|Home | About | Journals | Submit | Contact Us | Français|
Retrograde axonal transport of cellular signals driven by dynein is vital for neuronal survival. Mouse models with defects in the retrograde transport machinery including the Loa mouse (point mutation in dynein) and the Tgdynamitin mouse (overexpression of dynamitin) exhibit mild neurodegenerative disease. Transport defects have also been observed in more rapidly progressive neurodegeneration, such as that observed in the SOD1G93A transgenic mouse model for familial ALS. Here we test the hypothesis that alterations in retrograde signaling lead to neurodegeneration. In-vivo, in-vitro and live cell imaging motility assays show mis-regulation of transport and inhibition of retrograde signaling in the SOD1G93A model. However, similar inhibition is also seen in the Loa and Tgdynamitin mouse models. Thus, slowing of retrograde signaling leads only to mild degeneration and cannot explain ALS etiology. To further pursue this question, we used a proteomics approach to investigate dynein-associated retrograde signaling. These data indicate a significant decrease in retrograde survival factors including P-Trk and P-Erk1/2, and an increase in retrograde stress factor signaling, including P-JNK, Caspase-8 and p75NTR cleavage fragment in the SOD1G93A model; similar changes are not seen in the Loa mouse. Co-cultures of motor neurons and glia expressing mutant SOD1 (mSOD1) in compartmentalized chambers indicate that inhibition of retrograde stress signaling is sufficient to block activation of cellular stress pathways and to rescue motor neurons from mSOD1-induced toxicity. Hence, a shift from survival-promoting to death-promoting retrograde signaling may be key to the rapid onset of neurodegeneration seen in ALS.
Neurons are highly polarized cells, with extended processes that depend on active intracellular transport. The molecular motor kinesin drives anterograde transport from the cell body outward along the axon. The dynein/dynactin complex drives retrograde axonal transport, moving signaling molecules like neurotrophins from cell periphery to cell body
Neurotrophins released by postsynaptic targets promote neuronal survival by activating Trk receptors (Chao, 2003). Activated Trk receptors, together with effector molecules like Erk1/2 and Erk5, are retrogradely transported along the axon to the cell body to evoke changes in gene expression required for cell survival (Yano et al., 2001; Delcroix et al., 2003; Ye et al., 2003; Heerssen et al., 2004). Neurotrophins can also bind to the p75 neurotrophin receptor (p75NTR ), to mediate retrograde signaling. The activation of p75NTR may promote survival or trigger death, depending on differential interactions with ligands and co-receptors (Lee et al., 2001; Chao, 2003; Barker, 2004; Lu et al., 2005); cleavage of p75NTR may also cause cell death (Lee et al., 2001; Kenchappa et al., 2006). Both the Trk and p75NTR signaling pathways from cell periphery to cell body depend on the active transport of signaling endosomes by dynein (Delcroix et al., 2003; Heerssen et al., 2004; Deinhardt et al., 2006).
Defects in dynein function have been linked to neurodegeneration (Chevalier-Larsen and Holzbaur, 2006). Mice with either targeted disruption (LaMonte et al., 2002) or a specific mutation affecting dynein function (Hafezparast et al., 2003; Chen et al., 2007) display degeneration of motor and/or sensory neurons. In humans an autosomal dominant mutation in the dynein activator dynactin results in slowly progressive motor neuron (MN) degeneration (Puls et al., 2003; Puls et al., 2005).
Axonal transport defects have also been suggested to contribute to the pathogenesis of more rapidly progressive neurodegenerative disease such as Amyotrophic Lateral Sclerosis (ALS). Delays in slow axonal transport occur just prior to onset of clinical symptoms in a mouse model of familial ALS expressing mutant SOD1 (mSOD1) (Zhang et al., 1997; Williamson and Cleveland, 1999). However, defects in retrograde transport occur much earlier in disease pathogenesis (Kieran et al., 2005; Ligon et al., 2005). Accumulating data from other disease models also suggest that inhibition of axonal transport leads to neurodegeneration (Chevalier-Larsen and Holzbaur, 2006), presumably through inhibition of neurotrophic signaling.
Here we investigate the relative contributions of the slowing of axonal transport and changes in retrograde signaling along the axon to the pathogenesis of neuronal degeneration. We find that impairment in the efficiency of retrograde neurotrophic factor transport is not sufficient to induce the pronounced neuronal loss seen in severe neurodegenerative diseases such as ALS. However, pronounced changes in dynein-associated cargo from survival-promoting to death-promoting signaling pathways are observed as an early event in neurodegeneration. Further, we demonstrate that inhibition of retrograde stress signaling is sufficient to inhibit activation of markers of nuclear stress and to rescue motor neurons from mSOD1-induced toxicity. Thus, a change in the nature of retrograde signaling may be key in the rapid onset of neurodegeneration seen in ALS.
SOD1G93A (B6SJL-TgN[SOD1G93A]1Gur) mice (Gurney et al., 1994) from Jackson Labs were maintained as hemizygotes in a B6SJL F1 background. wtSOD1 (B6SJL-TgN[SOD1]2Gur) were a kind gift from Dr. Piera Pasinelli. Loa/+ hetrozygote mice (Hafezparast et al., 2003) were generously provided by Dr. Elizabeth Fisher (Institute of Neurology, University College London ,UK) and maintained at the University of Pennsylvania. Dynamitin transgenic (M21) mice (Tgdynamitin) were described previously (LaMonte et al., 2002; Ross et al., 2006). For vesicle assays, GFP-dynamitin (M30) transgenic mice (Ross et al., 2006) were crossed with SOD1G93A mice. For all studies, littermates negative for transgene insertion or point mutation were used as age-matched controls. The IACUC committee at the University of Pennsylvania approved all animal protocols.
DRG cultures from adult mice were dissociated with 100 U of papain followed by 1 mg/ml collagenase-II and 1.2 mg/ml dispase. The ganglia were then triturated in Hank’s Buffered Saline Solution (HBSS), (10 mM Glucose, and 5 mM HEPES ; pH 7.35). Neurons were purified through 20% percoll, plated on laminin, and grown in F12 medium.
Motor neurons were purified from Sprague-Dawley rat embryos at embryonic day 14 -15 by immunopanning (Henderson et al., 1993; Fryer et al., 1999). After 2 days in culture, cover slips with neurons were moved into dishes with glial cells and infected with herpes simplex virus (HSV) encoding either LacZ, wtSOD1 or the G85R mutant form of SOD1 (Neve et al., 1997). The cells were fixed 2 day later and immunostained. Analysis of the cultures by immunofluorescence indicated that >97% of infected neurons expressed SOD1 both in the soma and throughout the neuronal processes. For the compartmental cultures, dishes consisted of a Teflon divider (CAMP10, Tyler Research Instruments, Edmonton, AB, Canada) seated with silicone grease (Dow Corning, Midland, MI) into a poly-l-lysine and laminin-coated 35-mm Nunc tissue culture dishes. MN were plated at 4 × 104 cells in a 1.5 mm-wide center compartment. Axons and dendrites elongate across 1 mm-wide partitions to side compartments as previously described (Campenot, 1992). Motor neurons in the center chamber were maintained in Neurobasal medium with 2% horse serum, 5% B27 (Invitrogen), 0.5 mM Glutamax (Invitrogen), 2-mercaptoethanol (50 μM, Invitrogen), 1% penicillin-streptomycin, 1 ng/ml of CT-1, GDNF, CNTF, BDNF, and 5 μM 5-Fluoro-5′-deoxyuridine (Sigma). Medium in the side compartments was supplemented with 50ng/ml BDNF to promote neurite outgrowth. After 2 weeks in culture, BDNF was reduced to 1 ng/ml throughout the experiments. Motor neurons were labeled by addition of chloromethyl Dill (CM-Dill; 1 μM, Invitrogen) to the side chamber overnight at 37°C. In some experiments, 200 × 103 glial cells isolated from neonatal rat brain were plated in the side chamber with neurites. An inhibitory cocktail: 100 nM of SP600125 to inhibit JNK; 2 μM Z-IETD-FMK to inhibit Caspase-8; 1:100 antibody against the extra-cellular domain of p75NTR (AB8874, Abcam) and HSV-p50 was added to the distal chamber as noted.
Adult DRG cultures were observed 2-3 days after plating in glass bottom microwell dishes (FluoroDish; World Precision Instruments, Inc.) containing Hibernate-A low-fluorescence medium (Brain Bits, Springfield, IL) supplemented with 2% (v/v) B27 supplement (Invitrogen), 2 mM glutaMAX (Gibco), 0.4% D-glucose. The imaging chamber and objective were maintained at 37°C; cells remained viable for several hours. All imaging was performed using a Leica DM IRBE microscope with a 63 plan-apo objective. Digital images were taken with a Hamamatsu MTI RC300 CCD camera using OpenLab acquisition and analysis software. For time-lapse, images were acquired every 1-3 sec. ImageJ was used to analyze vesicles velocity and to make kymographs.
NGF-Qdot imaging was performed with biotinylated NGF (kind gift from Alomone labs), which was incubated on ice for 30 min with Qdot655, conjugated with streptavidin (Invitrogen), at a 10:1 molar ratio. 2 nM NGF-Qdots were added to the DRG cultures for 30 min; cells were washed 3 times prior to imaging. To ascertain that the coupling of NGF to Qdots did not reduce or inhibit biological function of the growth factor, we tested for activity using PC12 cells. There was no difference in PC12 cell differentiation or P-Trk activation in response to incubation with NGF-Qdots as compared to NGF treatment alone (100 ng/ml) (Fig. S4). Most of the internalized NGF-Qdots showed forward and backward movements with no net directionality over the time course of the experiment. Only vesicles that showed clear net transport towards the cell body for retrograde and from the cell body for anterograde, moving at least 10 μm at more than 0.2 μm/sec average velocity, and that could be unambiguously tracked over the time course were analyzed for live cell imaging experiments.
Ligation experiments were conducted with 85-day-old SOD1G93A (pre symptomatic), 6 month Loa/+, and 12-month Tgdynamitin mice and their age-matched controls. Mice were anesthetized with ketamine and xylazine, and sciatic nerves were double ligated with 4.0 silk for 3 hr. Segments immediately distal and proximal to the ligation site were subjected to western blot analysis. Distal or proximal ligation segments were pooled from 6 sciatic nerves from 3 mice; the experiment was replicated at least 3 times for each line. Axoplasm was obtained from ligation segments by gentle squeezing into phosphate-buffered saline (PBS) containing protease inhibitors (Roche) and the phosphatases inhibitors: 30 mM sodium fluoride; 40 mM ß-glycerophosphate; 20 mM sodium pyrophosphate; 1 mM sodium orthovanadate. Multiple SDS-PAGE gels (10 μg/lane) were run from each ligation pool and probed with the indicator antibodies. Western blots were quantified and normalized to GAPDH using ImageJ. Because Erk and JNK pathways are activated minutes after sciatic nerve injury such as ligation, we focused on changes in total levels of these proteins rather than assessing relative levels of phosphorylated forms of these signaling molecules.
His-tagged full length recombinant dynein intermediate chain was expressed in E. coli and purified by Ni2+ affinity chromatography, then coupled to activated CH-Sepharose 4B beads (Karki and Holzbaur, 1995). Axoplasm (10 mg pooled from spinal cord and sciatic nerve for the Kinex antibody microarray or 1 mg from sciatic nerve for verification pull downs), was obtained from dissected nerves after gentle compression in phosphate-buffered saline (PBS) containing protease inhibitors (Roche) and the phosphatases inhibitors: 30 mM sodium fluoride; 40 mM ß-glycerophosphate; 20 mM sodium pyrophosphate; 1 mM sodium orthovanadate. The axoplasm was precleared for 1 hr with control activated CH-Sepharose 4B beads, followed by overnight incubation with DIC beads. The beads then were washed extensively and eluted with 0.1% trifluoroacetic acid followed by neutralization, dialysis (Slide-A-Lyzer, Pierce) and centricon (Millipore) concentration (for the microarray) or by boiling in SDS-PAGE sample buffer before loading on gels for Western blot analyses (for verification pull downs).
The Kinexus microarray chip has 600 signaling antibodies in duplicate. The screen was repeated twice and the 4 sets of data were normalized and analyzed. We focused on signals showing at least a 30% difference consistently in both screens. As a secondary analysis, we compared the ratio of signal from mSOD1 to n.Tg for each protein on the two chips. Finally we create a consensus list of proteins that showed more than 30% difference in both methods of analysis.
Vesicles from spinal cord and sciatic nerve were purified as described (Caviston et al., 2007). Briefly, 100,000 × g spinal cord and sciatic nerve axoplasm pellet was fractionated by flotation through a sucrose step gradient with steps of 0.6, 1.5, and 2.0 M sucrose. The vesicles were isolated from the 0.6-M to 1.5-M interface. The experiment was repeated twice for Loa and mSOD mice and littermate controls, and one time with wtSOD1 mice.
The following antibodies were from Chemicon International: DIC-dynein intermediate chain, (MAB1618); ß-gal, (MAB3468); kinesin heavy chain (KHC), (MAB1614); MAP2, (AB5622); NGF, (AB1526); BDNF, (AB1779). The following antibodies were from Cell Signaling: Bim, (#4582); cleaved caspase-8, (#9496); caspase-8, (#9746); ERK5, (#3372); P-ERK5, (#3371); ERK1/2, (#9102); P-ERK1/2, (#9101); JNK, (#9258); P-JNK, (#9251); p38, (#9212); P-p38, (#9211); P-TRK, (#9141) P-cJun, (#9261). The following antibodies were from BD Transduction Laboratories: GM130, (#610823); p150, (#610474); p50, (#611003). The following antibodies were from Abcam: GAPDH, (#ab9484); p75NTR extracellular domain, (#ab8874); Rab5, (#ab18211); Rab7, (#ab50533). Intracellular domain antibody for p75NTR was from Upstate cell signaling solutions (#07-476). The SOD1 antibodies were from Santa Cruz (#sc-11407) and Sigma (S-2147). Antibodies to active caspase-3 was from Biovision, (#3015-100), to NFH N52 from Sigma, (N0142), and to neurofilaments from Sternberger (SMI31 and SMI32). Synapthotagmain antibody (#SYA-130) was from StressGene.
Western blots and immunostainings were carried out according to standard protocols. For immunofluorescence, DRG neurons were fixed with cold 3% paraformaldehyde containing phosphatase inhibitors for 20 min, while motor neurons were fixed in 4% paraformaldehyde for 10 min. Hoechst staining was used on fixed cells, at 0.025 μg/ml final concentration. Fluorescence intensity quantification was performed using ImageJ. A line was drawn along the axons and the total intensity and peaks along the line profile were measured per μm. To measured caspase-3 activation, percentage cells positive for activated caspase-3 from >100 cells from three independent experiments were analyzed.
Cytoplasmic dynein associated with dynactin was purified from mouse brain by microtubule-affinity ATP extraction and sucrose gradient centrifugation as described (Ross et al., 2006). Gliding assays comparing purified dynein from 125-day-old mSOD1 and wild-type mice were done as described (Ross et al., 2006). For in-vitro vesicles assays, polarity-marked microtubules were generated from 5mg/ml tubulin polymerized at a ratio of one rhodamine-labeled tubulin dimer per 5 unlabeled dimers then sheared with a syringe, and further incubated at 37°C with 1.7 mg/ml tubulin with one rhodamine-labeled tubulin dimer per 50 tubulin dimers. Polarity-marked microtubules were stabilized with 20 μM taxol. A flow chamber was made from two cover slips and double-sided tape. Microtubules were allowed to bind to the coverslip for 5 minutes. Unbound microtubules were washed out with wash buffer (5 mg/ml casein, 10 mM DTT, and 20 μM taxol in Motility Assay Buffer). Next, vesicle solution [50% vesicles (v:v), 100 μM ATP, glucose oxidase, glucose, and catalase in wash buffer] was flowed into the chamber. Vesicles were imaged with total internal reflection fluorescence microscopy. Image sequences were recorded at ~ 2-8 frames per second.
To obtain green vesicles we crossed mSOD1 mice to a line of transgenic mice (M30) that express low levels of a GFP-tagged dynamitin transgene (Ross et al., 2006). The labeled dynamitin is stoichiometrically incorporated into dynactin, and stably associates with vesicles purified from mouse brain or spinal cord, allowing the visualization of dynein-based vesicular movement along microtubules with high temporal and spatial resolution using total internal reflection fluorescence microscopy (Caviston et al., 2007). These vesicles bound to and translocated along polarity-marked microtubules in-vitro, and they released from microtubules at high ATP concentration, indicating that microtubule based motor proteins associated with the vesicles dictated their attachment to the microtubule.
In order to examine the contribution of axonal transport defects and inhibition of neurotrophic factor signaling to neuronal degeneration, we quantitatively compared transport in the SOD1G93A mouse model of neurodegeneration (Gurney et al., 1994) to both nontransgenic (n.Tg) littermate controls and transgenic mice expressing wild type human SOD1 (wtSOD1). Double ligation assays were performed, comparing the accumulation of motor subunits both proximal to and distal to the ligation site in sciatic nerves. We noted a significant inhibition of retrograde transport (~30%; P<0.01), as measured by accumulation of dynein intermediate chain (DIC) and dynactin subunits (p50 and p150) distal to the block, in mSOD1 mice compared to n.Tg and wtSOD1 controls lines (Fig. 1A). Inhibition of anterograde transport as measured by accumulation of kinesin heavy chain (KHC) proximal to the block, was less marked but also significant (P<0.05) in the mSOD1 mouse model compared to controls (Fig. 1A). There was no significant difference in the total levels of the tested motor subunits in sciatic nerve axoplasm (data not shown). Thus, there is a significant impairment of the motors driving the fast axonal transport in-vivo in the mSOD1 model. This impairment in retrograde transport is similar in magnitude to that observed in models with specific disruption in dynein function including the Loa mouse [point mutation in dynein (Hafezparast et al., 2003)] and the Tgdynamitin mouse [over-expression of dynamitin leading to disruption of the dynein/dynactin complex (LaMonte et al., 2002)] (see Fig. S1).
To characterize axonal transport defects at the cellular level, we established live cell imaging methods that allowed us to monitor both general vesicular transport and transport of a specific signaling molecule. First, we performed live cell-imaging of embryonic rat MN (motor neuron) cultures and assessed general vesicular transport by monitoring the motility of ~500 nm axonal vesicles as seen in phase contrast images. We compared vesicular motility in embryonic rat MN cultures that were infected with either SOD1G85R mSOD1 or wtSOD1 (Supplementary movies S1 and S2). In motor neurons expressing mSOD1 the average velocity of retrograde transport (0.41 μm/sec ± 0.02 n=43) was significantly slower (P<0.01) than that observed in neurons expressing wtSOD1 (0.76 μm/sec ± 0.03 n=85) or in uninfected control cells (0.77 μm/sec ± 0.02 n=81) (Fig. 1B). A similar slowing of transport in neurons expressing mSOD1 was also seen in studies using an exogenous tracer (Kieran et al., 2005) or EGFP tagged marker (De Vos et al., 2007) consistent with a generalized inhibition of retrograde vesicular transport in this model. We also analyzed the distribution profiles of instantaneous velocities (n>2000) for vesicles moving in either the retrograde or anterograde directions (Fig. 1C). We observed no difference in the maximum velocity of vesicular transport in these primary cultures, but did see a clear shift towards slower velocities in neurons expressing mSOD1 in comparison to either neurons expressing wtSOD1 or uninfected controls. Further analysis of vesicle movement shows that vesicles from MN expressing mSOD1 paused more frequently and moved significantly less (P<0.001) than vesicles in MN expressing wtSOD1 or in uninfected controls (73%, 88% and 91% time moving respectively). Anterograde transport was also inhibited in neurons expressing mSOD1 compared to neurons expressing wtSOD1, although the effect on anterograde transport was not as pronounced (Fig. 1B).
Because neurodegenerative diseases are age-related, and it is problematical to culture adult MN we also used dorsal root ganglion (DRG) neurons cultured from adult mice expressing mutant or wild type SOD1 as a model for late-onset neurodegeneration, using an 85 day time point which corresponds to the presymptomatic phase of disease. The mSOD1 cultures are viable but grow shorter neurites (the average neurite length after two days in culture was 994 μm±96 for neurons from mSOD1 mice and 1214 μm ±66 for n.Tg neurons; n>100; P<0.05). DRG neurons expressing mSOD1 also display significant Golgi fragmentation, a marker for dynein dysfunction that is also seen both in-vivo and in-vitro in MN expressing mSOD1. (Gonatas et al., 2006) (Fig. S2). Similar to our observations from embryonic MN cultures, vesicles in DRG neurons cultured from 85-day mSOD1 mice displayed significantly (P<0.01) slower retrograde and anterograde velocities (Fig. 1D) relative to DRG neurons from n.Tg and wtSOD1 controls. Thus, changes in retrograde transport are not specific to MN in the mSOD1 model, as sensory neurons are also affected.
The inhibition of retrograde transport that we observed both in-vivo (ligation assays) and in live cell imaging assays may be due either to defects in motor function or to misregulation of motor activity. In particular an association between mSOD1 but not wtSOD1 and dynein has been reported (Ligon et al., 2005; Zhang et al., 2007); although this interaction is relatively low affinity and may be indirectly (Fig. S3A-B). The interaction between dynein and mSOD1 is disrupted by mild detergent or further purification of dynein through a sucrose gradient.
We purified dynein from mSOD1 and littermate n.Tg control mice and compared in-vitro motor activity using microtubule gliding assays (Fig. 2A-B). There was no significant difference in the average instantaneous velocity of microtubule gliding for dynein purified from mSOD1 mice (0.27 μm/sec; n=3338), as compared to dynein purified from age matched n.Tg controls (0.26 μm/sec; n=3293). This observation suggests that expression of mSOD1 does not affect dynein function directly. Instead, expression of mSOD1 may lead to differential regulation of the motor, leading to an overall decrease in the efficiency of transport.
To further test our hypothesis that mSOD1 affects dynein-mediated vesicular transport, we conducted in-vitro vesicular motility assays using GFP-dynactin-labeled vesicles purified from mice expressing mSOD1 (M30/mSOD1) and from littermate control mice (M30) that express the GFP-dynactin protein but not the mSOD1 (See Methods). We noted significant differences in the net directional movement of vesicles purified from mSOD1 or M30 mice in in-vitro transport assays. Vesicles purified from M30/mSOD1 mice showed less processive motion (motion in the same direction >300 nm) than vesicles purified from M30 control mice (Fig. 2C-D). The presence of mSOD1 on vesicles (Fig. S3C) did not decrease the instantaneous velocity of movement (340.59±284.13 μm/sec and 386.68±191.11 μm/sec; P=0.4872). However, the average time that vesicles spent moving towards the minus-end or plus-end of the microtubule was always shorter for mSOD1-associated vesicles than for vesicles isolated from M30 control mice. Both the length of time spent moving and the distance moved in a single direction were decreased in mSOD1-purified vesicles, resulting in an overall decrease in directional determination (Fig. S3D). These in-vitro observations suggest that the expression of mutant SOD1 alters the overall regulation of motor function during vesicle transport along the axon. The regulation may be at the level of cargo binding or may affect the coordination of multiple motors attached to a single transport vesicle (Gross, 2004).
In order to determine how retrograde axonal transport defects can lead to neurodegeneration we tested specifically if there is inhibition in neurotrophic factor signaling. We compared the axonal transport of specific survival signaling molecules in the rapidly progressive mSOD1 model compared to n.Tg and wtSOD1 controls. We performed double ligation assays on sciatic nerve, and compared protein accumulation in segments immediately proximal and distal to the site of ligation (Fig. 3). We saw that the net retrograde transport of survival molecules like P-Trk, Erk1/2, Erk5 and the neurotrophic factors NGF and BDNF are significantly inhibited in-vivo by the expression of mSOD1 (about a 2-fold decrease compared to controls), (P<0.01) (Fig. 3A-E). The total levels of those factors did not change in total axoplasm extracts (data not shown).
To examine this inhibition of transport at higher resolution, we monitored the movement of a specific survival cargo, NGF conjugated to Quantum dots (NGF-Qdots), in primary DRG cultures from 85 day old mSOD1 and n.Tg control mice. NGF linked to Qdot retains its biological function (Fig. S4). We noted a significant inhibition of retrograde transport relative to control (P<0.01), (Fig. 3 F-I and Supplementary movies S3, S4, S5). Retrogradely moving NGF-Qdots in neurons expressing mSOD1 often exhibited brief directional reversals and more frequent pauses, while moving in a net retrograde direction, compared to the smoothly processive motility observed for NGF-Qdots in wild type neurons (Fig. 3G). The average velocity of NGF-Qdots in mSOD1 neurons (0.52 μm/sec ± 0.13 n=5) was about half that observed in n.Tg neurons (0.97 μm/sec ± 0.14 n=7) (Fig. 3I). Although there was no significant difference in the maximum velocity (2.36 μm/sec ± 0.13 in mSOD1 neurons and 2.05 μm/sec ± 0.27 in n.Tg), the distribution profiles for instantaneous velocities (n.Tg n=127; mSOD1 n=233) indicated a clear shift to slower velocities in neurons cultured from mSOD1 mice (Fig. 3H).
In order to find if inhibition of retrograde trophic factor signaling can explain the severe neurodegeneration seen in the mSOD1 model, we also examined neurotrophic signaling along the axon in mouse models with impaired dynein function: Loa and Tgdynamitin. These mouse models show a similar decrease in the retrograde transport of trophic factors to that seen in the mSOD1 model (Fig. 3 A-E), but develop only mild neurodegeneration. Thus, retrograde transport impairment leading to neurotrophic factor deprivation is not sufficient to induce the pronounced neuronal loss seen in severe neurodegenerative diseases such as ALS.
The changes in motor regulation we observed in the mSOD1 mouse model suggest that there may be alterations in intracellular signaling pathways, and in particular in the dynein-associated cargo, the signaling endosome. In order to test this possibility, we used a proteomics approach with protein microarrays, to identify and compare changes in signaling endosomes being retrogradely transported by dynein in mSOD1 and littermate n.Tg control mice.
We isolated the dynein-associated compartment from 85 day old mSOD1 (pre-symptomatic stage) and age-matched n.Tg mice (Fig. 4) by conducting dynein intermediate chain (DIC) pull-downs assays from an axoplasm-enriched fraction. The DIC pull-down approach allows us to isolate the cargos that interact with dynein directly, such as dynactin (Karki and Holzbaur, 1995), but also enriches for proteins that interact indirectly with dynein, because they are in the same vesicular compartment (Fig. 4B). After verifying the specificity of the enrichment of signaling endosomes components in our dynein pull-downs (Fig. 4B and data not shown), the extracts were subject to protein microarray analysis (Kinexus Bioinformatics Corporation).
Overall, we found that ~74 % of the dynein-associated proteins identified in the screen did not show a significant change (<30% level of difference) in the extent of their association with dynein in extracts purified from either mSOD1 or n.Tg mice. However, ~9% (52 proteins) showed an increased association, and ~17% (102 proteins) showed a decreased association with dynein in axoplasmic extracts from mSOD1 mice as compared to n.Tg controls (Supplementery Table 1). Many of the proteins that showed a decreased association with dynein in the mSOD1 model were proteins from survival pathways (47 proteins, 46% of the total) (Fig. 4C) as defined by the Swiss-Prot database. In contrast, many of the proteins that showed an increased association with dynein in axoplasm from mSOD1 mice were proteins related to stress/death pathways (23 proteins, 44% of the total) (Fig. 4D). Therefore, these results suggest that dynein-associated cargos in mSOD1 mice are changing, and that the balance between survival and death factors is altered.
Of the signaling molecules that were identified in our screen, we focused on several key pathways that we thought might play a role in ALS pathogenesis for further characterization. First we verified the association of these proteins with dynein by performing specific DIC pull-downs experiments using sciatic nerve extracts from mSOD1 or n.Tg mice (Fig. 5). The signaling molecules P-Trk, P-Erk1/2, and P-Erk5 are previously identified components of the signaling endosome that is actively transported in the retrograde direction along the axon (Delcroix et al., 2003). Interestingly, we found that these factors showed decreased association (~60%; P<0.05) with dynein in sciatic nerve extracts from mSOD1 mice (Fig. 5D and F). Furthermore, we found that the interaction of dynein with the pro-apoptotic protein Bim was also decreased (about 70%; P<0.01), potentially releasing Bim to act to promote cell death in response to stress (Puthalakath et al., 1999).
There were also factors that did not show any change in association with dynein in pull-downs from mSOD1 as compared to WT mice (Fig. S5B). We did not see any interaction of Fas receptor with dynein in either mSOD1 or n.Tg mice. Since Fas was shown to play a role in motor neuron death in a pathway which includes p38 and caspase-8 (Raoul et al., 2002), we also looked at p38 and caspase-8 specifically. We saw that phospho-p38 (P-p38) was significantly up-regulated in mSOD1 extracts, however the association of P-p38 with dynein was not altered (P>0.5, Fig. S5B).
We observed increased caspase-8 activation (18 kDa cleavage product), in sciatic nerve axoplasm from mSOD1 mice as compared to n.Tg control. In addition, activated caspase-8 showed a significant increase in its association with dynein (P<0.001) suggesting that this protein may function as a retrograde death signal (Fig. 5G and I). A significant increase in association with dynein was also seen for the stress-activated protein kinase P-JNK (70%; P<0.01) (Fig. 5G and I). Finally, we found that p75NTR receptor is cleaved in the mSOD1 mouse, and its cleavage fragments (~25 kDa) are significantly enriched in dynein pull downs (P<0.001) (Fig. 5G and I). Caspase-8 was shown to retrogradely transfer a death signal in a p75NTR -dependent pathway (Carson et al., 2005). Thus, there are significant changes in dynein-associated cargos in 85 day old, pre-symptomatic mSOD1 mice, indicating that these alterations in signaling pathways are occurring relatively early in the disease process. As changes in retrograde transport may become significant as early as 50 days after birth in the SOD1G93A line (Ligon et al., 2005), we also performed dynein pull downs at an earlier stage. In 50 days old mice as well, we noted decreased association of both P-Trk and P-Erk with dynein, and increased association of P-JNK and p75 cleavage fragment with dynein (Fig. S5A). Thus, the dynein-associated changes we have observed occur early in pathogenesis and are therefore may contribute to the development of frank degeneration of the axon observed later in the disease.
For comparison, we also performed dynein pull down assays on sciatic nerve extracts from Loa mice (point mutation in dynein; (Hafezparast et al., 2003) and age-matched littermate n.Tg controls to see if the nature of the signaling compartment is also changing in a mild model of neurodegeneration. There was no significant change in the extent of association of positive signals associated with neurotrophic signaling pathways, such as P-Trk, P-Erk1/2, and P-Erk5 with dynein in pull downs from 12 month old Loa+/- mice as compared to wild type littermate controls (Fig. 5E, F). Nor did we see an increased association of stress signaling molecules with dynein in Loa heterozygotes, in contrast to the observations in the mSOD1 model (Fig. 5H, I). Thus the changes observed in the nature of the dynein-associated cellular fraction are only seen in mSOD1 mice, which exhibit more rapidly progressive neurodegeneration.
We also looked at whether the change in association with dynein is due to a change in the overall levels of expression in the axoplasm or to a more specific alteration in association with a cellular compartment transported by dynein along the axon. When comparing the input fractions of sciatic nerve extracts isolated from either mSOD1 or n.Tg control mice (Fig. 5 and Fig. S5B), we do not see appreciable differences in the overall levels of proteins, such as P-JNK, P-ERK, although we do see significant changes in the association of these factors with dynein (P<0.01). In contrast, we do see significant changes in the p75NTR cleavage fragment and activated caspase-8 (Fig. 5G) (P<0.01), which are significantly up-regulated in axoplasm and also show an enhanced association with dynein. Thus, changes in the total levels of the different factors in axoplasm can not explain the differential association with dynein observed in these assays.
In order to further verify changes in the signaling endosomes, we isolated axonal transport vesicles (Caviston et al., 2007) from mSOD1, Loa and littermate control mice. Reduction in P-Trk and increase in p75NTR cleavage fragment were shown to associated with vesicles isolated from mSOD1 mice; these changes are not seen in vesicles purified from heterozygous Loa mice (Fig. 5J-K). Thus, the changes we see in dynein-associated signaling molecules in the mSOD1 mouse are due to alterations in the composition of the vesicular cargo that is transported by dynein along the axon and may be a mechanism to control the fidelity of specific retrograde signals.
The dynein pull-down experiments described above showed significant changes in the interaction of the signaling molecules with the retrograde transport machinery. In order to test if those signaling endosomes also move differentially in the nerve we performed ligation assays on sciatic nerve (Fig. 6A-C). Our results show that the net retrograde transport of stress factors such as p75NTR cleavage fragments, JNK, and activated caspase-8 were only seen in the mSOD1 model (~ 3-fold increase compared to n.Tg and wtSOD1 controls; P<0.02) (Fig. 6A-C). Transport of p38, Fas and PI3K were not appreciably changed between the neurodegeneration models (P>0.5) (S 10B).
In order to further analyze the change in the balance of survival and death receptors in degenerating neurons we examined the localization of signaling molecules in primary neuronal cultures expressing either wild type or mutant SOD1. Embryonic rat MN grown on glial cells were infected with HSV encoding either SOD1G85R mSOD1 or wtSOD1, and then stained with antibodies directed against the phosphorylated form of the Trk receptor or the intracellular domain of p75NTR (Fig. 6D-E). We quantified our observations by measuring the fluorescence intensity along the neurite per unit distance, and by quantifying the peaks of intensity along the neurite per mm (Fig. 6G). We observed significantly higher levels of p75NTR receptor and lower levels of P-Trk receptor in motor neurons expressing mSOD1 compared to wtSOD1 control neurons (> 2-fold difference compared to controls, P<0.01) neurites. There were also less P-Trk puncta (> 7-fold difference), and more p75NTR (> 3-fold difference) puncta along the neurite (Fig. 6E) in mSOD1-expressing neurons (P<0.01). These puncta may represent signaling endosomes that mediate signal transport from the periphery to the cell body; line scan analysis is also shown (Fig. 6F). In order to correlate these cellular observations with in-vivo changes, we looked at motor neurons innervating sternomastaid (STM) muscles in mice expressing either wild type or mutant SOD1. As shown in Fig. 6H-J, changes induced by mSOD1 expression in-vivo were very similar to results obtained from primary rat MN cultures. Motor neurons from SODG93A mSOD1 mice showed a striking loss of punctate P-Trk reactivity along the axon, and a clear enhancement of p75NTR immunostaining.
So although overall levels of these proteins in sciatic nerve extract are not altered, their specific association with a vesicular fraction moving along the axon appears to be changed significantly, consistent with the results from the biochemical characterization of the vesicles. Interestingly, rat neurons expressing mSOD1 grown without the presence of glial cells (Fig. 7) or on glia expressing wtSOD1 (data not shown), display no significant changes in P-Trk (Fig. 7A,C-D) and in p75 (Fig. 7B, E-F) levels and puncta along the neurite compare to wtSOD1. Furthermore, co-cultured neurons and glia expressing mSOD1 (Fig. 7G-H) shows a pronounced activation of caspase-3. This activation was not seen in neurons cultured in the absence of glia. Thus, the changes in survival/stress receptors like P-Trk and p75NTR are likely to be non-cell autonomous and may activate neuronal cell death.
In order to provide additional mechanistic insight into the role of the observed shift from survival-promoting to death-promoting retrograde signaling in the mSOD1 mice model, we have developed a cellular assay with motor neurons grown in compartmentalized chambers, with cell bodies in the center chamber and glial cells in the distal chamber (Fig. 8A). These compartmentalized cultures allow us to more explicitly test our hypothesis of the role of retrograde transport in non-cell autonomous stress signaling. Both cells type were infected with mSOD1 or wtSOD1. CM-Dill tracer was added to the distal side to label MN cell bodies that extend axons into the distal chamber. CM-Dill positive cells that appeared vital and healthy were counted after 2 and 6 days. The percent of MN survival was scored after 6 days. CM-Dill positive MN expressing mSOD1 cells shows significant (P<0.01) cell death (~50%) as compared to cells expressing wtSOD1 or to uninfected controls (Fig. 8B and data not shown). Addition of specific inhibitors to JNK, caspase-8 and p75NTR to the distal chambers in the presence of dynamitin overexpression to inhibit retrograde stress signaling led to a significant increase in cell survival (86%; P<0.01; more then 1000 cells were scored in 7 different chambers for each condition). In order to assess the cellular response to retrograde stress signaling, we looked at the down-stream target of JNK activation, the pro-apoptotic transcription factor; phospho-c-Jun. P-cJun is up-regulated in-vivo in spinal cords of mSOD1 mice model compare to n.Tg control (Fig. 8E), (Jaarsma et al., 1996). Immunostaining against P-cJun shows activation in mSOD1-expressing MN (Fig. 8D).
We then used this model to compare the effects of a slowing of retrograde transport via dynamitin overexpression (Fig. 8C) to an inhibition of retrograde signaling from the distal compartment, using a cocktail of inhibitors to JNK, caspase-8 and p75NTR (Fig. 8D). We found that overexpression of dynamitin (p50), which leads to a slowing but not a complete block in retrograde transport (LaMonte et al., 2002) led to a delay in stress signaling to the nucleus, as measured by the activation of c-Jun, which was significantly lower after 3 days in motor neurons expressing mSOD1 and dynamitin (32%) as compared to control cells that express only mSOD1 (69%; p<0.001). However, by 5 days in culture there was no significant difference (Fig. 8C), as with time stress signals arrive at the cell body and activate c-Jun. In contrast, addition of the cocktail of inhibitors to the distal chamber is sufficient to prevent the up-regulation of P-cJun after 5 days (Fig. 8B and D), (p<0.001) in parallel with the rescue of MN cell death.
These observations as well as the results described above support our hypothesis that inhibition of retrograde neurotrophic factor signaling leads to mild neurodegeneration. However, in the mSOD1 mouse model there are also changes in the nature of dynein cargo from survival to stress signaling likely mediated by glia, that lead to rapid, severe neurodegeneration (Fig. 9).
Defects in axonal transport are implicated in a range of neurodegenerative diseases, including ALS, Huntington’s Disease, and Alzheimer Disease. Here we describe for the first time a complete mechanism for how axonal transport defects may lead to severe neurodegeneration. This mechanism shows how extracellular signaling from mSOD1-expressing glia acts through neuronal receptors to activate intracellular stress signals causing downstream activation of stress responses in the neuronal nucleus.
We used in-vivo, in-vitro and live cell imaging assays to fully characterize the axonal transport defects in the SOD1G93A model of familial ALS. We also found that mouse models with impaired dynein function, Loa and Tgdynamitin show similar decreased efficiency of retrograde transport but unlike the mSOD1 model develop only mild neurodegeneration. Therefore, the slowing of neurotrophic factor signaling is not sufficient to induce pronounced neuronal loss. Instead, in the mSOD1 model in the early pre-symptomatic stage, there are profound changes in the nature of the cargo being transported by dynein along the axon; there is an overall shift from survival signaling to stress/death factor signaling. Inhibition of retrograde stress signaling is sufficient to delay activation of cellular stress pathways, as assessed by activation of c-Jun, and to rescue motor neurons from mSOD1-induced toxicity. These results support the model that inhibition of retrograde transport efficiency including the slowed transport of neurotrophic factors leads to only slowly progressive and mild neurodegeneration whereas alterations in the nature of the signals being retrogradely transported, from survival to stress, may leads to severe neuronal dysfunction and cell death as seen in ALS (Fig. 9). Still, it is not clear if axonal transport alterations are the cause for the diseases or are secondary symptoms effect of it. This reason/result question is not easy to answer. Because the axonal transport alterations are an early event, it is reasonable to speculate that these changes play a causative role. However, we cannot exclude the possibility that protein aggregation or mitochondrial dysfunction as well as other factors also play contributing roles in disease pathology. Further studies should lead to more comprehensive understanding of the machinery involved in retrograde signaling along the axon.
Here we examined the mechanisms of inhibition of retrograde transport observed in mSOD1 mice. In-vitro microtubule gliding assays indicate that the dynein motor itself is not impaired. Instead, in-vitro assays for vesicular motility indicate the effects of mutant SOD1 expression may be either direct or indirect, altering the regulatory balance of vesicle-associated motor proteins. For example expression of mSOD1 on the vesicle may interfere with motor-cargo binding and this may disrupt the motor regulation (Friedman and Vale, 1999). Furthermore, mSOD1 may also activate or repress one or more signaling pathways that regulate activity of motors in-vivo through coordination of motor activity. We have shown by live-cell imaging and in-vitro vesicle assays that there is an increase in bidirectional movement and a reduction in unidirectional, processive movement, which may be interpreted as unregulated motor function. These observations are consistent with the suggested role of signaling molecules such as P-JNK in regulating motor activity (Morfini et al., 2006; Horiuchi et al., 2007). JNK interacts with kinesin and the dynein/dynactin complex via JIP and sunday driver (syd) scaffolding proteins (Bowman et al., 2000; Verhey et al., 2001; Yano and Chao, 2004; Cavalli et al., 2005). JNK is downstream of p75NTR effectors and may have a role as a stress/death signal that leads to apoptosis. JNK may also be involved in the regulation of p75NTR cleavage, suggesting the possibility that there is a feedforward loop involved in the local regulation of JNK activation in the axon, leading to slowed transport and thus a further enhancement of localized signaling.
Inhibition of retrograde axonal transport means that vital factors associated with the signaling endosome (Delcroix et al., 2003), including neurotrophic factors and signaling molecules such as P-Trk, Erk1/2 and Erk5, are not effectively reaching the cell body. Impaired communication between target tissues and the neuron cell body may lead with time to the activation of programmed cell death and slow neurodegeneration. Here for the first time we show that neurotrophin inhibition is not sufficient to activate severe and rapidly progressive degeneration such as that seen in the mSOD1 model, and that activation of retrograde death/stress pathways are likely to be key.
Recent progress has suggested that the motor neuron cell death observed in the mSOD1 model results from a non-cell autonomous process (Lino et al., 2002; Clement et al., 2003; Boillee et al., 2006; Di Giorgio et al., 2007; Nagai et al., 2007). In contrast, if expression of mSOD1 is limited to motor neurons only slow axonal degeneration is observed (Jaarsma et al., 2008). Expression of mSOD1 may either activate or repress signaling pathways controlling cell fate. Glia cells expressing mSOD1 may release factors that will differentially trigger neuronal receptors, leading to alterations in dynein associated cargo and neurodegeneration. Indeed, a critical initiating event for the mechanism outlined above may be the differential activation of receptors like P-Trk and p75NTR due to the expression of mSOD1 in the surrounding glial cells. We also found activation of caspase- 8 in mSOD1 neurons, as well as an increased association of caspase-8 with dynein. Interestingly, axoplasmic caspase-8 was shown to be transported back to the cell body by the dynein/dynactin complex in a p75NTR -dependent manner causing cell death, after target removal (bulbectomy) or synaptic instability (Carson et al., 2005). Furthermore, the Fas/p38 signaling pathway was shown to activate caspase-8 leading to cell death in ALS motor neurons (Raoul et al., 2002). Thus, the cellular environment contributes significantly to cell death, leading to a change in the balance between survival and death receptors that enhance the progress of neurodegenerative disease as seen in the mSOD1 model.
We used a proteomics approach to examine changes in dynein cargo more broadly, focusing on changes in cellular signaling pathways. Our results suggest that there is a set of signaling factors that are altered, rather than a single pathway that is affected. These observations may explain the limited effects observed when crossing mSOD1 mice with knockouts in pathways that are thought to be important to disease pathology at the cellular level, such as p75 (Küst et al., 2003) FasL−/− (Petri et al., 2006) or Bax−/− (Gould et al., 2006) null mice. In contrast, crossing Loa mice with mSOD1 mice showed some rescue effects both in axonal transport defects and in the life span of the mice (Kieran et al., 2005). It is possible that in this cross, the mutation in dynein resulted in a delayed arrival of stress/death signals at the cell body, resulting in an extension of life span.
Partitioning of signals into the endosomal compartment may represent a key mechanism contributing to the specificity of a signal transduction pathway (Schenck et al., 2008). The retrograde transport of the signaling endosomes may also ensure spatial and temporal regulation that controls the fidelity of specific signals. Therefore, the specific balance between positive and negative signaling in time and place may mediate the regulation of cell survival, and a change in this balance toward stress/death signaling may lead to aggressive neuronal degeneration and cell death.
This work was supported by ALSA and MDA Postdoctoral Fellowships to EP and grants from ALSA and NIH (R01 NS060698) to ELFH. We are grateful to Elizabeth Fisher for the generous gift of the Loa mice and to Piera Pasinelli for the wtSOD1 mice. We thank Mariko Tokito for invaluable assistance; Phyllis Dan (Alomone labs) for the kind gift of biotin-NGF, Yale Goldman assistance with TIRF microscopy, Rachael Neve for packaging the recombinant HSV, Subhojit Roy for advice on live-cell imaging, and the Mobley lab for advice regarding NGF-Qdot tracking.