Axonal Transport of Mutant Superoxide Dismutase1 and Focal Axonal Abnormalities in the ProximalAxons of Transgenic Mice

David R. Borchelt,*,1 Philip C. Wong,*,1 Mark W. Becher,*,1

Carlos A. Pardo,*,† Michael K. Lee,* Zuo-Shang Xu,‡

Gopal Thinakaran,* Nancy A. Jenkins,§ Neal G. Copeland,§

Sangram S. Sisodia,*,¶ Don W. Cleveland,00 Donald L. Price,*,†,¶

and Paul N. Hoffman†,¶,***Department of Pathology, **Department of Ophthalmology, ¶Department of Neuroscience, and†Department of Neurology, The Johns Hopkins School of Medicine, Baltimore, Maryland 21205;‡Worchester Foundation for Experimental Biology, Shrewsbury, Massachusetts 01545;§The Mammalian Genetics Laboratory, The National Cancer Institute, Frederick, Maryland;and 00Department of Medicine and Department of Neuroscience, Ludwig Institute,University of California, San Diego, La Jolla, California 92093

Received October 13, 1997; revised January 30, 1998; accepted for publication February 20, 1998

Superoxide dismutase 1 (SOD1), a ubiquitously expressed enzyme, detoxifies superoxide radicals andparticipates in copper homeostasis. Mutations in this enzyme have been linked to a subset ofautosomal dominant cases of familial amyotrophic lateral sclerosis (FALS), a disorder characterized byselective degeneration of motor neurons. Transgenic mice expressing FALS mutant human (Hu) SOD1at high levels develop a motor neuron disease, indicating that mutant Hu SOD1 gains properties thatare particularly toxic to motor neurons. In this report, we demonstrate that transgenic mice expressingHu SOD1 with the G37R FALS mutation, but not mice expressing wild-type enzyme, develop focalincreases in immunoreactivity in the proximal axons of spinal motor neurons. This SOD1 immunoreac-tivity and immunoreactivity to hypophosphorylated neurofilament H epitopes are found adjacent tosmall vacuoles in axons. Using metabolic radiolabeling methods, we show that mutant G37R HuSOD1as well as endogenous mouse SOD1 are transported anterograde in slow component b in motor andsensory axons of the sciatic nerve. Together, these findings suggest that anterogradely transportedmutant SOD1 may act locally to damage motor axons. r 1998 Academic Press

Key Words: axonal transport; superoxide dismutase 1; slow component b; familial amyotrophiclateral sclerosis; mutant.

INTRODUCTION

Amyotrophic lateral sclerosis (ALS) manifests asweakness and muscular atrophy due to the selectivedegeneration of motor neurons. Recent studies havelinked ,20% of familial, autosomal dominant ALS(FALS) to mutations in Cu/Zn superoxide dismutase 1(SOD1). Highly abundant in all tissues, including thenervous system (Tsuda et al., 1994; Pardo et al., 1995;Wong et al., 1995), SOD1 catalyzes the dismutation of

superoxide anions to hydrogen peroxide and oxygenvia electron exchange with tightly bound Cu21 (Fridov-ich, 1986).

The seemingly critical function of SOD1 led investi-gators to speculate initially that FALS-linked muta-tions may compromise SOD1 function, leading to·O2-mediated damage of motor neurons (Deng et al.,1993). Although some FALS mutations reduce theactivity of SOD1, many mutant enzymes possess near-normal levels of enzyme activity/stability (Borchelt etal., 1994b). Moreover, mutant SOD1 subunits do notappear to alter the metabolism/activity of wild-type1These authors made equal contributions.

Neurobiology of Disease 5, 27–35 (1998)

Article No. NB980178

27

0969-9961/98 $25.00Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

SOD1 in a dominant negative fashion (Borchelt et al.,1995). Thus, heterozygous SOD1-linked FALS individu-als would be predicted to possess no less than 50% ofthe normal levels of SOD1 activities. Finally, althoughfacial motor neurons in SOD1 null mice are susceptibleto retrograde degeneration after axotomy, these ani-mals do not develop a FALS-like disease (Reaume etal., 1996). Thus, there is little evidence to substantiatethe notion that MND arises in SOD1-linked FALS casesas the sole result of reduced superoxide scavengingactivity.

Studies in transgenic mice have demonstrated thatenzymatically active G37R and G93A FALS-mutanthuman (Hu) SOD1 (but not wild-type enzyme) inducemotor neuron disease (Gurney et al., 1994; Wong et al.,1995). In animals expressing the mutant protein, levelsof SOD1 activity increase, leading to the conclusionthat mutant enzymes cause disease as a result ofgaining a toxic property. In mice expressing thesemutant SOD1, vacuolar pathology in the axons, den-drites, and cell bodies of motor neurons precedesaxonal degeneration, muscle atrophy, and motor neu-ron death (Wong et al., 1995; Dal Canto & Gurney,1994).

In the present study, we examined the distributionsof SOD1 polypeptides in the lumbar spinal cords ofpresymptomatic G37R Hu SOD1 transgenic mice (2–3months prior to clinical weakness). In multiple lines ofmice, the proximal axons of motor neurons were foundto contain intense, focal SOD1 and nonphosphorylatedNF-H immunoreactivity. These focal concentrations ofantigen, which colocalized with small vacuoles that wehave previously described in G37R mice (Wong et al.,1995), were absent in transgenic mice expressing highlevels of wild-type Hu SOD1 and nontransgenic mice.Using metabolic radiolabeling methods, we demon-strated that mutant Hu SOD1 and endogenous mouseenzyme are transported anterogradely in axons as partof the slow component b (SC b). Together, thesefindings demonstrate that mutant SOD1 is antero-gradely transported into axons and that these axonsshow signs of damage, consisting of focal increases inSOD1 and hypophosphorylated NF-H immunoreactiv-ity adjacent to small vacuoles, very early in the courseof disease.

MATERIALS AND METHODS

Transgenic Mice

All transgenic mice described in this study weremaintained as heterozygotes in hybrid strains of C3H/

HeJ and C57BL/6J. The G37R Hu SOD1 mutant trans-genic mouse lines (line 42 and 106) and the wild-typeHu SOD1 transgenic mouse line (line 76) have beendescribed previously (Wong et al., 1995). Transgenicanimals were identified at 4 weeks of age by polymer-ase chain reaction amplification of a defined segmentof the transgene as previously described (Wong et al.,1995).

Histopathological Studies

Mice were deeply anesthetized with intraperitonealinjections of 4.0% chloral hydrate or sodium pentobar-bital and perfused transcardially with 0.1 M PBS (pH7.4) followed by ice-cold 4.0% paraformaldehyde in 0.1M PBS (pH 7.4) for 10–15 min. Brains, spinal cords, andother organs were removed and postfixed overnight inthe same fixative, rinsed with 0.1 M PBS, and embed-ded in paraffin. Sections (10 µm) were deparaffinizedand stained with hematoxylin and eosin, cresyl violet,and Bielschowski silver. Immunocytochemical proce-dures were applied to additional sections using thefollowing well-characterized antibodies: Hu/MoSOD1(Borchelt et al., 1994b; Pardo et al., 1995), glial fibrillaryacidic protein (Dako, Carpenteria, CA), phosphory-lated NF-H (SMI-31; Sternberger Monoclonals, Inc,Baltimore, MD), nonphosphorylated NF-H (SMI-32;Sternberger Monoclonals), ubiquitin (Chemicon, Tem-ecula, CA), and microtubule-associated protein 2(MAP2; Sigma Chemical Company, St. Louis, MO).Standard peroxidase-anti-immunoglobulin procedureswith a light Mayer’s hematoxylin counterstain wereused with the following minor modifications: metha-nol/3.0% hydrogen peroxide (1/5) for 30 min, thor-ough rinse, and microwave to near boiling for 6–8 min(not used for SMI-31) prior to a blocking step withnormal serum. Sections were exposed to primaryantibodies overnight (16 h) at room temperature in ahumidity chamber incubated with secondary antibod-ies, then reacted with diaminobenzidine and preparedfor analysis as previously described (Pardo et al., 1995).

Immunoprecipitation of Radiolabeled SOD1 fromDorsal Root Ganglia (DRG) and Lumbar MotorNeurons

Proteins transported in motor and sensory axons ofthe sciatic nerve in young presymptomatic (2-month-old) nontransgenic and transgenic G37R (106) HuSOD1 mice (Wong et al., 1995) were labeled by injecting2–4 µl of [35S]methionine or [35S]cysteine (200 µCi/µl)into the L5 DRG or ventral horn of the lumbar spinalcord, respectively, following previously established

28 Borchelt et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

protocols (Hoffman et al., 1975). Similar injections of[35S]methionine into the L5 DRG of 2-month-old ratswere used to examine the rates of transport. To analyzepulse-labeled proteins transported in the slow compo-nents of axonal transport, animals were sacrificed 3 to7 days after labeling. To examine fast transport, aligature was placed on the sciatic nerve 2 cm from theDRG immediately following the injection of [35S]me-thionine. Eight hours after injection, the nerve wasdissected and cut into three segments: the DRG, thenerve between the DRG and 0.5 cm proximal to theligature, and a segment in proximity to the ligature.

Using a Hu/MoSOD1 antiserum raised against asynthetic peptide of identical sequence in Hu andmouse (Mo) SOD1 (Borchelt et al., 1994b, 1995), SOD1polypeptides were immunoprecipitated from nervesegments (0.5 cm) after homogenization in 200 µl of 50mM Tris (pH 8.0), 150 mM NaCl, 5 mM EDTA, 0.5%NP-40, 0.5% deoxycholate, and 0.2% sodium dodecylsulfate (SDS) (IP buffer). Extracts were boiled for 5 minand then centrifuged for 5 min at 10,000g. Aftercooling, 2 µl of the Hu/MoSOD1 peptide antiserumwas added to supernatants and incubated overnight at4°C. Antibody/SOD1 complexes were then absorbedto protein A agarose (Pierce Chemical Company, Rock-ford, IL), washed in IP buffer for 30 min (4°C),transferred to a clean tube, and washed twice in IPbuffer before analysis by SDS–polyacrylamide gelelectrophoresis (PAGE) (Laemmli, 1970) and autoradi-ography (Amplify, Amersham, Inc., Arlington Heights,IL). The NF proteins (NF-H, -M, and -L) and tubulinand actin (recognized by characteristic size and abun-dance) served as markers for SC a and SC b, respec-tively. As a marker for fast transport, the amyloidprecursor protein (APP) was immunoprecipitated fromextracts with the polyclonal antiserum CT15, as previ-ously described (Borchelt et al., 1994a).

RESULTS

Axonal Abnormalities Occur Very Early in G37RHuSOD1 Transgenic Mice

Using a previously described polyclonal antiserum(anti-Hu/Mo SOD1) (Borchelt et al., 1994b; Pardo et al.,1995), the distribution of SOD1 immunoreactivity wasexamined in preparations of lumbar spinal cord fromG37R mice (line 42) (Wong et al., 1995) (age 5–7weeks—clinical weakness appears in these animals at4–5 months of age). In these tissues, abbreviatedreactions of immunocomplexed horseradish peroxi-dase with diaminobenzidine substrates revealed focal

SOD1 immunoreactivity in proximal motor axons inthe root exit zone of the ventral horn (Figs. 1A–1C).Rarely, neuronal cell bodies also showed intense SOD1immunoreactivity under these staining conditions (Fig.1B, thick arrow). Under higher magnification, theSOD1 immunoreactivity in the proximal axons ofmotor neurons appeared to be localized around smallvacuoles (Fig. 1B, inset). Although we cannot followthese structures through time in a single animal, wepresume that these small vacuoles are progenitors ofthe larger vacuoles that are one of the primary patho-logical changes in both proximal axons and cell bodiesof older clinically weak G37R mice (Fig. 1F) (Wong etal., 1995). In contrast to the older animals (Fig. 1F),silver stains of the presymptomatic animals werelargely unremarkable except for occasional small clus-ters of axonal vacuoles (Fig. 1E, arrow).

In contrast to mice expressing G37R Hu SOD1, wecould not demonstrate focal axonal SOD1 immunoreac-tivity (under identical conditions used in stainingmutant mice) in age-matched animals expressing highlevels of wild-type Hu SOD1 (line 76) (Wong et al.,1995) (Fig. 1D) or nontransgenic littermates (notshown). Notably, under these conditions, the expectedpattern of cell body immunostaining was absent (Pardoet al., 1995), suggesting that the focal SOD1 immunore-activity in the axons of the mutant transgenic micemay contain greater quantities of antigen than nor-mally found in cell bodies. When peroxidase reactionswere allowed to develop longer than the imagesshown in Fig. 1, cytoplasmic SOD1 immunoreactivitybecame detectable (Fig. 2). Interestingly, the mostsignificant increases in SOD1 immunoreactivity in thespinal cords of wild-type Hu SOD1 transgenic mice(presumably reflecting the expression of the transgene)occurred in the neuropil and white matter tracts (Fig.2A) compared to nontransgenic animals (Fig. 2B).Notably, even under these conditions, the focal axonalSOD1 immunoreactivity seen in the mutant transgenicmice was not observed in these control animals (Fig. 2).

To determine whether mutant mice showed focalincreases in other antigens, tissues were immuno-stained with antisera recognizing ubiquitin (a markerof protein degradation that is normally not seen inneuronal cytoplasm or processes), phosphorylatedNF-H (normally present in axons), nonphosphorylatedNF-H (normally present in cell bodies but not axons),and MAP2 (normally present in dendrites). Of theseantisera, only antisera to nonphosphorylated NF-H(SMI-32) showed focal increases adjacent to the smallvacuoles in proximal axons (Figs. 1G and 1H, arrows)and neuropil (Fig. 1H, arrowhead and thick arrow).

Axonal Transport of Mutant SOD1 and Early Pathology 29

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

30

With this information on the pattern of SOD1 andnonphosphorylated NF-H immunoreactivity in G37R(line 42) mice, we examined tissues from severaldifferent lines of G37R mice. In both G37R (line 106)and G37R (line 50) mice [which express five and two to

three times the endogenous level of enzyme and whichdevelop disease at 6–8 and 10–12 months, respectively(Wong et al., 1995)], proximal axons showed focalincreases in SOD1 and nonphosphorylated NF-H im-munoreactivity around small vacuoles. Like the line 42

FIG. 1. SOD1 and nonphosphorylated NF-H immunoreactivity discloses early vacuolar changes in motor axons. Immunocytochemicalpreparations from 5- and 7-week-old (G37R–line 42) mice were stained with Hu/MoSOD1 antibody (A–C), showing prominent vacuolarchanges in the ventral motor axon exit zones of lumbar spinal cord (thin arrows) from 5- (A and B) and 7- (C) week-old animals. Althoughoccasional neuronal cell bodies showed increased SOD1 immunoreactivity (B, thick arrow), the majority of motor neurons in the ventral horns (Aand C, asterisks) were largely free of focal increases in SOD1 immunoreactivity. High magnification (inset B) shows the SOD1 antibodyimmunoprecipitate adjacent to and surrounding axonal vacuoles. Contemporaneous staining of mice expressing high levels of wild-type HuSOD1 (lines 76) were unremarkable (D). Conventional silver stains of young (5-week) animals (E) appear normal, except for very subtle (arrow)distortions. In contrast, 18-week-old animals (F) have extensive vacuolar changes in ventral root exit zones (arrows) and ventral horn neuropil(asterisk). Nonphosphorylated NF-H (SMI-32) antibody (G and H) colocalizes with small vacuoles (H, arrowhead) in dendrites, cell bodies (H,thick arrow), and proximal ventral root axons (G and H, thin arrows). (A–D) SOD1 with hematoxylin counterstain, (E and F) Bielschowski silver,(G and H) SMI-32 with hematoxylin counterstain. Bars: A, C, and G, 50 µm; B, D–F, and H, 25 µm; inset B, 10 µm.

FIG. 2. SOD1 is distributed uniformly in spinal cord tissues from transgenic mice expressing high levels of wild-type Hu SOD1.Immunocytochemical preparations from 2-month-old transgenic mice expressing wild-type Hu SOD1 (line 76) stained with Hu/MoSOD1antibody (A) revealed the pattern of Hu SOD1 distribution in spinal cord. In these animals, the overall levels of neuropil and axonal tractstaining (lower right) is disproportionately increased and overshadows the increase in immunoreactivity in the cell bodies and axons comparedto nontransgenic littermates (B). Vacuoles were absent in both preparations.

Axonal Transport of Mutant SOD1 and Early Pathology 31

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

G37R mice, these structures began to appear 2–3months prior to the onset of clinical weakness (datanot shown). In our hands, mice expressing wild-typeSOD1 (n 5 20) show normal life spans with nosigns of motor neuron loss or muscle weakness. Morethan 10 wild-type transgenic animals have been ana-lyzed neuropathologically at various ages and in nocase have we observed vacuolar pathology or the focalincreases in SOD1 and nonphosphorylated NF-H im-munoreactivity described here in the G37R mice. Col-lectively, our data indicate that the distribution ofSOD1 and nonphosphorylated NF-H epitopes is al-tered very early in the course of disease in the G37Rmice.

SOD1 Is Axonally Transported in SC b

Previous investigations (Pardo et al., 1995) as well asimmunocytochemical studies presented here indicatethat SOD1 is an axonal protein. To determine the rateof SOD1 transport into axons, we chose initially toexamine the anterograde movement of radiolabeledpolypeptide in sensory neurons of 2-month-old ratsbecause the rates of axonal transport have been bestcharacterized in these animals (Hoffman et al., 1975,1982, 1985, 1992). The L5 DRG was injected with[35S]methionine and 3 days later the sensory axonswere harvested. Immunoprecipitation demonstratedradiolabeled wild-type rat SOD1 in segments ,2 cmfrom the DRG (Fig. 3, top); a distance slightly fartherthan labeled NF proteins (NF-H, -M, and -L) (Fig. 3,bottom), which are transported exclusively in SC a andcoincident with components of SC b (i.e., tubulin andactin) (Hoffman et al., 1975, 1982, 1985, 1992).

To ask whether SOD1 may also be transported in thefast component as the result of an association withmembranous organelles, we injected the L5 DRG of thesciatic nerve of young mice with [35S]methionine,ligated the nerve approximately 2 cm distal to theDRG, and then, after an 8-h interval, harvested thenerve by cutting it into 0.5-cm segments. Althoughpulse-labeled SOD1 could be immunoprecipitated fromthe DRG, the protein had not yet migrated into thenerve, an observation indicating that SOD1 is notrapidly transported (Fig. 4). As a positive control forrapidly transported proteins, APP was also immunopre-cipitated from the same extracts. Nerve segmentsproximal to the ligature contained labeled APP, asexpected for this rapidly transported protein (Koo etal., 1990; Sisodia et al., 1993; Borchelt et al., 1994a).

Axonal Transport of Hu SOD1 in G37R Hu SOD1Transgenic Mice

To examine the axonal transport of G37R Hu SOD1proteins, we injected [35S]cysteine into the ventral hornof the lumbar spinal cord of G37R (line 106) mice.Cysteine was used because the N-terminal methionineof Hu SOD1 is removed posttranslationally, leavingthe polypeptide with no other methionine. Labeledprotein was immunoprecipitated from distal nervesegments 7 days after injection [the rate of slow

FIG. 3. SOD1 is transported in the SC b of sensory neurons. TheL4/L5 dorsal root ganglion of 2-month-old rats was injected with[35S]methionine, and SOD1 was immunoprecipitated (top) as de-scribed under Materials and Methods. One-twentieth of the totalextract was analyzed by SDS–PAGE (bottom) to examine thetransport of SOD1 relative to other axonal constituents. The NFproteins, tubulin, and actin were identified on the basis of abun-dance, size, and comparison with previous experience with slowlytransported proteins (Hoffman et al., 1975, 1982, 1985, 1992). ACoomassie blue stain of total extracts reveals a prominent 30-kDaprotein in both DRG and nerve segments (position marked by anasterisk); this species does not incorporate radiolabel and is notdetected by autoradiography.

32 Borchelt et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

transport ranges between 1 and 6 mm per day (Lasek etal., 1976; Vallee et al., 1991) ]. Immunoprecipitation ofradiolabeled polypeptides with the Hu/MoSOD1 pep-tide antiserum followed by SDS–PAGE analysis re-vealed that G37R Hu SOD1 was anterogradely trans-ported, apparently at a rate identical to that ofendogenous mouse SOD1 polypeptide (Fig. 5) [humanand murine SOD1 polypeptide can be resolved bySDS–PAGE (Wong et al., 1995; Borchelt et al., 1995;Elroy-Stein et al., 1987; Epstein et al., 1987)].

DISCUSSION

Our investigations of slow axonal transport in young(2-month-old) presymptomatic G37R Hu SOD1 miceindicate that this mutant protein is transported antero-gradely in SC b in both sensory and motor axons.Although previous immunocytochemical investiga-tions of nontransgenic mice had identified SOD1 epi-topes in axons (Pardo et al., 1995), it is importantto characterize the rate of axonal transport becausethe biology of proteins transported axonally in theslow component is much different than that of rapidlytransported proteins. Proteins transported in thefast component are generally associated with membra-nous organelles, whereas proteins transported inSC b are usually soluble cytosolic proteins or cyto-skeletal proteins (Brady et al., 1981; Vallee et al., 1991).Our data indicate that SOD1, in axons, does not

associate with rapidly transported vesicles to any greatextent.

Several lines of evidence indicate that the radiola-beled SOD1 that was immunoprecipitated from distalnerve segments was not from nonneuronal cells sur-rounding the axon (which may have acquired radiola-beled amino acids by diffusion), but rather was de-rived from actively transported protein. First,radiolabeled neuronal cytoskeletal proteins were themajor constituents of total nerve extracts, indicatingselective transport of subsets of proteins (see Fig. 3B).Second, labeled SOD1 could not be detected in nervesegments 8 h after label injection, an ample time fordiffusion to distribute radiolabeled amino acids tononneuronal cells. Finally, these new findings areentirely consistent with previous electron microscopystudies, which demonstrated SOD1 immunoreactivityin distal axons (Pardo et al., 1995). Thus, we concludethat both sensory and motor axons transport SOD1anterogradely in SC b.

FIG. 4. Endogenous Mo SOD1 in sensory axons is not transportedin the fast component. The L4/L5 dorsal root ganglion of 2-month-old mice was injected with [35S]methionine and SOD1 was immuno-precipitated as described under Materials and Methods. Unlikehuman SOD1, murine SOD1 possesses an internal methionineresidue and is labeled by radioactive methionine. As a positivecontrol for rapidly transported protein, APP was immunoprecipi-tated from nerve extracts.

FIG. 5. SOD1 harboring the FALS G37R mutation is axonallytransported in motor axons. Lumbar spinal cords of 2-month-oldG37R (line 106) mice were injected with [35S]cysteine, and SOD1polypeptides were extracted and immunoprecipitated as describedunder Materials and Methods. The section most proximal to thespinal cord (lane 1) was ,2 cm from the injection site.

Axonal Transport of Mutant SOD1 and Early Pathology 33

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

The amount of radiolabeled mutant Hu SOD1 in theaxons of G37R (106) mice appeared to be ,1.5- to2-fold greater than the amount of endogenous protein.However, Hu SOD1 possesses three cysteine residues,whereas endogenous Mo SOD1 possesses only two;thus the levels of each transported protein are virtuallyequal. Interestingly, previous immunoblot quantifica-tion of SOD1 levels in the spinal cords of G37R (line106) Hu SOD1 mice demonstrated that the level ofmutant HuSOD1 is 5-fold greater than that of endog-enous MoSOD1 (Wong et al., 1995). However, ourimmunocytochemical analyses of mice expressing wild-type Hu SOD1 demonstrated a significant increase inneuropil and axonal tract staining with less robustincreases in the staining of cell bodies. Thus, weconclude that in these transgenic animals a significantportion of the added Hu SOD1 polypeptides may notbe expressed in motor neurons, explaining why thelevel of mutant labeled protein in axons is onlyequivalent to the level of endogenous SOD1 polypep-tides.

In cultured cells, the half-life of mutant enzyme isshorter than that of wild-type enzyme (t1/2 5 13 h forG37R Hu SOD1 vs 30 h for wild-type) (Borchelt et al.,1994b). If the rate of synthesis of mutant enzyme inmotor neurons is fivefold greater than that of wild-type enzyme [as predicted by immunoblot analyses ofprotein SOD1 polypeptide levels in spinal cords (Wonget al., 1995)] and if the half-life measurements incultured cells accurately reflect the in vivo half-life ofthe mutant enzymes during axonal transport, then by 7days postlabeling the radiolabeled endogenous pro-tein should be 50 times more abundant in the axonthan the mutant protein. Clearly, our data do notconfirm these estimates. Thus, SOD1, particularly mu-tant enzyme, that is being axonally transported in SC bis much longer lived than estimated from cell cultureexperiments. Whether the motor axon is less activelyengaged in degrading proteins than the cell body, orother types of cells, is unknown. If mutant SOD1 wereselectively longer lived in motor axons compared toother tissues, then one might be able to explain whyneurons are more vulnerable to the toxic activity of themutant.

SOD1 and Nonphosphorylated NF-H Accumulatein Proximal Motor Axons Early in the Disease

Our immunocytochemical investigations of multiplelines of mice expressing G37R Hu SOD1 demonstratefocal increases in SOD1 and nonphosphorylated NF-Himmunoreactivity in the spinal cords of presymptom-

atic animals. Most of the immunoreactivity appearedto surround small vacuoles that we believe are theprogenitors of larger vacuoles that we previouslydescribed in older mice expressing G37R Hu SOD1(Wong et al., 1995). Although we cannot be certain thatall of the focal SOD1 immunoreactivity found in theproximal axons of young G37R mice is mutant polypep-tide, because our antisera recognize Hu and Mo SOD1with equal avidity (Pardo et al., 1995), our transportstudies argue that the mutant protein is present inaxons. In mice expressing Hu SOD1 harboring theG85R FALS variant, neuronal and glial inclusionscontain both the mutant Hu enzyme and the endog-enous mouse enzyme (Bruijn et al., 1997). Thus it ispossible that endogenous enzyme is recruited into theSOD1-immunoreactive structures observed in our G37Rmice.

The appearance of nonphosphorylated NF-H immu-noreactivity in the axons of the G37R mice was unex-pected, because the majority of axonal NF-H is nor-mally phosphorylated (Lee et al., 1996). Thus, themetabolism/phosphorylation of neurofilament pro-teins may be impaired early in the disease. Whetherthis defect occurs in all FALS cases is unclear, becausemice expressing the G85R variant of Hu SOD1 do notshow axonal vacuoles or axonal accumulations ofSOD1 or nonphosphorylated NF-H immunoreactivityat any stage of disease (Bruijn et al., 1997). Whetherother more subtle changes in axonal integrity occur inthese animals is unknown. In the G37R mice, however,our transport and immunocytochemical studies pro-vide evidence of motor axon injury, which may becaused locally by mutant SOD1. On the basis of thesefindings, it appears likely that axonal injury is in-volved in the pathogenesis of disease in at least asubset of SOD1-linked FALS cases.

ACKNOWLEDGMENTS

The authors thank Drs. Jeffrey Rothstein, Lucie Bruijn, and ValeriaCulotta for helpful discussions and Ms. Marilyn Peper for excellenttechnical assistance. This work was supported by grants from theNIH: NS 27036 (DWC, Z-SX), NS 32724 (PNH), NS 01719, AG 05146,and NS 20471. We also acknowledge grants from the AmyotrophicLateral Sclerosis Association (D.R.B., P.C.W., M.K.L.), the Metropoli-tan Life Foundation, and the Cal Ripken/Lou Gehrig Fund for ALSResearch. Drs. Price, Borchelt, and Wong are the recipients of aLeadership and Excellence in Alzheimer’s Disease (LEAD) awardfrom the NIH (AG 07914); Dr. Price is the recipient of a JavitsNeuroscience Investigator Award from the NIH (NS 10580).

34 Borchelt et al.

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.

REFERENCES

Borchelt, D. R., Koliatsos, V. E., Guarnieri, M., Pardo, C. A., Sisodia,S. S., & Price, D. L. (1994a) Rapid anterograde axonal transport ofthe cellular prion glycoprotein in the peripheral and centralnervous systems. J. Biol. Chem. 269, 14711–14714.

Borchelt, D. R., Lee, M. K., Slunt, H. H., Guarnieri, M., Xu, Z.-S.,Wong, P. C., Brown, R. H., Jr., Price, D. L., Sisodia, S. S., &Cleveland, D. W. (1994b) Superoxide dismutase 1 with mutationslinked to familial amyotrophic lateral sclerosis possesses signifi-cant activity. Proc. Natl. Acad. Sci. USA 91, 8292–8296.

Borchelt, D. R., Guarnieri, M., Wong, P. C., Lee, M. K., Slunt, H. S.,Xu, Z.-S., Sisodia, S. S., Price, D. L., & Cleveland, D. W. (1995)Superoxide dismutase 1 subunits with mutations linked to famil-ial amyotrophic lateral sclerosis do not affect wild-type subunitfunction. J. Biol. Chem. 270, 3234–3238.

Brady, S. T., & Lasek, R. J. (1981) Nerve-specific enolase and creatinephosphokinase in axonal transport: Soluble proteins and theaxoplasmic matrix. Cell 23, 515–523.

Bruijn, L. I., Becher, M. W., Lee, M. K., Anderson, K. L., Jenkins, N. A.,Copeland, N. G., Sisodia, S. S., Rothstein, J. D., Borchelt, D. R.,Price, D. L., & Cleveland, D. W. (1997) ALS-linked SOD1 mutantG85R mediates damage to astrocytes and promotes rapidly pro-gressive disease with SOD1-containing inclusions. Neuron 18,327–338.

Dal Canto, M. C., & Gurney, M. E. (1994) Development of centralnervous system pathology in a murine transgenic model of humanamyotrophic lateral sclerosis. Am. J. Pathol. 145, 1271–1280.

Deng, H.-X., Hentati, A., Tainer, J. A., Iqbal, Z., Cayabyab, A., Hung,W.-Y., Getzoff, E. D., Hu, P., Herzfeldt, B., Roos, R. P., Warner, C.,Deng, G., Soriano, E., Smyth, C., Parge, H. E., Ahmed, A., Roses,A. D., Hallewell, R. A., Pericak-Vance, M. A., & Siddique, T. (1993)Amyotrophic lateral sclerosis and structural defects in Cu,Znsuperoxide dismutase. Science 261, 1047–1051.

Elroy-Stein, O., Bernstein, Y., Avraham, K. B., Dafni, N., Levanon, D.,Danciger, E., Neer, A., & Groner, Y. (1987) Overexpression of thehuman CuZnSOD gene in transfected cells—Implications to Downsyndrome. In: Model Systems in Neurotoxicology: Alternative Ap-proaches to Animal Testing (A. M. Goldberg and A. Shahar, Eds.), pp373–392. A. R. Liss, New York.

Epstein, C. J., Avraham, K. B., Lovett, M., Smith, S., Elroy-Stein, O.,Rotman, G., Bry, C., & Groner, Y. (1987) Transgenic mice withincreased Cu/Zn-superoxide dismutase activity: Animal model ofdosage effects in Down syndrome. Proc. Natl. Acad. Sci. USA 84,8044–8048.

Fridovich, I. (1986) Superoxide dismutases. Adv. Enzymol. Relat.Areas Mol. Biol. 58, 61–97.

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) Motorneuron degeneration in mice that express a human Cu,Zn superox-ide dismutase mutation. Science 264, 1772–1775.

Hoffman, P. N., & Lasek, R. J. (1975) The slow component of axonaltransport: Identification of major structural polypeptides of the

axon and their generality among mammalian neurons. J. Cell Biol.66, 351–366.

Hoffman, P. N., Griffin, J. W., & Price, D. L. (1982) The transport ofactin and tubulin and the rate of axonal regeneration. J. Neuro-pathol. Exp. Neurol. 41, 370.

Hoffman, P. N., Thompson, G. W., Griffin, J. W., & Price, D. L. (1985)Changes in neurofilament transport coincide temporally withalterations in the caliber of axons in regenerating motor fibers. J.Cell Biol. 101, 1332–1340.

Hoffman, P. N., Lopata, M. A., Watson, D. F., & Luduena, R. F. (1992)Axonal transport of class II and III b-tubulin: Evidence that theslow component wave represents the movement of only a smallfraction of the tubulin in mature motor axons. J. Cell Biol. 119,595–604.

Koo, E. H., Sisodia, S. S., Archer, D. R., Martin, L. J., Weidemann, A.,Beyreuther, K., Fischer, P., Masters, C. L., & Price, D. L. (1990)Precursor of amyloid protein in Alzheimer disease undergoes fastanterograde axonal transport. Proc. Natl. Acad. Sci. USA 87,1561–1565.

Laemmli, U. K. (1970) Cleavage of structural proteins during theassembly of the head of bacteriophage T4. Nature (London) 227,680–685.

Lasek, R. J., & Hoffman, P. N. (1976) The neuronal cytoskeleton,axonal transport and axonal growth. Cold Spring Harbor Conf. CellProlif. 3, 1021–1049.

Lee, M. K., & Cleveland, D. W. (1996) Neuronal intermediatefilaments. Annu. Rev. Neurosci. 19, 187–217.

Pardo, C. A., Xu, Z., Borchelt, D. R., Price, D. L., Sisodia, S. S., &Cleveland, D. W. (1995) Superoxide dismutase is an abundantcomponent in cell bodies, dendrites, and axons of motor neuronsand in a subset of other neurons. Proc. Natl. Acad. Sci. USA 92,954–958.

Reaume, A. G., Elliott, J. L., Hoffman, E. K., Kowall, N. W., Ferrante,R. J., Siwek, D. F., Wilcox, H. M., Flood, D. G., Beal, M. F., Brown,R. H., Jr., Scott, R. W., & Snider, W. D. (1996) Motor neurons inCu/Zn superoxide dismutase-deficient mice develop normallybut exhibit enhanced cell death after axonal injury. Nature Genet.13, 43–47.

Sisodia, S. S., Koo, E. H., Hoffman, P. N., Perry, G., & Price, D. L.(1993) Identification and transport of full-length amyloid precur-sor proteins in rat peripheral nervous system. J. Neurosci. 13,3136–3142.

Tsuda, T., Munthasser, S., Fraser, P. E., Percy, M. E., Rainero, I., Vaula,G., Pinessi, L., Bergamini, L., Vignocchi, G., McLachlan, D. R. C.,Tatton, W. G., & St George-Hyslop, P. (1994) Analysis of thefunctional effects of a mutation in SOD1 associated with familialamyotrophic lateral sclerosis. Neuron 13, 727–736.

Vallee, R. B., & Bloom, G. S. (1991) Mechanisms of fast and slowaxonal transport. Annu. Rev. Neurosci. 14, 59–92.

Wong, P. C., Pardo, C. A., Borchelt, D. R., Lee, M. K., Copeland, N. G.,Jenkins, N. A., Sisodia, S. S., Cleveland, D. W., & Price, D. L. (1995)An adverse property of a familial ALS-linked SOD1 mutationcauses motor neuron disease characterized by vacuolar degenera-tion of mitochondria. Neuron 14, 1105–1116.

Axonal Transport of Mutant SOD1 and Early Pathology 35

Copyright r 1998 by Academic PressAll rights of reproduction in any form reserved.