Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative condition characterized by progressive motor neuron degeneration and muscle paralysis. Genetic evidence from man and mouse has indicated that mutations in the dynein/dynactin motor complex are correlated with motor neuron degeneration. In this study, we have generated transgenic mice with neuron-specific expression of Bicaudal D2 N-terminus (BICD2-N) to chronically impair dynein/dynactin function. Motor neurons expressing BICD2-N showed accumulation of dynein and dynactin in the cell body, Golgi fragmentation and several signs of impaired retrograde trafficking: the appearance of giant neurofilament swellings in the proximal axon, reduced retrograde labelling by tracer injected in the muscle and delayed expression of the injury transcription factor ATF3 after axon transection. Despite these abnormalities, BICD2-N mice did not develop signs of motor neuron degeneration and motor abnormalities. Interestingly, the BICD2-N transgene increased lifespan in ‘low copy’ SOD1-G93A ALS transgenic mice. Our findings indicate that impaired dynein/dynactin function can explain several pathological features observed in ALS patients, but may be beneficial in some forms of ALS.
Amyotrophic lateral sclerosis (ALS) is a clinically and genetically heterogeneous disease characterized by late-onset progressive degeneration of motor neurons resulting in paralysis of limb, facial and respiratory muscles ( 1 , 2 ). Pathologically, the disease in most instances is characterized by protein aggregates that contain TDP-43, a protein involved in mRNA metabolism ( 3 ). A minority of ALS patients (10%) show Mendelian inheritance, a subset of who have mutations in the Cu/Zn superoxide dismutase (SOD1) gene resulting in SOD1 aggregates in motor neurons and glia ( 1 , 2 , 4 ). More recently, mutations in a variety of other genes have been identified in patients with ALS and ALS-like diseases ( 2 , 5–7 ). In a family with a slowly progressive motor neuron disease, a missense (G59S) mutation was found in the p150Glued subunit of dynactin (DCTN1) ( 8 ) and subsequently other p150Glued mutations have been reported in ALS patients ( 9 ). Recently, it was reported that heterozygous knock-in and transgenic mice expressing mutant p150Glued-G59S develop motor neuron abnormalities and degeneration ( 10–12 ). Dynactin is a multiprotein complex that regulates microtubule-based motility of the cytoplasmic dynein motor complex by increasing processivity and efficiency of the motor ( 13 , 14 ). A direct link between impaired dynactin/dynein function and motor neuron disease was first demonstrated by the overexpression of the dynactin subunit p50, also named dynamitin, which disrupts the dynactin/dynein complex and causes a late-onset motor neuron disease in transgenic mice ( 15 ). In addition, heterozygous missense mutations in the cytoplasmic dynein heavy chain 1 gene were found in two mouse models with late-onset motor neuron degeneration, Legs at odd angles ( Loa) and Cramping 1 (Cra1) ( 16 , 17 ), suggesting that abnormalities in both dynein and dynactin may play a role in pathogenesis of ALS ( 18 , 19 ).
Given the role of the dynein/dynactin complex in retrograde transport of cargoes such as endosomes, signalling complexes, degradation products and neurofilaments ( 19–25 ), it is likely that alteration in dynactin/dynein function could influence several critical cellular processes within various compartments of the motor neuron. Potential consequences of disrupted dynein/dynactin function in ALS pathology are the reported abnormalities in axonal neurofilament distribution ( 10 , 15 , 24 , 26–28 ), fragmentation of the Golgi apparatus ( 29–31 ) and impaired retrograde trafficking ( 15 , 18 , 19 ). However, the precise relationship between disrupted dynein/dynactin function and the pathological features observed in ALS patients is unclear. It is not known which pathological abnormalities in motor neurons are directly related to impaired dynein/dynactin and whether a loss- or gain-of-function mechanism is the primary cause of motor neuron degeneration.
To determine the cellular and pathological effects of dynein/dynactin inhibition in motor neurons, we have generated transgenic mice with neuron-specific expression of the N-terminus of Bicaudal D2 (BICD2-N). Previous studies have shown that Bicaudal D is an evolutionarily conserved motor-adaptor protein, which is involved in dynein-mediated transport in Drosophila and mammals by linking the dynein motor complex to various cargoes ( 32–36 ). When deleting the C-terminal cargo-binding region, the N-terminus of BICD2 strongly binds the dynein/dynactin complex and impairs dynein–dynactin function ( 32 , 33 ). Thus, BICD2-N overexpression is a powerful tool for dissecting the roles of dynein and dynactin in motor neurons.
Here we show that the expression of BICD2-N in motor neurons impairs dynein/dynactin function and causes Golgi fragmentation, axonal neurofilament swellings and reduced retrograde transport. Despite these changes, we found no evidence of motor neuron degeneration up to 2 years of age. Furthermore, we show that impaired dynein/dynactin function increases the lifespan of transgenic mice that express an ALS-linked SOD1 mutation. We also observed the accumulation of dynein and p150Glued in SOD1 aggregates in SOD1-G93A ALS mice, suggesting that dynein/dynactin trapped in intracellular inclusions might be beneficial to the disease phenotype in SOD1-linked ALS.
BICD2-N causes accumulation of dynein/dynactin in motor neuron cell bodies
To examine the effect of BICD2-N expression in motor neurons in vivo , we generated transgenic mice by cloning GFP-BICD2-N into the Thy1.2 expression cassette (Fig. 1 A), which drives postnatal transgene expression in motor neurons and other neuron populations throughout the brain ( 37 , 38 ). Transgenic lines were screened on the basis of GFP-BICD2-N levels as identified by in situ hybridization (Fig. 1 B), western blot analysis (Fig. 1 C) and GFP signal in motor neurons as identified by ChAT-immunostaining (Fig. 1 D). Three lines with different transgene expression levels in motor neurons were selected for further study. None of the lines showed transgene expression in all motor neurons. Quantitative analysis revealed that lines BN1 and BN4 have relatively high GFP-BICD2-N expression in ∼50–60 and 30% of the spinal motor neurons, respectively, whereas line BN3 has relatively low GFP-BICD2-N expression in ∼70% of the motor neurons (Fig. 1 D). Furthermore, all lines show transgene expression throughout the brain, predominantly in the deep lamina of the cortex and the hippocampus (data not shown). Onset of transgene expression was between post-natal days 4–6. Most analyses were performed with multiple lines, but unless otherwise stated the data presented are from line BN1.
We first tested whether Thy1.2-GFP-BICD2-N expression affects dynein/dynactin expression. Western blot analysis showed increased levels of dynein heavy chain in the spinal cord of BN1 and BN3 mice compared with non-transgenic mice, whereas other dynein/dynactin components were unaltered (Fig. 1 C). Confocal immunofluorescence analysis revealed a robust increase of p150Glued and dynein heavy chain immunoreactivity in the cell bodies and proximal dendrites in BICD2-N-expressing motor neurons (Fig. 2 A, B and D). GFP-negative motor neurons in Thy1.2-GFP-BICD2-N mice showed the same levels of dynein/dynactin immunoreactivity as neurons from non-transgenic mice, indicating that the accumulation of dynein/dynactin components in the cell body is specifically caused by the presence of BICD2-N. Increased dynein/dynactin immunoreactivity was evident at the onset of transgene expression at post-natal days 4–6. The increased staining was specific for dynein/dynactin, as no change in labelling was observed for other markers examined, such as KIF5A (Fig. 2 C and D) and KIF5C (not shown), neurofilament proteins (NF-M, Fig. 4 C; SMI32, not shown), neuron-specific class III beta-tubulin (TuJ1, not shown), MAP2 (not shown), ChAT (Fig. 1 D) and the small heat shock protein Hsp25 (not shown).
These data were confirmed by experiments in cultured hippocampal neurons; GFP-BICD2-N expression increases dynein and dynactin staining in the cell body and proximal dendrites ( Supplementary Material, Fig. S1A and C ). In contrast, overexpression of p50, which also inhibits dynein- dependent transport ( 29 ), shows a normal distribution of dynein/dynactin in neurons ( Supplementary Material, Fig. S1A and C ). These data suggest that p50 and BICD2-N inhibit dynein/dynactin via distinct mechanisms: overexpression of p50 disassembles the dynactin complex ( 39 ), whereas BICD2-N binds dynein/dynactin and causes the accumulation of the complex in the cell body.
Detailed analysis of GFP-BICD2-N expression indicates that it is diffusely expressed over the somato-dendritic compartment of motor neurons, but absent in motor axons in the ventral roots and sciatic nerve. Also, axons of other populations of neurons expressing GFP-BICD2-N such as layer V cortical neurons were devoid of GFP-BICD2-N. Remarkably, a subset of neurons that express GFP-BICD2-N at relatively high levels were surrounded by spherical structures, 0.5–2 µm in diameter, that displayed high levels of GFP signal (Fig. 2 ). These structures were also intensely immunoreactive for p150Glued and dynein heavy chain, but not for other investigated markers, such as KIF5A and MAP2 (Fig. 2 ; data not shown). Confocal analysis with an antibody against the muscarinic M2-receptor that outlines the cell membrane of motor neurons indicated that in most instances these spheres were attached to the motor neuron via thin processes ( Supplementary Material, Fig. S2A ). This was confirmed by electron microscopic analysis of serial sections ( Supplementary Material, Fig. S2B and C ). Electron microscopy also showed that the membrane protrusions were filled with electron dense material, but did not contain synaptic specializations such as post-synaptic densities, and were not contacted by presynaptic boutons. Although the significance of the membrane protrusions is not clear, we hypothesize that the structures may result from high dynein/dynactin concentration that may alter the cortical cytoskeleton, or drive outward movement of microtubules to initiate protrusions as reported recently ( 40 ); accordingly, microtubules were found in the neck of the protrusions ( Supplementary Material, Fig. S2C ).
BICD2-N causes Golgi fragmentation in motor neurons
Previous studies showed that the loss of dynein function or BICD2-N overexpression causes fragmentation of the Golgi apparatus ( 32 , 33 , 41 ). Accordingly, GFP-BICD2-N and GFP-p50 overexpression caused Golgi fragmentation in cultured hippocampal neurons ( Supplementary Material, Fig. S1B and D ). Analysis of Golgi apparatus in spinal cord of Thy1.2-GFP-BICD2-N mice with an antibody against the cis -Golgi protein GM130 ( 42 ) revealed motor neurons with fragmented Golgi in BN1 and BN4, but not BN3 mice (Fig. 3 ). The fragmentation was characterized by the transformation of the Golgi apparatus from a network of linear profiles into dispersed smaller elements (Fig. 3 C). Double-labelling with antibodies against CGRP, a peptide that is present in the trans -Golgi and secretory granules of most large motor neurons ( 43 ), indicated that the fragmented Golgi consisted of mini-stacks containing all Golgi elements (Fig. 3 A and C). In BN1 mice, the percentage of GFP-BICD2-N-expressing motor neurons with fragmented Golgi was 1–2% at 20 weeks of age and 7–15% at 100 weeks of age, indicating that the frequency of motor neurons with fragmented Golgi increased with ageing. Golgi fragmentation was predominantly present in high-GFP-BICD2-N-expressing motor neurons and was never observed in GFP-BICD2-N-negative motor neurons (Fig. 3 A and B), nor in motor neurons of non-transgenic mice of any age. All motor neurons with fragmented Golgi showed a normal appearance. No consistent change in the densities of mitochondria (SOD2 staining) or endoplasmic reticulum (calreticulin staining) was observed in GFP-BICD2-N motor neurons ( Supplementary Material, Fig. S3 ).
BICD2-N causes giant proximal neurofilament swellings in motor axons
Impaired dynein/dynactin function in cultured neurons may lead to neurofilament accumulations in the proximal or distal axon ( 28 ). Therefore, in Thy1.2-GFP-BICD2-N mice, we studied the distribution of neurofilament proteins as well as peripherin, an intermediate filament protein that is expressed at high levels in motor axons. No major change in neurofilament-M and peripherin immunoreactivity occurred in motor nerve endings at the neuromuscular junctions. However, analysis of spinal cord sections revealed intense peripherin and neurofilament-M immunoreactive structures in the most ventral aspect of the grey matter, where the motor axons enter the white matter to course to the ventral roots (Fig. 4 A–C). In some instances, these structures also were identified in motor axons crossing the white matter, as well as in the proximal aspect of the ventral roots. The structures were also intensely labelled with antibodies against the SMI31 and SMI32 epitopes representing phosphorylated and non-phosphorylated neurofilament, respectively. Further analysis by electron microscopy showed that these structures consisted of swollen myelinated axons with a diameter reaching up to 20 µm that were filled with filamentous material (Fig. 4 E), and strongly resembled proximal giant filamentous axonal swellings reported in ALS patients ( 26 , 27 ). Comparison of spinal cord sections from different transgenic lines showed that the giant axonal neurofilament swellings occurred in both BN1 and BN4 lines, but not in mice from the low expressing BN3 line. Comparison of BN1 mice over different ages (4, 20 and 104 weeks) revealed that axonal swellings did not occur at 4 weeks, but were present at 20 and 104 weeks in equal number (Fig. 4 D). These data indicate that axonal swellings are a relatively early phenomenon after dynein/dynactin impairment.
BICD2-N expression reduces retrograde axonal transport
To examine whether motor neurons expressing BICD2-N show reduced retrograde axonal transport, we performed tracing experiments with the retrograde tracer fluorogold ( 15 ). With our transgenic lines, we can take advantage of the chimeric expression of the GFP-BICD2-N transgene, in particular, the BN1 line that expresses the transgene in 50–60% of the motor neurons, with the remaining motor neurons serving as controls. Fluorogold was injected in the gastrocnemic muscle of BN1 mice, and 48 h post-injection, lumbar L4–L5 sections were examined for fluorogold and GFP signals. As shown in Figure 5 , the amount of fluorogold labelling negatively correlated with the level of GFP signal, indicating that BICD2-N expression diminishes retrograde fluorogold transport.
Axonal injury activates a number of retrograde signalling pathways to reorganize gene expression and initiate repair programmes in the injured neuron ( 25 ). To examine whether BICD2-N expression has an effect on retrograde injury signalling, we have studied ATF3 expression after sciatic nerve transection in BN1 mice and non-transgenic littermates, as ATF3 is one of the transcription factors that is strongly induced in axotomized motor neurons ( 44 ). No nuclear ATF3 labelling is present in lumbar spinal cord sections of non-axotomized BN1 and control mice. Twelve hours post-axotomy of the left sciatic nerve, weak nuclear ATF3 staining was observed in ipsi-lateral sciatic nerve motor neurons, which are localized in the dorso-lateral aspect of the L4–L5 spinal cord (Fig. 6 A–C), and more intense labelling occurred at later time points. Quantitative analysis of ATF3 levels in axotomized BN1 mouse spinal cord showed that at 12 h post-axotomy, high-level BICD2-N-expressing motor neurons showed reduced ATF3 expression compared with BICD2-N-negative motor neurons (Fig. 6 D and E). No difference in ATF3-labelling intensities was observed between BICD2-N-positive and negative sciatic nerve motor neurons at 24 h post-axotomy (Fig. 6 D and E). These data show that inhibition of dynein/dynactin delays retrograde injury signalling.
BICD2-N mice do not develop motor abnormalities and motor neuron loss
To determine whether impaired dynein/dynactin function influences viability of motor neurons, BN1 mice were tested for the development of motor abnormalities up to the age of 2 years and subsequently analysed for degenerative changes in the neuromuscular system. BN1 mice did not show evidence of weight loss (Fig. 7 A) or reduced muscle strength, as determined by a hanging wire test (Fig. 7 B) and grip strength measurement (not shown). Furthermore, the size of the cell bodies and nuclei was the same in GFP-BICD2-N and control motor neurons (Fig. 7 C). Counting the number of motor neurons showed that 2-year-old BN1 mice contain the same amount of L4 spinal cord motor neurons as 2-year-old non-transgenic mice and 20-week-old BN1 mice (Fig. 7 D), and that the number of GFP-BICD2-N motor neurons was the same in 20-week- and 2-year-old BN1 mice (Fig. 7 D). In addition, a silver staining method that visualizes degenerating neurons and their processes, did not produce argyrophilic staining in spinal cord from 2-year-old BN1 mice. Consistent with the absence of neurodegenerative changes, 2-year-old BN1 mice did not show evidence of increased astrocytosis and microgliosis compared with non-transgenic controls (Fig. 7 F and G). Also, the neuromuscular junctions of the gastrocnemic muscle of 2-year-old BN1 mice did not show increased levels of denervated neuromuscular junctions compared with controls (Fig. 7 E). Together these data indicate that impairment of dynein/dynactin by BICD2-N at levels that cause neurofilament and Golgi abnormalities does not necessarily cause the premature loss of motor neurons.
BICD2-N expression increases survival of SOD1-G93A ALS mice
The pathological mechanisms that cause motor neuron degeneration in SOD1-ALS suggest that disruptions in axonal transport may play a significant role ( 18 , 24 , 45–47 ), possibly via a direct interaction of mutant SOD1 with dynein ( 48 ). To examine whether the inhibition of dynein/dynactin function by BICD2-N affects disease progression and lifespan in SOD1-ALS, we crossed our BICD2-N transgenic mice with G1del mice, a transgenic ALS mouse model that expresses human SOD1 with the G93A mutation and develops a fatal progressive motor neuron disease ( 49 , 50 ). Unexpectedly, double-transgenic BN1/G1del mice showed a delayed onset of motor symptoms compared with G1del mice (225 ± 8, versus 189 ± 6 day, respectively), and increased survival (271 ± 8 versus 237 ± 5 day, respectively; Fig. 8 A and B). Accordingly, also, crossing of G1del mice with the BN3 line, which express BICD2-N at lower levels though in a higher percentage of motor neurons (as described earlier), resulted in increased survival (256 ± 8 versus 236 ± 5 day; Fig. 8 A and B). Western blot analysis of spinal cord homogenates showed that expression levels of the mutant SOD1 protein in each BN1/G1del mice was the same as in G1del mice (Fig. 8 C), indicating that the difference in disease phenotype cannot be explained by altered mutant SOD1 expression levels. End-stage BN1/G1del mice showed similar levels of motor neuron loss compared with G1del mice. Furthermore, the loss of GFP-BICD2-N- expressing motor neurons was proportional to the loss of motor neurons labelled for CGRP (a peptide that is expressed in a subset of predominantly large motor neurons), indicating that BICD2-N expression delayed but not prevented motor neuron degeneration.
Dynein/dynactin accumulates in dendritic ubiquitinated SOD1 aggregates in SOD1-G93A ALS mice
Dendritic ubiquitinated SOD1 aggregates represent an early pathological feature preceding motor neuron loss in G1del mice ( 4 , 50 ). Ultrastructurally, these aggregates consist of disorganized filaments, amorphous electron dense material and vesicular structures ( 4 , 50 ). Systematic ultrastructural analysis of pre-embedding ubiquitin immunoperoxidase-stained ultrathin sections from BN1/G1del spinal cord showed that all ubiquitinated dendritic aggregates identified in this material ( n = 27) had the same morphological features as aggregates previously identified in G1del mice (Fig. 8 E). Confocal immunofluorescence further showed that ubiquitinated dendritic aggregates occurred in both GFP-BICD2-N-expressing and GFP-BICD2-N-negative motor neurons in double-transgenic G1del/BN1 mice. Consistent with the notion that mutant SOD1 interacts with dynein/dynactin ( 48 ), we observed that the dendritic aggregates were immunoreactive for p150Glued (Fig. 8 F) and dynein heavy chain (not shown). Aggregates in GFP-BICD2-N-positive dendritic profiles showed higher levels of p150Glued and dynein heavy chain immunoreactivity compared with GFP-BICD2-N-negative profiles (Fig. 8 F). Furthermore, also, GFP-BICD2-N was present in the ubiquitinated SOD1 aggregates. The increased dynein/dynactin levels in dendritic SOD1 aggregates in GFP-BICD2-N-expressing neurons raise the possibility that dynein/dynactin trapped in intracellular inclusions might be beneficial to the disease phenotype in SOD1-linked ALS.
In the present study, we have generated a new mouse model with impaired dynein/dynactin function by taking advantage of the properties of dynein/dynactin-interacting protein BICD2 ( 32 , 33 ). We show that BICD2-N causes accumulation of the dynein motor complex in the neuronal cell body and impairs retrograde axonal transport. Accordingly, the BICD2-N mice develop giant neurofilament swellings in the proximal axon, a feature that is consistent with reduced dynein/dynactin function ( 19 , 28 ). Despite the accumulation of dynein/dynactin components in the perykaryon and proximal dendrites, motor neurons expressing BICD2-N also develop abnormalities in this compartment, i.e. fragmentation of the Golgi apparatus. Golgi fragmentation is a well-established consequence of dynein/dynactin inhibition ( 29 , 41 , 51 ), but here we show for the first time that it can be induced via dynein/dynactin inhibition in neurons in vivo . Our data indicate that only motor neurons expressing relatively high levels of BICD2-N showed Golgi fragmentation, and that the frequency of neurons with fragmented Golgi increased with ageing. These data suggest that Golgi fragmentation is a phenomenon that requires a certain threshold of dynein/dynactin inhibition to occur. Golgi abnormalities have not been reported for other mutant mouse models with dynein/dynactin abnormalities. However, embryonic fibroblasts from mice homozygous for the Loa mutation in the dynein heavy chain gene show impaired Golgi restoration after Golgi fragmentation induced by the microtubule depolymerizing agent nocodazole ( 16 ).
We have generated the Thy1.2-GFP-BICD2-N mice as a dynein/dynactin loss-of-function mouse model to study the pathological aspects of ALS and related motor neuron diseases. Accordingly, our mice develop well-established features of ALS motor neurons, i.e. giant proximal neurofilamentous axonal swellings ( 26 , 27 ) and Golgi fragmentation ( 30 , 31 ). However, despite these abnormalities, our mice up to the age of 2 years did not develop motor abnormalities or signs of motor neuron degeneration. Even the motor neurons with fragmented Golgi did not show signs of illness. Thus, our data indicate that axonal neurofilament abnormalities and Golgi fragmentation are ALS phenomena that can be explained by impaired dynein/dynactin function, but that are not necessarily linked to neuronal degeneration. In contrast to our data, LaMonte et al . ( 15 ) have shown that disruption of dynein/dynactin function in neurons by dynamitin p50 overexpression causes a late-onset progressive motor neuron disease that has been linked to dynein-based axonal transport deficits ( 15 , 19 ). The motor neuron disease phenotype in these mice may be explained by a higher level of dynein/dynactin inhibition owing to very high levels of p50 expression. Accordingly, ‘low’ expressor p50 mice show a much milder phenotype ( 15 ). Alternatively, the inhibition of dynein/dynactin function by p50 is caused by dynactin disruption ( 39 ), which may differentially affect long-term motor neuron survival. Data from other dynein/dynactin mouse models have indicated that subtle specialized defects may underlie motor neuron abnormalities in these mice. For instance, comparison of mouse models carrying p150Glued with the G59S mutation has shown that the development of progressive motor neuron degeneration correlates with the formation of p150Glued aggregates ( 10–12 ), which is in accord with data from patients ( 52 ) and cultured motor neurons ( 53 ). In addition, data from another transgenic line indicate that G59S-p150Glued may induce abnormalities in motor neuronal lysosomal pathways, axonal calibre and neuromuscular junction morphology in the absence of retrograde axonal transport defects ( 12 ). Retrograde axonal transport as measured by the movements of a fluorescent tetanus toxin fragment was also reported to be normal in embryonic motor neurons from heterozygous Loa mice that carry a dynein heavy chain mutation and develop mild motor neuron degeneration at a progressed age ( 16 , 54 ). Recently, it has been shown that heterozygous Loa mice, as well as mice carrying another mutant allele of dynein heavy chain, termed Sprawling ( Swl) , exhibit significant prenatal degeneration of sensory proprioceptive dorsal root ganglion neurons ( 55 ). However, Swl mice, in contrast to Loa and Cra1 mice (i.e. another dynein heavy chain mutant), do not show late-onset motor neuron loss, indicating that this phenotype depends on a feature shared by the Loa and Cra1 dynein heavy chain mutants but not the Swl mutant and supporting the notion that motor neurons are vulnerable to specific abnormalities in dynein/dynactin function ( 16 , 17 , 55 ). Precise comparison of dynein/dynactin dependent-transport defects and pathological abnormalities between dynein/dynactin mouse models, including our BICD2-N mice, may help uncovering dynein/dynactin defects that contribute to motor neuron pathology.
In this study, we also show that the BICD2-N transgene increased lifespan of ‘low-copy’ G93A-SOD1 mice that develop an ALS-like motor neuron disease. This finding is consistent with the demonstration that also G93A-SOD1 mice carrying the Loa and Cra1 dynein heavy chain alleles show increased lifespan ( 54–56 ). Furthermore, the disease phenotype of SOD1-ALS mice that were heterozygous for G59S-mutant p150Glued or Swl -mutant dynein heavy chain was unaltered ( 10 , 55 ), indicating that interfering with dynein/dynactin function in SOD1-ALS mice is either beneficial or neutral. These data challenge the notion that the inhibition of dynein/dynactin-dependent processes is a contributing factor in SOD1-ALS pathogenesis ( 18 , 57 ). It has been proposed that Loa and Cra -mutant dyneins compensate or counteract axonal transport abnormalities triggered by mutant SOD1 ( 17 , 18 , 54 ). Alternatively, dynein/dynactin inhibition may attenuate functions that are potentially harmful to mutant SOD1-expressing motor neurons such as retrograde transport of axonal debris or protein aggregates ( 22 , 23 ), or deleterious retrograde signalling ( 25 , 58 , 59 ). Motor neurons in SOD1-ALS mice show early expression of axonal injury factors, including ATF3 ( 4 , 50 , 60 ). As BICD2-N-expressing motor neurons show attenuated axonal injury ATF3 response, a delay of ATF3 expression could contribute to the beneficial effect of BICD2-N in SOD1-ALS mice. We also observed an increased accumulation of dynein heavy chain and p150Glued in SOD1 aggregates in BICD2-N/SOD1-ALS mice, suggesting that trapping dynein/dynactin complexes at intracellular inclusions could restrain deleterious retrograde signalling.
Together the data from distinct dynein/dynactin mouse models, including our BICD2-N mice, indicate that partial inhibition of dynein/dynactin functions does not necessarily lead to motor neuron death, but is rather beneficial in some forms of motor neuron disease, such as SOD1-linked ALS. On the other hand, specific mutations in dynein/dynactin components may trigger preferential degeneration of motor neurons via specific gained properties such as the formation of aggregates.
MATERIALS AND METHODS
BICD2-N and SOD1-G93A transgenic mice
Animals were housed and handled in accordance with the Principles of Laboratory Animal Care (NIH publication No. 86-23) and the guidelines approved by the Erasmus University animal care committee. To generate Thy1.2-GFP-BICD2-N mice, a GFP-BICD2-N construct ( 32 ) was cloned into the Xho I site of the Thy1.2 expression vector and injected into fertilized oocytes, using standard techniques. Three lines, BN1, BN3 and BN4, were selected for further study. Transgenic lines were maintained into FVB background by crossing hemizygote males with non-transgenic females. Transgenic offspring was genotyped by PCR. A selected group of all lines was allowed to age for 2 years. These mice were weighed and inspected for signs of muscle weakness once a week, using a set of simple tests: mice were examined for their ability to extend their hind limbs when suspended in the air by their tail and their ability to hang upside down on a grid for 60 s. In addition, at specific ages, grip strength was measured using a grid attached to a force gauge (Bioseb, Chaville, France).
Other mice used in this study were G1del mice that carry a genomic hSOD1 construct with the G93A mutation and that develop progressive muscle weakness from age 24–30 weeks, reaching end-stage disease 3–10 weeks after the first symptoms ( 49 , 61 ). Double-transgenic mice carrying the G1del and the GFP-BICD2-N transgenes were generated by crossing hemizygous G1del mice with hemizygous BN1 or BN3 mice. The onset of symptoms was determined on the basis of the onset weight loss, the inability to normally extend one of the hind limbs or the inability to normally perform in the grid hanging test. Mice reached end-stage disease when they could not right themselves within 5 s when placed on their back, lost more than 30% of their maximal weight or developed infection of one of the eyes ( 4 ).
Primary antibodies [supplier; applications (IHC, immunohistochemistry; IF, immunofluorescence; WB, western blot) and dilutions] reported in this study are mouse-anti-actin (Millipore, WB 1:10 000), mouse-anti-Arp1 (Sigma, WB 1:2000), rabbit anti-ATF3 (Santa Cruz; IHC and IF 1:1000), rabbit-anti-BICD2 [( 32 ), WB 1:1000], goat-anti-choline acetyltransferase (ChAT, Chemicon, IF 1:500), rabbit-anti-CGRP (Calbiochem, IF 1:10 000), rat anti-CR3 receptor (clone 5C6; Serotec, IHC 1:500), rabbit-anti-dynein heavy chain (Santa Cruz, IF 1:500; WB 1:1000), mouse-anti-DIC74 (Millipore, WB 1:1000), rabbit anti-GFAP (DAKO, IF 1:5000), rabbit-anti-GFP (Abcam, WB 1:1000), mouse-anti-GM130 (BD Biosciences, IF 1:1000), rabbit anti-Hsp25 (Stessgen, IF 1:2000), rabbit-anti-KIF5A (Abcam, IF 1:2000), rat-anti-muscarinic M2-receptor (Millipore, IF 1:200), chicken-anti-NF-M (Millipore, IF 1:4000), mouse- anti-NF-M (Sigma, IF 1:10 000), mouse-anti-p150Glued (BD Biosciences, WB 1:1000; IF 1:500), rabbit-anti-p150glued (Santa Cruz, WB 1:1000), mouse-anti-p50 (BD Biosciences, WB, 1:1000), rabbit-anti-peripherin (Millipore, IHC 1:1000), rabbit-anti- SOD1 (AbSOD100, Stressgen, WB: 1:1000), rabbit-anti-murine SOD1 (AbSOD101, Stressgen, WB 1:1000), sheep-anti-SOD2 (Calbiochem, IF 1:5000), rabbit anti-ubiquitin (Dako; IHC and IF 1:2000); mouse anti-ubiquitin (clone FK2, Affiniti; IF 1:2000); goat anti-VAChT (Chemicon, IF 1:1000).
Secondary antibodies: For avidin–biotin–peroxidase immunocytochemistry, biotinylated secondary antibodies from Vector Laboratories diluted 1:200 were used. FITC-, Cy3- and Cy5-conjugated secondary antibodies raised in donkey (Jackson Immunoresearch, USA), and Alexa488, 568 or 633 conjugated antibodies raised in goat, were used for immunofluorescence. For western blots, HRP-conjugated goat-anti- mouse or goat-anti-rabbit IgG (DAKO) was used at 1:5000.
GFP-BICD2-N and GFP-p50 expression constructs
GFP-BICD2-N and GFP-p50 constructs have been described before ( 32 ). For expression in hippocampal neurons, GFP-BICD2 and GFP-p50 were subcloned into pGW1 expression vectors.
Primary neuron cultures and transfection
Primary rat hippocampal neurons were plated at a density of 75 000 on 15 mm glass coverslips and transfected at DIV13 with GFP, GFP-BICD2-N or GFP-p50 using Lipofectamine- 2000 (Qiagen) as described previously ( 62 ). After 2 days of transfection, neurons were fixed and stained with the antibodies indicated. Representative cells were imaged using a confocal microscope. The appearance of the Golgi apparatus was investigated. p150Glued fluorescence intensities were measured with Metamorph software and differences between control and transfected neurons were analysed using Student’s t -test.
Spinal cord tissue was homogenized in 10 volumes of PBS containing 0.5% Nonidet P-40 and 1× protease inhibitor cocktail (Complete, Roche), sonicated and centrifuged at 800 g for 15 min, and protein concentrations of the supernatants (S1) were determined using the BCA method (Pierce, Rockford, IL, USA). For the preparation of detergent-insoluble extracts, S1 supernatants were centrifuged at 15 000 g for 15 min. After the collection of supernatants (S2), pellets were thoroughly washed five times with PBS-0.5% Nonidet P-40 and then resuspended in sample buffer for SDS–PAGE electrophoresis and western-blotting. Samples containing 1–10 µg protein were electrophoresed on SDS–PAGE gels and blotted on PVDF membranes (Millipore). The membranes were blocked with 5% non-fat dry milk (Bio-Rad) in PBS with 0.05% Tween20 (PBST), incubated in primary antibody, diluted in PBST with 1% dry milk followed by an incubation in secondary antibody, incubated in chemiluminescence reagent (ECL, Amersham), exposed to film or a Kodak Image station and analysed with ImageQuant 2.2 software ( 4 ).
Immunohistochemical and histopathological procedures
For immunocytochemistry and immunofluorescence, mice were anaesthetized with pentobarbital and perfused transcardially with 4% paraformaldehyde. The lumbar and cervical spinal cord were carefully dissected out and post-fixed overnight in 4% paraformaldehyde. Routinely, spinal cord tissue was embedded in gelatin blocks, sectioned at 40 µm with a freezing microtome and sections were processed, free floating, employing a standard avidin–biotin–immunoperoxidase complex method (ABC, Vector Laboratories, USA) with diaminobenzidine (0.05%) as the chromogen, or single-, double- and triple-labelling immunofluorescence ( 4 ). In addition, a selected number of frozen sections were processed for a silver staining procedure that selectively labels dying neurons and their processes. Immunoperoxidase-stained sections were analysed and photographed using a Leica DM-RB microscope and a Leica DC300 digital camera. Sections stained for immunofluorescence were analysed with a Zeiss LSM 510 confocal laser scanning microscope. Quantitative analyses of motor neurons were performed as described before ( 50 ) on serial lumbar 4 (L4) sections immunoperoxidase-stained for ChAT or CGRP.
For the analysis of neuromuscular denervation, medial gastrocnemius muscle from 4% paraformaldehyde-fixed mice was dissected, embedded into gelatin blocks and sectioned at 80 µm with a freezing microtome. Sections were immunolabelled, free floating, for goat-anti-VAChT and chicken- anti-NFM, and motor endplates were labelled with FITC-bungarotoxin (1:500, Molecular Probes). Sections were examined for neuromuscular denervation under a Leica DM-RB fluorescence microscope as described ( 4 ).
mRNA in situ hybridization
In situ hybridization was performed on 30 µm-thick free-floating sections using standard methods with digoxigenin-labelled cRNA probes ( 63 ). Sense and antisense digoxigenin-labelled cRNAs were transcribed from linearized plasmids containing BICD2-cDNA.
Fluorogold retrograde tracing
To determine retrograde axonal transport, we have used retrograde tracing with fluorogold (FluoroChrome, Denver, CO, USA) ( 15 ). Briefly, anaesthetized 30-week-old non-transgenic and BN1 mice received four microinjections of fluorogold (1 µl, 2% in 0.9% saline) into the gastrocnemic muscle. After 48 h, mice were perfused transcardially with 4% paraformaldehyde, sectioned, mounted and analysed with Leica DM-RB epifluorescence microscope and a Zeiss LSM 510 META confocal laser scanning microscope with 63× Plan-apo oil immersion objective. Fluorogold signal was detected using a 405 nm laser and a META detector.
Axotomy of the sciatic nerve
Twenty-week-old BN1 and non-transgenic animals were anaesthetized, the sciatic nerve was exposed, bound with suture and cut just above the division of the sciatic nerve into the tibial and common peroneal nerves. A 2 mm piece of the nerve was removed. Animals were left to recover for 12 or 24 h. Following transection, animals were perfused transcardially with 4% paraformaldehyde and processed for immunohistochemistry as described before with antibodies against ATF3.
Analysis of immunofluorescence signal intensities
Analyses of GFP-BICD2-N, fluorogold or immunofluorescence signal intensities were performed with sections from spinal cord specimen from non-transgenic and transgenic mice embedded in a single-gelatin block to minimize variability due to sectioning and staining procedures. Images were taken using a Zeiss LSM 510 confocal laser scanning microscope using 40× or 63× Plan apo oil immersion objectives. Laser and detector settings were chosen to avoid saturation of the signal. Fluorescent intensities were determined using Metamorph image analysis software. For some analyses, GFP-BICD2-N-positive motor neurons were grouped according to GFP signal intensity into low (25–100), medium (100–175) and high (175–250) intensities. Therefore, all GFP-BICD2-N images were taken using the same confocal and laser settings in all material. Statistical analyses were done with GraphPad Software (Prism, San Diego, CA, USA).
For electron microscopy, mice were perfused transcardially with 4% paraformaldehyde with 0.1% (pre-embedding immunoperoxidase electron microscopy) or 1% glutaraldehyde (standard electron microscopy). Specimens were sectioned with a Vibratome and further processed using standard methods as described before ( 50 ). Vibratome sections (50–60 µm thick) were post-fixed in 1% osmium, dehydrated and embedded in Durcupan. Ultrathin (50–70 nm) sections were contrasted with uranyl acetate and lead citrate and analysed in a Phillips CM100 electron microscope at 80 kV.
This work is supported by Prinses Beatrix Fonds and Hersenstichting Nederland grants to C.C.H and D.J. Work in the laboratory of C.C.H. is supported by the Netherlands Organization for Scientific Research (NWO-VIDI), European Science Foundation [European Young Investigators (EURYI) Award] and ALS Association (ALSA).
We thank S.A. Spangler and N. Keijzer for preparing primary neuronal cultures, Dr J.C. Holstege for assisting with fluorogold tracing and A. Hossaini for assisting with in situ hybridization experiments.
Conflict of Interest statement . None declared.