Proc. Natl. Acad. Sci. USAVol. 93, pp. 5472-5477, May 1996Neurobiology

The Golgi apparatus of spinal cord motor neurons in transgenicmice expressing mutant Cu,Zn superoxide dismutase becomesfragmented in early, preclinical stages of the diseaseZISSIMOS MOURELATOS*, NICHOLAS K. GONATAS*, ANNA STIEBER*, MARK E. GURNEYt,AND MAURO C. DAL CANTOt§*Department of Pathology (Neuropathology), University of Pennsylvania Medical Center, Philadelphia, PA 19104; tCNS Diseases Research Unit, PharmaciaUpjohn, Inc., Kalamazoo, MI 49001; and lDepartment of Pathology (Neuropathology), Northwestern University Medical School, Chicago, IL 60611

Communicated by James Sprague, University of Pennsylvania Medical Center, Philadelphia, PA, February 5, 1996 (received for reviewAugust 21, 1995)

ABSTRACT Dominant mutations of the SOD] gene en-coding Cu,Zn superoxide dismutase have been found inmembers of certain families with familial amyotrophic lateralsclerosis (ALS). To better understand the contribution ofSODI mutations in the pathogenesis of familial ALS, wedeveloped transgenic mice expressing one of the mutationsfound in familial ALS. These animals display clinical andpathological features closely resembling human ALS. Earlychanges observed in these animals were intra-axonal anddendritic vacuoles due to dilatation of the endoplasmic retic-ulum and vacuolar degeneration of mitochondria. We havereported that the Golgi apparatus of spinal cord motorneurons in patients with sporadic ALS is fragmented andatrophic. In this study we show that spinal cord motor neuronsof transgenic mice for an SODI mutation display a lesion ofthe Golgi apparatus identical to that found in humans withsporadic ALS. In these mice, the stacks of the cisternae of thefragmented Golgi apparatus are shorter than in the normalorganelle, and there is a reduction in Golgi-associated vesiclesand adjacent cisternae of the rough endoplasmic reticulum.Furthermore, the fragmentation of the Golgi apparatus oc-curs in an early, presymptomatic stage and usually precedesthe development of the vacuolar changes. Transgenic miceoverexpressing the wild-type human superoxide dismutaseare normal. In familial ALS, an early lesion of the Golgiapparatus of motor neurons may have adverse functionaleffects, because newly synthesized proteins destined for fastaxoplasmic transport pass through the Golgi apparatus.

Amyotrophic lateral sclerosis (ALS) is a fatal degenerativedisease, which in the classical sporadic form is characterized bythe involvement of both upper and lower motor neurons andpyramidal tract degeneration (1, 2). About 10% of cases ofALS are familial and these are characterized by additionalpathological features that include involvement of sensoryfibers in the posterior columns as well as cerebellar projectionsin the lateral columns (2, 3). Until recently there were no cluesconcerning the etiology and pathogenesis of the degenerationand death of motor neurons in ALS. Now, a percentage ofpatients with familial ALS has been identified as carrying oneof several missense mutations in the gene for Cu,Zn superox-ide dismutase (SOD) (4-6). Cu,Zn SOD is a metalloenzymethat catalyzes the dismutation of superoxide anion (O-) tohydrogen peroxide (H202) (7, 8). The gene is located onchromosome 21 and consists of five exons. No deletions of theCu,Zn SOD gene have been identified in any of the familiesexamined so far. This strongly suggests that the expression ofthe mutant enzyme is necessary for the pathogenesis of thedisease.

To test this hypothesis, in 1994, Gurney et al. (9) generatedtransgenic mice expressing a mutant form of Cu,Zn SOD,which was identified in one of the families with familial ALS.The selected mutation was in exon 4 of the SOD gene andproduced a Gly -> Ala change at position 93 (G93A) of theenzyme. Multiple lines of transgenic mice carrying this muta-tion have now developed severe motor impairment. Clinicaland pathological studies have shown a pattern of disease that,in many respects, and especially in its late phases, is very similarto that in familial ALS (9-12). In fact, whereas initial alter-ations in these mice are mainly characterized by vacuolarchanges in axons, dendrites, and cell bodies of motor neurons,later the alterations consist of anterior horn neuronal loss("50%), neuronal atrophy, axonal swellings with filamentousaccumulations, and the presence of cytoplasmic inclusions,closely reminiscent of the Lewy-like bodies described in hu-mans with familial ALS (3, 10, 11). Importantly, mice trans-genic for the wild-type form of human SOD and expressingcomparable levels of protein and enzyme activity to that seenin mice transgenic for the mutant form never developedclinical disease, and, pathologically, changes remained verysubtle and were limited to mild cytoplasmic vacuolization,mainly in motor axons (11). The late pattern of disease,characterized by neuronal loss, never developed in mice over-expressing the wild-type human transgene. These data stronglysupport the hypothesis that pathological changes in this murinemodel of familial ALS are due to a new toxic function exertedby the mutant enzyme, rather than to loss of enzyme activitydue to a dominant negative effect (9-12).The cellular target or targets of the presumed toxic effect of

the SOD mutations are not known, but significant insights havebeen gained from ultrastructural studies of degenerating neu-rons in these transgenic mice. Among the ultrastructuralchanges observed in mice expressing the mutant SOD trans-gene, the most readily notable were dilatation of the cisternaeof the endoplasmic reticulum and peculiar alterations inmitochondria leading to swelling and linearization of theirarchitecture to form long, curled membranous profiles (10, 11,13). These changes accounted for most of the vacuolar changesthat were observed in the first phase of the disease. In additionto these changes, however, there were also more subtle alter-ations of the Golgi cisternae, mainly consisting of their dila-tation and apparent disorganization (11).

In patients with sporadic ALS, the Golgi apparatus of motorneurons, studied with an organelle-specific antiserum, is frag-mented (14-18). This organelle-specific antiserum reacts withMG160, a conserved membrane sialoglycoprotein of the me-dial cisternae of the Golgi apparatus (19). MG160 has beencloned and sequenced, and the gene for MG160 has been

Abbreviations: ALS, amyotrophic lateral sclerosis; SOD, superoxidedismutase; RER, rough endoplasmic reticulum.§To whom reprint requests should be addressed.

5472

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

Proc. Natl. Acad. Sci. USA 93 (1996) 5473

assigned to the long arm of chromosome 16 (20, 21). Further-more, developmental studies in chicken embryos have shownthat MG160 is a primordial protein of the Golgi apparatus,appearing very early in development (22).The fragmentation of the Golgi apparatus of motor neurons

in ALS resembles a similar change of the organelle induced bydrugs that disrupt microtubules (23, 24). Furthermore, allneurons containing ubiquitin-positive inclusions, thought to bepathognomonic for ALS (25), had fragmented Golgi apparatus(17). For these reasons it was proposed that a valid animalmodel for ALS should be associated with a fragmented Golgiapparatus of motor neurons. Such Golgi pathology may beimportant to explain, at least in part, the process leading toneuronal death in these patients, because the experimentalfragmentation of the organelle has been associated withsignificant impairment of secretion (26-28). Furthermore, theGolgi apparatus is involved in numerous important functionssuch as the transport, processing, and targeting of virtually allproteins synthesized in the rough endoplasmic reticulum(RER) and is destined for the secretory pathways, the plasmamembrane, or lysosomes (29, 30). Specifically in neurons, allnewly synthesized proteins destined for fast axoplasmic trans-port pass through the Golgi apparatus (31).

In the present study we asked whether human SOD trans-genic mice show an alteration of the Golgi apparatus similarto that seen in patients with sporadic ALS. In addition, weexamined whether changes in the Golgi apparatus might occurearly in the disease process, thus possibly being implicated inthe pathogenesis of the neuronal dysfunction. Such questionscannot be answered from human studies, because examinationof the central nervous system from ALS patients is generallydone at the terminal stages of the disease. Finally, we asked thequestion whether alterations of the Golgi apparatus weredependent on the expression of the mutant enzyme or theycould be produced by the overexpression of the wild-type formof human SOD.

MATERIALS AND METHODSThe production of transgenic mice, assessment of levels of

SOD protein and enzyme activity, and clinical assessment ofneurological disease have already been reported (9-12). Alsoreported have been both clinical and pathological changes intransgenic mice at different ages. Briefly, the first pathologicalalterations in the cytoplasm of motor neurons of mutanttransgenic animals were seen at '60 days, while the firstclinical abnormalities were observed at "80 days. In miceoverexpressing the wild-type human transgene, clinical signsnever developed, and only subtle alterations in the motor axonswere observed at later time points.

Guided by those earlier findings, in this study, we selectedmice expressing the mutant transgene and mice overexpressingthe wild-type transgene, both at different ages from 31 days to289 days. At least two animals were sacrificed per time point.For morphometric analysis, four mutant and four overexpres-sor transgenic animals were selected at -100 days of age.

Preparation of Tissues for Light Microscopy Immunohis-tochemistry. Mice were perfused with phosphate-bufferedphysiologic saline (PBS), pH 7.35, at room temperature.Perfusion was continued with -50 ml of freshly prepared 1%paraformaldehyde in PBS. The spinal cord was removed, cutinto 2- to 3-mm cross sections, kept in the same fixative at roomtemperature for 6-8 hr, and processed as previously described(32).Morphometry and Statistical Methodology. Morphometric

studies were done with a CUE-2 image analyzer using theplanomorphometry program according to a previously de-scribed protocol (Olympus, Lake Success, NY) (17). Meansurface area of neurons, their mean nuclear area, the area ofthe immunostained Golgi apparatus per neuron, the number of

Golgi elements per neuron, and the percentage of cytoplasmicarea occupied by the Golgi apparatus were measured. Tenmotor neurons from each of four mutant SOD transgenic miceand 10 motor neurons from each of four overexpressors ofwild-type SOD were analyzed. The statistical significance ofvalues obtained were evaluated by Student's t test.

Preparation of Tissues for Ultrastructural Immunohisto-chemistry. Spinal cords were fixed by perfusion and immersionin 4% paraformaldehyde and 0.05% glutaraldehyde in PBS,pH 7.35. Frozen sections were permeabilized for 30 min in0.05% saponin, incubated overnight in a polyclonal antiserumagainst MG160, prepared as described in ref. 32, and incubated2 hr each in biotinylated anti-rabbit IgG and ABC Elite(Vector Laboratories), all in PBS with 2% fish gelatin. Sectionswere post-fixed for 30 min in 1.5% glutaraldehyde, stained with0.05% 3,3'-diaminobenzidine tetrahydrochloride with 0.03%hydrogen peroxide with 10mM imidazole and 7.5% sucrose in50 mM Tris buffer, pH 7.4, post-fixed for 1 hr on ice in 1%osmium tetroxide with 1% potassium ferrocyanide in 0.1 Msodium cacodylate buffer, pH 7.4, and embedded in Araldite(33).

RESULTSLight Microscopy. All animals expressing the mutant trans-

gene for Cu,Zn SOD showed a severe fragmentation of theGolgi apparatus, which was identical to that described in ALSpatients (Fig. 1). It is important to note that the fragmentationof the Golgi was present as early as 31 days of age, the earliesttime point examined, which is -2 months before the firstobservable signs of clinical dysfunction. At that time point (31days), the severity of the fragmentation of the Golgi apparatusin any affected motor neuron was comparable to that observedat later time points. In contrast, the neurons of mice transgenicfor the wild-type human SOD did not show fragmentation ofthe Golgi apparatus, even as late as 289 days of age.

t 4.

FIG. 1. Mutant Cu,Zn SOD transgenic animal at 104 days of age.Section of mouse spinal cord immunostained with polyclonal anti-serum against MG-160, a Golgi-specific membrane protein, is shown.Arrow points to a neuron with fragmented or dispersed Golgi appa-ratus. For comparison note the immediately adjacent neuron, whichcontains a normal network of immunostained elements of the Golgiapparatus. All neurons in mice overexpressing the wild-type humanenzyme had a similar normal appearance. Arrowhead points to theaxon of the neuron with the dispersed Golgi apparatus; the initial thinsegment of the axon is followed by the enlarged segment that probablycorresponds to the described spheroids in these mice and in ALS.(x750.)

Neurobiology: Mourelatos et al.

5474 Neurobiology: Mourelatos et al.

As previously reported, the first changes observed in motorneurons of transgenic mice at both light and ultrastructurallevels consist of vacuolization of motor axons, followed by theneuronal cytoplasm and dendrites (10, 11). Vacuoles derivefrom swellings of both the cisternae of the endoplasmicreticulum and, especially, mitochondria, which eventuallytransform into linear or curved membranous arrays.

In paraffin sections from all animals examined in this study,we did not observe vacuolization of the neuronaI cytoplasmwithout fragmentation of the Golgi apparatus, whereas frag-mentation of this organelle was frequently observed in neuronswithout apparent cytoplasmic vacuolization. As shown inTable 1, in symptomatic animals of -100 days of age, there wasvariation in the percentage of motor neurons with fragmentedGolgi. In animals with greater Golgi fragmentation (8290 and8291), there was greater vacuolization, and only 10% ofneurons with Golgi fragmentation were not vacuolated. Inanimals with lesser Golgi fragmentation (9042 and 9043),vacuolization was also less severe, and 70-80% of motorneurons with fragmented Golgi were not vacuolated. However,by electron microscopy we detected a few neurons with a fewcytoplasmic vacuoles and an apparently intact Golgi apparatus(data not shown), suggesting that fragmentation of the Golgiapparatus and cytoplasmic vacuolization occur independently,although the Golgi fragmentation probably occurs earlier. Asmore neurons become affected, more neurons will presentboth Golgi changes and vacuolar pathology. In neurons wherevacuoles and altered Golgi apparatus coexist, vacuoles appearto be surrounded by dispersed round elements of the immu-nostained fragmented Golgi apparatus (Fig. 2).

Additional morphometric analyses were done in the abovemutant transgenic mice and in mice overexpressing wild-typehuman SOD, at comparable ages (Table 2). Significant dif-ferences were observed between the motor neurons of themutant transgenic mice with fragmented Golgi apparatus andthe motor neurons of the overexpressors with normal Golgi.The mutant mice showed smaller motor neurons with smallernuclei and more numerous Golgi elements, indicative offragmentation, but they showed less total "amount" of Golgiapparatus. These findings are similar to previous observationsin human ALS.

Electron Microscopy. The immunoperoxidase localizationof the anti-MG160 antiserum appeared predominantly in one

FIG. 2. Spinal cord motor neuron of a transgenic mouse withmutant Cu,Zn SOD, immunostained for MG-160. Note cytoplasmicvacuoles surrounded by dispersed round elements of the fragmentedimmuinostained Golgi apparatus. Arrowhead points to one of thesevacuoles. (x 1650.)

to two medial cisternae of the Golgi stacks, flanked by 1-3unstained cistemae (Figs. 3-5). The Golgi apparatus fromnormal-appearing spinal cord motor neurons was found in theproximity of the RER, and consisted of the usual stacks of 5-7cisternae, which were surrounded by numerous unstainedsmooth vesicles (Figs. 3 and 4). The elements of the frag-mented Golgi apparatus were detected at the margins ofcytoplasmic vacuoles. The stacks of the cisternae were shorterthan in the normal-appearing Golgi, and there was a reductionof Golgi-associated unstained vesicles, while neighboring RERwas scanty or absent (Fig. 5). The adjacent cytoplasmicvacuoles were partially enclosed by remnants of membranes,the origins of which could not be determined (Fig. 5).

DISCUSSIONWe have suggested that fragmentation of the Golgi apparatusis an early response of motor neurons to injury during the

Table 1. Fragmentation of the Golgi apparatus of spinal cord motor neurons in mutant SODtransgenic mice

PercentageTotal percentage of totalof neurons with neurons withfragmented fragmented

Golgi apparatus Golgi apparatusMutant SOD Age, (with or without but no Total number of

transgenic mice days vacuolization) vacuolization neurons counted8290

Cervical 20 2.0 55108

Lumbar 25 2.5 998291

Cervical 27 2.7 9894

Lumbar 29 2.9 1029042

Cervical 13 9.1 99104

Lumbar 15 10.5 1539043

Cervical 8 6.4 98104

Lumbar 11 8.8 162

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 5475

Table 2. Morphometric studies in spinal cord motor neurons in SOD transgenic mice

Mean neuron Mean nuclear Golgi Golgiarea area area/neuron elements/neuron % Golgi/neuron

Normal SOD transgenic mice(overexpressors) 1138 ± 492.8 222.8 ± 72.6 213.2 ± 62.5 28.1 ± 7.1 18.7 ± 3

Mutant SOD transgenic mice 899.9 ± 248.5 122.4 ± 39.1 76.6 ± 29.3 48.5 ± 13.8 8.5 ± 2.3P <0.005 <0.005 <0.005 <0.005 <0.005

Areas are in square micrometers. Pvalues are based on Student's t-test. Forty motor neurons from each category were analyzed; 10 motor neuronsfrom each of four overexpressors (mice 8730, 8731, 8986, and 8987) and 10 motor neurons from each of four SOD mutants (mice 8290, 8291, 9042,and 9043). Five-micron sections of cells containing nuclei were analyzed. To avoid analyzing a cell twice, of the serial sections, only every 10th sectionwas included in the study. Statistics were based on the number of animals with 7 degrees of freedom.

development of ALS and in certain motor neuropathies (14-18). However, it was difficult to discern in those human studieswhether such Golgi alterations represented a terminal event ina dying neuron or earlier changes contributing significantly tothe process of neuronal degeneration. Time-course studies toexplore whether changes of the Golgi apparatus are indeedearly events in ALS and related diseases are, obviously, notpossible in humans, although it would be possible in anappropriate animal model. Our transgenic model with expres-sion of a mutant form of Cu,Zn SOD, identified in one of thefamilies with familial ALS, gave us the perfect opportunity tolook into this important question.Cu,Zn SOD is an important enzyme in the cascade of events

leading to the neutralization of injurious oxygen radicals intissues. Its main function is to scavenge superoxide anion (O°)into hydrogen peroxide (H202), which is then decomposed towater by glutathione peroxidase. How mutant human Cu,ZnSOD damages motor neurons is unclear, but most hypothesespropose that damage is mediated by free radicals. Hypothesesinclude the generation of hydroxyl radical (34), the catalysis ofperoxynitrite-mediated nitration of proteins (35), and therelease of free copper, which might then catalyze oxidativedamage to lipids and proteins. For example, mutation mayalter the capacity of the enzyme to catalyze free radicalformation or may promote nitration reactions. Superoxidereacts with nitric oxide (NO) to form peroxynitrite anion(ONOO-). The reaction of mutant Cu,Zn SOD with ONOO-may promote formation of toxic nitronium intermediates,which would produce unwanted nitration of proteins (35, 36).Thus, either lipid peroxidation promoted by free radical

A

FIG.!3. Normal-appearingnern Te gp of th Glg

apparatus show one immunostained medial cisterna in a stack ofunstained 5 to 7 cisternae. Note that the stacks of Golgi cisternae aresurrounded by unstained vesicles. A few cisternae of the RER are seenat the lower left of the elecron micrograph. (X 16,000.)

species or protein nitration may initiate the pathogenesis ofALS.

This study shows fragmentation of the Golgi apparatus ofspinal cord neurons in mutant transgenic mice with an iden-tical pattern to that observed in patients with ALS. In addition,this study strongly suggests that the Golgi apparatus may beone of the earliest targets of the SOD mutation. Two featuressupport this contention. First, the fragmentation of the Golgiapparatus was seen in asymptomatic animals. Second, whilevirtually all neurons with vacuolar changes contained frag-mented Golgi apparatus, many neurons with fragmented Golgihad no vacuolar changes. These data strongly suggest that thealterations of the Golgi apparatus usually precede the vacuolarchanges in motor neurons of transgenic animals. The discoveryof lesions in the Golgi apparatus at very early times in themutant transgenics, before the development of any otherpathological or clinical evidence of disease, is potentiallyimportant. The Golgi apparatus is a crucial component of thecell, because it is involved in the transport and processing ofpolypetpides as well as in their targeting to different destina-tions, such as plasma membranes and lysosomes, and to theirsecretion (29, 30). Previous studies with ALS patients hadsuggested that fragmentation of a structure with such crucialroles in the cellular handling of proteins was likely to beassociated with significant impairment of function. The newimmunoultrastructural data of this study support this hypoth-esis. Specifically, the paucity of vesicles in the vicinity of theshortened Golgi cisternae (Fig. 5) imply an impairment ofmolecular traffic between Golgi cisternae and from the Golgiapparatus to the periphery-i.e., along dendrites, axons, andpresynaptic terminals. It is quite possible that membranes ofthe Golgi apparatus, or key proteins mediating the anchorage

.- .& ; W. T>,K s. ' tw S 's.No*"t'M %"0Fi$°;"Bo '' .0 W.;L '>''.'§s~~~~~~~~~~~~~~~4

I

9A

:. ~~~~~i >E' . s~~~~~~~~~~~~~~~~~~~~~~~Z,{w. >t.,>v , # . f ej.,;~~~~~~~~

$v,., ¢. ,,* :. , F.' .,:.

FIG. 4. One stack of Golgi cisternae surrounded by numerous clearvesicles from a normal-appearing motor neuron. Note immunostainedmedial cisternae of the organelle and abundance of the RER in theproximity of the Golgi apparatus. (x 16,000.)

Neurobiology: Mourelatos et al.

5476 Neurobiology: Mourelatos et al.

FIG. 5. In contrast to Figs. 3 and 4, this electron micrograph is froma motor neuron with cytoplasmic vacuoles surrounded by elements ofthe fragmented Golgi apparatus. The stacking of the immunostainedcisternae of the fragmented Golgi apparatus is preserved but (i) thecisternae are shorter than in normal-appearing neurons, (ii) there ispaucity of clear vesicles in the vicinity of the Golgi apparatus, and (iii)there is a virtual absence of adjacent RER. (X16,000.)

of Golgi membranes with microtubules, may be early targets ofthe toxic function of the mutated SOD protein (23, 24, 37). Thefirst experimental observation on the role of microtubules inthe structural organization of the Golgi complex, made morethan 30 years ago, has been confirmed and followed by manystudies from several laboratories (23, 38). However, the mech-anism or mechanisms for the interdependence between theGolgi apparatus and microtubules are still poorly understood.Turner and Tartakoff (24) suggested that the dispersion of theGolgi apparatus in colchicine-treated cells may be caused bythe disruption of the normal tubular interconnections betweenthe groups of Golgi stacks. A close comparison between thefine structure of the fragmented Golgi apparatus in the SODmouse and reports on the effects of microtubule-disruptingagents on the Golgi apparatus in other cells or of the mitoticform of the Golgi apparatus reveal certain similarities but alsosome differences that may be significant. For example, incolchicine-treated pancreatic acinar cells and mitotic FT210cells, the cisternae of the Golgi stacks are shorter than innormal cells, as in Fig. 5. However, mutant SOD mice did notcontain the large groups of small vesicles near the Golgielements observed in colchicine-treated and mitotic cells (39,40). An apparent fragmentation of the Golgi apparatus doesnot necessarily imply depolymerization of microtubules. Themicrotubules may be intact, but the molecular linkages be-tween the membranes of the Golgi apparatus and the micro-tubules may be impaired.An apparent fragmentation of the Golgi apparatus may also

be the result of an accentuated retrograde flow of Golgimembranes from the Golgi apparatus to the RER, while theorthograde flow of membranes from the RER to the Golgiapparatus is blocked or diminished. This hypothesis is sup-

ported by several experiments that suggest that there exists a

physiologic retrograde traffic of membranes from distal toproximal components of the Golgi apparatus or into the RER(40-43). It is conceivable, therefore, that the inhibition ofanother microtubule-associated motor molecule promoting aRER-to-Golgi apparatus transport might result in the accu-mulation of Golgi molecules into the RER, thus producing theimmunocytochemical image of fragmented or dispersed Golgiapparatus.

It is unlikely that the fragmentation of the Golgi apparatusis due to a Brefeldin A-like effect of the mutant proteins (44,45). In Brefeldin A-treated cells, the Golgi stacks are dis-rupted, while they are present here (Fig. 5); furthermore, inBrefeldin A-treated cells, proteins of the Golgi apparatus areredistributed in the RER including the nuclear envelope,which was never observed in sporadic ALS or in these trans-genic mice (14-18, 40, 46).Whereas vacuoles in the cytoplasm of motor neurons and

processes have been found in human SOD transgenic mice,these changes have not been reported in human ALS, althoughthey have been observed in motor neurons in tetanus, in whichoxidative stress has been postulated as a causative factor (47).However, vacuoles are predominantly a feature of the earlyphase of disease in the Gl line of mice which have relativelyhigh levels of expression of the human transgene. Animals withlower expression of the transgene, such as the G20 and G5 linesof animals, only show minimal vacuolar changes. These ani-mals-i.e., the G20 (9-11) and G5 lines (unpublished data)-survive long enough to develop a pattern of disease which ischaracterized by neuronal loss, atrophy of anterior horns, andfilamentous inclusions. These changes are essentially identicalto those in human ALS, particularly the familial form (48, 49).Therefore, it is quite possible that in humans with ALS, if thereis indeed an early phase of vacuolar changes, these are modestand more similar to those in transgenic mice with intermediatelevels of expression of the mutant transgene. Furthermore, itis likely that the vacuolar phase would go undetected because,in ALS patients, the examination of the central nervous systemis generally performed at the end stage of the disease process.Another significant result of this study is that mice overex-

pressing the wild-type human SOD fail to develop fragmen-tation of the Golgi apparatus as late as 289 days after birth.This observation is consistent with the conclusion that thefragmentation of the Golgi apparatus in SOD transgenic miceis dependent on the presence of the mutant enzyme. Thissupports the gain of function hypothesis for the pathogenesisof mutant SOD-induced cellular injury in this animal model ofALS.

The authors thank Mr. Jada Prasad for his expert technical assis-tance. This work was supported in part by Grants NS-13011, NS-32248,and NS-05572 (a Senator Javitz Neuroscience Award to N.K.G.) fromthe U.S. Public Health Service and by a grant from the MuscularDystrophy Association.

1. Charcot, J.-M. (1874) Prog. Med. 2, 325-327.2. Oppenheimer, D. R. & Esiri, M. M. (1992) in Greenfield's Neu-

ropathology, eds. Adams J. H. and Duchen L. W. (University,New York), pp. 988-1045.

3. Hirano A., Kurland L. T. & Sayre G. F. (1967) Arch. Neurol. 16,232-243.

4. Siddique, T., Pericak-Vance, M. A., Brooks, B. R., Roos, R. P.,Hung, W.-Y., Antel, J. P., Munsat, T. L., Phillips, K., Warner, K.,Speer, M., Bias, W. B., Siddique, N. A. & Roses, A. D. (1989)Neurology 39, 919-925.

5. Siddique, T., Figlewicz, D. A., Pericak-Vance, M. A., Haines,J. L., Rouleau, G., Jeffers, A. J., Sapp, P., Hung, W.-Y., Bebout,J., McKenna-Yasek, D., Deng, G., Horvitz, H. R., Gusella, J. F.,Brown, R. H. & Roses, A. D. (1991) New Engl. J. Med. 324,1381-1384.

6. Rosen, D. R., Siddique, T., Patterson, D., Figlewicz, D. A., Sapp,P., et al. (1993) Nature (London) 362, 59-62.

7. Crapo, J. D., Oury, T., Rabouille, C., Slot, J. W. & Chang, L.-Y.(1992) Proc. Nat. Acad. Sci. USA 89, 10405-10409.

8. Tainer, J. A., Getzoff, E. D., Beem, K. M., Richardson, J. S. &Richardson, D. C. (1982) J. Mol. Biol. 160, 181-217.

9. Gurney, M. E., Pu, H., Chiu, A. Y., Dal Canto, M. C., Polchow,C. Y., Alexander, D. D., Caliendo, J., Hentati, A., Kwon, Y. W.,Deng, H-X., Chen, W., Zhai, P., Sufit, R. L. & Siddique, T. (1994)Science 264, 1772-1775.

10. Dal Canto, M. C. & Gurney, M. E. (1994) Am. J. Pathol. 145,1271-1280.

Proc. Natl. Acad. Sci. USA 93 (1996)

Proc. Natl. Acad. Sci. USA 93 (1996) 5477

11. Dal Canto, M. C. & Gurney, M. E. (1995) Brain Res. 676, 25-40.12. Chiu, A. Y., Zhai P., Dal Canto, M. C., Peters, T. M., Kwon

Y. W., Prattis S. M. & Gurney, M. E. (1995) Mol. Cell Neurosci.6, 349-362.

13. Wong, P. C., Pardo, C. A., Borchelt, D. R., Lee, M. K., CopelandN. G., Jenkins N. A., Sisodia, S. S., Cleveland, D. W. & Price,D. L. (1995) Neuron 14, 1105-1116.

14. Mourelatos, Z., Adler, H., Hirano, A., Donnenfeld, H., Gonatas,J. 0. & Gonatas, N. K. (1990) Proc. Natl. Acad. Sci. USA 87,4393-4395.

15. Gonatas, N. K., Stieber, A., Mourelatos, Z., Chen, Y., Gonatas,J. O., Appel, S. H., Hays, A. P., Hickey, W. F. & Hauw, J.-J.(1992) Am. J. Pathol. 140, 731-737.

16. Mourelatos, Z., Yachnis, A., Rorke, L., Mikol, J. & Gonatas,N. K. (1993) Ann. Neurol. 33, 608-615.

17. Mourelatos, Z., Hirano, A., Rosenquist, A. & Gonatas, N. K.(1994) Am. J. Pathol. 144, 1288-1300.

18. Gonatas, N. K (1994) Am. J. Pathol. 145, 751-761.19. Gonatas, J. O., Mezitis, S. G. E., Stieber, A., Fleischer, B. &

Gonatas, N. K. (1989) J. Biol. Chem. 264, 646-653.20. Gonatas, J. O., Mourelatos, Z., Stieber, A., Lane, W. S., Brosius,

J. & Gonatas, N. K.(1995) J. Cell Sci. 108, 457-467.21. Mourelatos, Z., Gonatas, J. O., Nycum, L. M., Gonatas, N. K. &

Biegel, J. A. (1995) Genomics 28, 354-355.22. Stieber, A., Mourelatos, Z., Chen, Y.-J., Le Douarin, N. &

Gonatas, N. K. (1995) Exp. Cell Res. 219, 562-570.23. Robbins, E. & Gonatas, N. K. (1964) J. Histochem. Cytochem. 12,

704-711.24. Turner, J. R. & Tartakoff, A. M.(1989) J. Cell Biol. 109, 2081-

2088.25. Matsumoto, S., Goto, S., Kusaka, H., Imai, T., Murakami, N.,

Hashizume, Y., Okazaki, H. & Hirano, A. (1993) J. Neurol. Sci.115, 208-213.

26. Lacy, P. E., Howell, S. L., Young, D. A. & Fink, C. J. (1968)Nature (London) 219, 1177-1179.

27. Antoine, J. C., Maurice, M., Feldman, G. & Avrameas, S. (1980)J. Immunol. 125, 1939-1949.

28. Redman, C. M., Banerjee, D., Howell, K. & Palade, G. E. (1975)J. Cell Biol. 66, 42-59.

29. Farquhar, M. G. & Palade, G. (1981) J. Cell Biol. 91 (Suppl.),77s-103s.

30. Farquhar, M. G. (1985) Annu. Rev. Cell. Biol. 1, 447-488.31. Hammerschlag, R., Stone, G. C., Bolen, F. A., Lindsay, J. D. &

Ellisman, M. H. (1982) J. Cell Biol. 93, 568-575.32. Croul, S., Mezitis, S. G. E., Stieber, A., Chen, Y.-J., Gonatas,

J. O., Goud, G. & Gonatas, N. K. (1990)J. Histochem. Cytochem.38, 957-963.

33. Brown, W. J. & Farquhar, M. G. (1989) Methods Cell Biol. 31,553-569.

34. Yim, M. B., Chock, P. B. & Stadman, E. R. (1990) Proc. Natl.Acad. Sci. USA 87, 5006-5010.

35. Beckman, J. S., Carson, M., Smith, C. D., Koppenol & W. H.(1993) Nature (London) 364, 584.

36. Ischiropoulos, H., Zhu, L., Chen, J., Tsai, M., Martin, J. C.,Smith, C. D. & Beckman, J. S. (1992) Arch. Biochem. Biophys.298, 431-437.

37. Karecla, P. I. & Kteis, T. E. (1992) Eur. J. Cell Biol. 57, 139-146.38. Pavelka, M. & Ellinger, A. (1983) J. Cell Biol. 97, 737-748.39. Misteli, T. & Warren, G. (1995) J. Cell Sci. 108, 2715-2727.40. Johnston, P. A., Stieber, A. & Gonatas, N. K. (1994) J. Cell Sci.

107, 529-537.41. Martinez, O., Schmidt, A., Salamero, J., Hoflack, B., Roa, M. &

Goud, B. (1994) J. Cell Biol. 127, 1575-1588.42. Miesenbock, G. & Rothman, J. E. (1995) J. Cell Biol. 129,

309-319.43. Lippincot-Schwartz, J., Cole, N. B., Marotta, A., Conrad, P. A. &

Bloom, G. S. (1995) J. Cell Biol. 128, 293-306.44. Doms, R. W., Russ, G. & Yewdell, J. W. (1989) J. Cell Biol. 109,

893-899.45. Lippincot-Scwartz, J., Yuan, L. C., Bonnifacino, J. S. & Klausner,

R. D. (1989) Cell 56, 821-836.46. Robbins, E. & Gonatas, N. K. (1964) J. Cell Biol. 21, 429-463.47. Komai, K., Wang, H. S., Bauserman, S. C. & Chou, S. M. (1995)

J. Neuropathol. Exp. Neurol. 54, 454 (abstr.).48. Hirano, A. (1991) inAdvances in Neurology:Amyotrophic Lateral

Sclerosis and Other Motor Neuron Diseases, ed. Rowland, L. P.,(Raven, New York), pp. 91-101.

49. Hirano, A. & Kato, S. (1992) in Handbook ofAmyotrophic LateralSclerosis, ed. Smith, R. A., (Dekker, New York), pp. 183-192.

Neurobiology: Mourelatos et al.