Proc. Nati. Acad. Sci. USAVol. 88, pp. 10850-10854, December 1991Neurobiology

Abnormal expression of two microtubule-associated proteins (MAP2and MAP5) in specific subfields of the hippocampal formationin schizophrenia

(cytoskeleton/subiculum/entorhinal cortex)

STEVEN E. ARNOLD*, VIRGINIA M.-Y. LEEt, RAQUEL E. GUR*, AND JOHN Q. TROJANOWSKIttDepartments of *Psychiatry and Neurology and tPathology and Laboratory Medicine, The University of Pennsylvania School of Medicine,Philadelphia, PA 19104

Communicated by Eliot Stellar, September 16, 1991

ABSTRACT A variety of cytoarchitectural disturbanceshave been described in limbic regions in postmortem studies ofschizophrenia, many of which suggest a developmental distur-bance of normal neuronal geometry. This geometry is estab-lished and maintained by elements of the neuronal cytoskele-ton. Immunohistochemistry with a panel of 15 monoclonalantibodies was used to monitor the presence of neuronalcytoskeletal proteins in the hippocampal formations of sixpatients with schizophrenia, six normal controls, and six withneurodegenerative disorders. In five of the six subjects withschizophrenia, prominent and specific alterations were foundin the distribution of two microtubule-associated proteins,MAP2 and MAP5, which were anatomically selective for thesubiculum and entorhinal cortex. In contrast, the immunore-activity of other cytoskeletal proteins (i.e., tau, tubulins, andselected neuroframent protein phosphoisoforms) was similarfor all subjects. Defects in the expression ofMAP2 and MAPS,two proteins that contribute to the establishment and mainte-nance of neuronal polarity, could underlie some of the cytoar-chitectural abnormalities described in schizophrenia and im-pair signal transduction in the affected dendrites. The subic-ulum and entorhinal cortex interconnect the hippocampalformation with widespread cortices and subcortical nuclei andplay important roles in higher cognitive functions. Hence,pathologic lesions that distort the polarized geometry of neu-rons could play a role in the emergence of aberrant behavior inschizophrenia.

Although a variety of cytoarchitectural abnormalities inlimbic structures of patients with schizophrenia have beendescribed, the specificity of these alterations for schizophre-nia and their etiology remain unknown. Most pathologicstudies of schizophrenia have used conventional chemicalstains, which lack molecular specificity because the stainsexhibit an affinity for a number of cellular components.Nevertheless, these methods have revealed a variety ofabnormalities in neuronal polarity (1-3) as well as in thequantity (2, 5-9), laminar positions (4, 8), or spatial arrange-ment (10, 11) of selected neurons of limbic areas. Many ofthefindings have been ascribed to abnormal cortical develop-ment, rather than to a neurodegenerative process. Accord-ingly, the failure to establish a highly ordered cortical cyto-architecture during development would most likely coincidewith disturbed connectivity between the hippocampus andrelated limbic and non-limbic regions. The reduced fidelity ofneuronal circuits may lead to behavioral disturbances.The molecular basis of these various and subtle perturba-

tions of limbic cortical cytoarchitecture is obscure. Wehypothesized that some of the cytoarchitectural abnormali-

ties described could result from a developmental defect in theestablishment of the highly asymmetric three-dimensionalgeometry of neurons, which distinguishes them from all othermammalian cells. The structural and molecular basis ofneuronal polarity is determined in part by a well-charac-terized family of neuronal cytoskeletal proteins, many ofwhich are expressed exclusively in functionally and anatom-ically distinct domains (axons, dendrites) of neurons (12-17).These proteins could be the targets of neuropathologicalevents that ultimately result in aberrant behavior. For thesereasons, we used monoclonal antibodies (mAbs) to investi-gate the distribution of a key group of developmentallyregulated neuronal cytoskeletal proteins in the hippocampalregion of postmortem brain samples from patients withschizophrenia. Among the large group of proteins examinedhere, prominent abnormalities were noted in the immunore-activity of only two microtubule-associated proteins (MAPs),MAP2 and MAPS (alternatively known as MAP1B), andthese abnormalities were largely confined to specific sub-fields of the hippocampal formation in patients with schizo-phrenia.

MATERIALS AND METHODSThe cases studied are listed with clinical and neuropatholog-ical findings in Table 1. Brains were obtained at autopsy fromsix chronically institutionalized psychiatric patients (fivewith schizophrenia, one with schizoaffective disorder, ages68-86), six neurologically normal controls (ages 16-91), andsix controls with a neurodegenerative disease (ages 58-81).Psychiatric diagnoses were made by applying DSM-III-Rcriteria (31) on chart review. All cases had complete autop-sies and neuropathological examinations. The brains of fiveof the six patients with schizophrenia had no diagnosticneuropathologic lesions, whereas the sixth brain had a suf-ficient number of neocortical senile plaques to meet histo-pathologic criteria for AD. However, the hippocampal for-mation from this patient exhibited no AD pathology.

Blocks from at least four rostral-caudal levels of thehippocampus and parahippocampal gyrus from each brainwere sectioned at a thickness of 6 ,um and processed forimmunohistochemistry using the peroxidase/anti-peroxidasemethod as described (18, 19). Approximately 30 differentmAbs, specific for each of the major brain MAPs (MAP2,MAPS, tau) as well as a- and 83-tubulin and diverse neuro-filament (NF) protein phosphoisoforms, were used in pre-liminary studies. Fifteen mAbs were then selected for adetailed analysis ofthe polarized arrangement of cytoskeletalproteins in hippocampal neurons in the schizophrenic and

Abbreviations: EC, entorhinal cortex; mAb, monoclonal antibody;MAP, microtubule-associated protein; NF, neurofilament.tTo whom reprint requests should be addressed.

10850

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Dow

nloa

ded

by g

uest

on

May

5, 2

020

Proc. Natl. Acad. Sci. USA 88 (1991) 10851

Table 1. Subject data

Neuropsychiatric Neuropathologic Psychotropic MAP2 MAP5Case Age/sex diagnosis diagnosis treatment EC SUB (SUB)

1 (L) 86/M Schizophrenia Normal Neuroleptics - - -2 (R) 68/M Schizophrenia Normal Neuroleptics; +

lithium3 (L) 75/F Schizophrenia; Normal Neuroleptics -*

tardive dyskinesia4 (L) 76/F Schizoaffective; Normal Neuroleptics;

tardive dyskinesia lithium; ECT5 (L) 78/M Schizophrenia "Plaque-only" AD; few Neuroleptics -*

cortical Lewy bodies6 (R, L) 82/M Schizophrenia; Normal Neuroleptics; + + +

tardive dyskinesia reserpine7 (R) 62/F Normal Normal None + + +8 (R) 64/M Normal Normal None + + +9 (R) 16/M Normal Normal None + + +10 (R) 76/F Normal Normal None + + +11 (L) 91/F None Normal None + + +12 (L) 85/F None Normal None + + +13 (R) 75/M Dementia; PD; AD; PD ECT + + ND

depression14 (L) 68/M Dementia; AD None + + +

alcoholism15 (L) 80/M Dementia AD None _ _ _16 (L) 71/M Dementia; PD AD; PD None ±t + +17 (L) 58/M Dementia DLBD Neuroleptics + + +18 (L) 81/F Dementia AD None + + +

The left (L) or right (R) hippocampus examined in each case is indicated in parentheses. The presence (+) or absence(-) of MAP2 and/or MAP5 staining in neurons of subiculum (SUB) or entorhinal cortex (EC) is shown in the last threecolumns (ND, not done). AD, Alzheimer disease; PD, Parkinson disease; DLBD, diffuse Lewy body disease; ECT,electroconvulsive therapy.*Diminished immunoreactivity only in superficial layers.tAD case with severe neuron loss.

control cases (Table 2; refs. 20-25). Sections adjacent tothose probed by immunohistochemistry were stained withcresyl violet for standard cytoarchitectural assessment. Tis-sue sections were blindly rated as normal or abnormal by twoof the authors and fully characterized.

RESULTS

Deficits in the normal pattern of immunolabeling of MAP2and MAP5 in the hippocampal region were found in five ofthesix psychiatric patients. No deficits were found for tau,tubulins, or NF proteins.

Table 2. Specificity of mAbs for cytoskeletal proteinsCytoskeletalprotein(s) mAb(s)*

MAP2A/B M12, AP14MAP2A/B/C T24, T34, T39, T42, AP18MAP5 Anti-MAP5, 3G5tau T14NFM- RMDO20NFH+ TA51NFH+++ RM024a-Tubulin Anti-a-tubulin,/-Tubulin Anti-13-tubulin

The designations A, B, and C refer to the three major MAP2isoforms. NFM and NFH refer to the middle and high molecularweight NF subunits, while -, +, and +++ refer to the poorly,mildly, and heavily phosphorylated isoforms of these subunits.*AP14, AP18, 3G5, and MAP5 were kindly provided by L. Binderand A. Matus. mAbs to a and 8 tubulins were purchased fromAmersham.

Normal Pattern of MAP2 and MAP5 Immunoreactivity inthe Hippocampal Region. Neurons in the subiculum of thenormal controls exhibited the most robust MAP2 and MAP5immunoreactivity of any hippocampal subfield (Fig. 1 a andb; Fig. 2 b and c). This was especially intense in the dendritesof vertically oriented pyramidal cells and the fusiform neu-rons and their processes adjacent to the angular bundle. Inthe EC, moderate numbers of MAP2-immunoreactive neu-rons were present in layers II, V, and VI, while occasionalMAP2-positive neurons were seen in layers III and IV. Theimmunoreactive neurons in layer II often appeared in clus-ters. In contrast with MAP2, MAP5 immunoreactivity inneurons in the EC was weaker and MAP5-positive neuronswere rare. Notably, except for two controls with very severeAD pathology, the distribution and intensity of MAP2 andMAP5 immunoreactivity in the hippocampal region of theneurodegenerative controls was indistinguishable from thatof the normal controls. Specifically, one of the severe ADcases (case 15) showed diminished MAP2 labeling of the ECand subiculum, while the other case (case 16) showed de-creased MAP2 labeling only in the EC. Since these alter-ations were accompanied by extensive neuron loss, gliosis,and neurofibrillary pathology, the abnormal expression ofMAP2 and MAP5 was distinct from the abnormalities de-scribed below in the patients with schizophrenia.MAP2 and MAP5 Immunoreactivity Pattern in Schizophre-

nia. In contrast to the controls, there was a marked paucityof somatodendritic MAP2 immunoreactivity that selectivelyaffected neurons in the subiculum in five ofthe six psychiatriccases and the EC in four of the six. This diminution in MAP2staining was so dramatic that the hippocampal sections ofthese four schizophrenia cases could be distinguished fromthe control sections by macroscopic inspection (compare Fig.

Neurobiology: Amold et al.

Dow

nloa

ded

by g

uest

on

May

5, 2

020

Proc. Natl. Acad. Sci. USA 88 (1991)

S U B

DG

SUBFC

SUB

EC

b

DG

*;..-,

SUB

dFIG. 1. Low-power photographs demonstrate the macroscopically evident differences in MAP2 (labeled with M12) and MAP5 (labeled with

anti-MAP5) immunoreactivity between hippocampus of a normal elderly control (case 7) (a and b) and hippocampus of a schizophrenic (case3) (c and d). The control parahippocampal gyrus (a) shows especially robust MAP2 labeling in deep and superficial layers of the EC and in thesubiculum (SUB), while the schizophrenic parahippocampal gyrus (c) shows a paucity of neuronal immunoreactivity that is most evident in theEC and subiculum. Note the selectively preserved neuronal and neuropil labeling in the deeper layers of the EC in this case. Strong MAP5immunoreactivity is seen in the control subiculum (c), while the subiculum of the patient with schizophrenia is almost negative (d). DG, dentategyrus. Sections were counterstained with hematoxylin. (x6.5.)

1 a and c). In two cases (cases 1 and 4) only rare immuno-reactive neurons were present in all layers of EC, while in theother two (cases 3 and 5), the diminished MAP2 immunore-activity was confined to the superficial layers of the EC. Inanother case (case 2), MAP2 immunoreactivity was normal inthe EC but diminished in the subiculum. The reduction insubicular MAP2 staining was paralleled by a distinct andmarked diminution of MAP5 immunoreactivity in subicularneurons in the same cases (Fig. 1 b and d).

In sharp contrast to the two AD cases which had extensiveneuron loss along with reduced MAP2 and MAPS immuno-reactivity, the hippocampal formations of the patients withschizophrenia contained abundant neurons and did not showevidence of gliosis or neurofibrillary changes in preparationsexamined using conventional stains and mAbs (i.e., mAbs toglial fibrillary acidic protein, tau, and ubiquitin; data notshown). However, Nissl staining revealed subtle cytoarchi-tectural alterations. These included poorly formed clusters oflayer II neurons in the EC and dysmorphic subicular neurons(compare Fig. 2 a and e). However, these findings were muchless dramatic than the alterations in the expression of MAP2(Fig. 2 b andf) and MAP5 (Fig. 2 c and g) in these regions.Other Cytoskeletal Proteins. Immunohistochemistry with

antibodies directed at other protein elements of the neuronalcytoskeleton revealed no differences between schizophrenicand control cases. For example, RMDO20, a mAb specificfor the poorly phosphorylated mid-sized NF protein (which,like MAP2 and MAP5, is largely confined to the somatoden-dritic domain of neurons), labeled neurons in the hippocam-pal formation of control cases and all but one schizophreniccase in the same manner (compare Fig. 2 dand h). In this case(case 3), RMDO20 immunoreactivity appeared abnormallysparse in the superficial layers of the EC. Immunoreactivityfor tau, a- and B-tubulin, and other NF protein phosphoiso-forms in the hippocampal formation of the schizophreniacases was indistinguishable from that in the controls (data notshown).

DISCUSSIONAlthough the sample size was relatively small, there wereprofound abnormalities in five of the six psychiatric patientswith a lack of similar alterations in the neurologically normalsubjects or in the neurodegenerative disease controls (exceptfor two severe AD cases which had extensive neuron loss).Thus, the results may warrant the speculation that distur-bances in the expression of MAP2 and MAP5 representanatomically selective and highly specific molecular corre-lates of psychiatric dysfunction in at least a subset of patientswith schizophrenia. Another consideration in interpretingour findings is that the diminished MAP2 and MAPS immu-noreactivity in the schizophrenia sample could have been aresult of long-term psychotropic medications and other so-matic treatments (e.g., electroconvulsive therapy). How-ever, the absence of similar MAP2 and MAP5 abnormalitiesin a control patient who was chronically treated with thio-ridazine (for behavioral problems associated with diffuseLewy body disease) or in another control who receivedelectroconvulsive therapy (for depression) argues againstthis.Taken together, these data, as summarized in Table 1,

suggest that the marked reduction in MAP2 and MAP5immunoreactivity in patients with schizophrenia does notreflect a generalized abnormality of the neuronal cytoskele-ton. Rather, we have identified highly selective abnormalitiesof two cytoskeletal proteins that specifically affect particularpopulations of neurons in the hippocampal formation (i.e.,those in subiculum and EC).While it is unclear whether the abnormalities described

here are mechanistically linked to the emergence of the fullschizophrenic phenotype, it is plausible that they account forsome of the cytoarchitectural alterations that have beenobserved in the hippocampal formation in schizophrenia.Both MAP2 and MAP5 promote the polymerization oftubulinin vitro (12, 13) and stabilize microtubules (14-16), which arethought to play an important role in the establishment and

10852 Neurobiology: Amold et al.

.4%.,

,I':t,&F, l.. '.

IVI::s

Dow

nloa

ded

by g

uest

on

May

5, 2

020

Proc. Natl. Acad. Sci. USA 88 (1991) 10853

* In 'h~~~~*.1 *w,

t;,Iti^"#-;;¢?¢%^& Zigs~~~~~'

Z.,.¢h4

4 *

.. t.,&

*;e= '4twiZSs * ' *

* * .44 {e X;

Jo W,.-*' ! '; Ab' -' ;

j, ,O, ,,4' q,.#'. r4;

e :* e ; A.4' .4*S.... ..

fI

X.I

C

9f

(.1I A

lii

d

hran. 2. Photomicrographs of subiculum in case 7 (a-d) and case 3 (e-h). Nissl staining in the normal subiculum (case 7) showed uniform

pyramids with slender apical dendrites (a), while in the schizophrenic subiculum (case 3), neurons appeared smaller, more variably shaped, andchaotic in orientation (e). Immunolabeling for MAP2 (with M12) revealed robust somatodendritic labeling in the control (b) and a near absenceofMAP2 immunoreactivity in the schizophrenic (I). No differences could be detected with mAbs that are specific for MAP2A and MAP2B versusthose that are specific for MAP2A, -2B, and -2C. Immunolabeling for MAP5 (with anti-MAP5) also demonstrated reduced MAP5 staining incase 3 (g) compared with case 7 (c). Immunolabeling with the anti-NF mAb RMDO20 highlighted the presence of dendrites with preserved NFprotein expression in the schizophrenic (h) as well as the control (d). Sections in b-d and f-h were counterstained with hematoxylin. (x 100.)

maintenance of neuronal polarity (17). In addition, MAPs arebelieved to mediate interactions of microtubules with eachother and with other organelles (16). Hence, it is attractive tospeculate that reduced expression or posttranslationallymodified forms of MAP2 and MAP5 in schizophrenia couldaffect the equilibrium between polymerized microtubules andunpolymerized tubulins. If this affects the stability of den-drites or the ability of dendrites to maintain and remodelsynaptic contacts, signal transduction across these special-ized neuronal processes could be impaired.Given the unique connectional neuroanatomy of the hip-

pocampal formation and its role in higher cognitive functions(26,27), faulty synaptic transmission across unstable, MAP2-and MAP5-deficient dendrites that extend from neurons inthe EC and subiculum could have profound behavioral im-plications. The EC and subiculum have widespread connec-tions with higher order multimodal and sensory-specificassociation cortices in all four lobes as well as with subcor-tical limbic structures, including the amygdala, thalamus,hypothalamus, basal forebrain (26, 27), and dopaminergicventral tegmentum (28,29). Ifimpaired synaptic transmissionis one of the consequences of diminished MAP2 and MAPSexpression in dendrites of neurons in the subiculum and EC,this might affect the fidelity of neuronal circuits or even serveto functionally disconnect the hippocampus from much of thebrain. It may also alter limbic dopaminergic activity. Fur-thermore, if such disturbances occur early in brain develop-ment, the normal neural circuitry of the hippocampal regioncould be reorganized, as has been described in other neural

systems (30). This process could fail to reconstitute normallyfunctioning networks of interconnected neurons in schizo-phrenic patients.

Several mechanisms might account for the diminishedimmunoreactivity ofMAP2 and MAP5 in selected regions ofthe brains of schizophrenics. There could be a developmentalfailure to induce the transcription of MAP2 and MAP5 genesin subiculum and the EC, a genetic mutation, or selectiveproteolytic cleavage of normally translated protein. Alterna-tively, there may be extreme posttranslational modification(e.g., by phosphorylation) that would make MAP2 and MAP5unrecognizable to our antibodies. There are many steps in thetranscription, translation, and metabolism of MAP2 andMAP5, and disruption of any of these steps could lead todecreased immunoreactivity. If reduced levels of hippocam-pal MAP2 and MAP5 are phenotypic markers of at least asubset of patients with schizophrenia, it will be important toestablish the etiology of these abnormalities, the nature oftheir anatomically selective occurrence, and how they con-tribute to the behavioral manifestations of schizophrenia.

Appreciation is expressed to Ms. A. O'Brien, Mr. P. Newman, andMs. K. Fisher for technical assistance. Drs. G. W. Van Hoesen,B. T. Hyman, and R. C. Gur provided critical comments on themanuscript. The residents and staff of the Division of AnatomicPathology assisted in the acquisition of postmortem tissues. Thiswork was supported by grants from the National Institutes of Health.

1. Altshuler, L. L., Conrad, A., Kovelman, J. A. & Scheibel, A.(1987) Arch. Gen. Psychiatry 44, 1094-1098.

Neurobiology: Amold et al.

II

I

I.j

n. ik

I

f bI

i

Dow

nloa

ded

by g

uest

on

May

5, 2

020

Proc. Natl. Acad. Sci. USA 88 (1991)

2. Kovelman, J. A. & Scheibel, A. B. (1984) Biol. Psychiatry 19,1601-1621.

3. Scheibel, A. B. & Kovelman, J. A. (1981) Biol. Psychiatry 16,101-102.

4. Arnold, S. E., Hyman, B. T., Van Hoesen, G. W. & Damasio,A. R. (1991) Arch. Gen. Psychiatry 48, 625-632.

5. Brown, R., Colter, N., Corsellis, J. A. N., Crow, T. J., Frith,C. D., Jagoe, R., Johnstone, E. C. & Marsh, J. (1986) Arch.Gen. Psychiatry 43, 36-42.

6. Falkai, P. & Bogerts, B. (1986) Eur. Arch. Psychiatr. Neurol.Sci. 236, 154-161.

7. Falkai, P., Bogerts, B. & Rozumek, M. (1988) Biol. Psychiatry24, 515-521.

8. Jakob, H. & Beckmann, H. (1986) J. Neural Transm. 65,303-326.

9. Jeste, D. V. & Lohr, J. B. (1989) Arch. Gen. Psychiatry 46,1019-1024.

10. Benes, F. M. & Bird, E. D. (1987) Arch. Gen. Psychiatry 44,608-616.

11. Benes, F. M., Davidson, J. & Bird, E. D. (1986) Arch. Gen.Psychiatry 43, 31-35.

12. Riederer, B. & Matus, A. (1985) Proc. Natl. Acad. Sci. USA 82,6006-6009.

13. Sloboda, R. D., Dentler, W. L. & Rosenbaum, J. L. (1976)Biochemistry 15, 4497-4505.

14. Binder, L. I., Frankfurter, A. & Rebhun, L. I. (1986) Ann.N. Y. Acad. Sci. 466, 145-166.

15. Lewis, S. A., Ivanov, I. E., Lee, G. H. & Cowan, N. J. (1989)Nature (London) 342, 498-505.

16. Matus, A. (1988) Annu. Rev. Neurosci. 11, 29-44.17. Black, M. M. & Baas, P. W. (1989) Trends Neurosci. 12,

211-214.

18. Carden, M. J., Trojanowski, J. Q., Schlaepfer, W. W. & Lee,V. M.-Y. (1987) J. Neurosci. 7, 3789-3504.

19. Lee, V. M.-Y., Carden, M. J., Schlaepfer, W. W. & Trojan-owski, J. Q. (1987) J. Neurosci. 7, 3474-3488,

20. Blose, S. H., Meltzer, D. I. & Feramisco, J. R. (1982) J. CellBiol. 95, 229 (abstr.).

21. Lee, V. M.-Y., Otos, L., Jr., Schmidt, M. L., Trojanowski,J. Q. (1988) Proc. Natl. Acad. Sci. USA 85, 7384-7388.

22. Trojanowski, J. Q., Schuck, T., Schmidt, M. L. & Lee, V. M.-Y. (1989) J. Neurosci. Methods 29, 171-180.

23. Trojanowski, J. Q., Schuck, T., Schmidt, M. L. & Lee, V. M.-Y. (1989) J. Histochem. Cytochem. 37, 209-215.

24. Tucker, R. P., Binder, L. I. & Matus, A. (1988) J. Comp.Neurol. 271, 44-55.

25. Viereck, C., Tucker, R. P., Binder, L. I. & Matus, A. I. (1988)Neuroscience 26, 893-904.

26. Rosene, D. L. & Van Hoesen, G. W. (1987) in Cerebral Cortex,eds. Jones, E. G. & Peters, A. (Plenum, New York), Vol. 6,345-456.

27. Van Hoesen, G. W. & Hyman, B. T. (1990) in Progress inBrain Research, eds. Storm-Mathisen, J., Zimmer, J. & Ot-tersen, 0. P. (Elsevier, Amsterdam), Vol. 83, 445-457.

28. Nieuwenhuys, R. (1985) Chemoarchitecture of the Brain(Springer, Berlin).

29. Simon, H., Le Moal, M. & Calas, A. (1979) Brain Res. 178,17-40.

30. Sur, M., Garraghty, P. E. & Roe, A. W. (1988) Science 242,1437-1441.

31. Stitzer, R., ed. (1987) The Diagnostic and Statistical Manual ofMental Disorders (3rd Ed.-Revised) (Am. Psychiatr. Assoc.,Washington).

10854 Neurobiology: Amold et al.

Dow

nloa

ded

by g

uest

on

May

5, 2

020