Human Dysbindin (DTNBP1) Gene Expression inNormal Brain and in Schizophrenic Prefrontal Cortex and Midbrain

Context  The schizophrenia-susceptibility gene dysbindin (DTNBP1 on 6p22.3) encodes a neuronal protein that binds to β-dystrobrevin and may be part of the dystrophin protein complex. Little is known about dysbindin expression in normal or schizophrenic brain.

Objectives  To determine whether brain regions implicated in schizophrenia express dysbindin and whether abnormal levels of dysbindin messenger RNA (mRNA) may be found in this disorder and to test whether sequence variations in the dysbindin gene in the promoter region, 5′ and 3′ untranslated regions, or introns would affect dysbindin mRNA levels.

Methods  In patients with schizophrenia and controls, we compared dysbindin, synaptophysin, spinophilin, and cyclophilin mRNA levels in the dorsolateral prefrontal cortex and dysbindin mRNA levels in the midbrain by in situ hybridization. We genotyped brain DNA at 11 single nucleotide polymorphisms to determine whether genetic variation in the dysbindin gene affects cortical dysbindin mRNA levels.

Main Outcome Measures  Quantitative assessment of dysbindin mRNA levels across various brain regions and comparative studies of dysbindin mRNA levels in brains of patients with schizophrenia compared with normal controls.

Results  Dysbindin mRNA was detected in the frontal cortex, temporal cortex, hippocampus, caudate, putamen, nucleus accumbens, amygdala, thalamus, and midbrain of the adult brain. Patients with schizophrenia had statistically significantly reduced dysbindin mRNA levels in multiple layers of the dorsolateral prefrontal cortex, whereas synaptophysin, spinophilin, and cyclophilin mRNA levels were unchanged. Dysbindin mRNA levels were quantitatively reduced in the midbrain of patients with schizophrenia, but not statistically significantly. Cortical dysbindin mRNA levels varied statistically significantly according to dysbindin genotype.

Conclusions  Dysbindin mRNA is expressed widely in the brain, and its expression is reduced in schizophrenia. Variation in dysbindin mRNA levels may be determined in part by variation in the promoter and the 5′ and 3′ untranslated regions. These data add to the evidence that dysbindin is an etiologic factor in schizophrenia risk.

Inheritance of certain forms of genes can increase the risk of developingschizophrenia^1,2; however, itis unclear how these "vulnerability" alleles contribute to aberrant functioningof the brain and lead to the symptoms associated with schizophrenia.³ One possibility is that alleles that increase riskfor schizophrenia may directly affect the development, maturation, and adultfunction of the dorsolateral prefrontal cortex (DLPFC),⁴ anarea where cellular and molecular abnormalities are found in the schizophrenicbrain.^5-9 Recently,alleles of the gene that encodes dysbindin (β-dystrobrevin binding proteinor DTNBP1 on 6p22.3), whose protein product has beenlocalized to neurons in the central nervous system,^10,11 havebeen shown to increase the risk of developing schizophrenia.^12-14 However,to our knowledge, little information is currently available on the brain regionsthat contain dysbindin messenger RNA (mRNA) or protein in either healthy orpsychiatrically compromised human brain. This lack of information limits ourability to delineate anatomic brain areas and molecular pathways that maybe affected by the inheritance of dysbindin risk alleles in schizophrenia.One of the central challenges in the genetics of complex human traits, suchas schizophrenia, is uncovering how cis-acting polymorphisms in a susceptibilitygene may lead to altered gene regulation.¹⁵ Therefore,in this study, we also examine whether known single nucleotide polymorphisms(SNPs) affect dysbindin mRNA levels.

Dysbindin is a 40- to 50-kDa protein that is ubiquitously expressedin rodent tissues and that binds to α- and β-dystrobrevin in muscleand brain, respectively.¹¹ The dystrobrevinsassociate with dystrophin, the product of the gene mutated in Duchenne musculardystrophy, which, together with syntrophins and dystroglycan, form the coreof the dystrophin protein complex (DPC) (for a review, see Watkins et al¹⁶). The DPC has been studied primarily in muscle, whereit stabilizes the postsynaptic membrane, is involved in cytoskeletal rearrangement,and perhaps facilitates the transduction of extracellular signals.^17,18 Because patients with Duchenne musculardystrophy have cognitive dysfunction and occasionally mental retardation,¹⁹ it has been proposed that the DPC may be involvedin the maintenance of structure and function of the postsynaptic membranesof central nervous system neurons.²⁰ Componentsof the DPC that have been localized to cortical neurons and more specificallyto neuronal postsynaptic densities include dystrophin, syntrophin, and dystrobrevin.^19,21,22 Dysbindin proteinhas also been localized to postsynaptic sites, particularly in Purkinje cellsoma and dendrites in the mouse cerebellum.¹⁰ Somaldysbindin may play a role in organelle biogenesis by binding to proteins suchas pallidin and muted, which are involved in organelle trafficking.²³ However, another established binding partner fordysbindin, β-dystrobrevin, has been associated with presynaptic terminalsin photoreceptors in the avian retina, implying that dysbindin may also belocalized in axon terminals.²⁴ Available neuroanatomicevidence^10,11 is consistent witha synaptic localization for dysbindin, as mossy fiber terminal fields in thehippocampus and cerebellum and neuropil areas of the cortex and substantianigra contain dysbindin protein in the rodent brain. Thus, dysbindin may belocated presynaptically and postsynaptically in the central nervous system.

In this study, we characterized the anatomic distribution of dysbindinmRNA in 12 distinct areas of the human brain, with special emphasis on areasmost often implicated in schizophrenia.²⁵ Wethen tested whether dysbindin expression was altered in the DLPFC and midbrainof patients with schizophrenia. In brain imaging studies and cognitive assessments,the DLPFC has consistently been found to be dysfunctional in patients withschizophrenia.^26,27 Underlyingthese behavioral and functional deficits, molecular and cellular abnormalitiesexist, including reduced neuronal size, synaptic connectivity, and plasticityof cortical neurons.^28-37 Giventhe evidence of a generalized "synaptic pathology" in the brains of patientswith schizophrenia, we hypothesized that a reduction in dysbindin mRNA levelsmight be predicted owing to the presence of fewer presynaptic and postsynapticsites (where dysbindin can be located). In this scenario, a decrease in dysbindinexpression could simply be secondary to changes in synaptic structure. Tomonitor the magnitude of synaptic change, we also measured the expressionof synaptic-associated gene products used as proxies for the number of synapticterminals. We hypothesized that levels of synaptophysin and spinophilin mRNA,markers of the presynaptic and postsynaptic terminals, respectively, may bereduced in the DLPFC in patients with schizophrenia and that these changesmay correlate with changes in dysbindin mRNA levels.

For the normal anatomic distribution studies of dysbindin mRNA, 5 to7 healthy individuals were used (2 sections per case, a subset of the controlsgiven in Table 1). Postmortembrains were obtained as previously described elsewhere.^28,38 Forcryostat sectioning of the normal brain, tissue blocks were dissected fromthe following regions: (1) the DLPFC, blocked at the middle third of the middlefrontal gyrus anterior to the corpus callosum; (2) the basal ganglia, blockedat the level of the nucleus accumbens; (3) the midbrain, blocked at the exitof the oculomotor nerve; (4) the rostral mesial temporal lobe, blocked atthe amygdala/rostral hippocampus; and (5) the more caudal mesial temporallobe, blocked at the genu to midbody level of the hippocampus. Eight sectionsfrom the postmortem DLPFC of each subject (14 schizophrenic patients and 15controls [detailed demographics are available in Table 1 and were previously published²⁹])were used in this study (2 sections per case per probe). Sections from themidbrain at the level of the red nucleus were also used to compare dysbindinexpression in schizophrenic patients (n = 7) with that in normal brain (n= 13).

Patients with schizophrenia and controls were matched for age, tissuepH (determined for each case as previously described elsewhere²⁸),postmortem interval (PMI) (defined as the time between death and brain freezing),sex, race, and brain hemisphere. Diagnosis was determined by independent reviewof clinical records by 2 board-certified psychiatrists who used the Diagnostic Evaluation After Death (DEAD)^39,40 asa guide to review the material available on each case as described.²⁸ Only cases that met DSM-IV criteriafor schizophrenia were included, with 3 patients of the chronic undifferentiatedsubtype, 9 of the chronic disorganized subtype, and 2 of the chronic paranoidsubtype. The mean ± SD age at disease onset was 23 ± 7 years.All patients diagnosed as having schizophrenia had a history of auditory hallucinationsand paranoid delusions. IQ data were available for approximately half of thepatients with schizophrenia (full-scale IQ score, mean ± SD, 76 ±14). The total dose of neuroleptic medication given to the patients was calculatedas described in a previous publication,²⁸ anddetailed clinical information on these schizophrenia cases can be found ina previous publication from our group.²⁹

T7/T3 promoter–tagged riboprobe templates were generated usingreverse transcription polymerase chain reaction. To avoid nonspecific amplification,a nested amplification scheme was used. We amplified a 290–base pair(bp) template from exons 6, 7, and 8 (positions 490-779, GenBank accessionNo. BC011912). Under the polymerase chain reaction conditions described ina previous publication,⁴¹ a single dysbindincomplementary DNA (cDNA) product of expected size (412 bp) was amplified usingthe primers 5′-GAGGCGAGTTTTGAGGAGGT-3′ (sense) and 5′-CAGAGTTCAGGAAGACGTCCA-3′(antisense) from human brain cDNA.⁴¹ We thenused the first cDNA fragment as a template and T7/T3 promoter–taggedprimers 5′-CAGAGATGCATAATACGACTCACTATAGGGAGA AGGAGGTAGAGAACAACCTGC-3′(sense, artificial T7 promoter) and 5′-CCAAGCCTTCATTAACCCTCACTAAAGGGAGATCGCCGCTCTGCAATCTGCA-3′ (antisense, artificial T3 promoter). A singleband of expected size (355 bp, including promoters) was amplified and sequenced(identical to GenBank accession No. BC011912). Similarly, a single spinophilin/neurabinII product of expected size (608 bp) was amplified using the outer primers5′-CCTGGAGAATGGCAG CAC-3′ (978-994 bp, forward) and 5′-CGGTCTTGACGAAGATACCC-3′ (1566-1585 bp, reverse). In the next round of amplification,we used the first cDNA fragment as a template and T7/T3 promoter–taggedprimers 5′-CAGAGATGCA TAATACGACTCACTATAGGGAGAGGTAGATGAATCCAAGA AGGA-3′(sense, artificial T7 promoter) and 5′-CCAAGC CTTCATTAACCCTCACTAAAGGGAGACAGGGAACAGCTCCAACCT-3′ (antisense, artificial T3 promoter). A single band of expectedsize (421 bp, including promoters) was amplified and sequenced (identicalto GenBank accession No. AJ401189, positions 1128-1483). We subcloned humansynaptophysin cDNA in the pCR2.1 (InVitrogen Corp, Carlsbad, Calif) vectorin the sense and antisense directions from a human synaptophysin cDNA clone(provided by Thomas C. Sudhof, MD, Howard Hughes Medical Institute, Universityof Texas, Southwestern Medical Center, Dallas).^42,43 Ourinsert spans 591 to 1084 bp of GenBank accession No. X06389 Y00507 and wassequenced for confirmation. The cyclophilin template used was purchased fromAmbion Inc, Austin, Tex.

Sense and antisense riboprobes for dysbindin, spinophilin, synaptophysin,and cyclophilin were generated from templates using a T7 or T3 polymeraseand an in vitro transcription kit as recommended by the manufacturer (PromegaCorp, Madison, Wis). The ³²P-UTP (uridine triphosphate) antisenseriboprobes (Northern blotting) and ³⁵S-UTP antisense and senseriboprobes (for in situ hybridizations) were labeled to a specific activityof 1 to 2 × 10⁹ cpm/µg by addition of radiolabeledUTP and were purified by ethanol precipitation.

Northern blotting was performed as previously described elsewhere.²⁸ Multiple tissue blots (Clontech, Palo Alto, Calif)containing polyA+ RNA from several brain regions of adults (amygdala, caudate,corpus callosum, hippocampus, whole brain, and thalamus) were used to verifythe specificity of our dysbindin and spinophilin riboprobes; for the synaptophysinriboprobe, we used Northern blot, normalized by amount of mRNA (catalog No.N3234410; BioChain Institute Inc, Hayward, Calif), containing total RNA fromadults (frontal lobe, temporal lobe, parietal lobe, occipital lobe, cerebellum,and lung). The blots were exposed to autoradiography film (BioMax; EastmanKodak, Rochester, NY) for 2 hours to overnight.

Fresh frozen tissue sections were processed as described in a previousarticle.⁴⁴ Sections were hybridized with 200µL of hybridization cocktail containing radiolabeled probe, 5 ng/mL,and were exposed to film for 14 days (dysbindin), 2 days (spinophilin andsynaptophysin), or 1 day (cyclophilin). For the anatomic survey of dysbindinmRNA levels, areas were sampled by outlining the region of interest. Quantitationof mRNA levels in the DLPFC was performed as previously described elsewhere.²⁸ We analyzed dysbindin mRNA expression in the parscompacta (most cases were bilaterally sampled) of the midbrain by outliningthe region of robust tyrosine hydroxylase immunohistochemical staining onadjacent tissue slices and by drawing a matching region of interest over thescanned films from the in situ hybridization for dysbindin mRNA.

Tissue from the lateral hemisphere of postmortem cerebelli were pulverizedand weighed while frozen. DNA was isolated by following the general protocolsupplied by PUREGENE (Gentra Systems, Minneapolis, Minn). We determined thegenotype at 11 SNPs using the TaqMan 5′ exonuclease allelic discriminationassay.⁴⁵ The information for each SNP is givenin Table 2, in the following columnorder: our laboratory number, the intermarker distances, the distance fromthe first SNP, the SNP identification number (either the rs [reference SNPcluster] number from the National Center for Biotechnology Information, Bethesda,Md, or the human Celera Variant number [hCV] from Celera Discovery Systems,Celera Genomics, Rockville, Md), the allelic nucleotides with the common alleleshown first, the rare allele frequency, and the SNP location relative to theintron/exon structure of the common 10-exon transcript BC011912, followedby the chromosomal position from the April 2003 freeze (the version of thegenomic sequence that this analysis was based on) at the University of Californiaat Santa Cruz (available at: http://genome.ucsc.edu/).

Statistical comparisons between unaffected controls and patients withschizophrenia were made using an analysis of variance or analysis of covariancefollowed by post hoc least significant difference or t tests,with diagnosis (between-group factor) and cortical layer (within-group factor)as independent variables and mRNA level as the dependent variable. Sampleswith mRNA levels more than ±2 SD from the mean were considered outliersand were omitted from the analysis. Correlations between dysbindin, spinophilin,and synaptophysin mRNA levels and pH, PMI, and age were tested using a Pearsonproduct moment correlation. Either Spearman correlations or Pearson productmoment correlations were performed between the estimated neuroleptic exposurelevel in patients with schizophrenia and mRNA measurements of dysbindin, spinophilin,and synaptophysin. Differences in dysbindin mRNA associated with genotype,regardless of diagnostic category, were analyzed using separate Mann-Whitney U tests, with individuals heterozygous and homozygous forthe rare allele grouped together. To avoid potential stratification problemsdue to unequal representation of ethnic groups, only brains from the predominantethnic group (African American) were analyzed for association between genotypeand expression. Therefore, 2 white individuals were dropped from the controlgroup and 3 were dropped from the schizophrenia group (Table 1).

By Northern blotting, we detected a band at 1.4-kilobase (kb) transcriptsize, which probably corresponds to GenBank accession No. BC011912 of dysbindin,in all gray matter areas surveyed: amygdala, caudate nucleus, hippocampus,and thalamus (Figure 1). We alsodetected a band at approximately 1.2 kb in subcortical gray matter areas derivedfrom the telencephalon (caudate and amygdala) and the diencephalon (thalamus),the identity of which is not clear. It is likely that there are several splicevariants of dysbindin mRNA of varying abundance that have not been characterizedin different brain regions and cell types. Based on the current availabledata, our probe is predicted to monitor all known common dysbindin mRNA variantsin the DLPFC, and it is possible that some rarer transcripts would not bedetected. The Northern blot for spinophilin revealed a major band at approximately4.6 kb in all brain regions examined, consistent with transcript size.⁴⁶ Spinophilin mRNA levels were similar across subcorticaland cortical areas studied. We detected one major synaptophysin transcriptat the expected 2.5-kb size^42,43 inall brain regions examined, with no apparent band in total RNA from the lung.

We detected dysbindin mRNA in multiple regions of the adult brain byusing in situ hybridization. No detectable image was observed from sectionshybridized with the dysbindin sense strand control riboprobe (data not shown).Overall, dysbindin mRNA was expressed much more prominently in most gray matterareas relative to white matter areas, consistent with dysbindin localizationpredominantly to neurons and not glia. However, the low white matter dysbindinmRNA signal that we detect could reflect either a low dysbindin mRNA expressionby glia or a low cell density of dysbindin-positive subcortical interstitialwhite matter neurons. In cortical areas, dysbindin mRNA signal was most robustin the DLPFC (Figure 2E), followedby the temporal neocortex (Figure 2B),the entorhinal cortex (Figure 2Aand B), and the orbital frontal cortex. The dysbindin mRNA signal was fairlystrong in the entorhinal cortex, where the hybridization varied accordingto cortical layer, with an intense signal in the superficial cell clustersin layer II, a prominent signal in pyramidal neuronal layer III, no signalin the lamina dissecans, and a strong signal in layers V and VI (Figure 2A). The caudal portion of the entorhinalcortex (Figure 2B) seemed to expressless dysbindin mRNA relative to more rostral levels (Figure 2A). The laminar expression of dysbindin mRNA is distinctin the temporal neocortex found lateral to the collateral sulcus and is expressedmore abundantly in the superficial cortical layers (layer II) as opposed todeeper cortical layers (Figure 2B).Of the noncortical regions sampled, dysbindin mRNA was most highly expressedin the substantia nigra of the human midbrain (Figure 2D). A moderate dysbindin hybridization signal was notedin the periaqueductal gray area, the superior colliculus, and the midbrainreticular formation. Dysbindin was clearly expressed in subcortical telencephalicregions and was found at about equal intensities in the amygdala (Figure 2A), hippocampus (Figure 2B), and caudate nucleus (Figure 2C). In the basal ganglia, the nucleus accumbens containedmore dysbindin mRNA than the caudate nucleus, whereas the putamen had intermediatelevels (Figure 2C). The opticaldensity corresponding to dysbindin mRNA expression varied significantly acrossanatomic regions (F = 5.37; P<.01) (Figure 3). The gray matter of the medial frontal gyrus (DLPFC) hadsignificantly more dysbindin mRNA than did the frontal white matter (P<.001) and the gray matter of the orbital frontal cortex(P<.01) (Figure3A). However, dysbindin mRNA levels were not significantly higherin the DLPFC compared with the other cortical areas sampled (P>.05 for all). In the mesial temporal lobe, the temporal neocortexhas higher dysbindin mRNA levels than the hippocampus (P<.01), amygdala (P<.05), and caudalentorhinal cortex (P = .01) but not the rostral entorhinalcortex (Figure 3B). Dysbindin mRNAwas moderately expressed in the nucleus accumbens, where levels were higherthan in the caudate (P = .01) (Figure 3C), and highly expressed in the substantia nigra, wherelevels were higher than in the red nucleus and cerebral peduncles (P<.05) (Figure 3D).

Dysbindin mRNA is found in the gray matter of the middle frontal gyrusin patients and controls, with increased signal in the middle to deep cortex(layers IV to VI) (Figure 4A andB). We found clearly distinguishable dysbindin mRNA-positive pyramidal neuronsin the DLPFC (data not shown). We detected a significant 15% to 20% reductionin dysbindin mRNA in the brains of patients with schizophrenia compared withcontrols (F_1,24 = 4.51; P = .04) (Figure 5). Furthermore, dysbindin mRNA levelswere found to vary according to cortical depth in both groups (main effectof layer: F_1,120 = 110.84; P<.001).Each cortical layer had a significantly different amount of dysbindin comparedwith the other layers (least significance difference: P<.05 for all) except for layers IV and V, which had similar levelsof dysbindin mRNA. We also found a significant interaction effect betweendiagnostic group and cortical layer (F_5,120 = 6.64; P = .03). Post hoc t test analysis revealedthat dysbindin mRNA levels were reduced in patients in cortical layers withprominent pyramidal neurons, that is, layers II (t₂₄ = −2.05; P = .05), III (t₂₄ = −2.06; P = .05), V(t₂₄ = −2.26; P = .03), and VI (t₂₄ = −2.14; P = .04), whereas they were not significantly changed inlayers I (t₂₄ = 0.37; P = .72) and IV (t₂₄ = −1.59; P = .13). Dysbindin mRNA levels did not correlate significantlywith age, PMI, or pH in the DLPFC in any cortical layer. Dysbindin mRNA levelsdid not correlate significantly with any measure of neuroleptic exposure inthe DLPFC (average r = −0.10; P>.16 for all).

We found fairly robust dysbindin mRNA expression in the midbrain ofpatients with schizophrenia, and the dysbindin expression pattern seemed tobe similar to that of controls (Figure 6Aand B). On quantifying, patients showed a 26% mean decrease in dysbindin mRNAlevels in the substantia nigra (Figure 6C).This mean difference, although similar in magnitude to that found in the DLPFC,showed only a weak trend toward statistical significance using analysis ofcovariance (F_1,16 = 2.73; P = .12, witha nondirectional hypothesis). In the midbrain, dysbindin mRNA levels correlatednegatively with PMI (r = −0.51; P<.05) and positively with brain pH (r =0.76; P<.001). Hence, these factors were usedas covariates in the analysis of covariance for this region. Dysbindin mRNAlevels did not significantly correlate with measure of neuroleptic exposurein the midbrain (last dose: r = −0.27, P = .56; daily dose, r = −0.22, P = .63; total lifetime dose: r =0.56, P = .20).

Spinophilin (Figure 4C andD) and synaptophysin (Figure 4Eand F) mRNA hybridization signals in the DLPFC were robust and had a laminarappearance, with increased intensity at the middle cortex level (layer effect,F>57; P<.001 for both). In contrast to the reductionin dysbindin mRNA density, we did not detect a significant difference in spinophilinor synaptophysin mRNA levels in the DLPFC of patients with schizophrenia comparedwith controls (F_1,27 = 0.66; P = .42 andF_1,25 = 0.09; P = .77, respectively) (Figure 5B and C), and there was no evidenceof an alteration in lamina-specific patterns between the groups (interactioneffect: spinophilin, F_5,135 = 1.14; P =.34 and synaptophysin, F_5,125 = 0.32; P =.90). We found no diagnostic difference in cyclophilin mRNA levels, used asa "housekeeping gene" control for overall mRNA levels (Figure 5D).

Fairly consistent negative correlations between age and spinophilinand synaptophysin mRNA levels were found in controls in layers III, IV, V,and VI (r = −0.45 to −0.69; P≤.05 for all, except for spinophilin mRNA level in layer III, where P = .07). No statistically significant correlations betweenPMI and spinophilin mRNA level or PMI and synaptophysin mRNA level were detected.Spinophilin and synaptophysin mRNA levels correlated with tissue pH in layersIII, IV, V, and VI (all-around r = −0.40; P<.05 for all). Analyses of covariance with age andpH (spinophilin and synaptophysin) as covariates did not alter the statisticalsignificance of the main effect of diagnosis or the interaction effect. SpinophilinmRNA levels in the DLPFC showed negative correlations with neuroleptic doseestimates (average r = −0.44; P>.05 for all). This suggests that neuroleptics could be down-regulatingspinophilin mRNA levels, consistent with data⁴⁷ showingthat long-term treatment with haloperidol significantly decreased spinophilinprotein levels in the primate PFC. In general, synaptophysin mRNA levels didnot correlate with any measure of neuroleptic exposure (overall average r = −0.26; P>.05 for all).

Synaptophysin mRNA levels positively correlated with dysbindin mRNAlevels in most cortical layers (II-VI), as did spinophilin mRNA levels; however,the latter correlations reached statistical significance in layer II only(Table 3). The positive correlationsbetween dysbindin and synaptophysin were not found when controls were analyzedseparately but were found to be strong and statistically significant whenpatients with schizophrenia were analyzed separately. We also noted that incortical layers II to VI, statistically significant positive correlationswere found between synaptophysin and spinophilin mRNA levels.

We grouped individuals heterozygous and homozygous for the rare alleles(ie, 1/2 and 2/2 genotypes, or 2-allele carriers) together to obtain a largeenough sample size to allow comparison with individuals homozygous for thecommon allele (ie, 1/1 genotype). Also, we omitted white individuals fromthe analysis to avoid comparing groups of mixed ethnicity. Dysbindin mRNAlevels in the PFC varied significantly according to genotype at 4 of the 11SNPs where analysis was possible (Table2). P3230 is in the 3′ UTR, and the G allelecarriers showed a 17% increase in expression; P2 555 is in intron 3,and the A allele carriers showed a 22% increase inexpression; P3521 is in the 5′ UTR, and the G allelecarriers showed a 34% decrease in expression; and P3587 is in the 5′flanking region of the gene, and the C allele carriersshowed a 39% increase in expression.

We detected abundant and widespread expression of dysbindin mRNA inthe adult brain. The presence of dysbindin in cortical and subcortical regionsof the normal brain has recently been reported at the protein level as well.⁴⁸ We found that the dysbindin mRNA level is reducedin the DLPFC and possibly in the substantia nigra of patients with schizophrenia.No change in spinophilin, synaptophysin, or cyclophilin mRNA levels in theDLPFC of patients with schizophrenia was detected. These findings suggestthat the reduction in dysbindin mRNA levels is not secondary to an overallreduction in mRNA abundance or quality. Neither is it likely to be simplythe result of a generalized loss of synaptic contacts. Our preliminary resultsof a reduction in dysbindin mRNA levels in the substantia nigra, althoughthis did not reach statistical significance, suggest that reduction in dysbindinsynthesis in patients may not be restricted to the DLPFC, a fact that is alsosupported by another recent finding of reduced dysbindin protein levels inthe hippocampus of patients with schizophrenia relative to controls.⁴⁹

Dysbindin mRNA was expressed in temporal and frontal cortical associationareas involved in declarative memory (the hippocampus and entorhinal cortex)and working memory (the DLPFC). Abnormalities in hippocampal-based memoryare commonly found in patients with schizophrenia, and abnormalities in DLPFCfunction are considered one of the core cognitive problems in patients withschizophrenia.²⁶ Furthermore, ample evidence^25,50-52 ofcellular and molecular pathology exists in the temporal and frontal corticesin brains of patients with schizophrenia. Dysbindin mRNA is prominently expressedin the substantia nigra and basal ganglia, areas of origination and terminationof dopamine neurons, suggesting that dysbindin dysfunction may affect regionswith prominent dopamine neurotransmission. Dysbindin gene expression was alsodetected in the amygdala, a brain area that has received less attention inpathology research in schizophrenia but that is implicated because of aberrantemotional responses in patients.^53-55 Overall,we conclude that some parallels between the anatomic expression pattern ofdysbindin and the vulnerability of these regions to functional pathology inschizophrenia can be drawn.

Reduced dysbindin expression may directly relate to the synaptic pathologyin the schizophrenic DLPFC; however, it is not clear if it lies upstream ordownstream of the putatively altered synaptic communication. Certainly, somemodels of cortical pathology in schizophrenia include altered synaptic communicationamong cortical neurons as a component of disease etiology.³⁵ Incontrast, we and other researchers^56-58 havefailed to find a reduction in synaptophysin or spinophilin mRNA or proteinlevels in the DLPFC of patients with schizophrenia, emphasizing that a generalizeddecrease in the number or density of synaptic contacts in the frontal cortexmay not be readily detectable in all cohorts. However, we found a statisticallysignificant positive correlation between dysbindin and synaptophysin mRNAlevels in the DLPFC of patients with schizophrenia, suggesting that therecould be a link between dysbindin and synaptophysin mRNA levels in the pathologicalstate. Our results suggest that synaptic pathology may be difficult to detectby gross measurements in overall group comparisons with small numbers of subjects,and, in agreement with other researchers in the field, we suggest that synapticpathology may be subtle and variable and may be due to subject-specific etiologicfactors in the DLPFC of patients with schizophrenia.³⁵

Although specific SNPs in dysbindin have been associated with schizophreniathrough genetic analysis, it is unlikely that any of the causative "mutations"have been identified. In 2 of the published studies with positive results,^12,13 different alleles are overtransmittedto affected individuals at some SNPs. These differences indicate that theSNPs tested are in linkage disequilibrium with the unknown causative variants.Our observation that the dysbindin mRNA level is reduced in patients withschizophrenia may be due, in part, to such causative variants in the dysbindingene. Dysbindin mRNA may also be down-regulated in the patients studied hereinby the action of other schizophrenia-susceptibility genes or by environmentalvariables related to schizophrenia. In this study, sequence variations inthe 3′ UTR, in intron 3, in the 5′ UTR, and in the 5′ flankingregion (a putative promoter region) of the dysbindin gene all significantlyimpacted dysbindin mRNA levels in the PFC. Most of our results concerningthe effects of genotype on expression should be viewed as preliminary, aswe had a small number of subjects in many of the comparisons. However, ourfinding with SNP P3230 (called 1580740, a→g, in Bray et al⁵⁹),where inheritance of the G allele predicts increaseddysbindin mRNA levels, is consistent with a previous study of the same SNPby Bray and colleagues⁵⁹ using allele-specificexpression analysis. In their study, using heterozygous individuals, dysbindinmRNA transcripts containing the G allele at thisSNP are increased compared with dysbindin mRNA transcripts with the A allele, but to a variable degree depending on the individual.For the other SNP (called 15643772, t→c, in Bray et al⁵⁹),which corresponds to our P3236, a smaller difference in allele-specific expressionwas found, but we did not find any effect of this SNP on dysbindin mRNA levelsin our overall group analysis.

The function of dysbindin in the mammalian brain is not well understood,so it is difficult to predict how, if at all, a modest reduction in dysbindinmRNA would be detrimental to the brain. Dysbindin binds to dystrobrevins,key components of the DPC, a protein complex that links the extracellularmatrix to the intracellular cytoskeleton.¹⁹ TheDPC can contain dystrophins, syntrophins, dystroglycans, and α- and β-dystrobrevins,as well as other proteins that are localized to pyramidal neurons in the rodentcerebral cortex and hippocampus.^21,60,61 Manyproteins of the DPC are enriched in postsynaptic densities of inhibitory synapsesand are not detected in postsynaptic sites of excitatory glutamate synapses.⁶² In cell culture, dystrophin has been shown to extensivelyco-localize with α-2–γ-aminobutyric acid (GABA)(A) receptorsopposite presynaptic terminals containing GABA.⁶² OtherDPC components, syntrophin, and β-dystroglycan also cluster at GABAergicsynapses, suggesting that the entire DPC may assemble at postsynaptic inhibitorysites. It is possible that the DPC complex plays a unique role at these inhibitorysynaptic sites and that this role may be altered in schizophrenia. Indeed,ample evidence for altered GABAergic neurotransmission in schizophrenia canbe found.⁵¹ However, further work localizingdysbindin mRNA and protein at the cellular and subcellular levels is neededto determine whether dysbindin is localized to certain subsets of neuronsor terminals in the primate cortex. Also, additional studies aimed at determiningthe function(s) of dysbindin in brain cells are needed. Indeed, a recent study²³ shows that dysbindin binds to proteins of a multimericcomplex distinct from the DPC—the BLOC-1 (biogenesis of lysosome-relatedorganelles complex 1)—and that dysbindin is mutated in a human diseaseassociated with impairment of organelle trafficking, stressing that our knowledgeof the cellular function of dysbindin in the brain is rudimentary.

Alterations in dysbindin should ultimately relate to the documentedabnormalities in excitatory neurotransmission of the DLPFC in schizophrenia,although direct evidence of this is currently lacking. Glutamate neuronalplasticity and activity in the schizophrenic PFC are altered, and molecularindications of these alterations include a reduction in N-acetyl aspartate levels,^63-66 areduction in BDNF mRNA and protein levels,²⁹ anda reduction in GAP-43 mRNA levels.²⁸ In addition,in the same set of brains studied herein, we found statistically significantreductions in mRNA encoding neurotrophin receptors, which localize to spines,which are specialized sites for excitatory signaling.⁶⁷ Infact, reductions in mRNA encoding spine-related postsynaptic proteins, suchas PSD95 and synapse-associated protein-97,^68,69 andreductions in spine density^70,71 havebeen found in the PFC of patients with schizophrenia by other research groups.These studies, taken together, with the possible lack of change in spinophilinlevels, suggest that many, but not all, aspects of the postsynaptic spinemay be dysfunctional in the brains of patients with schizophrenia. The possibilitythat abnormalities in glutamate-mediated excitatory neurotransmission maybe due to dysbindin-related pathology is the subject of ongoing research.

In summary, we report robust dysbindin mRNA expression in multiple brainareas implicated in the pathology of schizophrenia, including the frontaland temporal cortical regions and subcortical sites. We found reduced expressionof dysbindin in the DLPFC and possibly in the midbrain of patients with schizophrenia.We report preliminary observations on the relationship between dysbindin mRNAlevels and genetic variation in the dysbindin gene. Further molecular geneticanalysis of dysbindin gene variants may prove informative because the controlof dysbindin gene transcription or of dysbindin transcript stability seemsto be altered in the brains of patients with schizophrenia. Further studiesare required to determine whether a reduction in the dysbindin mRNA levelis specific to schizophrenia or extends to patients with other severe mentalillnesses. This study provides initial insights into a molecular mechanismof disease pathology in schizophrenia using the evidence of dysbindin as aschizophrenia-susceptibility gene as a starting point to determine where inthe brain the dysbindin gene is transcribed and to identify dysbindin reductionat mRNA levels as part of the molecular abnormalities associated with schizophrenia.

Corresponding author and reprints: Cynthia Shannon Weickert, PhD,Intramural Research Program, National Institute of Mental Health, NationalInstitutes of Health, 10 Center Dr, Bldg 10, Room 4N312, Mail Stop Code 1385,Bethesda, MD 20892-1385 (e-mail: shannowc@intra.nimh.nih.gov).

Submitted for publication September 2, 2003; final revision receivedDecember 12, 2003; accepted January 28, 2004.

This study was presented in part in abstract form at the American Collegeof Neuropsychopharmacology; December 2002; San Juan, Puerto Rico; and InternationalCongress of Schizophrenia Research annual meetings; April 2003; Colorado Springs,Colo.