Using a novel assay based on the sorting and transport of a fluorescent fragment of tetanus toxin, we have investigated the cytoskeletal and motor requirements of axonal retrograde transport in living mammalian motor neurons. This essential process ensures the movement of neurotrophins and organelles from the periphery to the cell body and is crucial for neuronal survival. Unlike what is observed in sympathetic neurons, fast retrograde transport in motor neurons requires not only intact microtubules, but also actin microfilaments. Here, we show that the movement of tetanus toxin-containing carriers relies on the nonredundant activities of dynein as well as kinesin family members. Quantitative kinetic analysis indicates a role for dynein as the main motor of these carriers. Moreover, this approach suggests the involvement of myosin(s) in retrograde movement. Immunofluorescence screening with isoform-specific myosin antibodies reveals colocalization of tetanus toxin-containing retrograde carriers with myosin Va. Motor neurons from homozygous myosin Va null mice showed slower retrograde transport compared with wild-type cells, establishing a unique role for myosin Va in this process. On the basis of our findings, we propose that coordination of myosin Va and microtubule-dependent motors is required for fast axonal retrograde transport in motor neurons.
Neuronal homeostasis relies on an efficient bidirectional transport system for the correct and timely targeting of cargoes to distant cellular domains. Axonal anterograde transport allows the delivery of newly synthesized proteins, RNA and other membranous and nonmembranous structures from the cell body to distal neuronal sites (Nakata et al., 1998; Goldstein and Yang, 2000; Brown, 2003). Retrograde transport is essential for organelles and ligands to reach the soma from nerve terminals and perform their biological functions (Goldstein and Yang, 2000; Ginty and Segal, 2002; Brown, 2003). Moreover, virulence factors and pathogens, including tetanus toxin (TeNT) and several neurotropic viruses, exploit this transport route to enter the central nervous system (Schiavo et al., 2000; Tomishima et al., 2001; Butowt and Von Bartheld, 2003).
Multiple motors have been associated with axonal transport. Biochemical, pharmacological and genetic approaches identified dynein as the major microtubule (MT)-based motor responsible for driving retrograde transport in neurons (Schnapp and Reese, 1989; Waterman-Storer et al., 1997; Goldstein and Yang, 2000; Vale, 2003). Additionally, minus-end-directed kinesins such as KIFC2 have been linked to retrograde transport, although functional evidence is still lacking (Hirokawa, 1998; Goldstein and Yang, 2000). Actin cytoskeleton and myosin motors have also been implicated in bidirectional axonal transport (Kuznetsov et al., 1992; Bearer et al., 1993). In squid axoplasm, endoplasmic reticulum-derived vesicles containing both myosin and kinesin move on actin filaments (Tabb et al., 1998). Moreover, axoplasmic organelles and mitochondria are transported on both F-actin and MTs (Kuznetsov et al., 1992; Bearer et al., 1993; Morris and Hollenbeck, 1995). These results point to the presence of both actin- and MT-based motors on the same organelle (Brown, 1999). This concept has been strengthened by work in other systems, such as Xenopus melanophores, in which melanosomes are transported bidirectionally on MTs and are dispersed throughout the cytoplasm by myosin V (Rogers and Gelfand, 2000; Langford, 2002; Gross et al., 2002a). Interestingly, myosin V has been shown to interact directly with conventional kinesin, indicating that MT- and actin-based motors could exist in multifunctional complexes transporting organelles along both cytoskeletal systems in a coordinated fashion (Huang et al., 1999; Langford, 2002).
These observations support a `dual filament' model of transport (Langford, 1995; Langford, 2002), predicting a functional cooperation between MTs and the actin cytoskeleton. MT-based motors would ensure long-range transport, whereas myosins could allow local movement of organelles on F-actin in areas devoid of MT as, for example, nerve terminals, growth cones and subcortical plasma membrane regions. Probable candidates for powering such short-range movements are unconventional myosins, such as Va, VI and VII (DePina and Langford, 1999; Wu et al., 2000). However, with the notable exception of myosin V, little is know about the localization and function of unconventional myosins and their role in axonal transport (Bridgman and Elkin, 2000). By contrast, myosin Va has been investigated in detail and is found associated with different nonmembranous cargoes, such as neurofilament NF-L subunit (Rao et al., 2002) and membrane-enclosed organelles. These include melanosomes (Rogers and Gelfand, 1998), endoplasmic reticulum (Tabb et al., 1998; Wöllert et al., 2002), synaptic vesicles (Prekeris and Terrian, 1997), secretory granules (Neco et al., 2002; Rose et al., 2003) and other axonal vesicles (Evans et al., 1998; Bridgman, 1999). Moreover, myosin V-driven motility of distinct organelles is differentially controlled during the cell cycle (Wöllert et al., 2002).
Despite the importance of axonal retrograde transport in neuronal physiology and disease, few studies have focused on the identification of the machinery responsible for this process in mammalian neurons. Furthermore, the direct visualization of retrograde transport and its quantitative analysis have been hampered by the lack of a reliable assay in living cells. We have recently established such an assay in motor neurons (MNs) by using a nontoxic fluorescent fragment of tetanus toxin (TeNT HC) (Lalli and Schiavo, 2002). TeNT HC binds with high affinity to mammalian MNs and colocalizes in nonacidic compartments with p75 neurotrophin receptor (p75NTR) (Herreros et al., 2001; Lalli and Schiavo, 2002). p75NTR undergoes retrograde axonal transport and is important for targeting neurotrophins to the central nervous system (Butowt and Von Bartheld, 2003). Because TeNT HC shares a retrograde route with p75NTR, it provides an ideal probe for studying this important transport process.
In the present paper, we investigate the cytoskeletal requirements and the identity of the molecular motors involved in the axonal transport of TeNT HC carriers. Kinetic analysis of this compartment shows that fast retrograde transport in MNs requires both MT- and actin-based motors. Our results suggest that dynein acts not only as a major retrograde motor but also as a coordinator of other MT-based motors. Furthermore, we find that MNs from myosin Va null mice display a slower, more discontinuous TeNT HC transport compared with wild-type cells, implying that myosin Va is necessary to ensure fast axonal retrograde transport in mammalian MNs.
Materials and Methods
All reagents were obtained from Sigma, unless otherwise specified. Alexa™ Fluor 488, 546, 594 maleimides, Alexa™ Fluor 488 phalloidin and Texas Red-conjugated goat anti-mouse and anti-rabbit secondary antibodies were from Molecular Probes. Monoclonal anti-β-tubulin antibody was from Roche Diagnostics. A polyclonal rabbit antiserum against rat myosin Va tail (aminoacids 1005-1830) (Evans et al., 1997) was a kind gift from P. C. Bridgman (Washington University School of Medicine, St Louis, MO).
Preparation and fluorescent labelling of recombinant TeNT HC
Phosphorylated oligonucleotides (5′-GAT CCG CAG AGG CAG CAG CAC GAG AGG CTT GTT GTC GAG AGT GTT GTG CAC GAG AGG CAG CAG CAC GAG CG-3′ and 5′-AAT TCG CTC GTG CTG CTG CCT CTC GTG CAC AAC ACT CTC GAC AAC AAG CCT CTC GTG CTG CTG CCT CTG CG-3′) were annealed and then ligated into the BamHI/EcoRI site of pGEX-4T3 (Amersham Pharmacia Biotech), resulting in the pGEX-4T3-Cys vector. TeNT HC was excised with SalI/EcoRI from the pGEX-6 vector (Lalli et al., 1999) and ligated into pGEX-4T3-Cys. The resulting plasmid (pGEX-4T3-Cys-TeNT HC) encodes glutathione S-transferase, which is fused to the peptide A EAAAR EACCR ECCAR EAAAR A and to TeNT HC (Fig. 1A). The boxed domains are predicted to form α-helices with the thiol groups properly oriented for labelling with fluorescein arsenical helix binder (FLASH) (Griffin et al., 1998). TeNT HC was purified following standard procedure (Lalli et al., 1999), concentrated through dialysis against 10 mM Hepes-NaOH, pH 7.4, 150 mM NaCl, 15% polyethylene glycol 20,000, 5 mM 2-mercaptoethanol and dialysed overnight against 20 mM Hepes-NaOH pH 7.4, 100 mM NaCl, 10% glycerol, 0.1 mM 2-mercaptoethanol.
Before labelling with FLASH, TeNT HC (1 nanomole) was reduced with 12.5 mM 2-mercaptoethanol for 30 minutes at 37°C. The labelling reaction was performed using 5 nmoles of FLASH-1,2-ethandithiol (Aurora Biosciences) and carried out for 90 minutes at room temperature in the dark. Labelling with the Alexa maleimides was performed according to the manufacturer's instructions.
Retrograde transport assay in living rodent MNs
Rat spinal cord MNs were purified from E14 Sprague-Dawley embryos and maintained in culture as previously described (Henderson et al., 1995). One week after plating, rat MNs were incubated with 40 nM TeNT HC-Alexa488 (unless otherwise indicated) in Neurobasal medium (GIBCO BRL) at 37°C for 25 minutes, washed and imaged after 20 minutes at 37°C with time-lapse low-light microscopy (Lalli and Schiavo, 2002), using an exposure time of 0.3 seconds for each frame. In competition experiments, MNs were pre-incubated at 37°C for 10 minutes with a 100× excess of native TeNT before adding the fluorescent TeNT HC.
Single embryo cultures enriched in spinal cord MNs were obtained by trypsinization of E13 ventral spinal cords of wild-type mice and mice homozygous for the dilute lethal mutation Myo5ad–l (strain DLS/Le a/a Myo5ad–l +/+ Bmp5se; Jackson Laboratories, Bar Harbor, ME) (Mercer et al., 1991). Cells were maintained in culture as described (Arce et al., 1999) and used one week after plating. Phenotyping of dilute lethal embryos was performed by western blot analysis of brain homogenates as previously described (Evans et al., 1997).
Twenty minutes after removing the fluorescent TeNT HC, MNs were incubated at 37°C with 0.5 μM latrunculin B [Lat B, Calbiochem; 1 mM stock in dimethylsulfoxyde (DMSO)] or 5 μM vincristine (Vin; 2 mM stock in methanol). Imaging was resumed 15 minutes after the addition of Lat B or 40 minutes after the addition of Vin. In control MNs, an equal amount of DMSO or methanol (final concentration 0.3% or less) was added and imaging was performed at similar timepoints used for the drug-treated samples. Cells were then fixed and stained for MTs and F-actin (see below).
In other sets of experiments, MNs were incubated with 40 nM TeNT HC-Alexa488 for 25 minutes at 37°C, washed and incubated with the following drugs after 20 minutes: 1 mM erythro-9-(2-hydroxy-3-nonyl)adenine (EHNA; from Calbiochem, 250 mM stock in water), 10 μM aurintricarboxylic acid (AA; 10 mM stock in DMSO) alone or in combination. Imaging resumed 15 minutes after addition of the drugs and stopped before any noticeable change in the morphology of cell bodies or the appearance of neurite blebbing.
Tracking and data quantification
Vesicle tracking was performed on time-lapse sequences using the Motion Analysis software (Kinetic Imaging) as previously described (Lalli and Schiavo, 2002). Speed and average speed values were determined by measuring the distance covered by each carrier between two consecutive frames or between the initial and the final tracking points, respectively. The speed distributions of TeNT HC alone or on treatment with motor inhibitors were analysed with KaleidaGraph (Synergy Software) by applying a multiple Gaussian curve fit. Preliminary analysis with unimodal and bimodal Gaussian profiles failed to satisfactorily describe the speed distributions of control and untreated samples. Optimal fitting was obtained by using the sum of three Gaussian profiles centred at the following speeds: for rat, 0, 0.53 and 1.15 μm/second; for mouse, 0, 0.53 and 0.95 μm/second. Datasets with a correlation coefficient R2<0.8 were excluded from the final analysis. The contribution of single speed components was obtained by calculating the integral of the corresponding Gaussian equation in the range of speeds determined experimentally. Each contribution was expressed as a percentage of the sum of the components. The incidence of reversal was calculated by dividing the number of changes of direction (over a distance larger than 0.2 μm) by the number of organelles analysed. Statistical analysis was performed using SigmaPlot 2000 (SPSS Science) and KaleidaGraph. 8-bit images were assembled into movies with the Kinetic Imaging software using the Microsoft Video 1, 100% quality compression algorithm.
Cells were cooled to 4°C, washed twice with 0.2% bovine serum albumin (BSA) in Hanks' buffer (Lalli et al., 1999) and incubated with 10 nm gold-conjugated TeNT HC (Aurion). After washing three times with 0.2% BSA in Hanks' buffer, cells were either fixed or warmed to 37°C for 5 or 20 minutes before fixing for 1 hour with 2.5% glutaraldehyde in 0.1 M sodium phosphate buffer, pH 7.4 (Sorensen's buffer). Cells were then washed and processed for electron microscopy (EM). Alternatively, cells were incubated at 4°C with 40 nM TeNT HC tagged with the vesicular stomatitis virus protein G epitope (VSVG) (Lalli et al., 1999) for 20 minutes, washed three times with 0.2% BSA in Hanks' buffer and blocked with 2% BSA, 0.25% porcine skin gelatin, 0.2% glycine, 15% fetal calf serum in PBS (blocking buffer) for 15 minutes at 4°C. Cells were then treated with mouse anti-VSVG antibody (1:80) (Lalli et al., 1999) diluted in PBS containing 1% BSA, 0.25% skin gelatin (antibody dilution buffer) for 30 minutes at 4°C. After rinsing in Hanks' buffer, cells were incubated with 10 nm gold-conjugated goat anti-mouse IgG (1:100; British BioCell International) for 25 minutes at 4°C, washed with Hanks' buffer and immediately fixed or warmed to 37°C for 5 or 20 minutes before fixation. MNs were post-fixed with 1% osmium tetroxide for 30 minutes, washed, dehydrated in ascending ethanol series and embedded in araldite over 2 days. Thin sections were stained with methanolic uranyl acetate and lead citrate for morphological examination or aqueous uranyl acetate for immunogold samples. Sections were then observed with a Jeol 1010 transmission electron microscope.
MNs were fixed with 4% paraformaldehyde, 20% sucrose in PBS for 15 minutes at room temperature, washed with PBS, incubated with 50 mM NH4Cl for 20 minutes, rinsed and permeabilized with blocking buffer containing 0.1% Triton X-100 for 15 minutes. Primary antibodies were applied in antibody dilution buffer for 1 hour. After rinsing with PBS, secondary antibodies were applied for 25 minutes. Cells were then washed and mounted with Mowiol 4-88 (Harco). For F-actin detection cells were incubated with Alexa Fluor 488 phalloidin (1:50) in PBS for 20 minutes after blocking. Imaging was performed with a Zeiss LSM 510 confocal microscope using a 63×, 1.40 NA Plan-Apochromat oil-immersion objective.
Rat MNs bind and internalize fluorescent TeNT HC through a specific mechanism
To visualize retrograde transport in living MNs we designed a probe based on a recombinant nontoxic fragment of TeNT (Lalli et al., 1999) tagged at the N-terminus with a cysteine-rich domain designed to covalently bind the biarsenical fluorescein derivative FLASH (Fig. 1A) (Griffin et al., 1998). This cysteine-rich tag is also the site of modification by other thiol-specific fluorescent reagents, such as the Alexa 488, 546 or 594 maleimides. Multiple fluorophores can be coupled to TeNT HC upstream of its coding sequence, avoiding the risk of perturbing the C-terminal portion, which is required for neurospecific binding (Schiavo et al., 2000). Fluorescent TeNT HC bound to purified rat MNs and was internalized in vesicular structures irrespective of the fluorophore used for labelling (Fig. 1B-E; arrowheads). These carriers are of endosomal origin, given that they were labelled by Texas Red-dextran, a general endocytic marker (Nakata et al., 1998) (data not shown). Incubation of MNs with fluorescent TeNT HC at 4°C followed by washing and warming to 37°C led to the appearance of vesicular carriers, suggesting that fluorescent TeNT HC internalization is due to a specific uptake mechanism and not to fluid phase endocytosis. Moreover, binding and internalization were prevented by pre-incubating MNs with an excess of TeNT (Fig. 1F,G), showing that fluorescent TeNT HC interacts with the same surface receptors as the native holotoxin in living motor neurons.
To characterize TeNT HC carriers at an ultrastructural level, we analysed their morphology by EM using VSVG-tagged TeNT HC (Lalli et al., 1999) followed by anti-VSVG primary antibody and 10 nm gold-conjugated secondary antibody. At 4°C TeNT HC bound to the axonal plasma membrane of MNs (Fig. 2A) and accumulated at coated invaginations (Fig. 2B), synaptic sites and neurite contacts (Fig. 2C,D), as observed with fluorescent TeNT HC in living cells (data not shown). When neurons were subsequently shifted to 37°C for 5 minutes, gold particles were found in deep pits and coated vesicles (Fig. 2E,F). After 20 minutes at 37°C, TeNT HC was detected in uncoated tubulo-vesicular and round endocytic structures (Fig. 2G,H), consistent with the pleiomorphic retrograde carriers observed in living MNs by time-lapse fluorescence microscopy (Lalli and Schiavo, 2002). By EM, the diameter of the vesicular compartments varied between 50 and 100 nm, whereas the length of tubular structures was 2-3 μm. Similar results were obtained by using gold-conjugated TeNT HC and are consistent with a previous study performed with TeNT in mixed spinal cord cells (Parton et al., 1987).
TeNT HC retrograde transport depends on both F-actin and microtubules
Fast axonal retrograde transport is dependent on MTs and MT-dependent motors (Hirokawa, 1998; Goldstein and Yang, 2000). However, other studies have shown the involvement of F-actin in organelle motility (Brady et al., 1984; Kuznetsov et al., 1992; Bearer et al., 1993; Morris and Hollenbeck, 1995; Tabb et al., 1998), leading to the hypothesis that the two cytoskeletal systems cooperate in vesicle transport (Langford, 1995; Goode et al., 2000). We therefore investigated the effect of F-actin or MT disruption on TeNT HC retrograde movement. Drugs were added to cells after TeNT HC internalization, when retrograde-moving compartments were already visible. Incubation of MNs with control medium (see Materials and Methods) did not affect the transport of TeNT HC carriers (Fig. 3A and video 1), whereas treatment with the F-actin-disrupting drug latrunculin B (Lat B) caused the majority of the vesicles to stop or oscillate (Fig. 3B and video 2). We quantified the inhibitory effect of Lat B by counting the moving and stationary/oscillating vesicles in untreated and drug-treated cells (Fig. 3C). In untreated MNs, 80.1±5.1% of the vesicles moved in retrograde fashion with speed ranging between 0.2 and 3.6 μm/second (Lalli and Schiavo, 2002), whereas 19.9±5.1% were stationary. Lat B treatment led to a significant increase in the number of resting organelles (76.0±15.0%). Moving carriers represented only 24.0±14.2% of the total vesicles observed and were characterized by a reduced average speed ranging from 0.1 to 0.7 μm/second. Similar results were obtained using another F-actin-disrupting agent, swinholide A (data not shown). In control MNs, F-actin stained by fluorescent phalloidin showed a homogeneous distribution along neurites with sparse bright puncta (Fig. 3E). This staining disappeared after Lat B treatment (Fig. 3F). In the timecourse of our analysis, Lat B did not alter neurite morphology nor MT distribution as revealed by immunofluorescence with an anti-β-tubulin antibody (Fig. 3D) and by quantitative EM analysis of neurite cross-sections (Fig. 4C). It is therefore very unlikely that in our experimental conditions the disruption of the actin cytoskeleton causes a generalized neurite collapse or re-organization of the cellular cytoskeleton, indirectly affecting other actin-independent processes. To further exclude potential problems linked to acute F-actin depolymerization, MNs were differentiated for 3 days in the presence of LatB or cytochalasin B, a protocol resembling the chronic treatment previously reported in the literature (Morris and Hollenbeck, 1995). In addition to minor morphological alterations and a reduction of TeNT HC internalization, long-term treatment with LatB caused an inhibition of retrograde transport with a significant increase of still/oscillating carriers, which is in agreement with the results shown in Fig. 3C. As previously reported (Morris and Hollenbeck, 1995), neurite morphology after chronic treatment with cytochalasin B was dramatically altered (data not shown), hampering any further analysis of retrograde transport under these conditions.
Vincristine (Vin) has been shown to specifically disrupt MTs in neurons (Allison et al., 2000). Acute treatment with this drug greatly reduced retrograde movement (Fig. 4A and video 3), as shown by a threefold increase of the number of stationary/oscillating vesicles compared with control cells (84.0±4.5%; Fig. 4B). Vin-treated cells showed a 99% decrease of MT density compared with untreated cells, on the basis of quantitative EM analysis of neurite cross-sections (Fig. 4C). In Vin-treated MNs, only tubulin paracrystals could be observed (Fig. 4D, compare with Fig. 3D), whereas F-actin distribution was similar to untreated controls (Fig. 4E). Although the short-term treatment with Vin is well tolerated, chronic treatment at low dosage is toxic for MNs, causing detachment and cell death (data not shown). Therefore, this condition has not been further used for quantitative analysis. Treatment of MNs with the MT-depolymerising agent nocodazole only reduced retrograde movement (data not shown). However, morphometric EM analysis of neurite cross-sections revealed the presence of nocodazole-resistant MTs (25% of control cells; data not shown), strongly suggesting that the partial inhibition seen with nocodazole is due to an incomplete depolymerization of MT rather than to MT-independent axonal transport. Taken together, these data show that retrograde transport of TeNT HC is dependent on the integrity of both MTs and actin cytoskeleton.
Efficient TeNT HC retrograde transport relies on multiple MT-based motors
We then chose a pharmacological approach to characterize the motors required for TeNT HC retrograde transport. To selectively impair MT-based motors, we used the well-established dynein inhibitor EHNA (Penningroth et al., 1982; Reynolds et al., 1998) and the kinesin inhibitor aurintricarboxylic acid (AA) (Hopkins et al., 2000). AA has been recently identified via a structure-based computer screening for kinesin-binding molecules using human conventional kinesin as template (Hopkins et al., 2000). It has been shown to have a broad specificity for members of the kinesin superfamily (conventional kinesin, ncd C-terminal kinesin and Unc104/KIF1) (Vale, 2003) and an IC50 in the low micromolar range (0.5-1.4 μM), which is comparable to that of the only other known kinesin inhibitor, adocia sulfate-2 (Sakowicz et al., 1998). At the concentration used in the assay (10 μM), AA inhibits kinesin specifically, as the F-actin-dependent activity of myosin II remains unchanged (Hopkins et al., 2000).
TeNT HC-positive carriers from control and treated cells were tracked, and the distribution of speed values calculated between two consecutive frames was plotted. Control cells consistently showed a speed distribution profile centred at 0.8-1 μm/second, extending up to 3.6 μm/second in the retrograde direction (Fig. 5A-C, black bars). This distribution is consistent with the rates calculated for TeNT retrograde transport in vivo, ranging between 0.8 and 3.6 μm/second (Schiavo et al., 2000). Interestingly, dynein inhibition with EHNA led to a decrease of fast movements but did not completely stop TeNT HC retrograde transport (Fig. 5A, white bars). Incubation with the kinesin inhibitor AA had a less pronounced effect on transport than dynein impairment. However, AA still caused a decrease of high speed values accompanied by a parallel increase of pauses and anterograde movements, represented conventionally by negative speed values (Fig. 5B, white bars). To ascertain whether different MT-based motors act synergistically in promoting efficient TeNT HC retrograde transport, we simultaneously incubated MNs with EHNA and AA. However, these drugs had an effect on the speed distribution similar to that observed with EHNA alone (Fig. 5C, white bars; compare with Fig. 5A).
The speed distributions of TeNT HC alone or after treatment with motor inhibitors were further analysed by applying a multiple Gaussian curve fit. For all experiments (n=7), the speed profile of TeNT HC carriers is best described by the sum of three distinct components identified by Gaussian distributions centred at 0±0.11, 0.53±0.31 and 1.15±0.70 μm/second (Fig. 5D) (R2=0.997). This trimodal representation allowed us to determine the relative contribution of each speed component to retrograde transport, with the fastest component (1.15 μm/second) responsible for more than 70% of the total movements (Fig. 5E, black bars and Table 1).
In EHNA-treated MNs, we observed a dramatic decrease of the fast (1.15 μm/second) speed component, whereas the intermediate (0.53 μm/second) was unaffected (Fig. 5E, white bars and Table 1). EHNA also caused an increase in the frequency of pauses and of slow, short anterograde movements (Fig. 5A), resulting in oscillation of TeNT HC carriers. In agreement with this observation, the incidence of reversal (number of changes of direction per organelle) in EHNA-treated cells increased more than tenfold compared with untreated samples (10.3±2.4; 122 and 281 tracked carriers in drug-treated and control cells, respectively; n=3 independent experiments). This result confirms and extends our recent report that the fast component of axonal retrograde transport is impaired in mice carrying mutations in dynein heavy chain (Hafezparast et al., 2003).
AA treatment also caused a decrease of the fast component, although less marked than the one observed with EHNA, whereas the contribution of the intermediate speed component remained similar to the control (Fig. 5E, hatched bars and Table 1). Simultaneous incubation with EHNA and AA led to an inhibition of the fast component similar to the one observed with EHNA alone. By contrast, the intermediate component remained unaffected (Fig. 5E, grey bars). This residual movement observed after inhibition of dynein and kinesins might be due to the involvement of actin-based motors, as predicted by the requirement for F-actin in TeNT HC transport observed in Fig. 3. To test this hypothesis, we used the drug 2,3 butanedione monoxime (BDM, 10 mM) as a broad-spectrum myosin inhibitor (Cramer and Mitchison, 1995; Lin et al., 1996; Prekeris and Terrian, 1997; Duran et al., 2003). Simultaneous treatment of MNs with EHNA, AA and BDM completely stopped TeNT HC carriers (data not shown). This result supports the idea that myosins play a role in retrograde transport and indicates that the residual movement observed after simultaneous treatment with EHNA and AA was not due to an incomplete inhibition of MT-dependent motors.
This evidence points to the presence on TeNT HC carriers of multiple types of motors contributing to fast axonal retrograde transport. Moreover, it suggests that TeNT HC transport relies mainly on dynein, but requires kinesins and possibly actin-based motors to achieve optimal efficiency.
Myosin Va is required for efficient retrograde transport in mammalian MNs
To identify myosins involved in retrograde movement of TeNT HC carriers, we performed an immunofluorescence screening in spinal cord MNs for different types of myosins previously reported to have a role in vesicular transport in neurons (DePina and Langford, 1999; Bridgman and Elkin, 2000; Wu et al., 2000; DeGiorgis et al., 2002; Brown, 2003). We detected colocalization of TeNT HC carriers only with myosin Va, a motor protein participating in the transport of several types of organelles in vitro and in vivo (Wu et al., 2000; Langford, 2002) (Fig. 6A-C, arrowheads).
To clarify whether myosin Va is functionally important for axonal retrograde transport, we tested MNs from wild-type and homozygous dilute lethal (myosin Va null) mice (Mercer et al., 1991) for their ability to transport TeNT HC. Retrograde movement of fluorescent TeNT HC in dilute lethal MNs was clearly impaired compared with wild-type cells (Fig. 6D,E, arrowheads, asterisks; see also video 4). Wild-type MNs displayed a speed distribution curve centred at 0.8 μm/second and reaching a maximum value of 3 μm/second (Fig. 6F, filled circles). Kinetic analysis of TeNT HC carriers in MNs from dilute lethal animals revealed both a decrease in the frequency of high-speed values and an increase of pauses and slow bidirectional movements (Fig. 6F, empty circles). This led to a substantial rise of oscillations, as further indicated by the ∼eightfold increase of the incidence of reversal of TeNT HC carriers in dilute lethal compared with wild-type MNs (7.78±2.3; 141 and 357 carriers in dilute lethal and wild-type cells, respectively; n=3). The speed distribution curve for TeNT HC transport in wild-type mice MNs can be best described by the sum of three distinct Gaussian profiles centred at 0, 0.53 and 0.95 μm/second, respectively (Fig. 6G). As observed in rat MNs, the intermediate (0.53 μm/second) and fast (0.95 μm/second) components account for more than 80% of the carriers. TeNT HC transport in dilute lethal MNs showed a severe reduction of the fastest (0.95 μm/second) speed component, with a corresponding increase in the frequency of pauses, whereas the intermediate speed carriers remain unaltered (Fig. 6G, white bars). These data reveal a novel role of myosin Va in axonal retrograde transport and suggest that this myosin is required for fast retrograde movement in mammalian MNs.
In this study, we have used a fluorescent fragment of TeNT, which is specifically recruited to a fast axonal retrograde transport pathway (Lalli and Schiavo, 2002), as a probe to analyse the cytoskeletal and motor requirements for this process in mammalian MNs. The retrograde transport of TeNT HC is both MT- and F-actin-dependent. Fast axonal transport is known to rely on an intact MT cytoskeleton (Goldstein and Yang, 2000). However, we found a dramatic effect of F-actin depolymerization on retrograde transport, under conditions preserving the integrity of MTs. After F-actin disruption by Lat B and other F-actin-depolymerising agents, the great majority of TeNT HC carriers were stationary or oscillated. This result indicates that actin microfilaments are required for the directional transport of axonal organelles, as previously suggested by experiments on isolated squid axoplasm (Brady et al., 1984). EM studies have revealed a tight association of MTs with actin microfilaments (Fath and Lasek, 1988; Bearer and Reese, 1999) that could be important not only for maintaining the structural integrity of the axon, but also for supporting fast transport. In sympathetic neurons, disruption of F-actin led to a significant increase in the speed of bidirectionally moving mitochondria or myosin V-associated organelles (Morris and Hollenbeck, 1995; Bridgman, 1999), suggesting that actin-based motility is not strictly necessary for rapid axonal movement. However, in MNs, fast retrograde transport requires the simultaneous presence of MTs and F-actin. Different neuronal systems might therefore use motors with distinct cytoskeletal requirements, depending on the type of organelle transported, its direction and the differentiation state of the neurons.
An important implication of our findings is that both MT- and actin-based motors power the movement of TeNT HC carriers in MNs. Dynein is probably the major motor responsible for TeNT HC retrograde transport, as its inhibition caused the largest decrease in the speed of TeNT HC carriers and an increase in the frequency of stationary pauses and incidence of reversal. The central role for dynein in axonal transport is underscored by the finding that point mutations in dynein heavy chain specifically impair fast retrograde transport in mice, generating pathological conditions closely resembling amyotrophic lateral sclerosis (Hafezparast et al., 2003). Moreover, a missense mutation in the dynein-interacting protein dynactin/p150glued causes lower motor neuron disease in humans (Puls et al., 2003), confirming the functional link between fast axonal retrograde transport, cytoplasmic dynein and neuronal degeneration.
Regardless of the central role played by dynein in fast retrograde transport, several results indicate that multiple motors are involved in this process. First, blocking dynein activity by a pharmacological approach or via specific mutations in its core subunit (Hafezparast et al., 2003) did not lead to a complete halt of TeNT HC carriers. Second, the speed distribution curve of TeNT HC endosomes shows multiple peaks. Third, a kinesin inhibitor also affected transport, although less dramatically. Moreover, cytoplasmic dynein supports long-range intracellular movements of cargo in vivo but does not appear to be a processive motor (Hirakawa et al., 2000; King and Schroer, 2000). It could therefore require other MT and actin-based motors to achieve maximal speed and processivity. Inhibition of kinesins slowed the movement of TeNT HC carriers in the presence of active dynein but did not produce any synergistic inhibitory effect when dynein was blocked by EHNA. The dynein complex could therefore act not only as a main retrograde motor, but also as a regulator of the activity of other motors associated with TeNT HC carriers. Studies in yeast and Drosophila point to a functional interaction of dynactin/p150glued with kinesins during mitosis and in fast axonal transport (Blangy et al., 1997; Martin et al., 1999) and have shown its involvement in the coordination of plus- and minus-end-directed motors (Gross et al., 2002b). In the case of mammalian MNs, it is tempting to speculate that the dynein/dynactin complex regulates the engagement of multiple retrograde motors, allowing the cargo to move more efficiently along the axon.
The residual movement observed after the inhibition of MT-based motors suggests that other molecules, such as one or more myosins, are also implicated in retrograde transport. By using our retrograde transport assay in dilute lethal mice we have shown that myosin Va is required for this process. Studies on the dynamics of organelles associated with myosin Va (Wu et al., 1998; Bridgman, 1999; Al-Haddad et al., 2001), together with the demonstration that myosin V is a processive motor (Mehta et al., 1999), support the hypothesis that this protein plays an important role in organelle transport. In sympathetic neurons, the movement of myosin Va-associated organelles is bidirectional. However, these organelles become anterogradely biased in dilute lethal neurons, implying the participation of myosin Va in retrograde transport (Bridgman, 1999). Recently, myosin Va has been shown to control neurofilament density, its ablation determining an increase in neurofilament number in vivo (Rao et al., 2002). This raises the possibility that dilute lethal neurons show a general impairment of axonal transport. However, the observed increase of the maximum speed of axonal vesicles in these neurons (Bridgman, 1999), together with their biased directionality, strongly argues against this scenario.
On the basis of this previous work (Bridgman, 1999), myosin Va has been proposed to dampen organelle movement, as seen in mouse melanosomes and phagosomes (Wu et al., 1998; Al-Haddad et al., 2001). This and other experimental findings have been recently summarized by the `cooperative/capture mechanism' (Wu et al., 1998). According to this model, myosin Va anchors melanosomes on the peripheral random network of actin microfilaments, preventing them from moving centripetally on MTs to reach the perinuclear region. In an extension of this model, Al-Haddad and colleagues proposed an `antagonistic/cooperative mechanism', whereby actin-dependent digressions powered by myosin Va antagonize dynein-dependent phagosome motility (Al-Haddad et al., 2001). A prediction of this model is that the absence of myosin Va in dilute lethal MNs should result in a faster retrograde transport compared with wild-type neurons. However, our findings argue against an inhibitory role of myosin Va, as fast retrograde axonal transport is impaired in myosin Va null MNs.
These apparent differences in myosin Va function can be reconciled by assuming that its contribution to organelle transport varies accordingly to the orientation of the actin cytoskeleton in each cell type. Myosin Va, a barbed-end-directed motor (Cheney et al., 1993), could contribute to fast retrograde transport in MNs by moving on actin filaments running co-axially with MTs and oriented with the barbed ends towards the cell body. Interestingly, mitochondria, which move bidirectionally, show a net retrograde transport in the absence of MTs in sympathetic neurons, implying the existence of F-actin with biased polarity in axons (Morris and Hollenbeck, 1995). Alternatively, these differences in myosin Va function could be attributed to a selective organelle-specific regulation, as shown for the myosin Va-dependent transport of endoplasmic reticulum in mitotic Xenopus egg extracts (Wöllert et al., 2002), the class V myosin Myo2p-driven movement of vacuoles (Ishikawa et al., 2003; Tang et al., 2003) and Myo4p-directed RNA transport (Kruse et al., 2002).
Our experimental data imply that myosin Va activity is coordinated with MT-based retrograde motors to ensure fast retrograde transport in mammalian MNs. In this system, myosin Va could power movement on F-actin, which serves as complementary tracks during MT-based transport. This might help carriers to dodge obstacles, similarly to what has been observed in Xenopus melanophores (Rogers and Gelfand, 2000; Gross et al., 2002a). In addition, myosin Va could contribute to fast transport by allowing movement on F-actin during transition from one MT to another, thus decreasing the probability of diffusional movements of organelles. Precise coordination between MT-based motors and myosin Va could be achieved through additional regulating factors or direct interaction. Both myosin Va and the intermediate chain of cytoplasmic dynein share an 8 kDa light chain (Mermall et al., 1998) able to dimerize (Benashski et al., 1997), suggesting that these two motors interact on the surface of a transport carrier. Deletion of myosin Va from a dynein-associated particle might impair the activity of the entire motor complex, providing the molecular explanation for the absolute requirement of both myosin Va and dynein for fast retrograde transport in this system. Discovering the molecular mechanisms responsible for the coordination between myosin Va and MT-based retrograde motors and the relationship between MTs and actin cytoskeleton in axons are challenging objectives for future studies.
We thank Drs P. C. Bridgman for the gift of anti-myosin Va antiserum, M. Seabra and R. Anders for the gift and handling of dilute lethal mice and J. Monypenny for help with video editing. We are grateful to Drs D. Ish-Horowicz, C. Reis e Sousa, S. Tooze and M. Way for critical reading of the manuscript. This work was supported by Cancer Research UK.
Movies available online
↵* Present address: MRC Laboratory for Molecular Cell Biology, University College London, Gower Street, London WC1E 6BT, UK
- Accepted June 27, 2003.
- © The Company of Biologists Limited 2003