The relationship between enzyme activity and

(1)

J. Neurol. Neurosurg. Psychiat., 1963, 26,479

The relationship between

enzyme

activity

and

neuroglia

in

the

prodromal

and

demyelinating

stages

of

cyanide

encephalopathy

in

the

rat

M. Z. M. IBRAHIM, PHILANDER B. BRISCOE, JR1.,

OLGA B. BAYLISS, AND C. W. M. ADAMS

From the Department of Pathology. Guy's HospitalMedicalSchool,London

In aprevious communication(IbrahimandAdams, 1963), we suggested that the increased enzymatic activity at the edge of the old yet active plaque of multiple sclerosis was due to the increase in oligo-dendrocyte population in this zone. Our obser-vations on the multiple sclerosis plaque gave no indication whether this increase in neuroglial population and enzyme activity was a primary or secondary event in the demyelinating lesion. For this reason we have studied the neuroglial and enzyme changes in the prodromal stage of the

experimental demyelinating diseaseproduced bythe injection of cyanide into therat.

MATERIAL AND METHODS

Inapilot experiment carried out by the second author, large doses of cyanide were injected intraperitoneally into rats for periods up to five days: repeated daily injectionsweregiven and the maximumdosageachieved in thefour-dayperiodwas105 mg.Althoughno demyeli-natinglesions wereproduced by this regime, the oligo-dendroglia of the corpus callosum showedconsiderably increased oxidative enzyme activity as seen in histo-chemical enzymepreparations (seebelow).Theverylarge dosesofcyanide administered tothe ratsresulted inan

impractically highmortality rate, so the following pro-cedure, modified from Lumsden(1950),wasused instead. Twelve female rats with an average body weight of 275 g.and 13maleratswithanaverageweightof350g. were injected on five days a week with cyanide by the subcutaneous route. Solutions wereprepared tocontain 4 to 8 mg. NaCN per ml. indistilled water. Theinitial dosewas2 to3mg.(0-8 mg./lOOg.bodyweight).This level ofdosagewasmaintaineduntilno animal diedfollowing injection; the nextday the survivors received an incre-mentof0 5to 1mg.intheircyanidedosage.Theabsence of injections on Saturday and Sunday led to increased cyanide sensitivity on Monday, so onthe latter daythe dosagewasreducedby 10mg. Injectionswerecontinued for threeweeks and, bythistime,themaximum dosage of cyanide was 6 0 mg. daily. All animals either died

'Present address: Johns Hopkins Hospital School of Medicine,

Baltimore,Md.,U.S.A.

withintwohoursofinjection or werekilled by chloroform anaesthesia. Five animals survived the three-week period ofinjections and werekilled at intervals from 25 to 92 days from the beginning of the experiment. Within 20 min. of death selected blocks from the brain were eitherplacedin 15%formol-2%ammoniumbromide or werefrozen ontoacryostat chuck in a Slee chuck freezer. Three to four transverse blocks, 2to 3 mm. thick, were usually prepared from each brain. Examination of control material established that autolytic and post-mortem changes had not occurred in the neuroglia during the 20-min. interval between death andfixation.

CYTOLOGICAL METHODS The upper andlower blocks of each brainwere fixed in formol-ammonium bromide at room temperaturefor 24to36 hours and were transferred tofreshfixativeat50°C. for 10min.before sectioning on thefreezing microtomeat20 to

251,.

These sections were then stained by Hortega's methods for oligodendroglia (1956a, republished) and microglia (1954, republished). Penfield's (1924) method was found especially valuable for staining oligodendroglia in the corpus callosum, where these cells are sometimes difficult to impregnate. Cajal's (1913;1916)gold-sublimatemethod and his(1920) modification ofBielschowsky's (1904; 1909) silver method were used todemonstrate astroglia, and the unmodified Bielschowsky methodwasusedfor staining neurones and axones.Somesections from these blocks were stained by Sudan black and by the osmium tetroxide-a-naphthy-lamine(O.T.A.N.) method (Adams, 1959) for triglyceride fats,cholesterol esters, and phospholipids.

HISTOCHEMICAL METHODS The middle block of each brain was usually used for enzyme histochemistry. Un-fixed sections were cuton the Slee cryostat at6a. These sections were stained by adenosine triphosphatase2 (PadykulaandHerman,1955a and b), diphosphopyridine nucleotide (DPNH2) tetrazolium reductase3, and lactic dehydrogenase4 methods. The dehydrogenase and re-ductase methods were slight modifications (Adams, Davison, and Gregson, 1963) of the methods of Nachlas, Walker, and Seligman (1958a and b), Hess and Pearse 2ATP phosphohydrolase (3.6.1.4).

3NADH,-tetrazolium reductase = DPN diaphorase = NADH,:

cytochrome oxidoreductase(1.6.2.3).

'L-lactate NADoxidoreductase(11.1.27).

479

Protected by copyright.

on July 4, 2021 by guest.

(2)

ESSENTIAL CHARACTERISTICS OF CYANIDE LESIONS

During the first few days of cyanide injectionssome animals develop necrotic lesions, usually seen in the corpus callosum, corpusstriatum, posterior part of the cerebral cortex, hippocampus, medulla, and basal ganglia. These necrotic lesions are character-ized by the relative absenceof cellularreaction in the early stages of their development, by generalized

softening,loss of neuroglialcytoplasmand processes, lossof neurones, pyknotic nuclei (Fig. 1), and early lossof axones (Fig. 2)throughoutthe lesion.

Demyelinating lesionsusually appear from about thefourteenthday of injection onwardsin thecorpus callosum, and, sometimes, also in the callosal radiation, anterior commissure, optic chiasm and tract, hippocampal commissure, fimbria, thalamus,

hypothalamus, substantia nigra, corpus striatum,

hippocampus, cerebellum, medulla, and cerebral cortex.Thesedemyelinatinglesionsarecharacterized by early oligodendroglial proliferation (Fig. 3),

*:;WiaAEs:. ;

FIG. 3

FIG. 1. Necroticlesionincorpusstriatum. Note necrotic centre containingfew cells andperiphery showing

neuro-glialreaction.Hortega'ssilver methodfor microglia x 47. FIG.2. Degenerating axons(cluband granularforms)in lighter-staining necrotic lesion (arrow). Bielschowsky's silvermethodforaxons x 470.

FIG.3. Early demyelinating lesion in corpus callosum. Note density of oligodendroglia in centre of lesion. Hortega's silvcr methodfor oligodendroglia x 118.

FIG. I

(3)

Enzyme activity andneuroglia intheprodromalanddemyelinating stages of cyanideencephalopathy 481 ..,.,.. ..,. g.'-'": . 5 FIG. S FIG. 4

FIG. 4. More advanceddemyelinatinglesions in corpus callosum (toparrow)and hippocampalcommissure(lower arrow). The dark area between the two is less affected white matter. Hortega's silver method for oligodendroglia x 47. FIG. 5. Partial preservation of axons in a demyelinating lesion of corpus callosum (white arrow). The rounded cells (black arrow) with darknuclei surrounded by pale haloes are gitter cells. Bielschowsky's silver method for axons x 470. followed by demyelination (Fig. 4) but withrelative

preservation ofaxones in the early stages (Fig. 5). The subsequent fate of these lesions is described below.

OLIGODENDROGLIA Ourmaininterestwastoobserve

changes in the oligodendroglia of white matter before the onsetof demyelinating lesions. Forthis purpose, enzymatically active oligodendrocytes were countedin atransversestrip across thecorpus callosum as far as the callosal radiations and in a further strip, atright angles to the first, across the

junction between the corpus callosum and callosal radiation. In enzyme preparations the oligodendro-cyte is recognized by its round, carmalum-stained nucleus and the crescentic or sometimes circular arrangement of enzymatically-active mitochondria around the nucleus (Adams, 1962; Friede, 1962; Yonezawa, Bornstein, Peterson, and Murray, 1962; Ibrahim and Adams, 1963).

The result of these counts is shown in Table I and they show some increase in the number of enzymatically-active oligodendrocytes in the corpus callosum during theprogress ofcyanide encephalo-pathy, even before demyelinating lesions appear. Furthermore, not only did the number of active cells increase, but the intensity ofstaining and size oftheir mitochondria also increased(Fig. 6). Like-wise, the number of enzymatically-active

mito-chondria in theneuropil,which have beententatively identified as those ofoligodendroglia (Ibrahim and Adams, 1963), also show a considerable increase (Fig. 6). The number of enzymatically-active cells increases still further in the early demyelinating lesion (Fig. 7) but, subsequently, the number of active cells declines. These increases in enzymatic activity are not seen in necrotic lesions (Table 1).

TABLE I

NUMBER OF ENZYMATICALLY-ACTIVE OLIGODENDROGLIA IN CORPUS CALLOSUM OF CYANIDE-INJECTED RATS'

Daysof CorpusCallosum

Cvanide

Injection Normal Necrotic Demyelinating

0(control) I 2 3 4 7 10 14 16 25 32 44 56 92 A reas 90, 89, 76, 76 55, 1122 90,105 91,87 1033, 86 178 151 1333 1192 892 Lesions Lesions 30 -86 -- 271, 179 - 302 - 151 - 29 - 121

'Numberof cellsstained forlacticdehydrogenaseper 0.1 sq.mm.

2=necroticlesion elsewhere inbrain

3 =demyelinatinglesion elsewhere inbrain

(4)

ru

FIG. 6a N4 FIG. 6b Nitro-BT x 470. FIG. 7. Marked increase inlactic

de-hydrogenase activity ofoligodengroglia andofbackgroundin earlydemyelinating lesion(arrow) of corpuscallosum. Nitro-BT x 470. FIG. 7 TABLE NUMBER OF OLIGODENDROGLIA IN SILVER-STAINED PREPARATIONS OF CORPUS CALLOSUM

IN CYANIDE-INJECTED RATS

Daysof CyanideInjection No. ofOtigodendroglia per 0-1 sq.mm.

0(control

3

16 Normalarea

Demyelinatinglesion

'Standarderrorofmean

169 ±9-81

237 ±8-3 275 ±9-0

About490

To assess whether this increase in

ezymatically-active cellswasduetoan actualincrease in the num-ber of

oligodendroglia

or due

only

to a greater

proportion

of cells

becoming enzymatically-active,

counts were made on the corpus callosum of two animalsafter three and 16

days

of

cyanide

injections

(Table II).

The

three-day animal,

which had no

apparentlesion

anywhere

inthe

brain,

showsa

slight

increase in

oligodendroglia

while the

16-day animal,

which had an

early demyelinating

lesion in another

partof thecorpus

callosum,

showsafurther increase of these cells in normal areas. Inthe

demyelinating

lesion in thecorpus callosum of this

16-day

animal the number of

oligodendroglia

is

strikingly

increased even

though

theresult is

probably

an underestimate

on account of

overlapping

ofcells. In silver-stained sections of late

demyelinating

lesions the number of

oligodendrocytes

declines but

they

do not

disappear

completely (Table

I).

In Table III is shown the averageof themean of maximum and minimumdiameters for20

randomly

selected

oligodendroglia

in normal areas of corpus TABLE III

MEAN DIAMETER OF OLIGODENDROGLIA IN SILVER-STAINED PREPARATIONS OF NORMAL AREAS OF

CORPUS CALLOSUM IN CYANIDE-INJECTED RATS

Daysof Cyanide Injection Mean Diameter of Oligodendroglial

0(control) 34-2±1-6'

3 43-2±5-0

16 56-2 ±5-3

'Mean of maximum and minimumdiameteraofperikaryon

'Standarderrorofmean

(5)

Enzymeactivity andneurogliainthe prodromaland

demyelinating

stages

of

cyanide

encephalopathy

483

1-1v1. S FIG.9

FIG. 8. Acute swelling (arrow) ofoligodendroglia in early demyelinating lesion of cerebral cortex. Hortega's silver methodforoligodendroglia x 470.

FIG. 9. Marchi-positivereaction in gitter cells of latedemyelinating lesion in corpus callosum. O.T.A.N.method x 118.

callosumincontrol, three-day, and 16-dayanimals.

(Oligodendroglial

diameters in the demyelinating lesionof the 16-day animal could not be accurately

measured.)These results show that the

oligodendro-glia undergohypertrophy, aswell ashyperplasia (see

above), in the pre-demyelinating phase of cyanide

encephalopathy. Apartfromhyperplasiaand

hyper-trophy, the oligodendroglia in silver-stained

pre-parations of the early demyelinating lesion show that form ofintracellular oedema which is known

asacuteswelling (Fig. 8).

MICROGLIA These cells do not show cytological evidence of activation in silver preparations (see

Ibrahim andAdams, 1963) until adefinite necrotic or advanced demyelinating lesion develops. Gitter cells, containing sudanophilic and Marchi-positive

(O.T.A.N.-black)

granules, appear early in necrotic

lesions but they are seen only in the later stages of

demyelinating

lesions (Fig. 9). An important differ-ence between these two types of lesion is that the activated and gitter forms of microglia are found in the edge of the early necrotic lesion but, in the

demyelinating lesion, they appear throughout the

plaque, including its edge. Inhistochemicalenzyme preparations, gitter cells show their usual character-istic oxidative activity (see Rubinstein and Smith, 1962; Smith and Rubinstein, 1962; Ibrahim and Adams, 1963).

ASTROCYTES Insilverpreparationsofearlynecrotic

lesions,

hypertrophic and multinucleate forms of astrocytes are seen; activation of these cells leads ontotheformation ofa glioticwalland, finally, to

glial

'scarring'.Silverpreparations of early

demyeli-natinglesionsnevershow much evidence of

astroglial

activation,

but hypertrophicand clasmatodendritic'

forms of astrocytes arepresent in theedgeofolder

demyelinating

lesions (Fig. 10). Gliosis and scar

formation

(Figs.

11 and

12)

areseenin the last

stage

of these self-limiting, cyanide-induced,

demyeli-nating lesions. Transformation of protoplasmic

astrocytes into fibrous types is evident in lesions

occurringingreymatter.Increased oxidative enzyme

activity

is seen in activated forms ofastrocytes in both necrotic anddemyelinating lesions(see

Friede,

1962;

Osterbergand Wattenberg, 1962;

Rubinstein,

Klatzo,

and Miquel, 1962; Smith and

Rubinstein,

1962).

SERUM BICARBONATE A small additional group of four animals was injected with a single dose of

cyanide to assess the extent of changes in serum bicarbonate duetothedrug.There wasasubstantial fall in the bicarbonate level (see Table IV) which

was

probably

aresult of

depressed

tissue

respiration.

'Swelling and fragmentation of astroglial processes together with

cytoplasmicgranularity.

(6)

lesion in corpus callosum. Cajal'sgold-sublimate method

for

astroglia

x 47.

FIG. 12. Higher powerfield of similar lesions to those seen inFig. 11 toshowastroglialfibres x 118.

FIG. 10

FI.G. 11-W ..

FIG. II1 FIG. 12

TABLE IV

CHANGES IN SERUM BICARBONATE

LEVEL AFTER INJECTION OF A SINGLE DOSE OF

CYANIDE IN THE RAT

HoursafterInjectionof2mg.NaCN SerumBicarbonate Level(mM)

0 (control) 1 2 4 24 22-5 5-5 8-5 10-5 20-5 DISCUSSION

CYTOLOGICAL AND ENZYMATIC FEATURESOFCYANIDE

LESIONS Our observations have shown that both necrotic and demyelinatinglesions occur duringthe course ofcyanide encephalopathy in the rat. The necrotic lesion results in complete loss of structure, while the demyelinating lesion is characterized by loss ofmyelin but partialpreservation ofaxons. In

(7)

Enzyme activity and neuroglia in theprodromal and demyelinating stagesofcyanide encephalopathy 485 the edge of the necrotic lesion, there is a moderate

proliferation of microglia and astrocytes but the oligodendrocyte is not the predominant cell. Before the onsetof demyelination,oligodendrogliaofwhite matterproliferate, undergohypertrophy, and become enzymatically more active; these changes become moreprominent throughout the established demyeli-nating plaque. Proliferation and activation of microglia and astrocytes appear to be a secondary response to demyelination, as these changes follow but do not precede formation of the plaque.

These observations indicate that increased meta-bolic activity of oligodendroglia is a feature of the prodromal stage of cyanide demyelination. The initial increase in enzyme activity of the oligoden-droglia does not appear to be due simply to increased activity in a greater proportion of these cells, for the'silver' counts and cellular dimensions show com-parable increasestothose in the 'enzyme' counts (see Tables I, IL, Ill). It is apparent, therefore, that the increased enzyme activity of the oligodendroglia is dueto both hyperplasia and hypertrophy of these cells. It should be noted, however, that we have assumed that the initial increase in the number of oligodendroglia is due to hyperplasia (see Lewis and Swank, 1953; Smart, 1960; Koenig, 1960; Smart and Leblond, 1961; Adrian and Walker, 1962; Koenig, Bunge, and Bunge, 1962) but the possibility that it is due to infiltration or migration of these cells has not been entirely excluded.

Although the oligodendroglia proliferate and become enzymatically active in the early demyeli-nating lesion (the stage associated with myelin pallor) their numbers and enzyme activity subsequently decline during the later Marchi-positive and gliotic stagesof the plaque. We are uncertain whether this subsequent fall in oligodendroglial population and enzymeactivity indicates that cyanide demyelination is a non-progressive and self-limiting lesion or whether it is that the withdrawal of cyanide allows thedemyelinating process to halt.

Previous studies have indicated that both necrotic and demyelinating lesions result from prolonged administration of cyanidetoanimals(Ferraro, 1933; Hurst, 1942; 1944; 1952; Lumsden, 1950; Levine and Stypulkowski, 1959) but a recent report has described lesions after brief exposure of animals to high concentrations of cyanide (van Houten and Friede,1961).From ourpilot studiesitwould appear that the latter technique produces predominantly necrotic and not demyelinating lesions. Necrotic lesions,whetherproduced byischaemia(Macdonald and Spector, 1963), freezing (Rubinstein, Klatzo, and Miquel, 1962), or by cyanide result in little

oligodendroglial proliferation and little increase in oxidative enzymeactivityattheedge of the softened

area. This circumstance may explain van Houten

and Friede's observation of increased oxidative enzyme activity in astrocytes, which are

character-istically seen at theedge of necrotic lesions andnot in theoligodendroglia of theircyanide-injected rats. These authors, however, noteda diffuse increase in the enzymaticstaining of white matter, which

prob-ably corresponds to the increased mitochondrial

activityobservedbyusinwhitematter.

MECHANISM OF CYANIDEDEMYELINATION Itis known thatcyanide is notacumulative poison(Hurst, 1940; Wyndham,1941)and it isnottheproductsofcyanide

metabolism that cause cyanide encephalopathy (Rose, Harris, and Chen, 1954). Van Houten and Friede(1961)andothers haveattributedthe

encepha-lopathy to the inhibition of cytochrome oxidase by cyanide but, if this explanation isaccepted, itis

difficulttoexplain the absence of lesions in the many other organs that depend upon oxidative meta-bolism. Moreover, cyanide inhibits a wide range of metal-dependant enzymes and it is, therefore, difficult to accept that cytochrome oxidase is the only enzyme to beimpaired by thispoison.

Gallagher, Judah, and Rees (1956) found that cyanideimpairs thecondensationofacetylcoenzyme A withoa-glycerophosphate, resulting in diminished

products ofphosphatidicacid,ametabolicprecursor of the phospholipids. However, failure of

phos-pholipid synthesis in brain would not necessarily

promote demyelination, for it has been shown that myelin phospholipids are metabolically rather inert and have, at most, only a slow turnover

(Davison

andDobbing, 1960).

It is tempting to speculate that cyanide damages

either vascular endothelium or the

oligodendrocyte

and that this results in local oedema of white matter (see Levine, 1960; Luse, 1960; van Houten and Friede, 1961). Such oedema has been observed in cyanide encephalopathy in the form of myelin pallor or reticulation and may itself, for reasons

unknown, cause demyelination

(Lumsden

and Pomerat, 1951;Lumsden, 1957; Bunge, Bunge, and Ris, 1960;vanHoutenandFriede, 1961).The oligo-dendroglial changes we have observed in

cyanide

encephalopathycould be

explained

as

hyperactivity

of these cells in response to oedema. In this con-nexion the oligodendrocyte has been

regarded

as an osmo-regulator between the blood vessel and brain (Lumsden, 1957; Katzman,

1961).

RELATIONSHIP OF CYANIDE DEMYELINATION TO MUL-TIPLE

SCLEROSIS

The main purpose of this

experi-mental study has been to establish whether the increased

oligodendroglial

population

and enzyme

activity in the multiple sclerosis

plaque

(Ibrahim

(8)

fore,essentially thesameinbothconditions butthe

subsequentprogress ofthelesions is often different.

SUMMARY

After a few days of cyanide injections, rats

com-monly develop necrotic lesions in the brain but, aftertwo weeks ofinjections, demyelinating lesions appear-predominantly in the corpus callosum. In theprodromal stages, before the onset of demyeli-nation, theoligodendroglia of the corpus callosum show increasedenzyme activity and undergo

hyper-plasia andhypertrophy. These changesbecomemore

marked in the early demyelinating lesion but they

regress asthe lesion maturesand becomes inactive.

The neuroglial reaction around necrotic lesions is predominantly astrocytic and microglial while the oligodendroglial element, described above, is not prominent.

Itissuggested that thehyperactivity of the oligo-dendroglia in the prodromal stage of cyanide encephalopathy isaprimaryeventindemyelination and is not a response tomyelin breakdown. From this experimental evidence it is inferred that oligo-dendroglial activity at the edge of the multiple sclerosis plaque is a primary feature of the disease

process in man.

REFERENCES

Adams, C. W. M. (1959). J. Path.Bact.,77, 648.

(1962) Develop. Med. Child Neurol., 4, 393.

,Davison, A. N.,andGregson,N. A. (1963). J. Neurochem.,

10, 383.

Hess, R.,andPearse, A. G.E.(1961).Exp.Cell.Res.,25,187.

Hortega, P. DelRio (1954). Arch. Histol. (B. Aires), 5, 105

(re-published).

(1956a).Ibid.,6, 132(republished).

Hurst, E. W.(1940).Aust. J.exp.Biol.med.Sci.,18,201.

--(1942). Ibid., 20, 297. (1944).Brain,67, 103. (1952).Amer. J.Med., 12, 547.

Ibrahim, M. Z. M., andAdams,C.W. M.(1963).J.Neurol.Neurosurg.

Psychiat.,26, 101.

Katzman, R.(1961). Neurology(Minneap.),11,27.

Koenig, H. (1960).J.Histochem.Cytochem.,8, 337.

,Bunge, M. B., and Bunge, R. P.(1962).Arch. Neurol.(Chic.),

6, 177.

Levine,S.(1960).J.Neuropath.exp.Neurol., 19,238.

,andStypulkowski, W. (1959). Arch. Path., 67, 306. Lewis,R.C.,andSwank, R.L.(1953).J.Neuropath.exp.Neurol.,

12,57.

Lumsden, C. E.(1950).J.Neurol. Neurosurg.Psychiat., 13, 1.

(1957).lIle Congres InternationaldeNeuropathologie,Bruxelles;

Rapports etDiscussions. "Actamedica Belgica," Brussels,

p.27.

andPomerat, C. M.(1951). Exp.CellRes.,2, 103.

Luse, S. A.(1960).Anat.Rec.,138,461.

Macdonald,M., andSpector,R.G.(1963).Brit.J.exp.Path.,44,11.

Nachlas, M. M., Walker, D. G., and Seligman, A. M. (1958a).

J. biophys.biochem.Cytol., 4,29.

(1958b).Ibid., 4,169.

Osterberg,K.A., andWattenberg,L.W.(1962).Arch.Neural.(Chic.),

7, 211.

Padykula,H.A.,andHerman,E.(1955a).J.Histochem.Cytochem., 3, 161.

(1955b).Ibid.,3, 170. Penfield,W.(1924). Brain, 47, 430.

Rose,C.L.,Harris,P.N.,andChen,K. K.(1954).Proc.Soc.exp.

Biol.(N. Y.), 87,632.

Rubinstein, L.J., andSmith, B.(1962). Nature(Lond.), 193, 895.

,Klatzo,I., andMiquel,J.(1962).J. Neuropath.exp. Neurol.,

21, 116.

Smart,I. (1960).Anat.Rec.,136,279.

-, andLeblond, C.P.(1961).J.comp.Neurol., 116,349.

Smith, B.,andRubinstein,L.J.(1962).J.Path.Bact.,83,572.

Thomas,E., and Pearse,A. G. E.(1961).Z.Zellforsch.Abt.Histochem., 2,266.

vanHouten, W.H.,andFriede,R. L.(1961). Exp. Neurol.,4,402.

Wyndham,R. A.(1941).Aust.J.exp. Biol. med.Sci.,19,243.

Yonezawa, T.,Bornstein,M.B., Peterson, E.R.,andMurray,M.R.

(1962).J.Neuropath.exp.Neurol., 21,479.