Nature Publishing Group

Altered axonal transport in ALS. Another charac-

teristic of ALS is the reduced activity of axonal trans-

port, described first in patients with ALS

116

 and more 

recently in transgenic mouse models of ALS

117

 

(FIG. 4)

. 

The transport of molecules and organelles is a fun-

damental cellular process that is particularly impor-

tant for the development, function and survival of 

neurons. This process is dictated by the highly polarized 

anatomy of neurons: axonal proteins are synthesized in 

the cell body and must be transported in an anterograde 

manner along the axons and dendrites to reach synapses, 

whereas substances such as peripherally located trophic 

factors must be transported centrally from the synaptic 

regions by retrograde transport. The molecular motors 

for anterograde and retrograde transport are 

kinesin

 and 

the dynein–dynactin complex, respectively. Transport 

is conventionally regarded as either slow or fast. 

Presumably, functional and efficient axonal transport 

is particularly important for motor neurons, which are 

among the largest and longest cells in the body.

Several findings indicate that defects in axonal trans-

port might contribute to the demise of motor neurons in 

ALS. First, both slow and fast anterograde transport are 

slowed in transgenic G93A-SOD1 and G37R ALS mice 

prior to disease onset. These deficits are exacerbated 

Figure 4 | Axonal transport is abnormal in ALS. The motors for anterograde and 

retrograde fast axonal transport are the kinesins and dynactin complex proteins, 

respectively; microtubules provide the tracks for these motors. Vesicles for transport are 

sorted and loaded onto transport motors both in the cell body and the distal nerve 

terminal. The former are transported not only into the axon but also into dendrites. Those 

in the distal nerve terminal permit uptake and axosomatic movement of substances such 

as trophic proteins. Mutations in dynactin (humans), dynein (mice) and three different 

forms of kinesin all provoke motor neuron degeneration. Perturbation of neurofilaments 

through mutations and changes in phosphorylation and the physical structure of the 

axon could also adversely affect axonal transport. In transgenic mouse models, mutant 

superoxide dismutase 1 (SOD1) impairs anterograde axonal transport. ER, endoplasmic 

reticulum.

R E V I E W S

716 

|

 SEPTEMBER 2006 

|

 VOLUME 7 

 

www.nature.com/reviews/neuro

©

 

200

6

 

Nature Publishing Group

 

 

Table 2 | Neuronal genes with defects that impair motor neuronal function or viability

Disease

Gene (gene symbol)

Inheritance

Class

Refs

ALS

ALS

Superoxide dismutase 1 (SOD1)

AD

Mainly cytosolic enzyme; also in mitochondrion

11

ALS

VAMP-associated protein B (VAPB)

AD

Golgi/vesicle transport and sorting factor

16

ALS-like

Slow ALS

Senataxin (SETX)

AD

DNA/RNA metabolism

15

jALS

Alsin (ALS2)

AR

GEF signalling

12,13

Slow ALS

Dynactin (DCTN)

AR

Transport motor

17

Peripheral neuropathy

CMT1F/2E

Neurofilament light chain (NEFL)

AD

Cytoskeletal filament

154

CMT2A1

Kinesin K1F1B

β (K1F1Bβ)

AD

Transport motor 

157

CMT2A2

Mitofusin (MFN2)

AD

Mitochondrial protein

188

CMT2B

RAB7 (RAB7)

AD

Golgi/vesicle transport and sorting factor

160

CMT2D

Glycyl-tRNA synthetase (GARS)

AD

DNA/RNA metabolism

162

CMT2F

Heat shock protein 27 (HSP27)

AD

Heat shock protein

166

CMT2L, scapuloperoneal 

neuropathy

Heat shock protein 22 (HSP22)

AD

Heat shock protein

167

DI-CMT 2B

Dynamin (DNM2)

AD

GTPase

189

DI-CMT, type C

Tyrosyl-tRNA synthetase (YARS)

AD

DNA/RNA metabolism

163

Hereditary sensory motor 

neuropathy

Rho GEF 10 (ARHGF10)

AD

GEF signalling

190

HSN1

Serine palmitoyl transferase (SPTLC1)

AD

Enzyme

191

Porphyria

Porphobilinogen deaminase (PBGD)

AD

Enzyme

192

AR-CMT2A (and other 

phenotypes)

Lamin A/C (LMNA)

AD, AR

Nuclear membrane protein

193,194

Andermann syndrome

Potassium chloride co-transporter (KCC3)

AR

Cation-chloride cotransporter

195

Giant axonal neuropathy

Gigaxonin (GAN)

AR

Cytoskeletal filament

155,156

HSN4

Tyrosine kinase A receptor/nerve growth factor 

receptor (TRKAR/NGFR)

AR

Growth factor receptor

196

Metachromatic 

leukodystrophy 

Arylsulphatase A (ASA)

AR

Enzyme

197

Refsum’s disease

Phytanoyl CoA hydroxylase (PhyH)

AR

Enzyme

198

Riley-Day (HSN3)

Inhibitor of 

κ light chain enhancer in B cells (IKBKAP) AR

Transcription factor

199

Tangier sensory motor 

neuropathy

ABC1 (ABC1)

AR

ABC transporter

200

Lower motor neuropathy

X-linked spinobulbar 

muscular atrophy 

Androgen receptor (AR)

AD

DNA/RNA metabolism

201

Congenital fibrosis of 

extraocular muscles

Kinesin 21A (KIF21A)

AD

Transport motor

202

Tay-Sachs disease

Hexoseaminidase A and B (HEXA, HEXB)

AR

Enzyme

203

SMARD1

Immunoglobulin 

µ-binding protein 2 (IGSMBP2)

AR

DNA/RNA metabolism

165

Spinal muscular atrophy

Survival motor neuron (SMN)

AR

DNA/RNA metabolism

153

Lower motor predominant 

ALS

Mitochondrial isolysinyl tRNA synthetase 

(MItRNAS)

M

Mitochondrial DNA/RNA metabolism

43

Hereditary spastic paraplegia

SPG3A

Atlastin (SPG3A)

AD

Dynamin-family GTPase; vesicle recycling

159

SPG17/Silver syndrome

Berardinelli-Seip congenital lipodystrophy type 2 

(BSCL2)

AD

Transcription factor

204

SPG13

Heat shock protein 60 (HSP60)

AD

Mitochondrial protein chaperone

205

SPG10

Kinesin KIF 5A (KIF5a)

AD

Microtubule transport motor

206

SPG6

NIPA1 (NIPA1)

AD

Possibly membrane protein

207

SPG4

Spastin (SPAST)

AD

AAA protein, associates with microtubules

208

Adrenomyeloneuropathy

Adrenoleukodystrophy protein (ABCD1)

AR

ATP transporter in peroxisomes

209

jALS, jHSP

Alsin (ALS2)

AR

GEF signalling, vesicle trafficking

12,13

SPG21

Maspardin (SPG21)

AR

Golgi/vesicle transport and sorting factor

210

SPG7

Paraplegin (SPG7)

AR

Mitochondrial AAA metalloprotease

211

SPG20/Troyer syndrome

Spartin (SPG20)

AR

Endosomal protein trafficking

161

Corticospinal predominant 

ALS

Cytochrome c (COXC)

M

Mitochondrial electron transport protein

42

AAA, ATPase associated with diverse cellular activities; AD, autosomal dominant; ALS, amyotrophic lateral sclerosis; AR, autosomal recessive; CMT, Charcot-Marie-

Tooth; GEF, guanine nucleotide exchange factor; HSN, hereditary sensory neuropathy; jALS, juvenile ALS; jHSP, juvenile hereditary spastic paraplegia; M, maternal; 

SMARD1, spinal muscular atrophy with respiratory distress; SPG, spastic paraplegia; VAMP, vesicle-associated membrane protein.

R E V I E W S

NATURE REVIEWS 

|

 

NEUROSCIENCE

 

 VOLUME 7 

|

 SEPTEMBER 2006 

|

 717

©

 

200

6

 

Nature Publishing Group

 

 

as the disease progresses

118–120

. Second, retrograde 

transport is also disrupted in ALS mice

121

. Third, 

although the molecular basis for this slowing is not fully 

elaborated, several authors suggest that aggregations of 

neuro filaments in the proximal axons (spheroids) might 

physically compromise the transport apparatus

116

, at least 

for anterograde traffic. Neurofilaments have also been 

incriminated as modulators of axonal transport because 

they regulate axonal calibre

122

. Diminution of retrograde 

transport in ALS mice has been attributed to the mis-

localization and disruption of dynein function

123

.

Human motor neuron disorders other than ALS arise 

because of perturbations in the structure and function of 

the motors that mediate axonal transport. Mutations in 

three different kinesin genes have been implicated 

in slowly progressive, motor-specific neuropathy, a 

congenital ocular motor neuropathy and one form 

of hereditary spastic paraplegia 

(TABLE 2)

. In two lines of 

mice (legs at odd angles and cramping), point mutations 

in the dynein heavy chain impair motor neuron func-

tion and viability

124,125

. Dynein is a major component 

of the dynein–dynactin retrograde transport motor in 

neurons and, as noted above, mutations in the p150 

subunit of dynactin elicit an unusual form of human 

lower motor neuron disease. Finally, in mice, disrup-

tion of the dynein–dynactin complex by the forced 

expression of dynamitin, a sub unit of dynactin, also 

produces a late onset motor neuron disease

126

. Among 

the most intriguing observations from the past two years 

is that introduction of the dynein mutations that gen-

erate either the legs at odd angles

127

 or the cramping

128

 

phenotypes into G93A-SOD1 ALS mice significantly 

ameliorates the motor neuron disease and the slowing 

of axonal transport initiated by mutant G93A-SOD1 

protein. How this occurs is not understood, although 

the implication is that some aspect of cytotoxicity 

derived from mutant SOD1 protein requires dynein-

based transport along microtubules, either in the axon 

or within the soma of the motor neuron.

These diverse observations favour the view that dis-

turbances and attenuation of axonal transport might be 

a unitary feature of all forms of ALS and potentially a 

primary pathogenic event in this disease. Challenging 

this concept is the finding that a genetic mutation 

(wlds) that slows the active process of Wallerian axonal 

degeneration does not alter the phenotype of G93A-

SOD1 mice

129

. However, this negative observation is 

not entirely surprising, as the wlds mutation operates 

early in development in mice and not at the later ages 

when motor neurons first degenerate in ALS mice. 

Two other considerations emphasize the importance 

of dysfunctional transport in ALS. First, axonal trans-

port is perturbed in mice with other neuro degenerative 

diseases (for example, Huntington’s disease) and 

in vitro  by proteins implicated in those diseases 

(for example, huntingtin and presenilin 1)

130–132

. 

Second, aberrant vesicular sorting, which is relevant to 

successful axonal transport, has been implicated both 

in the wobbler mouse (characterized by mutations 

in a vesicular sorting protein

133

  that  lead  to  motor 

nerve degeneration) and in a human pedigree with 

frontotemporal dementia

41

 (a bifronto-temporal degen-

eration that can overlap with ALS).

Excitotoxicity and glutamate transport. Another compo-

nent of neuronal degeneration in many neurodegenera-

tive disorders is excessive glutamate-induced stimulation 

of postsynaptic glutamate receptors. This activates mas-

sive calcium influxes that are potentially detrimental 

through calcium-activated processes and molecules 

(for example, proteases, nucleases and lipases). There 

is considerable evidence in support of this view, such as 

the observed threefold increase in glutamate levels in the 

cerebrospinal fluid of patients with ALS

134–136

 and the ben-

efits in ALS of the anti-glutamate drug riluzole. EAATs 

are present at most synapses in the CNS, and trans-

port glutamate from the synaptic space into astrocytes 

after glutamate release during neurotransmission

137–139

. In 

most cases (~65%) of SALS, and in transgenic rodent ALS 

models, there is a profound reduction in the expression 

and activity of EAAT2 in the cortex and spinal cord

140,141

. 

One suggestion has been that the loss of EAAT2 arises 

from splicing errors in the mRNA for EAAT2; indeed, 

abnormalities such as exon skipping and intron retention 

have been found in a subset of SALS patients

142

. EAAT2 

might be linked to ALS by other mechanisms. In one 

sporadic ALS case, a germline mutation in EAAT2 affected 

N-linked glycosylation and glutamate clearance

143,144

. 

Moreover, in SOD1-linked ALS, oxidative damage 

inactivated EAAT2 

(REF. 140)

. Similarly, in astrocytes, 

active CASP3 might also contribute to the reduction of 

EAAT2 expression and activity by cleaving the trans-

porter at the carboxyl terminus

85

. Because its activity is 

abnormal in both sporadic human ALS and ALS mice, 

EAAT2 is a molecular intersection point between these 

two forms of ALS, and therefore defines one element in 

a common pathogenetic pathway of ALS.

Glutamate-induced excitotoxicity provides another 

possible explanation for the selective vulnerability 

of motor neurons in ALS, because it is predicated on 

elevations of cytosolic calcium

168

 and because, rela-

tive to other types of neurons, motor neurons have a 

diminished capacity to buffer calcium

145

. Additionally, 

motor neurons express AMPA (

α-amino-3-hydroxy-

5-methyl-4-isoxazole propionic acid)/kainate recep-

tors for glutamate, whereas in most neurons glutamate 

toxicity is mediated by NMDA (N-methyl-d-aspartate) 

receptors. Typically, the GluR2 subunit regulates (and 

restricts) calcium permeability in AMPA/kainate recep-

tors. Because this subunit is absent in motor neurons, 

the motor neuronal AMPA/kainate receptor population 

is unusually calcium permeable and therefore vulner-

able to excessive glutamate stimulation

146

. Glutamatergic 

excitotoxicity can also reflect defective energy metabo-

lism; conceivably, subnormal energy production in ALS 

motor neurons can sustain glutamate-mediated toxicity 

without elevations in glutamate.

Non-neuronal cells affect motor neurons. The concept 

that populations of non-neuronal cells could affect the 

viability of motor neurons arose originally from the 

observation that substantial activation of microglial 

R E V I E W S

718 

|

 SEPTEMBER 2006 

|

 VOLUME 7 

 

www.nature.com/reviews/neuro

©

 

200

6

 

Nature Publishing Group

 

 

Table 3 | Non-neuronal genes that influence motor neuron viability

Gene (gene symbol)

Inheritance

Class

Refs

Peripheral myelin protein 22 

(PMP22)

AD

Myelin/lipid protein

169

Myelin protein P0 (P0)

AD

Myelin/lipid protein

170

Lipopolysaccharide-induced 

tumour necrosis factor-

α 

(LITAF/SIMPLE)

AD

Protein transport, degradation

171

Early growth response gene  

(EGR2)

AD

Transcription factor

172

SOX10 (SOX10)

AD

Myelin-specific transcription 

factor

212

Ganglioside-induced 

differentiation-associated 

protein (GDAP1)

AR

Mitochondrial protein

173

Myotubularin-related protein 

(MTMR2)

AR

Phosphatase

174

SET binding factor 2 (SBF2) 

(MTMR13)

AR

Myotubularin pseudo-

phosphatase

175

Protein assembly protein 

(SH3TC2) (KIAA1985)

AR

Cytosolic protein

176

N-myc downstream regulated 

gene (NDRG1)

AR

DNA/RNA metabolism

164

Periaxin (PRX)

AR

Neuronal membrane protein

177

Connexin 32 (GJB1)

XD

Gap junction protein

178

L1 cell adhesion molecule 

(L1CAM)

AR

Surface protein

213

Proteolipid protein (PLP)

AR

Myelin/lipid protein

214

AD, autosomal dominant; AR, autosomal recessive; XD, X-linked dominant.

and astroglial cells in ALS is one of earliest microscopic 

manifestations of this disease

147

. Nonetheless, it has been 

difficult to obtain decisive data to test this hypothesis 

or to discern whether the activation of astrogliosis or 

microgliosis is detrimental or beneficial.

Experiments with mice have shown that the specific 

expression of mutant SOD1 in motor neurons or glia 

fails to trigger motor neuron degeneration

148,149

. With the 

caveat that the levels of expression of mutant SOD1 in 

the targeted cells might have been insufficient to induce a 

phenotype, these experiments were interpreted to favour 

the view that motor neuron death in transgenic ALS mice 

is not cell autonomous. Analyses of chimeric mice with 

mixed populations of cells expressing either endogenous or 

transgenic mutant SOD1 were consistent with this view

150

. 

In these mice, motor neurons expressing transgenic G93A 

or G37R SOD1 failed to degenerate if they were adjacent 

to large numbers of supporting cells (such as astrocytes 

and glia) without the mutant protein. Reciprocally, motor 

neurons without the transgene demonstrated pathology 

if surrounded by non-neuronal cells with the mutant 

SOD1 transgene. The minimal sets of mutant-expressing 

cell types required for the development of motor neuron 

degeneration are still poorly defined. Recent studies with 

deletable transgenes have documented that expression 

of the mutant SOD1 transgene in microglia accelerates 

the late phase of murine ALS

151

. Parallel studies suggest 

that expression of the mutant SOD1 transgene within the 

macro phage lineage (presumably encompassing micro-

glial cells) is not required for the ALS phenotype

152

.

Other motor neuron disease genes. Mutations in more 

than 50 different human genes are implicated in the 

pathogenesis of motor neuron cell death

153

 

(TABLES 2,3;

 

Supplementary information S1

 (figure)). Five have 

been associated with ALS although only two (SOD1 

and VAPB) demonstrate the phenotype of ALS. This 

ensemble of gene defects sheds light on several aspects 

of ALS.

First, the majority of mutations produce slowly 

progressive phenotypes, with clinical, pathological and 

physiological features, suggesting that the burden of 

the pathology is axonal. That is, most of these disorders 

might be categorized as axonocentric, with an extremely 

slow evolution of cell death. By contrast, mutations in 

two genes (SOD1 and VAPB) trigger an adult onset, 

fulminant course of rapidly progressive paralysis leading 

to death, with pathological and physiological character-

istics indicating that the cell body is involved early in 

the disease. These disorders behave as if they are somato-

centric, with secondary axonal features.

Second, from the effects of the genetic mutations 

discussed above, it can be concluded that there are 

recurring themes in the pathogenesis of these dis-

orders. Fourteen of the genes directly or indirectly 

implicate aspects of axonal or organelle trafficking. 

Of these, seven are either cytoskeletal filaments

154,155

 

or micro tubule-based motors

17,157,202,206,208

, whereas the 

other seven are crucial for aspects of vesicle formation, 

recycling and trafficking (one is involved in GEF signal-

ling

12,13

, two are dynamin family GTPases

158,159

 and four 

are central to Golgi and endoplasmic reticular vesicle 

function

16,160,161,210

). Of note are six genes related to RNA/

DNA metabolism

15,153,162,163,165,201

 and three HSPs

166,167,205

. 

The paucity, so far, of other classes of gene is also 

instructive. So far, only one trophic factor is directly 

implicated (nerve growth factor and its receptor tyrosine 

kinase A

196

).

Third, the only defective genes that are not expressed 

in neurons but nonetheless impair motor neuron viability 

are expressed in cells that myelinate motor neurons

169–178

 

(TABLE 3)

. It has long been established that motor 

neurons can degenerate if subjected to severe, sustained 

demyelination. However, apart from the myelin-related 

genes, none of the other genes in 

TABLE 2

 is expressed 

exclusively in non-neuronal cells. This is a persuasive 

argument that non-neuronal cells almost certainly 

modulate the thresholds or set points for degeneration 

in motor neurons, although an absolute requirement for 

the development of mutation-initiated motor neuron 

disease is the expression of the mutant protein within 

the motor neuron.

Conclusion and comments on therapy

Genetic analyses of human motor neuron degenera-

tion have defined diverse molecular pathways in motor 

neuron cell death. Investigations of mutant SOD1 have 

illuminated crucial components of the death process 

including: a propensity for mutant SOD1 to be unstable; 

a multiplicity of mitochondrial defects that predict cel-

lular energy failure, enhanced glutamate sensitivity and 

activation of the machinery of programmed cell death; 

R E V I E W S

NATURE REVIEWS 

|