International Journal Of Neuropsychopharmacology

International Journal of Neuropsychopharmacology (2007), 10, 503511. Copyright f 2007 CINPdoi: 10.1017/S146114570600722XOligodendrocyte pathophysiology : a newview of schizophreniaS P EC I A L S EC T I O NCINPDevorah Segal, Jessica R. Koschnick, Linda H. A. Slegers and Patrick R. HofDepartment of Neuroscience, Mount Sinai School of Medicine, New York, NY, USAAbstractA recent focus of schizophrenia research is disruption of white-matter integrity as a key facet of thiscomplex disorder. This was spurred, partly, by new imaging modalities, magnetic transfer imaging anddiusion tensor imaging, which showed dierences in white-matter integrity and tract coherence inpersons with schizophrenia compared to controls. Oligodendrocytes, in particular, have been the subjectof increased study after gene microarray analyses revealed that six myelin-related genes specic tooligodendrocytes have decreased expression levels in schizophrenia. Oligodendrocytes have also beenshown to be decreased in number in the superior frontal gyrus of subjects with schizophrenia. The MAGknockout mouse, missing a myelin-related gene linked to schizophrenia, may prove to be a useful animalmodel for the dysmyelination observed in the human disease. Studies currently ongoing on this modelhave found changes in dendritic branching patterns of pyramidal cells in layer III of the prefrontal cortex.Further characterization of the pathology in these mice is underway.Received 8 June 2006, Reviewed 5 July 2006, Revised 2 August 2006, Accepted 9 August 2006,First published online 12 February 2007Key words: Myelin, myelin-associated glycoprotein (MAG), oligodendrocytes, schizophrenia.IntroductionA new focus of research has emerged recently in theeld of schizophrenia research. From the dopaminecentred hypotheses that long dominated the eld, anew, more complex view has emerged positing thataltered brain connectivity plays a critical role in thedevelopment of schizophrenia. Several studies haveshown increased neuronal density without increasedabsolute numbers of neurons in a number of brainareas of patients with schizophrenia, implying thatcortical volume is reduced in schizophrenia, possiblybecause of reduced neuropil (Selemon and GoldmanRakic, 1999). In addition, there is a notable absenceof a clear degenerative process in schizophrenia, contrasting sharply with the pathology of neurons seenin other neurological disorders, such as Parkinsonsdisease or Alzheimers disease. It has therefore beenproposed that it is disorganization of specic whitematter tracts that may result in the functional decitsseen in the disease, including impaired workingmemory, cognitive decits, and inappropriate aect.Address for correspondence : Dr P. R. Hof, Department ofNeuroscience, Mount Sinai School of Medicine, Box 1065,One Gustave L. Levy Place, New York, NY 10029, USA.Tel. : 212-659-5904 Fax : 212-849-2510E-mail : Patrick.Hof@mssm.eduIn fact, this hypothesis is supported by several studiesusing new brain-imaging modalities. A major advantage of such approaches is that they can be used tostudy changes in schizophrenia in vivo, allowinginvestigation of dierent stages of the disease andproviding correlates to post-mortem and animalstudies. Magnetic transfer imaging (MTI) measuresprotons bound to macromolecules included in cellmembranes and myelin, and studies using this technique have demonstrated decreased myelin or axonalmembrane integrity in the temporal and frontal lobesof subjects with schizophrenia (Foong et al., 2000,2001, Kubicki et al., 2005). This decrease is particularlycorrelated with severity of negative symptoms (Foonget al., 2001). Diusion tensor imaging (DTI) can beused to measure the vectors of water movement in thebrain. Because water molecules within white mattermove most freely in the direction parallel to the axons,DTI provides a measurement of white-matter tractdirection and, by measuring the strength of the direction vector, tract coherence (Davis et al., 2003). RecentDTI studies have demonstrated decreased anisotropy,implying decreased tract coherence, in several majorwhite-matter tracts in persons with schizophrenia(Buchsbaum et al., 1998, Kubicki et al., 2003, 2005, Limet al., 1999, Sun et al., 2003, Wang et al., 2004). In thethalamus, phosphatidylcholine, the main membrane504D. Segal et al.lipid, and sphingomyelin and galactocerebroside,major myelin membrane components, were also foundto be decreased in schizophrenia (Schmitt et al.,2004), providing additional evidence for a myelindysfunction in the disease. Comparison of schizophrenia with other demyelinating diseases such asmetachromatic leukodystrophy (MLD) and multiplesclerosis (MS) provides further evidence for the existence of dysmyelination in schizophrenia. When MLDis diagnosed in late adolescence and early adulthood,the age when schizophrenia symptoms often appear,this demyelinating disease can present with psychoticsymptoms sometimes indistinguishable from thoseof schizophrenia (Davis et al., 2003). Likewise, patientswith MS who display cognitive and psychiatricsymptoms frequently have white-matter lesions in thefrontal and temporal lobes, which are the brain regionsmost implicated in schizophrenia (Davis et al., 2003).This convergence of symptoms among demyelinatingdiseases aecting the age and brain regions targetedin schizophrenia suggests that a common pathologypossibly exists.The role of oligodendrocytesAs evidence mounted that white-matter tracts arein some way disorganized in schizophrenia, newstudies began to shed light on what that precise defectmight be. A ground-breaking study used gene microarray analysis to examine gene expression levelsin post-mortem dorsolateral prefrontal cortex (PFC)of patients with schizophrenia and matched controls(Hakak et al., 2001). The studys unexpected discoverywas that the expression of six myelin-related genes(MAG, MAL, CNP, HERR3, gelsolin, and transferrin)was signicantly decreased in post-mortem schizophrenia brains. These genes are all predominantlyexpressed in oligodendrocytes. These results, laterconrmed independently (Tkachev et al., 2003, Peirceet al., 2006 in the case of CNP) and extended to otherbrain areas (Dracheva et al., 2006), implied that there isa pathology of oligodendrocytes underlying schizophrenia. This pathology is probably region-specic ,reduced myelin-related gene expression in schizophrenia has been shown in the PFC (Sugai et al., 2004),hippocampus, superior temporal cortex, and cingulategyrus (Katsel et al., 2005), but not in the putamen(Dracheva et al., 2006). Intriguingly, the gene SOX10,an oligodendrocyte-specic transcription factor, tendsto be highly methylated in the brains of personswith schizophrenia, and this correlates with reducedexpression of both SOX10 and other oligodendrocyterelated genes (Iwamoto et al., 2005). These ndingsmay represent an epigenetic indication ofoligodendrocyte dysfunction in schizophrenia, although genetic variations in the SOX10 gene do notappear to aect susceptibility towards the disease(Iwamoto et al., 2006). The myelin oligodendrocyteglycoprotein gene (MOG) is another gene potentiallyinvolved in schizophrenia. One group reported a weakpositive association between MOG gene markers andschizophrenia in the Chinese Han population (Liu etal., 2005), while another group failed to nd any signicant evidence for the MOG gene as a susceptibilityfactor for schizophrenia in the general population(Zai et al., 2005).Still under debate is the issue of whether these geneexpression changes may be due in part to treatmentwith antipsychotic drugs. Drugs that target dopaminereceptors may play a role in altered gene expressionpatterns, because oligodendrocytes express D2 andD3 receptor mRNA and protein at dierent stagesof maturation (Bongarzone et al., 1998, Rosin et al.,2005). Agonists for these receptors provide protectionof oligodendrocytes against glutamate toxicity andinjury after oxygen/glucose deprivation, while antagonists limit this protection (Rosin et al., 2005).Oligodendrocytes also express N-methyl-D-aspartate(NMDA) receptors, which can mediate excitotoxicinjury (Matute, 2006). Indeed, administration to rats ofphencyclidine, an NMDA receptor antagonist that caninduce psychotic symptoms in humans, resulted inaltered expression of many genes, including severalrelated to oligodendrocyte lineage (Kaiser et al., 2004).However, several studies of gene expression in humantissue have not indicated a drug-induced eect ongene expression. Tissue from patients with bipolardisorder showed similar oligodendrocyte-related geneexpression changes to tissue from patients withschizophrenia, even though bipolar disorder is generally not treated with antipsychotic drugs (Tkachevet al., 2003). Additionally, separate analyses of patientswith schizophrenia who were treated and not treatedwith neuroleptic drugs showed no dierences in geneexpression changes (Hakak et al., 2001, Iwamoto et al.,2005, Tkachev et al., 2003). In one study, long-tailedmacaque monkeys treated with haloperidol for 3months did show alterations in glia-related genes, butthe specic genes aected were not all identical tothose altered in humans with schizophrenia (Sugaiet al., 2004).Recent electron microscopy studies have furtherbolstered the hypothesis of oligodendrocyte pathologyin schizophrenia. Studying post-mortem tissue fromthe PFC and caudate nucleus, Uranova and colleagues(2001) have demonstrated apoptotic oligodendrocytesand damaged myelin sheath lamellae formingOligodendrocyte pathophysiologyDensity mapSchizophrenicControlCoefficient of variationFigure 1. Voronoi diagrams of the spatial distribution ofoligodendrocytes in the white matter under area 9 in a controlcase and a schizophrenia case. Each polygon is associatedwith a CNPase-immunoreactive cell. The CV and the localdensities are displayed. The schizophrenia patient displaysless clustering than the control case. Clustering is revealedby more intense red areas and homogeneity by blue areas.The bottom panel shows a dierence in the CV of thespatial distribution of oligodendrocytes in the white matterof area 9 revealing higher homogeneity in schizophrenia.(From Hof et al., 2003.)concentric lamellar bodies in schizophrenia brains,along with irregularities of heterochromatin andmitochondria in oligodendrocytes. With these resultsin mind, our group and others undertook to analyseand quantify the pathology of oligodendrocytes andwhite matter in schizophrenia.Homogeneous cellular distributions are the exception rather than the rule in nervous tissue, an observation not reected in simple counts of specic celltypes in a given brain area. It is specically thecharacteristics of oligodendrocyte spatial distribution,however, that may be relevant to the pathophysiologyof schizophrenia. This is particularly signicant in thecontext of exploring the organizational structure ofwhite matter in order to provide a cellular correlate tothe decreased anisotropy in schizophrenia observedby DTI. To the end of clarifying possible dierencesin oligodendrocyte spatial distribution, we utilized atechnique using Voronoi tessellation maps (Figure 1)to represent oligodendrocyte distribution in postmortem material from the superior frontal gyrus ofpersons with schizophrenia and controls (Hof et al.,2003). Voronoi polygons enclose the region of spacethat is closest to an oligodendrocyte by drawinglines at mid-distance between neighbouring cells andconnecting those lines to form a tessellation of polygons. It is thus the coecient of variation (CV) ofpolygon areas that represents the spatial distribution505of oligodendrocytes. A homogeneous cellular distribution will yield polygons of similar sizes with a smallCV, while a more heterogeneous distribution willresult in very dierently sized polygons and a largerCV of polygon area. The CV thus provides an objectiveestimate of the degree of clustering of oligodendrocytes. When this method was applied to thesuperior frontal gyrus, the CV of polygon area incontrols was 30 % higher than in matched subjectswith schizophrenia. This indicated a more clusteredarrangement of oligodendrocytes in controls, whichmay perhaps contribute to the greater white-mattertract coherence in controls observed using DTI. Inaddition, the density of oligodendrocytes, as calculated from these maps, was 28.3 % lower in subjectswith schizophrenia compared to controls, and therewas a clear correlation between decreased CV ofspatial distribution and decreased local cell density inschizophrenia brains. Absolute numbers of oligodendrocytes were also signicantly decreased in schizophrenia (Figure 2), in both layer III and the whitematter underlying area 9, where the decrease reached27 % (Hof et al., 2003), which is comparable to theresults obtained from the tessellation analyses. In aseparate study of the anterior thalamic nucleus, thenumber of oligodendrocytes was again found to bereduced in schizophrenia, particularly in males (Byneet al., 2006). These results clearly support the notionthat pathology of oligodendrocytes is present inschizophrenia, at least in the frontal gyrus. This regionof the PFC has been shown to undergo changes inschizophrenia (Bouras et al., 2001, Rajkowska et al.,1998, Selemon et al., 1995, 1998). The ndings ofaltered oligodendrocytes in this same region maytherefore indicate that pathology of white matter contributes to the functional and morphometric changesobserved in the disease. We are currently extendingthese studies to other white-matter areas, includingthe anterior cingulate gyrus and temporal areas.Mouse knockout modelsAn acknowledged disadvantage of studying humanpost-mortem material is the generally poor qualityof the tissue, making it on the whole unsuitable forultrastructural studies and ne analyses on the singlecell level, unless materials can be xed within a veryshort post-mortem delay, as was the case in theUranova et al. (2001) study. Animal models of humandiseases provide a way to circumvent the limitationsinherent in studying human tissue. The mouse MAGknockout may serve as a putative animal model forschizophrenia. This particular model was chosen inD. Segal et al.Oligodendrocyte numbers (×106)(a)(b)807060**50403020*100Layer IIIOligodendrocyte densities (×103)5061101009080706050403020100White matter***Layer IIIWhite matterFigure 2. (a) A 28.3 % decrease in oligodendrocyte total numbers was noted in layer III of area 9 in schizophrenics (p&lt,0.01)and a dierence of 27 % (p&lt,0.01) in the white matter. (b) Similar decreases in oligodendrocyte densities were also noted.&amp,, Controls, %, schizophrenics. (From Hof et al., 2003.) Asterisks indicate p&lt,0.01.16Number of brancheslight of the gene microarray studies mentioned abovethat found decreased expression of MAG amongthe myelin-related genes implicated in schizophrenia(Hakak et al., 2001). MAG has also been found to be asusceptibility gene for schizophrenia in the ChineseHan population (Wan et al., 2005). In addition, MAGis known to interact with neuron membranes and isinvolved in initiation of myelination in the centralnervous system (Montag et al., 1994). MAG has alsobeen shown to enhance oligodendrocyte survivalwith trophic signals (Weiss et al., 2000). Although thephenotype of the MAG knockout mouse is admittedlymild, with only slight developmental decits (see, forexample, Li et al., 1994, Weiss et al., 2000, 2001), it isunclear how a mouse model of schizophrenia wouldbe expected to behave. Behavioural studies performedon these mice have shown several subtle abnormalities. Mice missing the MAG gene are less procientthan wild-type mice in maintaining balance on arotating cylinder, and they displayed hyperactivityand impaired hindlimb reex extension (Pan et al.,2005). However, the mutant mice showed no dierences in spatial learning and memory or in swimmingspeed, as demonstrated by a Morris water mazetest (Montag et al., 1994). A mouse model missing theCNPase gene has been developed which showsmarked pathology : although myelin ultrastructureappears normal, the mice develop axonal swellingsand diuse neurodegeneration resulting in hydrocephalus and early death (Lappe-Siefke et al., 2003).This model may shed light on several neurodegenerative disorders, but it clearly does not representthe primary processes occurring in schizophrenia.*14*121086*42012345678Branch order numberFigure 3. Dierences in the dendritic trees of a MAGknockout mouse (&amp,) and control (2). Data from thecingulate cortex with three animals in each group showfewer branch numbers throughout the tree, predominatingin the third order. Asterisks indicate p&lt,0.05.Schizophrenia is a disease of disordered thoughtprocesses and, as such, is a fundamentally humandisease. Although the MAG knockout mouse modelcannot be expected to show the behavioural decitsobserved in the human disease, it can still provide avehicle for studying the morphological and anatomical abnormalities that may result from a geneticabnormality known to be linked with schizophrenia.All research performed on mice was approved bythe Institutional Animal Care and Use Committee ofMount Sinai School of Medicine.We are currently examining the repercussions ofdecient myelination resulting from absence of theMAG gene on the morphology of pyramidal neuronsin the frontal cortex of mice at dierent developmentalOligodendrocyte pathophysiology5072 µmFigure 4. Dendritic spines from a wild-type (left) and MAG knockout mouse (right), with the knockout mouse showing anapparently lower spine density. Neurons are from the frontal cortex and are loaded with Lucifer Yellow. Images wereacquired on a confocal laser scanning microscope at a magnication of 100r and represent stacks of 100r100r0.1 mmimages that were tiled together.stages. These cortical neurons may be the targets ofaxonal pathways that are aected in schizophrenia,such as the cingulum bundle. Layer III cortical pyramidal neurons may be especially relevant in light ofevidence that these neurons are reduced in numberin persons with schizophrenia (Selemon et al., 1998).Using a cell-loading approach, we can reconstructthe 3D morphology of these neurons and analyse thegeometric complexity of their dendritic arborization,providing an indication of the functional integrityof these neurons. Neuronal morphology and spinedensities have been shown to be altered in schizophrenia (Garey et al., 1998, Glantz and Lewis, 2000,Kalus et al., 2000, Pierri et al., 2001), so these studiesmay help gain a fuller understanding of the diseaseprocess.In a preliminary study, we visualized layer IIIpyramidal cells in the PFC of MAG knockout andwild-type mice aged 3 months by intracellular loadingwith Lucifer Yellow and 3D reconstruction. We constructed dendrograms as a measure of neuronal complexity and calculated dendritic spine densities asan indicator of neuronal functional integrity. In youngMAG knockout mice, PFC layer III pyramidal cellsshowed a 25 % decrease in total basal dendritic length(D. Segal, unpublished data). The sixth-order dendriticbranches were even more aected, with a 65 % decrease in length. Also, the number of branches weredecreased in the third (23 %) and sixth (79 %) orders.However, basal dendrites of MAG knockout micehad 15 % more second-order branches (Figure 3).In addition, MAG knockout mice had 34 % fewerfth-order apical dendritic branches. These data implythat a disturbance in the organization of myelin dueto impaired expression of MAG results in alterationsof morphology of PFC layer III pyramidal cells,particularly with respect to basal dendritic integrity.Such alterations may lead to abnormalities of specicwhite-matter tracts and aect the prefrontal circuitsseverely, thereby playing a key role in the development of schizophrenia. Although no signicantdierences were noted in spine densities betweenknockout and control mice in the young animals,spine pathology may be more prominent in old mice(Figure 4) as recently demonstrated in a non-humanprimate model of ageing (Duan et al., 2003). Furtheranalyses at additional time-points will be necessary toassess the age-dependence of such changes.Planned future studies also include using electronmicroscopy to examine ultrastructural changes inoligodendrocytes at the single-cell level in this model.Electron microscopy allows for the direct visualizationof cell and myelin integrity. If these studies show thesame sorts of changes observed in tissue from patientswith schizophrenia, it may help validate the use of theMAG knockout mouse as a model for schizophrenia aswell as shed new light on the processes that may beoccurring in the disease.Another mouse model for schizophrenia researchthat has recently been suggested is the QKI (quaking)mouse. In the mouse, the Qki protein regulates akinase inhibitor which is involved in terminal dierentiation of oligodendrocytes, and may thereforesimilarly regulate maturation of oligodendrocytesin humans (Aberg et al., 2006). Indeed, QKI is potentially a regulator of oligodendrocyte-related genes inhumans, and has decreased expression levels inschizophrenia (Katsel et al., 2005). The two availablemouse mutants, the qkv and the qke5 mutants, bothdisplay the reduced expression of myelin-relatedgenes and dysmyelination that have been noted inschizophrenia (McInnes and Lauriat, 2006). The QKI508D. Segal et al.mice display severe body tremors and decreasedlifespan (McInnes and Lauriat, 2006), traits that are notassociated with schizophrenia, so they are probablynot accurate models for the disease itself. These micemay, however, provide another system in which tostudy the various eects of dysmyelination similar tothat in the human disorder.Future directionsIncreasingly, parallel evidence from very dierentlines of research supports the premise that pathophysiology of oligodendrocytes may play a criticalrole in the development of schizophrenia. It remains tobe seen what position oligodendrocytes hold in thecascade of malfunctions that results in the constellation of behavioural decits seen in the disease.For example, does oligodendrocyte pathophysiologyresult in the symptoms of schizophrenia, or is therea separate, common basis that has not yet beendescribed ? Studies examining the MAG knockoutmouse model may shed light on this particular issue,and it remains to conduct behavioural studies of thesemice to look for subtle changes that may parallel thoseseen in the human disease. One plausible possibilitythat links several areas of schizophrenia research isthat AMPA receptor-mediated excitotoxicity damagesoligodendrocytes. Decreased functioning of NMDAmay lead to compensatory glutamate release thatcould trigger excitotoxic damage by acting on severalionotropic receptor subtypes (Akbarian et al., 1996,Dracheva et al., 2001, Gao et al., 2000, Olney andFarber, 1995, Theberge et al., 2002). Indeed, oligodendrocytes have excitatory glutamate receptors, andexcitatory axons can induce fast AMPA receptormediated currents in the cells (Bergles et al., 2000). Ithas been shown that oligodendrocytes in the forebrainare particularly susceptible to excitotoxic damage(Levine et al., 2001, Matute et al., 1997, McDonaldet al., 1998). In addition, gene microarray analyses oftissue from subjects with schizophrenia (Hakak et al.,2001, Mirnics et al., 2000) have found abnormalitiesof several genes related to receptors and synapticfunctions, further bolstering the notion that excitotoxicity and may play an important role in the pathophysiology of schizophrenia, perhaps by damagingoligodendrocytes.Additionally, the evidence available at present hasnot indicated whether dysmorphic oligodendrocytesare the cause or result of the dysmyelination evidenced in MTI and DTI ndings. Studying dierentstages of the disease, as well as animal models atdierent ages, may help elucidate the answer to thisparticular question. These sorts of studies may alsohelp explain another central issue in schizophreniaresearch : why do the decits of schizophrenia rstcome to light at a peculiar developmental stage?Quantitative and ultrastructural studies may helpexplain what is actually being measured by DTI, on acellular level, in the observed decreased anisotropy inpersons with schizophrenia. Further investigation ofthe oligodendrocyte-related disturbances in schizophrenia may also shed light on several other psychiatric conditions. Studies have shown a decrease inmRNA transcripts for oligodendrocyte…

Originally posted 2018-07-16 02:53:17. Republished by Blog Post Promoter

International Journal Of Neuropsychopharmacology

International Journal of Neuropsychopharmacology (2007), 10, 503511. Copyright f 2007 CINPdoi: 10.1017/S146114570600722XOligodendrocyte pathophysiology : a newview of schizophreniaS P EC I A L S EC T I O NCINPDevorah Segal, Jessica R. Koschnick, Linda H. A. Slegers and Patrick R. HofDepartment of Neuroscience, Mount Sinai School of Medicine, New York, NY, USAAbstractA recent focus of schizophrenia research is disruption of white-matter integrity as a key facet of thiscomplex disorder. This was spurred, partly, by new imaging modalities, magnetic transfer imaging anddiusion tensor imaging, which showed dierences in white-matter integrity and tract coherence inpersons with schizophrenia compared to controls. Oligodendrocytes, in particular, have been the subjectof increased study after gene microarray analyses revealed that six myelin-related genes specic tooligodendrocytes have decreased expression levels in schizophrenia. Oligodendrocytes have also beenshown to be decreased in number in the superior frontal gyrus of subjects with schizophrenia. The MAGknockout mouse, missing a myelin-related gene linked to schizophrenia, may prove to be a useful animalmodel for the dysmyelination observed in the human disease. Studies currently ongoing on this modelhave found changes in dendritic branching patterns of pyramidal cells in layer III of the prefrontal cortex.Further characterization of the pathology in these mice is underway.Received 8 June 2006, Reviewed 5 July 2006, Revised 2 August 2006, Accepted 9 August 2006,First published online 12 February 2007Key words: Myelin, myelin-associated glycoprotein (MAG), oligodendrocytes, schizophrenia.IntroductionA new focus of research has emerged recently in theeld of schizophrenia research. From the dopaminecentred hypotheses that long dominated the eld, anew, more complex view has emerged positing thataltered brain connectivity plays a critical role in thedevelopment of schizophrenia. Several studies haveshown increased neuronal density without increasedabsolute numbers of neurons in a number of brainareas of patients with schizophrenia, implying thatcortical volume is reduced in schizophrenia, possiblybecause of reduced neuropil (Selemon and GoldmanRakic, 1999). In addition, there is a notable absenceof a clear degenerative process in schizophrenia, contrasting sharply with the pathology of neurons seenin other neurological disorders, such as Parkinsonsdisease or Alzheimers disease. It has therefore beenproposed that it is disorganization of specic whitematter tracts that may result in the functional decitsseen in the disease, including impaired workingmemory, cognitive decits, and inappropriate aect.Address for correspondence : Dr P. R. Hof, Department ofNeuroscience, Mount Sinai School of Medicine, Box 1065,One Gustave L. Levy Place, New York, NY 10029, USA.Tel. : 212-659-5904 Fax : 212-849-2510E-mail : Patrick.Hof@mssm.eduIn fact, this hypothesis is supported by several studiesusing new brain-imaging modalities. A major advantage of such approaches is that they can be used tostudy changes in schizophrenia in vivo, allowinginvestigation of dierent stages of the disease andproviding correlates to post-mortem and animalstudies. Magnetic transfer imaging (MTI) measuresprotons bound to macromolecules included in cellmembranes and myelin, and studies using this technique have demonstrated decreased myelin or axonalmembrane integrity in the temporal and frontal lobesof subjects with schizophrenia (Foong et al., 2000,2001, Kubicki et al., 2005). This decrease is particularlycorrelated with severity of negative symptoms (Foonget al., 2001). Diusion tensor imaging (DTI) can beused to measure the vectors of water movement in thebrain. Because water molecules within white mattermove most freely in the direction parallel to the axons,DTI provides a measurement of white-matter tractdirection and, by measuring the strength of the direction vector, tract coherence (Davis et al., 2003). RecentDTI studies have demonstrated decreased anisotropy,implying decreased tract coherence, in several majorwhite-matter tracts in persons with schizophrenia(Buchsbaum et al., 1998, Kubicki et al., 2003, 2005, Limet al., 1999, Sun et al., 2003, Wang et al., 2004). In thethalamus, phosphatidylcholine, the main membrane504D. Segal et al.lipid, and sphingomyelin and galactocerebroside,major myelin membrane components, were also foundto be decreased in schizophrenia (Schmitt et al.,2004), providing additional evidence for a myelindysfunction in the disease. Comparison of schizophrenia with other demyelinating diseases such asmetachromatic leukodystrophy (MLD) and multiplesclerosis (MS) provides further evidence for the existence of dysmyelination in schizophrenia. When MLDis diagnosed in late adolescence and early adulthood,the age when schizophrenia symptoms often appear,this demyelinating disease can present with psychoticsymptoms sometimes indistinguishable from thoseof schizophrenia (Davis et al., 2003). Likewise, patientswith MS who display cognitive and psychiatricsymptoms frequently have white-matter lesions in thefrontal and temporal lobes, which are the brain regionsmost implicated in schizophrenia (Davis et al., 2003).This convergence of symptoms among demyelinatingdiseases aecting the age and brain regions targetedin schizophrenia suggests that a common pathologypossibly exists.The role of oligodendrocytesAs evidence mounted that white-matter tracts arein some way disorganized in schizophrenia, newstudies began to shed light on what that precise defectmight be. A ground-breaking study used gene microarray analysis to examine gene expression levelsin post-mortem dorsolateral prefrontal cortex (PFC)of patients with schizophrenia and matched controls(Hakak et al., 2001). The studys unexpected discoverywas that the expression of six myelin-related genes(MAG, MAL, CNP, HERR3, gelsolin, and transferrin)was signicantly decreased in post-mortem schizophrenia brains. These genes are all predominantlyexpressed in oligodendrocytes. These results, laterconrmed independently (Tkachev et al., 2003, Peirceet al., 2006 in the case of CNP) and extended to otherbrain areas (Dracheva et al., 2006), implied that there isa pathology of oligodendrocytes underlying schizophrenia. This pathology is probably region-specic ,reduced myelin-related gene expression in schizophrenia has been shown in the PFC (Sugai et al., 2004),hippocampus, superior temporal cortex, and cingulategyrus (Katsel et al., 2005), but not in the putamen(Dracheva et al., 2006). Intriguingly, the gene SOX10,an oligodendrocyte-specic transcription factor, tendsto be highly methylated in the brains of personswith schizophrenia, and this correlates with reducedexpression of both SOX10 and other oligodendrocyterelated genes (Iwamoto et al., 2005). These ndingsmay represent an epigenetic indication ofoligodendrocyte dysfunction in schizophrenia, although genetic variations in the SOX10 gene do notappear to aect susceptibility towards the disease(Iwamoto et al., 2006). The myelin oligodendrocyteglycoprotein gene (MOG) is another gene potentiallyinvolved in schizophrenia. One group reported a weakpositive association between MOG gene markers andschizophrenia in the Chinese Han population (Liu etal., 2005), while another group failed to nd any signicant evidence for the MOG gene as a susceptibilityfactor for schizophrenia in the general population(Zai et al., 2005).Still under debate is the issue of whether these geneexpression changes may be due in part to treatmentwith antipsychotic drugs. Drugs that target dopaminereceptors may play a role in altered gene expressionpatterns, because oligodendrocytes express D2 andD3 receptor mRNA and protein at dierent stagesof maturation (Bongarzone et al., 1998, Rosin et al.,2005). Agonists for these receptors provide protectionof oligodendrocytes against glutamate toxicity andinjury after oxygen/glucose deprivation, while antagonists limit this protection (Rosin et al., 2005).Oligodendrocytes also express N-methyl-D-aspartate(NMDA) receptors, which can mediate excitotoxicinjury (Matute, 2006). Indeed, administration to rats ofphencyclidine, an NMDA receptor antagonist that caninduce psychotic symptoms in humans, resulted inaltered expression of many genes, including severalrelated to oligodendrocyte lineage (Kaiser et al., 2004).However, several studies of gene expression in humantissue have not indicated a drug-induced eect ongene expression. Tissue from patients with bipolardisorder showed similar oligodendrocyte-related geneexpression changes to tissue from patients withschizophrenia, even though bipolar disorder is generally not treated with antipsychotic drugs (Tkachevet al., 2003). Additionally, separate analyses of patientswith schizophrenia who were treated and not treatedwith neuroleptic drugs showed no dierences in geneexpression changes (Hakak et al., 2001, Iwamoto et al.,2005, Tkachev et al., 2003). In one study, long-tailedmacaque monkeys treated with haloperidol for 3months did show alterations in glia-related genes, butthe specic genes aected were not all identical tothose altered in humans with schizophrenia (Sugaiet al., 2004).Recent electron microscopy studies have furtherbolstered the hypothesis of oligodendrocyte pathologyin schizophrenia. Studying post-mortem tissue fromthe PFC and caudate nucleus, Uranova and colleagues(2001) have demonstrated apoptotic oligodendrocytesand damaged myelin sheath lamellae formingOligodendrocyte pathophysiologyDensity mapSchizophrenicControlCoefficient of variationFigure 1. Voronoi diagrams of the spatial distribution ofoligodendrocytes in the white matter under area 9 in a controlcase and a schizophrenia case. Each polygon is associatedwith a CNPase-immunoreactive cell. The CV and the localdensities are displayed. The schizophrenia patient displaysless clustering than the control case. Clustering is revealedby more intense red areas and homogeneity by blue areas.The bottom panel shows a dierence in the CV of thespatial distribution of oligodendrocytes in the white matterof area 9 revealing higher homogeneity in schizophrenia.(From Hof et al., 2003.)concentric lamellar bodies in schizophrenia brains,along with irregularities of heterochromatin andmitochondria in oligodendrocytes. With these resultsin mind, our group and others undertook to analyseand quantify the pathology of oligodendrocytes andwhite matter in schizophrenia.Homogeneous cellular distributions are the exception rather than the rule in nervous tissue, an observation not reected in simple counts of specic celltypes in a given brain area. It is specically thecharacteristics of oligodendrocyte spatial distribution,however, that may be relevant to the pathophysiologyof schizophrenia. This is particularly signicant in thecontext of exploring the organizational structure ofwhite matter in order to provide a cellular correlate tothe decreased anisotropy in schizophrenia observedby DTI. To the end of clarifying possible dierencesin oligodendrocyte spatial distribution, we utilized atechnique using Voronoi tessellation maps (Figure 1)to represent oligodendrocyte distribution in postmortem material from the superior frontal gyrus ofpersons with schizophrenia and controls (Hof et al.,2003). Voronoi polygons enclose the region of spacethat is closest to an oligodendrocyte by drawinglines at mid-distance between neighbouring cells andconnecting those lines to form a tessellation of polygons. It is thus the coecient of variation (CV) ofpolygon areas that represents the spatial distribution505of oligodendrocytes. A homogeneous cellular distribution will yield polygons of similar sizes with a smallCV, while a more heterogeneous distribution willresult in very dierently sized polygons and a largerCV of polygon area. The CV thus provides an objectiveestimate of the degree of clustering of oligodendrocytes. When this method was applied to thesuperior frontal gyrus, the CV of polygon area incontrols was 30 % higher than in matched subjectswith schizophrenia. This indicated a more clusteredarrangement of oligodendrocytes in controls, whichmay perhaps contribute to the greater white-mattertract coherence in controls observed using DTI. Inaddition, the density of oligodendrocytes, as calculated from these maps, was 28.3 % lower in subjectswith schizophrenia compared to controls, and therewas a clear correlation between decreased CV ofspatial distribution and decreased local cell density inschizophrenia brains. Absolute numbers of oligodendrocytes were also signicantly decreased in schizophrenia (Figure 2), in both layer III and the whitematter underlying area 9, where the decrease reached27 % (Hof et al., 2003), which is comparable to theresults obtained from the tessellation analyses. In aseparate study of the anterior thalamic nucleus, thenumber of oligodendrocytes was again found to bereduced in schizophrenia, particularly in males (Byneet al., 2006). These results clearly support the notionthat pathology of oligodendrocytes is present inschizophrenia, at least in the frontal gyrus. This regionof the PFC has been shown to undergo changes inschizophrenia (Bouras et al., 2001, Rajkowska et al.,1998, Selemon et al., 1995, 1998). The ndings ofaltered oligodendrocytes in this same region maytherefore indicate that pathology of white matter contributes to the functional and morphometric changesobserved in the disease. We are currently extendingthese studies to other white-matter areas, includingthe anterior cingulate gyrus and temporal areas.Mouse knockout modelsAn acknowledged disadvantage of studying humanpost-mortem material is the generally poor qualityof the tissue, making it on the whole unsuitable forultrastructural studies and ne analyses on the singlecell level, unless materials can be xed within a veryshort post-mortem delay, as was the case in theUranova et al. (2001) study. Animal models of humandiseases provide a way to circumvent the limitationsinherent in studying human tissue. The mouse MAGknockout may serve as a putative animal model forschizophrenia. This particular model was chosen inD. Segal et al.Oligodendrocyte numbers (×106)(a)(b)807060**50403020*100Layer IIIOligodendrocyte densities (×103)5061101009080706050403020100White matter***Layer IIIWhite matterFigure 2. (a) A 28.3 % decrease in oligodendrocyte total numbers was noted in layer III of area 9 in schizophrenics (p&lt,0.01)and a dierence of 27 % (p&lt,0.01) in the white matter. (b) Similar decreases in oligodendrocyte densities were also noted.&amp,, Controls, %, schizophrenics. (From Hof et al., 2003.) Asterisks indicate p&lt,0.01.16Number of brancheslight of the gene microarray studies mentioned abovethat found decreased expression of MAG amongthe myelin-related genes implicated in schizophrenia(Hakak et al., 2001). MAG has also been found to be asusceptibility gene for schizophrenia in the ChineseHan population (Wan et al., 2005). In addition, MAGis known to interact with neuron membranes and isinvolved in initiation of myelination in the centralnervous system (Montag et al., 1994). MAG has alsobeen shown to enhance oligodendrocyte survivalwith trophic signals (Weiss et al., 2000). Although thephenotype of the MAG knockout mouse is admittedlymild, with only slight developmental decits (see, forexample, Li et al., 1994, Weiss et al., 2000, 2001), it isunclear how a mouse model of schizophrenia wouldbe expected to behave. Behavioural studies performedon these mice have shown several subtle abnormalities. Mice missing the MAG gene are less procientthan wild-type mice in maintaining balance on arotating cylinder, and they displayed hyperactivityand impaired hindlimb reex extension (Pan et al.,2005). However, the mutant mice showed no dierences in spatial learning and memory or in swimmingspeed, as demonstrated by a Morris water mazetest (Montag et al., 1994). A mouse model missing theCNPase gene has been developed which showsmarked pathology : although myelin ultrastructureappears normal, the mice develop axonal swellingsand diuse neurodegeneration resulting in hydrocephalus and early death (Lappe-Siefke et al., 2003).This model may shed light on several neurodegenerative disorders, but it clearly does not representthe primary processes occurring in schizophrenia.*14*121086*42012345678Branch order numberFigure 3. Dierences in the dendritic trees of a MAGknockout mouse (&amp,) and control (2). Data from thecingulate cortex with three animals in each group showfewer branch numbers throughout the tree, predominatingin the third order. Asterisks indicate p&lt,0.05.Schizophrenia is a disease of disordered thoughtprocesses and, as such, is a fundamentally humandisease. Although the MAG knockout mouse modelcannot be expected to show the behavioural decitsobserved in the human disease, it can still provide avehicle for studying the morphological and anatomical abnormalities that may result from a geneticabnormality known to be linked with schizophrenia.All research performed on mice was approved bythe Institutional Animal Care and Use Committee ofMount Sinai School of Medicine.We are currently examining the repercussions ofdecient myelination resulting from absence of theMAG gene on the morphology of pyramidal neuronsin the frontal cortex of mice at dierent developmentalOligodendrocyte pathophysiology5072 µmFigure 4. Dendritic spines from a wild-type (left) and MAG knockout mouse (right), with the knockout mouse showing anapparently lower spine density. Neurons are from the frontal cortex and are loaded with Lucifer Yellow. Images wereacquired on a confocal laser scanning microscope at a magnication of 100r and represent stacks of 100r100r0.1 mmimages that were tiled together.stages. These cortical neurons may be the targets ofaxonal pathways that are aected in schizophrenia,such as the cingulum bundle. Layer III cortical pyramidal neurons may be especially relevant in light ofevidence that these neurons are reduced in numberin persons with schizophrenia (Selemon et al., 1998).Using a cell-loading approach, we can reconstructthe 3D morphology of these neurons and analyse thegeometric complexity of their dendritic arborization,providing an indication of the functional integrityof these neurons. Neuronal morphology and spinedensities have been shown to be altered in schizophrenia (Garey et al., 1998, Glantz and Lewis, 2000,Kalus et al., 2000, Pierri et al., 2001), so these studiesmay help gain a fuller understanding of the diseaseprocess.In a preliminary study, we visualized layer IIIpyramidal cells in the PFC of MAG knockout andwild-type mice aged 3 months by intracellular loadingwith Lucifer Yellow and 3D reconstruction. We constructed dendrograms as a measure of neuronal complexity and calculated dendritic spine densities asan indicator of neuronal functional integrity. In youngMAG knockout mice, PFC layer III pyramidal cellsshowed a 25 % decrease in total basal dendritic length(D. Segal, unpublished data). The sixth-order dendriticbranches were even more aected, with a 65 % decrease in length. Also, the number of branches weredecreased in the third (23 %) and sixth (79 %) orders.However, basal dendrites of MAG knockout micehad 15 % more second-order branches (Figure 3).In addition, MAG knockout mice had 34 % fewerfth-order apical dendritic branches. These data implythat a disturbance in the organization of myelin dueto impaired expression of MAG results in alterationsof morphology of PFC layer III pyramidal cells,particularly with respect to basal dendritic integrity.Such alterations may lead to abnormalities of specicwhite-matter tracts and aect the prefrontal circuitsseverely, thereby playing a key role in the development of schizophrenia. Although no signicantdierences were noted in spine densities betweenknockout and control mice in the young animals,spine pathology may be more prominent in old mice(Figure 4) as recently demonstrated in a non-humanprimate model of ageing (Duan et al., 2003). Furtheranalyses at additional time-points will be necessary toassess the age-dependence of such changes.Planned future studies also include using electronmicroscopy to examine ultrastructural changes inoligodendrocytes at the single-cell level in this model.Electron microscopy allows for the direct visualizationof cell and myelin integrity. If these studies show thesame sorts of changes observed in tissue from patientswith schizophrenia, it may help validate the use of theMAG knockout mouse as a model for schizophrenia aswell as shed new light on the processes that may beoccurring in the disease.Another mouse model for schizophrenia researchthat has recently been suggested is the QKI (quaking)mouse. In the mouse, the Qki protein regulates akinase inhibitor which is involved in terminal dierentiation of oligodendrocytes, and may thereforesimilarly regulate maturation of oligodendrocytesin humans (Aberg et al., 2006). Indeed, QKI is potentially a regulator of oligodendrocyte-related genes inhumans, and has decreased expression levels inschizophrenia (Katsel et al., 2005). The two availablemouse mutants, the qkv and the qke5 mutants, bothdisplay the reduced expression of myelin-relatedgenes and dysmyelination that have been noted inschizophrenia (McInnes and Lauriat, 2006). The QKI508D. Segal et al.mice display severe body tremors and decreasedlifespan (McInnes and Lauriat, 2006), traits that are notassociated with schizophrenia, so they are probablynot accurate models for the disease itself. These micemay, however, provide another system in which tostudy the various eects of dysmyelination similar tothat in the human disorder.Future directionsIncreasingly, parallel evidence from very dierentlines of research supports the premise that pathophysiology of oligodendrocytes may play a criticalrole in the development of schizophrenia. It remains tobe seen what position oligodendrocytes hold in thecascade of malfunctions that results in the constellation of behavioural decits seen in the disease.For example, does oligodendrocyte pathophysiologyresult in the symptoms of schizophrenia, or is therea separate, common basis that has not yet beendescribed ? Studies examining the MAG knockoutmouse model may shed light on this particular issue,and it remains to conduct behavioural studies of thesemice to look for subtle changes that may parallel thoseseen in the human disease. One plausible possibilitythat links several areas of schizophrenia research isthat AMPA receptor-mediated excitotoxicity damagesoligodendrocytes. Decreased functioning of NMDAmay lead to compensatory glutamate release thatcould trigger excitotoxic damage by acting on severalionotropic receptor subtypes (Akbarian et al., 1996,Dracheva et al., 2001, Gao et al., 2000, Olney andFarber, 1995, Theberge et al., 2002). Indeed, oligodendrocytes have excitatory glutamate receptors, andexcitatory axons can induce fast AMPA receptormediated currents in the cells (Bergles et al., 2000). Ithas been shown that oligodendrocytes in the forebrainare particularly susceptible to excitotoxic damage(Levine et al., 2001, Matute et al., 1997, McDonaldet al., 1998). In addition, gene microarray analyses oftissue from subjects with schizophrenia (Hakak et al.,2001, Mirnics et al., 2000) have found abnormalitiesof several genes related to receptors and synapticfunctions, further bolstering the notion that excitotoxicity and may play an important role in the pathophysiology of schizophrenia, perhaps by damagingoligodendrocytes.Additionally, the evidence available at present hasnot indicated whether dysmorphic oligodendrocytesare the cause or result of the dysmyelination evidenced in MTI and DTI ndings. Studying dierentstages of the disease, as well as animal models atdierent ages, may help elucidate the answer to thisparticular question. These sorts of studies may alsohelp explain another central issue in schizophreniaresearch : why do the decits of schizophrenia rstcome to light at a peculiar developmental stage?Quantitative and ultrastructural studies may helpexplain what is actually being measured by DTI, on acellular level, in the observed decreased anisotropy inpersons with schizophrenia. Further investigation ofthe oligodendrocyte-related disturbances in schizophrenia may also shed light on several other psychiatric conditions. Studies have shown a decrease inmRNA transcripts for oligodendrocyte…

International Journal of Neuropsychopharmacology

International Journal of Neuropsychopharmacology (2007), 10, 503511. Copyright f 2007 CINP
doi:10.1017/S146114570600722X

Oligodendrocyte pathophysiology : a new
view of schizophrenia

S P EC I A L S EC T I O N

CINP

Devorah Segal, Jessica R. Koschnick, Linda H. A. Slegers and Patrick R. Hof
Department of Neuroscience, Mount Sinai School of Medicine, New York, NY, USA

Abstract
A recent focus of schizophrenia research is disruption of white-matter integrity as a key facet of this
complex disorder. This was spurred, partly, by new imaging modalities, magnetic transfer imaging and
diusion tensor imaging, which showed dierences in white-matter integrity and tract coherence in
persons with schizophrenia compared to controls. Oligodendrocytes, in particular, have been the subject
of increased study after gene microarray analyses revealed that six myelin-related genes specic to
oligodendrocytes have decreased expression levels in schizophrenia. Oligodendrocytes have also been
shown to be decreased in number in the superior frontal gyrus of subjects with schizophrenia. The MAG
knockout mouse, missing a myelin-related gene linked to schizophrenia, may prove to be a useful animal
model for the dysmyelination observed in the human disease. Studies currently ongoing on this model
have found changes in dendritic branching patterns of pyramidal cells in layer III of the prefrontal cortex.
Further characterization of the pathology in these mice is underway.
Received 8 June 2006; Reviewed 5 July 2006; Revised 2 August 2006; Accepted 9 August 2006;
First published online 12 February 2007
Key words: Myelin, myelin-associated glycoprotein (MAG), oligodendrocytes, schizophrenia.

Introduction
A new focus of research has emerged recently in the
eld of schizophrenia research. From the dopaminecentred hypotheses that long dominated the eld, a
new, more complex view has emerged positing that
altered brain connectivity plays a critical role in the
development of schizophrenia. Several studies have
shown increased neuronal density without increased
absolute numbers of neurons in a number of brain
areas of patients with schizophrenia, implying that
cortical volume is reduced in schizophrenia, possibly
because of reduced neuropil (Selemon and GoldmanRakic, 1999). In addition, there is a notable absence
of a clear degenerative process in schizophrenia, contrasting sharply with the pathology of neurons seen
in other neurological disorders, such as Parkinsons
disease or Alzheimers disease. It has therefore been
proposed that it is disorganization of specic whitematter tracts that may result in the functional decits
seen in the disease, including impaired working
memory, cognitive decits, and inappropriate aect.
Address for correspondence : Dr P. R. Hof, Department of
Neuroscience, Mount Sinai School of Medicine, Box 1065,
One Gustave L. Levy Place, New York, NY 10029, USA.
Tel. : 212-659-5904 Fax : 212-849-2510
E-mail : Patrick.Hof@mssm.edu

In fact, this hypothesis is supported by several studies
using new brain-imaging modalities. A major advantage of such approaches is that they can be used to
study changes in schizophrenia in vivo, allowing
investigation of dierent stages of the disease and
providing correlates to post-mortem and animal
studies. Magnetic transfer imaging (MTI) measures
protons bound to macromolecules included in cell
membranes and myelin, and studies using this technique have demonstrated decreased myelin or axonal
membrane integrity in the temporal and frontal lobes
of subjects with schizophrenia (Foong et al., 2000,
2001; Kubicki et al., 2005). This decrease is particularly
correlated with severity of negative symptoms (Foong
et al., 2001). Diusion tensor imaging (DTI) can be
used to measure the vectors of water movement in the
brain. Because water molecules within white matter
move most freely in the direction parallel to the axons,
DTI provides a measurement of white-matter tract
direction and, by measuring the strength of the direction vector, tract coherence (Davis et al., 2003). Recent
DTI studies have demonstrated decreased anisotropy,
implying decreased tract coherence, in several major
white-matter tracts in persons with schizophrenia
(Buchsbaum et al., 1998; Kubicki et al., 2003, 2005; Lim
et al., 1999; Sun et al., 2003; Wang et al., 2004). In the
thalamus, phosphatidylcholine, the main membrane

504

D. Segal et al.

lipid, and sphingomyelin and galactocerebroside,
major myelin membrane components, were also found
to be decreased in schizophrenia (Schmitt et al.,
2004), providing additional evidence for a myelin
dysfunction in the disease. Comparison of schizophrenia with other demyelinating diseases such as
metachromatic leukodystrophy (MLD) and multiple
sclerosis (MS) provides further evidence for the existence of dysmyelination in schizophrenia. When MLD
is diagnosed in late adolescence and early adulthood,
the age when schizophrenia symptoms often appear,
this demyelinating disease can present with psychotic
symptoms sometimes indistinguishable from those
of schizophrenia (Davis et al., 2003). Likewise, patients
with MS who display cognitive and psychiatric
symptoms frequently have white-matter lesions in the
frontal and temporal lobes, which are the brain regions
most implicated in schizophrenia (Davis et al., 2003).
This convergence of symptoms among demyelinating
diseases aecting the age and brain regions targeted
in schizophrenia suggests that a common pathology
possibly exists.
The role of oligodendrocytes
As evidence mounted that white-matter tracts are
in some way disorganized in schizophrenia, new
studies began to shed light on what that precise defect
might be. A ground-breaking study used gene microarray analysis to examine gene expression levels
in post-mortem dorsolateral prefrontal cortex (PFC)
of patients with schizophrenia and matched controls
(Hakak et al., 2001). The studys unexpected discovery
was that the expression of six myelin-related genes
(MAG, MAL, CNP, HERR3, gelsolin, and transferrin)
was signicantly decreased in post-mortem schizophrenia brains. These genes are all predominantly
expressed in oligodendrocytes. These results, later
conrmed independently (Tkachev et al., 2003; Peirce
et al., 2006 in the case of CNP) and extended to other
brain areas (Dracheva et al., 2006), implied that there is
a pathology of oligodendrocytes underlying schizophrenia. This pathology is probably region-specic ;
reduced myelin-related gene expression in schizophrenia has been shown in the PFC (Sugai et al., 2004),
hippocampus, superior temporal cortex, and cingulate
gyrus (Katsel et al., 2005), but not in the putamen
(Dracheva et al., 2006). Intriguingly, the gene SOX10,
an oligodendrocyte-specic transcription factor, tends
to be highly methylated in the brains of persons
with schizophrenia, and this correlates with reduced
expression of both SOX10 and other oligodendrocyterelated genes (Iwamoto et al., 2005). These ndings
may represent an epigenetic indication of

oligodendrocyte dysfunction in schizophrenia, although genetic variations in the SOX10 gene do not
appear to aect susceptibility towards the disease
(Iwamoto et al., 2006). The myelin oligodendrocyte
glycoprotein gene (MOG) is another gene potentially
involved in schizophrenia. One group reported a weak
positive association between MOG gene markers and
schizophrenia in the Chinese Han population (Liu et
al., 2005), while another group failed to nd any signicant evidence for the MOG gene as a susceptibility
factor for schizophrenia in the general population
(Zai et al., 2005).
Still under debate is the issue of whether these gene
expression changes may be due in part to treatment
with antipsychotic drugs. Drugs that target dopamine
receptors may play a role in altered gene expression
patterns, because oligodendrocytes express D2 and
D3 receptor mRNA and protein at dierent stages
of maturation (Bongarzone et al., 1998; Rosin et al.,
2005). Agonists for these receptors provide protection
of oligodendrocytes against glutamate toxicity and
injury after oxygen/glucose deprivation, while antagonists limit this protection (Rosin et al., 2005).
Oligodendrocytes also express N-methyl-D-aspartate
(NMDA) receptors, which can mediate excitotoxic
injury (Matute, 2006). Indeed, administration to rats of
phencyclidine, an NMDA receptor antagonist that can
induce psychotic symptoms in humans, resulted in
altered expression of many genes, including several
related to oligodendrocyte lineage (Kaiser et al., 2004).
However, several studies of gene expression in human
tissue have not indicated a drug-induced eect on
gene expression. Tissue from patients with bipolar
disorder showed similar oligodendrocyte-related gene
expression changes to tissue from patients with
schizophrenia, even though bipolar disorder is generally not treated with antipsychotic drugs (Tkachev
et al., 2003). Additionally, separate analyses of patients
with schizophrenia who were treated and not treated
with neuroleptic drugs showed no dierences in gene
expression changes (Hakak et al., 2001; Iwamoto et al.,
2005; Tkachev et al., 2003). In one study, long-tailed
macaque monkeys treated with haloperidol for 3
months did show alterations in glia-related genes, but
the specic genes aected were not all identical to
those altered in humans with schizophrenia (Sugai
et al., 2004).
Recent electron microscopy studies have further
bolstered the hypothesis of oligodendrocyte pathology
in schizophrenia. Studying post-mortem tissue from
the PFC and caudate nucleus, Uranova and colleagues
(2001) have demonstrated apoptotic oligodendrocytes
and damaged myelin sheath lamellae forming

Oligodendrocyte pathophysiology
Density map

Schizophrenic

Control

Coefficient of variation

Figure 1. Voronoi diagrams of the spatial distribution of
oligodendrocytes in the white matter under area 9 in a control
case and a schizophrenia case. Each polygon is associated
with a CNPase-immunoreactive cell. The CV and the local
densities are displayed. The schizophrenia patient displays
less clustering than the control case. Clustering is revealed
by more intense red areas and homogeneity by blue areas.
The bottom panel shows a dierence in the CV of the
spatial distribution of oligodendrocytes in the white matter
of area 9 revealing higher homogeneity in schizophrenia.
(From Hof et al., 2003.)

concentric lamellar bodies in schizophrenia brains,
along with irregularities of heterochromatin and
mitochondria in oligodendrocytes. With these results
in mind, our group and others undertook to analyse
and quantify the pathology of oligodendrocytes and
white matter in schizophrenia.
Homogeneous cellular distributions are the exception rather than the rule in nervous tissue, an observation not reected in simple counts of specic cell
types in a given brain area. It is specically the
characteristics of oligodendrocyte spatial distribution,
however, that may be relevant to the pathophysiology
of schizophrenia. This is particularly signicant in the
context of exploring the organizational structure of
white matter in order to provide a cellular correlate to
the decreased anisotropy in schizophrenia observed
by DTI. To the end of clarifying possible dierences
in oligodendrocyte spatial distribution, we utilized a
technique using Voronoi tessellation maps (Figure 1)
to represent oligodendrocyte distribution in postmortem material from the superior frontal gyrus of
persons with schizophrenia and controls (Hof et al.,
2003). Voronoi polygons enclose the region of space
that is closest to an oligodendrocyte by drawing
lines at mid-distance between neighbouring cells and
connecting those lines to form a tessellation of polygons. It is thus the coecient of variation (CV) of
polygon areas that represents the spatial distribution

505

of oligodendrocytes. A homogeneous cellular distribution will yield polygons of similar sizes with a small
CV, while a more heterogeneous distribution will
result in very dierently sized polygons and a larger
CV of polygon area. The CV thus provides an objective
estimate of the degree of clustering of oligodendrocytes. When this method was applied to the
superior frontal gyrus, the CV of polygon area in
controls was 30 % higher than in matched subjects
with schizophrenia. This indicated a more clustered
arrangement of oligodendrocytes in controls, which
may perhaps contribute to the greater white-matter
tract coherence in controls observed using DTI. In
addition, the density of oligodendrocytes, as calculated from these maps, was 28.3 % lower in subjects
with schizophrenia compared to controls, and there
was a clear correlation between decreased CV of
spatial distribution and decreased local cell density in
schizophrenia brains. Absolute numbers of oligodendrocytes were also signicantly decreased in schizophrenia (Figure 2), in both layer III and the white
matter underlying area 9, where the decrease reached
27 % (Hof et al., 2003), which is comparable to the
results obtained from the tessellation analyses. In a
separate study of the anterior thalamic nucleus, the
number of oligodendrocytes was again found to be
reduced in schizophrenia, particularly in males (Byne
et al., 2006). These results clearly support the notion
that pathology of oligodendrocytes is present in
schizophrenia, at least in the frontal gyrus. This region
of the PFC has been shown to undergo changes in
schizophrenia (Bouras et al., 2001; Rajkowska et al.,
1998; Selemon et al., 1995, 1998). The ndings of
altered oligodendrocytes in this same region may
therefore indicate that pathology of white matter contributes to the functional and morphometric changes
observed in the disease. We are currently extending
these studies to other white-matter areas, including
the anterior cingulate gyrus and temporal areas.
Mouse knockout models
An acknowledged disadvantage of studying human
post-mortem material is the generally poor quality
of the tissue, making it on the whole unsuitable for
ultrastructural studies and ne analyses on the single
cell level, unless materials can be xed within a very
short post-mortem delay, as was the case in the
Uranova et al. (2001) study. Animal models of human
diseases provide a way to circumvent the limitations
inherent in studying human tissue. The mouse MAG
knockout may serve as a putative animal model for
schizophrenia. This particular model was chosen in

D. Segal et al.

Oligodendrocyte numbers (×106)

(a)

(b)
80
70
60

**

50
40
30
20

*

10
0
Layer III

Oligodendrocyte densities (×103)

506

110
100
90
80
70
60
50
40
30
20
10
0

White matter

**

*

Layer III

White matter

Figure 2. (a) A 28.3 % decrease in oligodendrocyte total numbers was noted in layer III of area 9 in schizophrenics (p<0.01)
and a dierence of 27 % (p<0.01) in the white matter. (b) Similar decreases in oligodendrocyte densities were also noted.
&, Controls; %, schizophrenics. (From Hof et al., 2003.) Asterisks indicate p<0.01.

16
Number of branches

light of the gene microarray studies mentioned above
that found decreased expression of MAG among
the myelin-related genes implicated in schizophrenia
(Hakak et al., 2001). MAG has also been found to be a
susceptibility gene for schizophrenia in the Chinese
Han population (Wan et al., 2005). In addition, MAG
is known to interact with neuron membranes and is
involved in initiation of myelination in the central
nervous system (Montag et al., 1994). MAG has also
been shown to enhance oligodendrocyte survival
with trophic signals (Weiss et al., 2000). Although the
phenotype of the MAG knockout mouse is admittedly
mild, with only slight developmental decits (see, for
example, Li et al., 1994, Weiss et al., 2000, 2001), it is
unclear how a mouse model of schizophrenia would
be expected to behave. Behavioural studies performed
on these mice have shown several subtle abnormalities. Mice missing the MAG gene are less procient
than wild-type mice in maintaining balance on a
rotating cylinder, and they displayed hyperactivity
and impaired hindlimb reex extension (Pan et al.,
2005). However, the mutant mice showed no dierences in spatial learning and memory or in swimming
speed, as demonstrated by a Morris water maze
test (Montag et al., 1994). A mouse model missing the
CNPase gene has been developed which shows
marked pathology : although myelin ultrastructure
appears normal, the mice develop axonal swellings
and diuse neurodegeneration resulting in hydrocephalus and early death (Lappe-Siefke et al., 2003).
This model may shed light on several neurodegenerative disorders, but it clearly does not represent
the primary processes occurring in schizophrenia.

*

14
*

12
10
8
6

*

4
2
0
1

2

3

4

5

6

7

8

Branch order number
Figure 3. Dierences in the dendritic trees of a MAG
knockout mouse (&) and control (2). Data from the
cingulate cortex with three animals in each group show
fewer branch numbers throughout the tree, predominating
in the third order. Asterisks indicate p<0.05.

Schizophrenia is a disease of disordered thought
processes and, as such, is a fundamentally human
disease. Although the MAG knockout mouse model
cannot be expected to show the behavioural decits
observed in the human disease, it can still provide a
vehicle for studying the morphological and anatomical abnormalities that may result from a genetic
abnormality known to be linked with schizophrenia.
All research performed on mice was approved by
the Institutional Animal Care and Use Committee of
Mount Sinai School of Medicine.
We are currently examining the repercussions of
decient myelination resulting from absence of the
MAG gene on the morphology of pyramidal neurons
in the frontal cortex of mice at dierent developmental

Oligodendrocyte pathophysiology

507

2 µm

Figure 4. Dendritic spines from a wild-type (left) and MAG knockout mouse (right), with the knockout mouse showing an
apparently lower spine density. Neurons are from the frontal cortex and are loaded with Lucifer Yellow. Images were
acquired on a confocal laser scanning microscope at a magnication of 100r and represent stacks of 100r100r0.1 mm
images that were tiled together.

stages. These cortical neurons may be the targets of
axonal pathways that are aected in schizophrenia,
such as the cingulum bundle. Layer III cortical pyramidal neurons may be especially relevant in light of
evidence that these neurons are reduced in number
in persons with schizophrenia (Selemon et al., 1998).
Using a cell-loading approach, we can reconstruct
the 3D morphology of these neurons and analyse the
geometric complexity of their dendritic arborization,
providing an indication of the functional integrity
of these neurons. Neuronal morphology and spine
densities have been shown to be altered in schizophrenia (Garey et al., 1998; Glantz and Lewis, 2000;
Kalus et al., 2000; Pierri et al., 2001), so these studies
may help gain a fuller understanding of the disease
process.
In a preliminary study, we visualized layer III
pyramidal cells in the PFC of MAG knockout and
wild-type mice aged 3 months by intracellular loading
with Lucifer Yellow and 3D reconstruction. We constructed dendrograms as a measure of neuronal complexity and calculated dendritic spine densities as
an indicator of neuronal functional integrity. In young
MAG knockout mice, PFC layer III pyramidal cells
showed a 25 % decrease in total basal dendritic length
(D. Segal, unpublished data). The sixth-order dendritic
branches were even more aected, with a 65 % decrease in length. Also, the number of branches were
decreased in the third (23 %) and sixth (79 %) orders.
However, basal dendrites of MAG knockout mice
had 15 % more second-order branches (Figure 3).
In addition, MAG knockout mice had 34 % fewer
fth-order apical dendritic branches. These data imply
that a disturbance in the organization of myelin due
to impaired expression of MAG results in alterations
of morphology of PFC layer III pyramidal cells,

particularly with respect to basal dendritic integrity.
Such alterations may lead to abnormalities of specic
white-matter tracts and aect the prefrontal circuits
severely, thereby playing a key role in the development of schizophrenia. Although no signicant
dierences were noted in spine densities between
knockout and control mice in the young animals,
spine pathology may be more prominent in old mice
(Figure 4) as recently demonstrated in a non-human
primate model of ageing (Duan et al., 2003). Further
analyses at additional time-points will be necessary to
assess the age-dependence of such changes.
Planned future studies also include using electron
microscopy to examine ultrastructural changes in
oligodendrocytes at the single-cell level in this model.
Electron microscopy allows for the direct visualization
of cell and myelin integrity. If these studies show the
same sorts of changes observed in tissue from patients
with schizophrenia, it may help validate the use of the
MAG knockout mouse as a model for schizophrenia as
well as shed new light on the processes that may be
occurring in the disease.
Another mouse model for schizophrenia research
that has recently been suggested is the QKI (quaking)
mouse. In the mouse, the Qki protein regulates a
kinase inhibitor which is involved in terminal dierentiation of oligodendrocytes, and may therefore
similarly regulate maturation of oligodendrocytes
in humans (Aberg et al., 2006). Indeed, QKI is potentially a regulator of oligodendrocyte-related genes in
humans, and has decreased expression levels in
schizophrenia (Katsel et al., 2005). The two available
mouse mutants, the qkv and the qke5 mutants, both
display the reduced expression of myelin-related
genes and dysmyelination that have been noted in
schizophrenia (McInnes and Lauriat, 2006). The QKI

508

D. Segal et al.

mice display severe body tremors and decreased
lifespan (McInnes and Lauriat, 2006), traits that are not
associated with schizophrenia, so they are probably
not accurate models for the disease itself. These mice
may, however, provide another system in which to
study the various eects of dysmyelination similar to
that in the human disorder.
Future directions
Increasingly, parallel evidence from very dierent
lines of research supports the premise that pathophysiology of oligodendrocytes may play a critical
role in the development of schizophrenia. It remains to
be seen what position oligodendrocytes hold in the
cascade of malfunctions that results in the constellation of behavioural decits seen in the disease.
For example, does oligodendrocyte pathophysiology
result in the symptoms of schizophrenia, or is there
a separate, common basis that has not yet been
described ? Studies examining the MAG knockout
mouse model may shed light on this particular issue,
and it remains to conduct behavioural studies of these
mice to look for subtle changes that may parallel those
seen in the human disease. One plausible possibility
that links several areas of schizophrenia research is
that AMPA receptor-mediated excitotoxicity damages
oligodendrocytes. Decreased functioning of NMDA
may lead to compensatory glutamate release that
could trigger excitotoxic damage by acting on several
ionotropic receptor subtypes (Akbarian et al., 1996;
Dracheva et al., 2001; Gao et al., 2000; Olney and
Farber, 1995; Theberge et al., 2002). Indeed, oligodendrocytes have excitatory glutamate receptors, and
excitatory axons can induce fast AMPA receptormediated currents in the cells (Bergles et al., 2000). It
has been shown that oligodendrocytes in the forebrain
are particularly susceptible to excitotoxic damage
(Levine et al., 2001; Matute et al., 1997; McDonald
et al., 1998). In addition, gene microarray analyses of
tissue from subjects with schizophrenia (Hakak et al.,
2001; Mirnics et al., 2000) have found abnormalities
of several genes related to receptors and synaptic
functions, further bolstering the notion that excitotoxicity and may play an important role in the pathophysiology of schizophrenia, perhaps by damaging
oligodendrocytes.
Additionally, the evidence available at present has
not indicated whether dysmorphic oligodendrocytes
are the cause or result of the dysmyelination evidenced in MTI and DTI ndings. Studying dierent
stages of the disease, as well as animal models at
dierent ages, may help elucidate the answer to this

particular question. These sorts of studies may also
help explain another central issue in schizophrenia
research : why do the decits of schizophrenia rst
come to light at a peculiar developmental stage?
Quantitative and ultrastructural studies may help
explain what is actually being measured by DTI, on a
cellular level, in the observed decreased anisotropy in
persons with schizophrenia. Further investigation of
the oligodendrocyte-related disturbances in schizophrenia may also shed light on several other psychiatric conditions. Studies have shown a decrease in
mRNA transcripts for oligodendrocyte…