ORIGINAL ARTICLE Molecular Evidence for Mitochondrial Dysfunction in Bipolar Disorder Christine Konradi, PhD; Molly Eaton, BA; Matthew L. MacDonald, BS; John Walsh, MS; Francine M. Benes, MD, PhD; Stephan Heckers, MD Background: The disease mechanism of bipolar disor- Results: The expression of nuclear messenger RNA cod- der remains unknown. Recent studies have provided evi- ing for mitochondrial proteins was significantly de- dence for abnormal gene expression in bipolar disorder. creased in the hippocampus in subjects with bipolar dis- order but not in those with schizophrenia. Subjects with Objective: To determine the expression of 12558 nuclear bipolar disorder were characterized by a pronounced and genes in the human hippocampus in healthy control sub- extensive decrease in the expression of genes regulating jects and those with bipolar disorder or schizophrenia. oxidative phosphorylation and the adenosine triphos- phate–dependent process of proteasome degradation. Design: We used gene arrays to study messenger RNA expression. Data were verified with a real-time quanti- Conclusions: These findings point toward a wide- tative polymerase chain reaction assay. spread dysregulation of mitochondrial energy metabo- lism and downstream deficits of adenosine triphosphate– Subjects: We studied 10 healthy control subjects, 9 dependent processes in bipolar disorder. subjects with bipolar disorder, and 8 subjects with schizophrenia. Arch Gen Psychiatry. 2004;61:300-308 IPOLAR DISORDER AFFECTS chondrial genes coding for mitochon- approximately 0.5% of the drial proteins.8 In this article, we report world population, often that the expression of nuclear messenger leading to recurrent illness RNA (mRNA) coding for mitochondrial and a marked decline in so- proteins is significantly decreased in the Bcial function.1 The clinical features of bi- hippocampus in bipolar disorder but not polar disorder (ie, recurrent episodes of de- in schizophrenia. pression and either full-blown mania with frank psychosis or milder bouts of hypo- METHODS mania) have long been recognized.2 How- ever, the etiologic and disease mecha- SUBJECTS nisms remain unknown. For example, We analyzed the expression of 12558 nuclear bipolar disorder shows a high degree of genes in 3 study groups: healthy controls, sub- heritability (approximately 0.8%), and sev- jects with bipolar disorder, and subjects with eral studies have reported linkage of bi- schizophrenia. Brain specimens were ob- polar disorder to chromosomal loci, but tained from the Harvard Brain Tissue Re- From the Department of not a single locus has repeatedly been source Center (McLean Hospital, Belmont, Psychiatry, Harvard Medical linked to bipolar disorder.3 Mass) and initially consisted of 10 subjects in School, Boston, Mass Recent spectroscopic studies have each diagnostic group. Each control subject was (Drs Konradi, Benes, and provided evidence for bipolar disorder as matched with 1 subject who had schizophre- Heckers); and the Laboratory of a disease of mitochondrial energy metabo- nia and 1 who had bipolar disorder for age and Neuroplasticity (Dr Konradi, lism,4 including decreased pH5 and de- postmortem interval to ensure homogeneity of Ms Eaton, and Mr MacDonald) 6,7 the groups. One subject with bipolar disorder and Laboratory for Structural creased high-energy phosphates in the and 1 subject with schizophrenia were ex- Neuroscience (Drs Benes and frontal and temporal lobes of these pa- cluded from the study because they did not pro- Heckers and Mr Walsh), tients. Such mitochondrial dysfunction in vide sufficient RNA quality, as assessed by the McLean Hospital, bipolar disorder could be due to an ab- 3Ј/5Ј ratio of glyceraldehyde-3-phosphate de- Belmont, Mass. normal expression of nuclear or mito- hydrogenase (Ͼ4), 3Ј/5Ј ratio of ␤-actin (Ͼ4), (REPRINTED) ARCH GEN PSYCHIATRY/ VOL 61, MAR 2004 WWW.ARCHGENPSYCHIATRY.COM 300 ©2004 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/25/2021 and percentage (Ͻ37%) of “gene-present calls” (the percent- GENE ARRAY DATA ANALYSIS age of genes on the array that were above the detection limit in a sample). Samples were analyzed in diagnostic groups using the dChip All diagnoses were established by 2 psychiatrists at the Har- program (http://www.dchip.org).11 Model-based expression was vard Brain Tissue Resource Center via retrospective review of performed on perfect match–only data. A control sample with all available medical records and extensive questionnaires about average intensity was chosen for normalization. We found no social and medical history completed by family members of the significant differences in the quality control criteria provided donors. We applied the criteria of Feighner et al9 for the diag- by the Data Mining Tool (Affymetrix) and dChip analyses (3Ј/5Ј nosis of schizophrenia and that of the DSM-III10 for the diag- ratios for glyceraldehyde-3-phosphate dehydrogenase and nosis of schizoaffective disorder and bipolar disorder. Pro- ␤-actin as well as scaling factor and background) or in the ra- bands with a documented history of substance dependence or tio of 28S/18S ribosomal RNA obtained with the bioanalyzer. neurological illness were excluded from the study. During our A significant difference was found in gene-present calls, which study it became evident that the documentation of 1 subject were lower (P=.04) in the bipolar disorder group (Table 1). with schizophrenia was not sufficient to verify the diagnosis, We explored expression profiles revealed by the dChip analy- and that case had to be excluded. sis further with the GenMAPP and MAPPfinder (http://www One hemisphere of each brain underwent a comprehen- .genmapp.org) programs. GenMAPP was used to draw maps of sive neuropathologic examination, which revealed no evi- genes in functionally related groups.12 The MAPPfinder pro- dence of stroke, tumor, infection, or neurodegenerative changes. gram was used to find regulation trends in groups of genes or- After exclusion of 3 cases (insufficient RNA quality in 2 cases ganized according to biological process, molecular function, or and insufficient documentation of the psychiatric history in 1 cellular component, as defined by the Gene Ontology Consor- case), the final sample sizes were 10 control subjects, 9 sub- tium (http://www.geneontology.org). The following criteria were jects with bipolar disorder, and 8 subjects with schizophrenia chosen for the MAPPfinder analysis: PϽ.02, with gene-present (Table 1). calls in more than 50% of the samples and at a fold induction higher than 1.1. For result verification, data were also com- TISSUE PREPARATION puted with Affymetrix Data Mining Tool software version 3.0. AND RNA EXTRACTION REAL-TIME QUANTITATIVE POLYMERASE All brains were transported on wet ice and dissected immedi- CHAIN REACTION ately on arrival by specially trained staff at the Harvard Brain Tissue Resource Center using a standard protocol (see http: For real-time quantitative polymerase chain reaction (PCR), //www.brainbank.mclean.org/ for details). A coronal block of complementary DNA was synthesized from 1 µg of total RNA the hippocampus was obtained at the level of the lateral genic- with a synthesis system (SuperScript First-Strand Synthesis Sys- ulate nucleus, frozen in liquid nitrogen vapor, and stored tem for real-time quantitative PCR; Invitrogen Corp) using oli- at −80°C. Mean±SD storage time was 31±14 months with no gonucleotide deoxythymidine as the primer. A primer set for significant difference between groups. Blocks were trimmed to each gene was designed with the help of Primer3 software include only the dentate gyrus and cornu ammonis sectors (http://www.genome.wi.mit.edu/cgi-bin/primer/primer3.cgi). Am- 1 through 4 without adjacent white matter of the parahippo- plicons were designed to be between 100 and 150 base pairs in campal gyrus. Twenty-five slices (10 µm thick) were cut from length. Melt curve analysis and polyacrylamide gel electropho- each hippocampal block in a cryostat and used for RNA ex- resis were used to confirm the specificity of each primer pair. The traction. real-time quantitative PCR reaction was performed in accor- Human hippocampal RNA was prepared according to the dance with described procedures13 (DNA Engine Opticon; Op- protocol provided by Affymetrix (Santa Clara, Calif). The RNA ticon Monitor Data Analysis Software version 1.4; MJ Research, was extracted from 50 to 100 mg of tissue with an extraction kit Waltham, Mass) with a PCR kit (DyNAmo SYBR Green real- (RNAgents kit; Promega, Madison, Wis). The total yield of RNA time quantitative PCR kit; Finnzymes, Espoo, Finland) in a vol- was the same in all 3 groups. The RNA quality was assessed us- ume of 25 µL, with 2.5 µL of 1:5 diluted complementary DNA ing an analytical gel and a bioanalyzer (Agilent Technologies, samples and 0.3-µM primers. The PCR cycling conditions were Palo Alto, Calif). We used 8 µg of total RNA for complementary initially 95°C for 10 minutes followed by 49 cycles at 94°C for DNA synthesis with a double-stranded complementary DNA syn- 30 seconds, 55°C for 30 seconds, and 72°C for 30 seconds. Data thesis kit (SuperScript; Invitrogen Corp, Carlsbad, Calif), and were collected between 72°C and 79°C depending on amplicon in vitro transcription was performed with an RNA transcript la- melting temperature. A melt curve analysis was performed at the beling kit (Enzo IVT kit; Enzo Biochem, Farmingdale, NY). Both end of each real-time quantitative PCR experiment. Dilution curves the schizophrenia group and the bipolar disorder group had a were generated for each primer in every experiment by diluting 20% lower yield of biotinylated RNA. Whereas the difference was complementary DNA twice from a control sample with a ratio of not significant in the bipolar disorder group, it reached signifi- 1:3, yielding a dilution series of 1.00, 0.33, and 0.11. The loga- cance in the schizophrenia group. Biotinylated RNA was frag- rithm of the dilution value was plotted against the cycle thresh- mented and hybridized to the HG-U95Av2 array (Affymetrix) old value.