Mutations in Dynein Link Motor Neuron Degeneration to Defects in Retrograde Transport

See allHide authors and affiliations

Science  02 May 2003:
Vol. 300, Issue 5620, pp. 808-812
DOI: 10.1126/science.1083129


Degenerative disorders of motor neurons include a range of progressive fatal diseases such as amyotrophic lateral sclerosis (ALS), spinal-bulbar muscular atrophy (SBMA), and spinal muscular atrophy (SMA). Although the causative genetic alterations are known for some cases, the molecular basis of many SMA and SBMA-like syndromes and most ALS cases is unknown. Here we show that missense point mutations in the cytoplasmic dynein heavy chain result in progressive motor neuron degeneration in heterozygous mice, and in homozygotes this is accompanied by the formation of Lewy-like inclusion bodies, thus resembling key features of human pathology. These mutations exclusively perturb neuron-specific functions of dynein.

Motor neuron disease (MND) is one of the most common neurodegenerative diseases (1). The molecular basis of MND is poorly understood. Defects in the SMN1 gene are responsible for some cases of SMA (2), and one form of SBMA has been linked to triplet repeat expansions in the androgen receptor (3). Mutations in the superoxide dismutase 1 (SOD1) gene account for only a minority of the familial forms of ALS (FALS) (1, 4). Thus, the underlying genetic defects in the vast majority (>98%) of ALS, other MND-type syndromes, and some SMA subgroups remain unknown.

Even in the cases where causative mutations have been identified, there is no widely accepted mechanistic model to explain how changes in ubiquitously expressed genes can cause selective death of motor neurons (1, 4). The histopathological lesions reported in sporadic ALS include neuronal loss; Lewy or Bunina body–like inclusions containing SOD1, CDK5, neurofilaments (NFs), and ubiquitin; and fragmentation of the Golgi apparatus of motor neurons.

Impaired axonal retrograde transport has been reported in SOD1 transgenic mouse models of FALS (5). A major molecular motor involved in retrograde transport is cytoplasmic dynein, a motor protein complex that moves along microtubules in the minus-end direction (6). It is composed of at least four classes of subunits (heavy, intermediate, light intermediate, and light chains) and mediates the intracellular movements of vesicles and protein complexes for nuclear motility, Golgi function, and axonal transport. The protein complex dynactin, consisting of several subunits, including dynamitin, tethers some cargoes by interacting with the latter and with the dynein light intermediate chains (7). In agreement with the housekeeping functions of dynein, mice homozygous for a targeted disruption of the cytoplasmic dynein heavy chain 1 (Dnchc1) gene die at an early stage of embryogenesis; heterozygotes showed no obvious abnormalities (8).

Recent evidence has suggested that inhibition of dynein-mediated axonal transport by postnatal transgenic overexpression of dynamitin in motor neurons causes neurodegeneration (9). We showed that subtle germline heterozygous missense mutations in Dnchc1 give rise to anterior horn cell death; in homozygotes, Lewy body–like inclusions were also seen. We showed that selective impairment of one type of axonal retrograde transport likely explains the neurodegeneration, and thus closely resembles the clinical situation. We also demonstrated that these mutations perturb neuronal development and migration but leave housekeeping functions of cytoplasmic dynein intact.

Two mouse mutants, Legs at odd angles (Loa) and Cramping 1 (Cra1), which arose in two independent mutagenesis experiments in the offspring of N-ethyl-N-nitrosourea (ENU)–treated male C3H mice (10, 11), manifest similar progressive locomotor disorders [supporting online material (SOM) text]. Both lines were initially identified by the unusual twisting of the body and clenching of the hindlimbs when suspended by the tail. Further analyses confirmed that Loa and Cra1 are autosomal dominant traits that give rise to an age-related progressive loss of muscle tone and locomotor ability in heterozygous mice (fig. S1) without major reduction in life-span (10). Homozygous Loa and Cra1 mice show a more severe phenotype with an inability to feed and move, and die within 24 hours of birth.

Consistent with the progressive motor impairment, histopathological analysis detected a significant decrease in the number of α motor neurons in the spinal cord anterior horn when comparing wild-type littermates to 3- and 16-month-old Cra1 heterozygotes. A similar result was observed in 19-month-old Loa heterozygotes (Fig. 1, A and B, a to f). Loss of spinal α motor neurons in 16-month-old Cra1/+ mice was accompanied by an altered composition of muscle fiber types with a predominance of large type I fibers (Fig. 1B, g and h).

Fig. 1.

Progressive impairment of muscle function and motor coordination is associated with decreasing numbers of α motor neurons. (A) Quantitative analysis of α motor neurons of aged Cra1/+ and Loa/+ mice (gray bars) demonstrates the significant loss of anterior horn neurons when compared to +/+ littermates (black bars). (B) Motor neuron degeneration in 16-month-old Cra1/+ heterozygote mice (b, d, and f) in comparison to wild-type littermates (a, c, and e). Immunohistochemistry using NeuN primary antibody [(a) to (d)], and using hematoxylin and eosin (H&E) staining [(e) and (f)]. Muscle fiber morphology is altered in Cra1/+ mice (h) as compared with +/+ littermates (g), with a predominance of large type 1 fibers in periodic acid–Schiff (PAS) staining. Scale bars, 150 μmin(a) and (b); 20 μm in (c) and (d); and 10 μm in (e) to (h).

Furthermore, we observed a loss of 80% of the embryonic spinal anterior horn cells (AHCs) in Cra1/Cra1 mice, and a 50% loss in Loa/Loa mice as compared with wild-type littermate controls at embryonic day 18.5 (E18.5) (Fig. 2H). A striking histological feature in these animals was a highly enhanced apoptotic rate in AHCs (Fig. 2, A and B, marked by arrows), confirmed by TUNEL staining (12), and the presence of round or ovoid eosinophilic Lewy-like intracellular inclusion bodies in surviving neurons. In Loa/Loa mice, inclusions were seen in anterior horn cells (Fig. 2C); in Cra1/Cra1 mice, large dorsal root ganglion neurons were affected (Fig. 2, D and E). An immunohistochemical analysis detected the deposition of ubiquitin, SOD1, CDK5, and NFs in the perinuclear inclusions (Fig. 2, F and G). Similar depositions, sometimes referred to as Bunina bodies (13), have been described as a characteristic feature of human ALS, especially in more slowly progressing subtypes.

Fig. 2.

Loss of anterior horn cells and inclusion body formation in surviving neurons in E18 homozygous Cra1/Cra1 and Loa/Loa embryos. Shown are neuronal cell loss and apoptosis in the anterior horn, marked by the dotted area and red arrow, using NeuN antibody (A and B), and in dorsal root ganglia, using H&E staining (D and E), in +/+ and Cra1/Cra1 embryos [(A), (D), (B), and (E), respectively]. (H) Quantitative analysis demonstrates a similar effect for both Cra1/Cra1 and Loa/Loa embryos. Black bars, +/+; gray bars, mutant/mutant. Inclusion body formation [(C), arrows, H&E staining], with positive immunoreactivity for NFs (F) and SOD1 (G), is observed in surviving motor neurons of Loa/Loa embryos. Scale bars, 50 μm in [(A) and (B)]; 10 μm in [(D) and (E)]; and 25 μm in [(C), (F), and (G)].

The Loa mutation had been mapped to distal mouse chromosome 12 (14). Further refinement of the critical region and analysis of candidate genes led to identification of a T-to-A transversion in the Dnchc1 gene that cosegregates with the Loa phenotype (SOM text). This mutation results in residue 580 changing from phenylalanine (TTC) to tyrosine (TAC). Independently, the Cra1 mutation was mapped to the same chromosomal region. Again, sequencing of candidate genes revealed a mutation in Dnchc1 that changes residue 1055 from wild-type tyrosine (TAC) to Cra1 cysteine (TGC). Residues Phe580 and Tyr1055 are both highly conserved (fig. S2A). Intercrossing heterozygous Cra1/+ with heterozygous Loa/+ mice yielded compound heterozygotes that displayed the same phenotype as the homozygotes and all died within 48 hours of birth, demonstrating that Cra1 and Loa are allelic (12).

Cytoplasmic dynein heavy chains homodimerize and bind to other proteins, including intermediate-chain dyneins, to form the cytoplasmic dynein complex, which is the most abundant minus-end–directed microtubule motor (8) and is active in all cell types. DNCHC1 is a >500 kD protein containing several domains for protein-protein interaction and other functions such as adenosine triphosphatase activity (fig. S2B) (15, 16). The Loa mutation occurs within the highly conserved region thought to be the binding site for the dynein intermediate chains (17) and the site of homodimerization (18) (SOM text and fig. S2). The Cra1 mutation lies within the potential homodimerization site (17, 18).

Western blot and immunoprecipitation studies did not reveal differences in the amount of dynein heavy-chain and intermediate-chain proteins when comparing Cra1 or Loa homozygous, heterozygous, or wild-type mice (SOM text). Similarly, immunohistochemistry showed no detectable difference in the localization of DNCHC1 or intermediate chains (SOM text). Thus, the mutant phenotype is not likely to be due to gross changes in DNCHC1 protein levels or intracellular distribution.

Furthermore, analysis of some of the ubiquitous functions of dynein [such as nuclear motility during cell division (19) and the formation and positioning of the Golgi apparatus] revealed that nuclear motility is not affected in the Loa mice and that steady-state Golgi morphology and positioning were normal (SOM text). However, after the Golgi complex was dispersed by the microtubule depolymerizing agent nocodazole and the microtubules were allowed to repolymerize, the rebuilding of the pericentrosomal Golgi complex was significantly impaired in the Loa/Loa embryonic fibroblasts (SOM text and fig. S3). This finding suggests that the Loa mutation reduces the performance of cytoplasmic dynein only in situations of cellular stress. Disruption of the Golgi complex is also a prominent lethal phenotype reported for homozygous null Dnchc1 embryos (8). In both human ALS and mutant SOD1 transgenic mice that model human ALS, the Golgi complex has been described as fragmented in a manner resembling that of nocodazole-treated cells (20).

In addition to its ubiquitous roles, cytoplasmic dynein has several functions specific to neurons, where it is involved in neuronal migration (19, 21); the growth and development of neurites; and axonal transport of microtubules, NFs, and organelles (6, 7, 22). In Drosophila, hypomorphic mutations in components of the dynactin complex have been reported to impair later stages of neuronal development (23). Therefore, we investigated the effects of the Dnchc1 mutations on the development of motor neurons. Analysis of the cranial and spinal nerve formation in Loa mutants and wild-type littermates at E10.5 showed no difference in the timing of motor neuron differentiation in the neural tube or in the pattern of nerve extension from the neural tube into the periphery (fig. S4) (12).

However, the migration of facial motor neuron cell bodies to their final destination in the hindbrain was altered in Loa homozygous mice (Fig. 3). Many facial motor neurons fail to follow the normal path in Loa/Loa mutants and turn away from the midline prematurely to condense in a more anterior position on the pial side of the hindbrain. In contrast, the migratory pattern of trigeminal motor neurons is not compromised by the Loa mutation, because they migrate in the direction of axon extension to assemble in their appropriate location near the axon exit point (Fig. 3, C and D). This compromised development of the facial nerve might contribute to the inability of Loa/Loa newborns to suckle and thus to their perinatal lethality. Facial motor weakness driven by impairment of brainstem lower motor neurons is a classic clinical and pathological feature of SBMA, as well as of some bulbar subtypes of ALS.

Fig. 3.

Abnormal facial motor neuron migration in Loa/Loa embryos. Facial motor neurons (labeled VIImn) normally migrate from their birthplace on the ventricular side of the anterior hindbrain (black arrowheads) (A)toa more posterior position on the pial face of the hindbrain (black arrowheads) (C), where they assemble into the paired facial motor nuclei (labeled VIIn) (E). During their migration, they assume a curved path away from the midline (black asterisks). Only a proportion of facial motor neurons migrates appropriately and condenses in the correct position in Loa/Loa mutants (B, D, and F), while a second population deviates from the normal migratory route (white arrowheads) (B) to assemble into slightly smaller nuclei more anteriorly (white stars) [(D) and (F)]. Consequently, the facial motor nuclei appear dumbbell-shaped rather than round [compare (E) and (F)]. In contrast, trigeminal motor neurons assemble in their appropriate location on the pial face of the hindbrain to form the paired trigeminal motor nuclei (Vn) both in wild type (+/+) (C) and Loa/Loa (D) littermates. Motor neurons were visualized by in situ hybridization with the motor neuron marker Isl1 at E13.5 [days post coitum (dpc)][(A) to (D)] and E14.5 [(E) and (F)].

As observed for other spinal nerves, target finding of the limb nerves was not perturbed in Loa mutants. However, branching and elongation of these nerves within the target area was impaired (SOM text and fig. S5). These defects were not caused by any instability of microtubules at nerve termini (fig. S6). We conclude that the decrease in branching of the limb nerves with the Loa mutation is not due to their inability to recognize the target area or to form terminal arbors, nor can it be explained by the degeneration of nerve branches that had initially formed. Rather, the process of branching and branch elongation within the target area was less efficient when cytoplasmic dynein function was compromised.

Because cytoplasmic dynein is the most important retrograde motor in neurons, we investigated whether axonal transport was affected in Dnchc1 mutants. We could detect no gross axonal transport abnormalities in a mixed peripheral nerve by performing a sciatic nerve ligation experiment (fig. S7). However, by using an assay for the visualization and quantitation of axonal retrograde transport based on a fluorescent fragment of tetanus toxin (TeNT HC) (24), we demonstrated the effect of the Dnchc1 mutations in spinal cord motor neurons.

Upon specific binding and uptake, TeNT HC is recruited to a retrograde transport pathway shared by nerve growth factor (NGF) and p75NTR (24). Kinetic analysis of retrograde transport in motor neuron cultures derived from E13 wild-type and Loa/Loa embryos revealed a marked alteration of the speed distribution of TeNT HC–positive axonal carriers (Fig. 4A). Although retrograde transport was not abolished in homozygous Loa/Loa neurons, mutants presented a reduction in frequency of the high-speed carriers (from 67.1 ± 4.1% in wild-type to 21.3 ± 3.6% in Loa/Loa motor neurons; P < 0.05) and an increase in the stationary pauses (from 11.1 ± 5.6% to 50.6 ± 4.0%; P < 0.05) (Fig. 4B). These results strongly indicate that Dnchc1 mutations can impair the ability of cytoplasmic dynein to sustain fast retrograde transport, of at least some cargoes, specifically in spinal cord motor neurons.

Fig. 4.

Kinetic analysis of TeNT HC carriers in wild-type and homozygous Loa spinal cord motor neurons. (A) Speed distribution of the TeNT HC–positive compartment in wild-type (+/+; n = 246) and Loa/Loa (n = 79) motor neurons. Retrograde transport is conventionally shown as positive. Error bars represent ±SEM. (B) Analysis of the different components contributing to the axonal retrograde transport of TeNT HC–positive carriers in wild-type and homozygous Loa spinal cord motor neurons. A marked reduction in the frequency of the high-speed carriers and an increase in the stationary pauses in homozygote Loa motor neurons are observed. Error bars represent ±SEM. n = 6 for wild-type and n = 5 for Loa/Loa embryos. Asterisks indicate P < 0.05.

Microtubule-based transport, a process essential for the highly specialized morphology and function of the motor neuron, has been implicated in the degeneration and specific pathological features of MND. Holzbaur and colleagues (9) have recently demonstrated that postnatal and motor neuron–specific disruption of the dynein/dynactin complex can recapitulate several features of MND in mice. This raises the question of whether naturally occurring mutations can cause similar disease phenotypes. Here, we provide direct evidence that subtle heterozygous missense mutations in DNCHC1 cause progressive motor neuron degeneration and separate its general housekeeping functions from its neuron-specific functions. These results suggest that a large spectrum of potential mutations in components of the dynein complex could be natural causes of ALS and related disorders.

The Loa and Cra1 mutations do not cause overt deficits across the range of known DNCHC1 functions but result in a specific defect in fast retrograde transport that appears to manifest only in α motor neurons. Comparison of the Loa/+ and Cra1/+ phenotypes with the published phenotype of the targeted null mutation (8) suggests that the point mutations result in either a dominant negative or hypomorphic effect on a subset of dynein functions, which could impair the recruitment of neuron-specific dynein subunits to the dynein/dynactin complex. This could explain the unique and specific phenotype observed in Cra1 and Loa mice (SOM text).

The impairment of retrograde transport in α-motor neurons may contribute to specific effects on neurodevelopment, because developmental signals may no longer be transported efficiently. This is supported by the striking similarity of the Dnchc1 phenotype to that of mice lacking components of the hepatocyte growth factor/Met signaling pathway (25, 26). In the adult organism, an analogous mechanism could explain the degenerative changes observed in heterozygote old mice. Although in young heterozygote Dnchc1 mutants, the clinical manifestation of the developmental phenotype is not substantial, the age-related progressive neurodegeneration could be the consequence of a constantly reduced supply of trophic factors (such as NGF) caused by impaired dynein-dependent retrograde axonal transport.

We also identified a specific abnormality in the migration of facial motor neurons. A missense mutation in the p150glued subunit of dynactin has recently been identified in a human kindred displaying a SBMA-like syndrome dominated by facial, bulbar, and distal extremity weakness (27). This further supports the hypothesis that subtle alterations of the dynein/dynactin system can lead to human MND; mutations in anterograde axonal transport proteins, including the microtubule motor kinesin (KIF1B), can lead to slow progressive motor neuronopathy (28), indicating the potential generality of the link between retrograde and anterograde axonal motor protein deficits and motor neuron degeneration.

The Loa and Cra1 mutations exhibit remarkable similarities to specific features of human pathology, including Lewy body–like inclusions containing SOD1, CDK5, NFs, and ubiquitin. Abnormal accumulation of NFs in the cell body and proximal axons is the most frequently seen early pathological characteristic of ALS (29). Retrograde transport of NFs is dynein-dependent, suggesting that the buildup of NFs is secondary to the defect in retrograde transport. SOD1 deposition, which may also be dynein-dependent (4), is an unexpected finding pointing to a link between dynein dysfunction and normal SOD1 function.

The mechanism by which SOD1 mutations bring about motor neuron degeneration is unknown. Recent data have demonstrated the slowing of axonal transport as an early event in the motor neuron toxicity of SOD1 mutants (30). The disruption of dynein function by interaction with mutant SOD1, creating an aberrant protein complex that interferes with the function of the wild-type complex, may also provide a link between axonal transport, NF accumulation, and cell death.

Supporting Online Material

Materials and Methods

SOM Text

Figs. S1 to S7

Table S1


References and Notes

Stay Connected to Science

Navigate This Article