CHROMOSOME 21 CHROMOSOME 22 p

13 13 12 = locus implicated in schizophrenia 12 11.2 11.2 11.1 11.1 11.1 11.2 11.1 11.2 21 12 22q11-q13 Complex V, OSC P 21q22.11 22q11.2 (2) 13 22 Carnitine palmitoyltransferase I, muscle 22q13.33 q 22q11.3 (3)

= locus implicated in schizophrenia

*Not a GenMAPP gene *Not a GenMAPP gene

X CHROMOSOME Y CHROMOSOME

p 22.3 22.2 p 22.1 11.3 21 11.2 = locus implicated in schizophrenia 11.4 Monoamine oxidase A 11.1 11.3 Xp11.4-p11.3 11.1 11.2 UDP-galactose transporter Xp11.23 11 11.1 11.2 12 13

21 = locus implicated in schizophrenia 12 22 Angiotensin receptor AT2 23 Xq22-q23 24 q 25 q 26 *Not a GenMAPP gene 27 28 *Not a GenMAPP gene Figure 2 Continued.

was much greater than that seen with gene expression chromosomes than would be expected by a random profiling, with the exception of the D2 receptor and distribution. Accordingly, we postulate that amygda- G3PDH, where no changes were observed either with lar activation of the hippocampus may induce gene expression profiling or qRT-PCR (Figure 3). changes in the transcriptional regulation of groups of genes, possibly through epigenetic mechanisms, and that these changes in regulation may affect Discussion specific topographic loci along Chromosomes 1, 4 In general, the hippocampal genes regulated by and 8. The presence of these clusters suggests that amygdalar activation were randomly distributed increased excitatory activity directed to the hippo- across the genome. This was expected since it is campus from the basolateral nucleus of the amygdala difficult to conceptualize a biological explanation for may potentially be associated with the activation or the activation of genes (either more or less than deactivation of epigenetic mechanisms operating at expected) on one particular chromosome rather than the chromosomal level. another. In this setting, the clustering identified on The presence of clusters of similarly regulated specific coding regions of chromosomes 1, 4 and 8 genes may be a likely indication of higher-order was unexpected. These three groups of genes have transcriptional regulation in eukaryotes. Imprinting been identified because they are physically distri- is one well-known example of this regulation, in buted closer to one another on their respective which epigenetic mechanisms cause only one par-

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 169 D2 CPT AT1A PKC delta M1 state of cytosine–guanine (CpG) islands in the 6 p=0.044 p=0.009 promoter of a gene or genes and/or changes in the acetylation state of histones, around which DNA is 4 p=0.021 wound.37–39 In the case of imprinting, epigenetic mechanisms seem to regulate clusters of genes.40 2 p=0.916 Transcriptional control of clusters of genes may have developed from the prokaryotic operon under evolu- 0 tionary pressure to keep related genes near one

2 another on the genome so they can be regulated and -2 G3PDH 1.5 p=0.001 quickly expressed together in response to environ- 41 1 mental change. This could explain observations that Normalized Fold Changes -4 0.5 genes which interact in biological pathways in 42 0 saline picro humans are clustered together, likely in order to -6 assist in coregulation. Genes located near one another Figure 3 A comparison of mRNA levels for clustered also tend to be coexpressed in other species, includ- 43 (carnitine palmitoyl transferase, CPT; angiotensin 1A, ing rats. The human genome itself seems to be AT1A) and unclustered (dopamine D2 receptor, protein organized into higher-order clusters of ubiquitously kinase C delta, PKC delta; M1 muscarinic receptor and expressed housekeeping genes with characteristically glyceraldehydes-3-phosphate dehydrogenase, G3PDH) high expression levels, known as ridges.44,45 The genes in the hippocampus of saline- and picrotoxin-treated opposite has also recently been discovered in the rat. G3PDH that showed no difference between the saline- form of antiridges, clusters of genes with low and picrotoxin-treated rats was used to normalize the data expression levels.34 Transcriptional regulation of for the other genes. The direction of the changes are ridges and antiridges has been postulated to occur at identical to those observed using gene expression profiling. both an individual-gene level and at the cluster level through an epigenetic or other unknown mechanism. A possible biological explanation for the gene Table 3 Sequence of primer sets and amplicon sizes clustering observed in this study may be that we have activated mechanisms involved in the regulation Gene Sequence of primer Amplicon of antiridges, since our clusters have very low (Assession forward and reverse size (bp) 0 0 expression levels and are mostly expressed in only Number) (5 –3 ) one or a few tissues, characteristic of genes in antiridges.34 However, further study is needed to Cpt-1a CATGTCAAGCCAGACGAAGA 111 determine if the regulation involved in epigenetic (NM 031559) TGGTAGGAGAGCAGCACCTT phenomena, ridges, ‘synexpression groups’,41 and the AT1A-R GGAAACAGCTTGGTGGTGAT 171 clusters identified here are the same or disparate (BC078810) ACATAGGTGATTGCCGAAGG phenomena. This is a particularly challenging area, given the dearth of information about clustering in the M1ach-R CCAAAAGCTCCCCAAATACA 265 rat genome. (M16406) CATAGCCAGTAGCCCAGCTC It is important to emphasize that changes in gene regulation do not necessarily imply epigenetic mecha- D2R CATTGTCTGGGTCCTGTCCT 154 nisms are operative, unless considered in the broadest (53278) GACCAGCAGAGTGACGATGA context. The individual gene clusters identified in the current report clearly show changes in regulation that PKC delta GGAAGCCAGAGACACCAGAG 176 (M18330) AAAGCTGCCTTTGCCAAGTA are independent of one another. It is noteworthy that some genes within a single cluster showed increased GAPDH CTACCCACGGCAAGTTCAAT 108 expression, while others showed decreased expres- (M17701) ATTTGATGTTAGCGGGATCG sion. For demethylation and methylation reactions, increases and decreases, respectively, are generally The sequences shown represent primer pairs that were observed;46,47 however, it is not clear whether both designed using web-based Primer-3 design program mechanisms can operate within a small topographical (www.genome.wi.mit.edu). The numbers in parentheses area of a single chromosome, although this sort of represent Genbank accession numbers for each gene. The two-level regulation at the cluster and individual length of the specific amplification products (amplicons) is gene level is postulated to occur in ridges.34 In the shown in the right column. All primer pairs were designed current study, it is not possible at present to know to span an intron to exclude potential genomic DNA what epigenetic mechanism(s) could theoretically contamination. account for the observed changes in gene expression within the three identified clusters. ental allele to be expressed.37 Epigenetic mechanisms It is noteworthy that the number of genes over- influence gene expression by making gene DNA more lapping genetic linkage susceptibility sites for schizo- or less accessible to transcriptional machinery. This is phrenia was nonsignificant. In assessing the accomplished through alterations in the methylation importance of this null finding, one must consider

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 170 that the genes showing changes in regulation in 15 Smith GN, Lang DJ, Kopala LC, Lapointe JS, Falkai P, Honer WG. response to picrotoxin infusion in the BLA may be Developmental abnormalities of the hippocampus in first-episode part of a nonspecific environmentally driven factor schizophrenia. Biol Psychiatry 2003a; 53: 555–561. 21,48 16 Joyal CC, Laakso MP, Tiihonen J, Syvalahti E, Vilkman H, Laakso A such as stress. Such a mechanism may enhance et al. The amygdala and schizophrenia: a volumetric magnetic the ability of the hippocampus to store relevant resonance imaging study in first-episode, neuroleptic-naive information.22 A strong epigenetic component could patients. Biol Psychiatry 2003b; 54: 1302–1304. play a significant role in controlling the regulation of 17 Smith GN, Lang DJ, Kopala LC, Lapointe JS, Falkai P, Honer WG. genes associated with learning and the stress re- Developmental abnormalities of the hippocampus in first-episode 49–51 schizophrenia. Biol Psychiatry 2003b; 53: 555–561. sponse. This could either occur independently of 18 Sanders SK, Shekhar A. Regulation of anxiety by GABAA 52–54 faulty genes or through a synergistic collaboration receptors in the rat amygdala. Pharmacol Biochem Behav 1995; with them.55 In the latter scenario, the genes showing 52: 701–706. a changes in regulation in response to amygdalar 19 Berretta S, Munno DW, Benes FM. Amygdalar activation alters the activation could be among those necessary for hippocampal GABA system: ‘partial’ modelling for postmortem changes in schizophrenia. J Comp Neurol 2001; 431: 129–138. ‘triggering’ the onset of schizophrenic illness in those 20 Benes FM, Burke R, Walsh JP, Berretta S, Konradi C, Minns M. who are genetically predisposed to develop it.56–58 Acute amygdalar activation induces an upregulation of multiple monoamine G protein coupled pathways in rat hippocampus. Mol Psychiatry 2004a; 9: 932–945. Acknowledgments 21 Akirav I, Richter-Levin G. Biphasic modulation of hippocampal plasticity by behavioral stress and basolateral amygdala stimula- This work was supported by grants from the National tion in the rat. J Neurosci 1999; 19: 10530–10535. Institutes of Health (MH42261, MH62822, MH/ 22 Roozendaal B, Nguyen BT, Power AE, McGaugh JL. Basolate- NS31862) and the generosity of Menachem and ral amygdala noradrenergic influence enables enhancement of memory consolidation induced by hippocampal glucocorticoid Carmella Abraham, John and Virginia Taplin and receptor activation. Proc Natl Acad Sci USA 1999; 96: 11642– Anne Allen. 11647. 23 Jaenisch R, Bird A. Epigenetic regulation of gene expression: how the genome integrates intrinsic and environmental signals. Nat References Genet 2003; 33(Suppl): 245–254. 24 O’Donovan MC, Williams NM, Owen MJ. Recent advances in the 1 Kopnisky KL, Cowan WM, Hyman SE. Levels of analysis in genetics of schizophrenia. Hum Mol Genet 2003; 12(Spec No 2): psychiatric research. Dev Psychopathol 2002; 14: 437–461. R125–R133. 2 Barondes SH, Alberts BM, Andreasen NC, Bargmann C, Benes F, 25 Li C, Hung Wong W. Model-based analysis of oligonucleotide Goldman-Rakic P et al. Workshop on schizophrenia. Proc Natl arrays: model validation, design issues and standard error Acad Sci USA 1997; 94: 1612–1614. application. Genome Biol 2001; 2, RESEARCH0032. 3 Bray NJ, Owen MJ. Searching for schizophrenia genes. Trends Mol 26 Li C, Wong WH. Model-based analysis of oligonucleotide arrays: Med 2001; 7: 169–174. expression index computation and outlier detection. Proc Natl 4 Gottesman II, Gould TD. The endophenotype concept in psychia- Acad Sci USA 2001; 98: 31–36. try: etymology and strategic intentions. Am J Psychiatry 2003; 160: 27 Dahlquist KD, Salomonis N, Vranizan K, Lawlor SC, Conklin BR. 636–645. GenMAPP, a new tool for viewing and analyzing microarray data 5 Sullivan PF, Kendler KS, Neale MC. Schizophrenia as a complex on biological pathways. Nat Genet 2002; 31: 19–20. trait: evidence from a meta-analysis of twin studies. Arch Gen 28 Liu G, Loraine AE, Shigeta R, Cline M, Cheng J, Valmeekam V et al. Psychiatry 2003; 60: 1187–1192. NetAffx: Affymetrix probesets and annotations. Nucleic Acids Res 6 Benes FM, Berretta S. Amygdalo-entorhinal inputs to the hippo- 2003; 31: 82–86. campal formation in relation to schizophrenia. Ann NY Acad Sci 29 Karolchik D, Baertsch R, Diekhans M, Furey TS, Hinrichs A, Lu 2000b; 911: 293–304. YT et al. The UCSC Genome Browser Database. Nucleic Acids Res 7 Ledoux J. The amygdala and emotion. In: Aggleton JP (ed). The 2003; 31: 51–54. Amygdala. Oxford University Press: Oxford, 2000, pp 289–310. 30 Adler D. In: U.o.W.D.o. (Ed.) Pathology 1994. http://www.patho- 8 Simpson MD, Slater P, Deakin JF, Royston MC, Skan WJ. Reduced logy.washington.edu/research/cytopages/idiograms/human. GABA uptake sites in the temporal lobe in schizophrenia. 31 Wheeler DL, Church DM, Federhen S, Lash AE, Madden TL, Neurosci Lett 1989; 107: 211–215. Pontius JU. et al. Database resources of the National Center for 9 Reynolds GP, Czudek C, Andrews H. Deficit and hemispheric Biotechnology. Nucleic Acids Res 2003; 31: 28–33. asymmetry of GABA uptake sites in the hippocampus in schizo- 32 NCBI Online Mendelian Inheritance in Man, OMIM (TM). Johns phrenia. Biol. Psychiatry 1990; 27: 1038–1044. Hopkins University. MIM Numbers: 181 500, 603 342, 181 510, 10 Benes FM, Khan Y, Vincent SL, Wickramasinghe R. Differences in 600 511, 603 013, 603 176, 605 419, 603 206, 600 850, 192 430, the subregional and cellular distribution of GABAA receptor 188 400, 604 906. Baltimore, MD, 2003; http://www.ncbi.nlm.nlh. binding in the hippocampal formation of schizophrenic brain. gov/omim/. Synapse 1996; 22: 338–349. 33 Sokal R, Rohlf FJ. Biometry: The Principles and Practice of 11 Todtenkopf MS, Benes FM. Distribution of glutamate decarboxy- Statistics in Biological Research. W.H. Freeman and Co.: New lase65 immunoreactive puncta on pyramidal and nonpyramidal York, 1995, pp 727–741. neurons in hippocampus of schizophrenic brain. Synapse 1998; 34 Versteeg R, van Schaik BD, van Batenburg MF, Roos M, Monajemi 29: 323–332. R, Caron H et al. The human transcriptome map reveals extremes 12 Benes FM, Berretta S. Amygdalo-entorhinal inputs to the hippo- in gene density, intron length, GC content, and repeat pattern for campal formation in relation to schizophrenia. Ann NY Acad Sci domains of highly and weakly expressed genes. Genome Res 2003; 2000a; 911: 293–304. 13: 1998–2004. 13 Benes FM, Berretta S. GABAergic interneurons: implications for 35 Zhong S, Li C, Wong WH. ChipInfo: Software for extracting gene understanding schizophrenia and bipolar disorder. PG-1-27. annotation and information for microarray analysis. Neuropsychopharmacology 2001; 25: 1–27. Nucleic Acids Res 2003; 31: 3483–3486. 14 Joyal CC, Laakso MP, Tiihonen J, Syvalahti E, Vilkman H, Laakso A 36 Pfaffl MW, Horgan GW, Dempfle L. Relative expression software et al. The amygdala and schizophrenia: a volumetric magnetic tool (REST) for group-wise comparison and statistical analysis of resonance imaging study in first-episode, neuroleptic-naive relative expression results in real-time PCR. Nucleic Acids Res patients. Biol Psychiatry 2003a; 54: 1302–1304. 2002; 30: e36.

Molecular Psychiatry Mapping hippocampal gene clusters RE Burke et al 171 37 Jones PA, Takai D. The role of DNA methylation in mammalian 50 Kovalchuk O, Burke P, Arkhipov A, Kuchma N, James SJ, epigenetics. Science 2001; 293: 1068–1070. Kovalchuk I et al. Genome hypermethylation in Pinus silvestris 38 Jenuwein T, Allis CD. Translating the histone code. Science 2001; of Chernobyl – a mechanism for radiation adaptation? Mutat Res 293: 1074–1080. 2003; 529: 13–20. 39 Dennis C. Epigenetics and disease: altered states. Nature 2003; 51 Tchantchou F, Graves M, Ashline D, Morin A, Pimenta A, Ortiz D 421: 686–688. et al. Increased transcription and activity of glutathione synthase 40 Verona RI, Mann MR, Bartolomei MS. Genomic imprinting: in response to deficiencies in folate, vitamin E, and apolipoprotein intricacies of epigenetic regulation in clusters. Annu Rev Cell E. J Neurosci Res 2004; 75: 508–515. Dev Biol 2003; 19: 237–259. 52 Tremolizzo L, Carboni G, Ruzicka WB, Mitchell CP, Sugaya I, 41 Niehrs C, Pollet N. Synexpression groups in eukaryotes. Nature Tueting P et al. An epigenetic mouse model for molecular and 1999; 402: 483–487. behavioral neuropathologies related to schizophrenia vulnerabil- 42 Lee JM, Sonnhammer EL. Genomic gene clustering analysis of ity. Proc Natl Acad Sci USA 2002; 99: 17095–17100. pathways in eukaryotes. Genome Res 2003; 13: 875–882. 53 Detich N, Bovenzi V, Szyf M. Valproate induces replication- 43 Fukuoka Y, Inaoka H, Kohane IS. Inter-species differences of co- independent active DNA demethylation. J Biol Chem 2003a; 278: expression of neighboring genes in eukaryotic genomes. BMC 27586–27592. Genomics 2004; 5:4. 54 Detich N, Hamm S, Just G, Knox JD, Szyf M. The methyl 44 Caron H, van Schaik B, van der Mee M, Baas F, Riggins G, van donor S-Adenosylmethionine inhibits active demethylation Sluis P et al. The human transcriptome map: clustering of highly of DNA: a candidate novel mechanism for the pharmacological expressed genes in chromosomal domains. Science 2001; 291: effects of S-Adenosylmethionine. J Biol Chem 2003b; 278: 20812– 1289–1292. 20820. 45 Lercher MJ, Urrutia AO, Hurst LD. Clustering of housekeeping 55 Petronis A. The origin of schizophrenia: genetic thesis, epigenetic genes provides a unified model of gene order in the human antithesis, and resolving synthesis. Biol Psychiatry 2004; 55: genome. Nat Genet 2002; 31: 180–183. 965–970. 46 Futscher BW, Oshiro MM, Wozniak RJ, Holtan N, Hanigan CL, 56 Day R, Nielsen JA, Korten A, Ernberg G, Dube KC, Gebhart J et al. Duan H et al. Role for DNA methylation in the control of cell type Stressful life events preceding the acute onset of schizophrenia: a specific maspin expression. Nat Genet 2002; 31: 175–179. cross-national study from the World Health Organization. Cult 47 Robertson KD. DNA methylation and chromatin – unraveling the Med Psychiatry 1987; 11: 123–205. tangled web. Oncogene 2002; 21: 5361–5379. 57 Benes FM. The role of stress and dopamine–GABA interactions 48 Cahill L, Vazdarjanova A, Setlow B. The basolateral amygdala in the vulnerability for schizophrenia. J Psychiatr Res 1997; 31: complex is involved with, but is not necessary for, rapid acquisition 257–275. of Pavlovian ‘fear conditioning’. Eur J Neurosci 2000; 12: 3044–3050. 58 Castine MR, Meador-Woodruff JH, Dalack GW. The role of life 49 Finnegan EJ. Epialleles – a source of random variation in times of events in onset and recurrent episodes of schizophrenia and stress. Curr Opin Plant Biol 2002; 5: 101–106. schizoaffective disorder. J Psychiatr Res 1998; 32: 283–288.

Molecular Psychiatry