|Home | About | Journals | Submit | Contact Us | Français|
Neurons transport and position mitochondria using a combination of saltatory, bidirectional movements and stationary docking. Axonal mitochondria move along microtubules (MTs) using kinesin and dynein motors, but actin and myosin also play a poorly-defined role in their traffic. To ascertain this role, we have used RNA interference to deplete specific myosin motors in cultured Drosophila neurons and quantified the effects on mitochondrial motility. We produced a fly strain expressing the C. elegans RNA transporter SID-1 in neurons to increase the efficacy of RNAi in primary cultures. These neurons exhibited significantly increased RNAi-mediated knockdown of gene expression compared to neurons not expressing this transporter. Using this system, we observed a significant increase in mitochondrial transport upon myosin V depletion. Mitochondrial mean velocity and duty cycle were augmented in both anterograde and retrograde directions, and the fraction of mitochondrial flux contained in long runs almost doubled for anterograde movement. Myosin VI depletion increased the same movement parameters, but was selective for retrograde movement, while myosin II depletion produced no phenotype. An additional effect of myosin V depletion was an increase in mitochondrial length. These data indicate that myosin V and VI play related but distinct roles in regulating MT-based mitochondrial movement: they oppose, rather than complement protracted MT-based movements and perhaps facilitate organelle docking.
Neurons transport mitochondria over long distances within the axon and also distribute them non-uniformly among sites with heightened need for ATP production or Ca2+ homeostasis such as growth cones, axon branches, synapses or nodes of Ranvier (Zhang et al., 2010; Berthold et al., 1993; Morris and Hollenbeck, 1993; Bristow et al., 2002; Ruthel and Hollenbeck, 2003; Miller and Sheetz, 2004; Hollenbeck and Saxton, 2005). This is accomplished by a rather baroque, inefficient-looking process of saltatory, bidirectional mitochondrial movement with frequent stops and, for some fraction of axonal mitochondria, persistent docking in specific regions (Hollenbeck and Saxton, 2005). It is clear that much of the total mitochondrial flux in axons is generated by MT-based movements using the anterograde motor protein kinesin (Nangaku et al., 1994; Hurd and Saxton, 1996; Hollenbeck and Saxton, 2005; Pilling et al., 2006) and the retrograde motor protein dynein (Pilling et al., 2006). However, actin and myosin also play a role that has remained uncertain (Morris and Hollenbeck, 1995; Ligon and Steward, 2000; Hollenbeck and Saxton, 2005). One possible role for actomyosin-based transport is that it augments or supplements MT-based movement, returning mitochondria that have transiently detached back to their MT tracks, or moving them through small regions of the axon that are MT-poor (Langford, 1995). This has also been suggested for other organelles in non-neuronal cells, such as endosomes or secretory vesicles that move along paths including both the MT-rich cytoplasm and actin-rich cell cortex (Ross et al., 2008). However, an alternative possibility is that the myosin-based movement of mitochondria actually contributes to their distribution by opposing protracted MT-based movements, as has been shown for melanosomes in Xenopus pigment cells (Rodionov et al., 1991; Rogers et al., 1999; Tuma and Gelfand, 1999; Gross et al., 2002) . This would not only help explain the complex, saltatory pattern of mitochondrial motility and their non-uniform distribution, but also the augmentation of mitochondrial movements in the absence of actin (Morris and Hollenbeck, 1995) and the requirement for actin in the induced halting and docking of mitochondria along the axon (Chada and Hollenbeck, 2003; Chada and Hollenbeck, 2004).
To test this hypothesis, we selectively depleted specific myosins in Drosophila primary neurons using RNAi and quantified the effects on the axonal transport of mitochondria. We observed an increase in movements in both directions after myosin V depletion and of retrograde movements after myosin VI depletion, consistent in its details with a myosin-MT motor competition that interrupts mitochondrial movement. We also observed a significant increase in mitochondrial length after myosin V depletion, suggesting that it may play a role in the mitochondrial fission-fusion equilibrium.
The strain w1118; +; D42-GAL4, UAS-mitoGFP (“mito-GFP”) expressing mitochondrially-targeted GFP in motor neurons (Pilling et al., 2006), was a gift from W. M. Saxton (UC Santa Cruz). For creating the UAS-sid-1 line, a sid-1::Flag cDNA fragment was excised from PAcPL-sid-1::Flag (donated by C. Hunter) using BamHI and was ligated to the Bgl II site of pUAST. This generated an 11.2 kb plasmid (pUASt map information kindly provided by Andrea Brand). This plasmid (Figure S3) was then transformed into a w1118 strain and mapped to the second chromosome. The UAS-sid-1 chromosome was crossed into the w1118, +, D42-GAL4, UAS-mitoGFP background using dominantly marked balancer chromosomes to obtain the working fly strain: w1118; UAS-SID-1; D42-GAL4, UAS-mitoGFP.
Flies were maintained in population cages at 25°C for collecting embryos. Prior to egg collection, they were placed on a 12 hr light/ dark cycle and fed with yeast paste streaked on apple agar plates. Flies were allowed to lay eggs for 2 hrs and the embryos were allowed to develop for 4-6 hrs at 25°C so as to obtain early gastrulae embryos.
Cultured cells were prepared by dissociating early gastrula embryos (4-6 hr after laying) using established methods (Salvaterra et al., 1987). This generates a cell population that differentiates into many cell types, including neurons. Motor neurons were identified by their morphology and their mitochondrial GFP expression. Cells were grown in Shields and Sang medium supplemented with 20% FBS, 100U/ml penicillin and streptomycin (Gibco BRL) and 200ng/ml insulin. One day after plating, serum-containing medium was removed and replaced with serum free medium. Cultures were then exposed to specific dsRNA at 40 μg/ml in serum-free medium for 2.5 hrs, followed by addition of serum-containing medium and incubation at 25°C for 2-3 days. These cells were then used for microscopy and/ or western blotting. For giant neuron preparation cultured cells were prepared as described above and exposed to cytochalasin B at 2 μg/ml for 24 hrs (Pilling et al., 2006).
Myosin dsRNA was generated using methods and primer sequences described previously (Goshima and Vale, 2003; Rogers et al., 2003). Briefly, mRNA was isolated from fly larvae using Qiagen RNeasy kit and reverse transcribed using specific primers for myosins II, V or VI (Qiagen One-Step RT PCR kit). The resulting cDNA was PCR amplified and this product was used as a template for in vitro transcription (Ambion MEGAscript T7 kit). Contaminating DNA was removed by DNAase treatment and dsRNA obtained was purified (Qiagen RNeasy kit) and its concentration determined before use. Off-target control dsRNA was prepared using the pC4βgal vector and primer sequences: 5'-TAATACGACTCACTATAGGCACCCGAGTGTGATCATCTG-3' 5'-TAATACGACTCACTATAGGATGACGGAACAGGTATTCGC-3'. For GFP dsRNA, pcDNA3.1/mtGFP was used as template and primer sequences: 5'-TAATACGACTCACTATAGGTCAGTGGAGAGGGTGAAGGT-3' 5'-TAATACGACTCACTATAGGGGCAGATTGTGTGGACAGG-3'
Cells were observed on a Nikon TE300 inverted microscope equipped with MicroMax cooled CCD camera (Princeton Instruments). Fluorescent images were gathered at 1 sec intervals for 200-250 sec. Movement of mitochondria was tracked and quantified using Image-J software. Isolated neurons that were clearly separated from neuronal clusters were imaged and every infocus mitochondrion within an axon was tracked to eliminate selection bias. To avoid stage drift and positioning errors that can lead to small erroneous measurements and distort actual transport values by repetition, we excluded all displacement values of <0.1 μm between consecutive frames (Pilling et al., 2006). Intervals with average velocities between 0.10 and -0.10 μm/s for at least three consecutive frames were designated as pauses. Uninterrupted motion between pauses and/ or direction reversal was considered a “run”. The sum total of all run lengths in a particular category gave an estimate of flux. The percent of time spent moving in a particular direction gave duty cycle values.
Lysates were prepared for SDS-PAGE from cultured cells obtained from normal and SID-1-expressing flies that had been treated with eGFP dsRNA. Samples were loaded in serial dilutions to yield immunoblotting signals in the linear range and adjusted volumes were subsequently used for quantification. Primary antibodies used were rabbit anti-GFP (1:1000; Clontech), mouse monoclonal anti-ELAV (1:500; Developmental Studies Hybridoma Bank), rabbit anti-non muscle myosin II (1:1000), rabbit anti-myosin V (1:1000) or mouse monoclonal anti-myosin VI (1:400) antibodies. Myosin antibodies used were generated against Drosophila proteins and are well characterized and specific (Kiehart and Feghali, 1986; Miller et al., 1989; Li et al., 2007). Blots were then probed with Alexa Fluor 680 goat anti-rabbit (Molecular Probes) and IRDye 800 CW conjugated goat anti-mouse antibody (Rockland). Bands were visualized using an Odyssey Infrared Imaging System (Li-Cor Biosciences) and band intensities and areas were measured using Metamorph software. The total intensity was quantified for GFP and ELAV bands and ratio of GFP/ELAV was used to obtain normalized values for each treatment.
For immunostaining of the endogenous myosins, cells were fixed with 4% paraformaldehyde in PBS, permeablized with 0.1% Triton X-100 for 10 mins, and blocked in 5% BSA/PBS/0.1% Tween-20. Primary antibodies used were rabbit anti-myosin V [1: 500], rabbit anti-non muscle myosin II [1: 500] or mouse monoclonal anti-myosin VI antibodies [1: 30] and secondary antibody used were Alexa Fluor 680 goat anti-rabbit and Alexa Fluor 680 goat anti-mouse. To obtain values for background subtraction, the primary antibody was omitted. Staining was visualized using epifluorescent illumination with FITC and Texas Red filter sets.
Quantitative imaging was done using Metamorph software. Images were thresholded, the integrated morphometry analysis function was used to measure fluorescence intensities for largest possible area of each axon, and the values obtained were divided by area and given as arbitrary fluorescent units/μm2. Non-cellular background fluorescence values were substracted from axonal fluorescence. Mitochondrial lengths were measured using the region measurement function in Metamorph. Multiple measurements of a single mitochondrion were made over the course of 250 frames and averaged.
RNA interference (RNAi) using double-stranded RNA (dsRNA) is a remarkably effective tool for suppressing specific gene expression in Drosophila cells (Goshima and Vale, 2003; Rogers et al., 2003; Sharma and Nirenberg, 2007). However neurons are generally thought to be less amenable to RNAi than other cell types. The C.elegans sid-1 gene encodes an RNA transporter protein found at the plasma membrane. Drosophila do not have a sid-1 ortholog and endocytic pathways are their only known means of dsRNA uptake. Thus to increase the uptake of dsRNA into Drosophila neurons for our experiments, we generated a fly strain expressing a C.elegans UAS-sid-1 cDNA transgene in the genetic background of the mitochondrial marker line D42-Gal,UASmito-GFP (Pilling et al., 2006) to yield the strain: w1118; UAS-sid-1; D42-GAL4, UAS-mitoGFP, referred to here as D42: mito-GFP, SID-1.
We tested the efficacy of this system by knocking down GFP expression with specific dsRNA in primary neurons derived from mito-GFP fly lines with or without the sid-1 gene and assessing the degree of GFP depletion using quantitative fluorescence measurements (Figure 1A-G). Consistent with other primary cell responses to dsRNA, treatment with GFP dsRNA led to a reduction in protein expression of around 50% in neurons from mito-GFP flies. However, neurons from the mito-GFP/SID-1 line responded to the same dsRNA treatment with an 80% reduction in GFP expression. These results were confirmed by immunoblotting, using the ratio of signal intensity for anti-GFP to that of an anti-ELAV (embryonic lethal, abnormal vision), in order to normalize for the neuronal content of mixed primary cultures (Figure 1H). Thus, expression of the sid-1 transgene leads to a substantial improvement in RNAi effects, suggesting that this is a strategy with significant implications for other dsRNA studies in primary cultures. We employed the D42: mito-GFP, SID-1 strain for all experiments described below.
The identities of the myosin motors involved in actin-based axonal transport of mitochondria have not been determined, but of Drosophila's thirteen myosin genes myosins V and VI are the likeliest candidates. The processive motor myosin V seems most likely to be involved, since it has not only been shown to associate with a variety of organelles in neurons (Prekeris and Terrian, 1997; Bridgman, 1999; Miller and Sheetz, 2000) but also participates in the transport and distribution of melanosomes (Wu et al., 1998; Rogers et al., 1999; Gross et al., 2002) and secretory granules (Rudolf et al., 2003; Desnos et al., 2007). Furthermore myosin V is also known to support movement in vitro at rates similar to axonal transport of mitochondria on actin filaments (Morris and Hollenbeck, 1995; Wolenski et al., 1995). In addition a class V myosin (Myo2) has been shown to play a direct role in mitochondrial transport and inheritance in budding yeast (Altmann et al., 2008). Myosin VI has been shown to transport cytoplasmic particles in Drosophila embryos (Mermall et al., 1994) and also plays a role in trafficking uncoated vesicles during clathrin mediated endocytosis (Buss et al., 2001). Myosin II has been reported to decorate the cytoplasmic surface of axoplasmic organelles via immunogold electron-microscopy (DeGiorgis et al., 2002), implicating it as yet another candidate organelle motor.
Preliminary studies in our lab have shown >80% co-localization between a GFP-tagged myosin V tail and axonal mitochondria in cultured vertebrate neurons (Figure S1). In Drosophila motor neurons, myosin II, V and VI immunostaining revealed a punctate pattern throughout the cell and partially co-localized with mitochondria. We observed that 82% mitochondria showed myosin II staining, 85% showed myosin V staining and 74% showed myosin VI staining (Figure S2). We treated neurons from Drosophila embryos with specific myosin dsRNA (see methods) and cultured them for 3-4 days before observation; this protocol was optimized using the dsRNA knock-down of GFP, an abundantly-expressed transgene. We confirmed the decrease of motor protein levels by immunocytochemistry using specific antibodies. Myosin V dsRNA treatment led to a ≈ 75% reduction in anti-myosin V immunostaining compared to cells treated with β-galactosidase (β-gal) dsRNA as an off-target control. For myosin II and VI, a >60% decrease in imunostaining compared to controls was seen (Figure 2A, B). To test the specificity of myosin knock-down, neurons treated with myosin II dsRNA were immunostained for myosin V and VI. Average fluorescence intensities obtained were very similar to those obtained for off-target controls, showing that the knock-down was specific (Figure 2B). Similarly, for myosin V and myosin VI dsRNA-treated cells, decreased fluorescence intensities were only seen for targeted myosins (data not shown). Western blots confirmed the sensitivity and specificity of myosin protein knock-down after dsRNA treatment (Figure 2C). Blotting with anti myosin II, V and VI antibodies revealed that exposure to specific dsRNA lead to robust protein knock-down. Myosin V depletion was only seen when cultures had been treated with myosin V dsRNA and not with either myosin II or myosin VI dsRNA treatment. Similarly for myosin VI, depletion was only seen when neurons were treated with specific dsRNA and not upon exposure to either myosin II or V dsRNA.
To test the roles of myosins in mitochondrial movement, we carried out detailed quantification of organelle transport in the axons of dsRNA-treated primary neurons. Myosin V depletion in neurons produced an obvious transport phenotype: an increase in mitochondrial transport. More mitochondria displayed movement than in controls, and the frequency of protracted movements was increased (Figure 3A-C, videos 1-3). A significant enhancement in average mitochondrial run length was seen in both directions upon myosin V knock-down but only in the retrograde direction upon myosin VI knock-down (Figure 3D). Myosin V and VI knock-downs also produced an increased frequency of longer runs compared to controls. The range of mitochondrial run lengths observed was very broad (0.3-20 μm, 3545 runs measured), and although shorter runs far outnumbered longer ones, the longer runs constituted a large fraction of the total mitochondrial flux in these axons. Thus, we compared the fraction of total movement (or flux) contained in these runs under the different knock-down treatments. We found that myosin V knock-down doubled the fraction of total anterograde movement found in the longer run class (>1μm) from 28% to 57%. For retrograde movement a modest increase (from 40% to 50%) was also seen (Figure 3E). Knock-down of myosin VI also led to an increase in the fraction of total retrograde movement in longer runs (Figure 3E), while no effect was seen on anterograde movement.
We also observed an increase in mitochondrial velocity upon myosin depletion (Figure 3 D). As seen for other transport measures, myosin V depletion significantly increased velocity in both anterograde and retrograde directions while myosin VI dsRNA treatment affected only the retrograde direction. The increased run lengths in myosin V and VI knock-downs indicate that, when present, these motors interfere with the capacity of mitochondria to undergo persistent, uninterrupted movement. The increased mitochondrial velocities in these knock-downs suggest further that myosins exert force on mitochondria that competes with MT-based motors. Myosin II knock-down did not affect velocity, run length or flux, ruling out a direct role for this motor in mitochondrial transport.
The interruption of protracted mitochondrial transport by myosins could regulate the transition to a stationary, docked state, or the direction of transport, or both. To test this, we quantified the effect of myosin knock-downs on stationary time versus direction changes. Myosin V depletion lead to a >25% decrease in the percentage of time mitochondria spent being stationary. (Figure 4A-C). The resulting increase in duty cycle occurred for both anterograde and retrograde directions (Figure 4B-C). Myosin VI knock-down had a more subtle effect, a modest but significant increase in retrograde duty cycle only (Figure 4B, C). Although the changes in transport parameters for retrograde movement seen with myosin VI depletion were similar to those we observed for myosin V (Figures 3D, E and and4C),4C), the myosin VI effect was distributed differently. Upon myosin VI knock-down a minority of mitochondria showed most of the duty cycle increase, whereas myosin V knock-down lead to a broad increase in transport for most organelles (Figures 4B, C; videos 3 and 4). The differential effect on transport seen upon myosin V and VI knock-downs may be due to interactions with kinesin or dynein that are specific to each myosin or it could be due to the organization of actin filaments in these neurons.
To determine whether myosin V regulates the direction of mitochondrial movement, we also quantified the frequency of direction change exhibited under control conditions and after myosin V knock-down. Each change in direction from either anterograde to retrograde or vice versa was counted as one event and the average number of events and distance covered between them was calculated. We observed that the average number of direction changes seen per mitochondrion per 200 sec observation interval almost doubled (24 vs. 13) upon myosin V depletion. However, since mitochondria covered twice as much distance in myosin V-depleted cells, the average distance moved between direction changes remained similar for both controls and myosin V knock-down cells.(Figure 4D). These results indicate that myosin acts to interrupt mitochondrial transport on MTs to decrease duty cycle and perhaps facilitate docking and that the observed interruption of transport does not mediate or regulate direction changes.
An additional, unexpected phenotype observed with myosin V knock-down was an obvious and significant increase in the length of axonal mitochondria (Figure 5A). The mean length in myosin V-depleted axons was 2.6 ± 0.3 μm, compared to 1.5 ± 0.1 μm in controls. Mitochondrial length was not affected by knock-down of myosin VI (mean 1.4 ± 0.2 μm) or myosin II (mean 1.5 ± 0.1 μm). More than 80% of mitochondria in controls had lengths ≤ 2.0 μm whereas in myosin V depleted cells only ≈40% mitochondria had lengths ≤ 2.0 μm (Figure 5B). This data was derived from multiple measurements made for each mitochondrion over a course of 250 frames (1 frame acquired per sec) in order to prevent inaccurately large length measurements due to transient overlaps of mitochondria. Equally unexpected was that these longer mitochondria showed motility that was in all respects equal or greater to that of their shorter counterparts in control cells (data not shown)
We studied the role of myosin motors in mitochondrial transport in cultured Drosophila neurons. In order to increase the neuronal responsiveness to RNAi, we constructed a fly strain expressing in neurons, the SID-1 dsRNA transporter of C. elegans. Expression of this protein markedly increased the efficacy of dsRNA mediated knockdown in primary neuronal cultures. We used this system to knockdown myosin II, V and VI and assessed mitochondrial transport. Myosin II knockdown did not lead to any phenotype. Upon myosin V depletion we observed a significant increase in mitochondrial transport and the fraction of mitochondrial flux contained in long runs almost doubled for anterograde movement. The mean velocity and duty cycle also increased in both directions; however the frequency of direction changes per unit distance travelled remained unaffected. Myosin VI depletion increased the same movement parameters as myosin V, but was selective for retrograde movement. In addition to the transport phenotypes, an increase in mitochondrial length in axons was also seen upon myosin V depletion.
If actomyosin-based transport were augmenting mitochondria motility on MTs, we would have seen a decrease in at least some transport parameters with myosin knock-downs. Instead, we observed increases in duty cycle and run length and a corresponding decrease in stationary phases, which are all consistent with the hypothesis that the normal role of myosins in mitochondrial axonal transport is to oppose protracted MT-based movement. This has a precedent in the hormone-regulated movement of pigment granules in Xenopus melanocytes. These organelles move away from the cell center upon exposure to melanocyte-stimulating hormone, but rather than traveling along MTs all the way to the periphery, they become dispersed throughout the cell. They achieve this distribution via bidirectional movements on the radial MT array, driven by kinesin and dynein, in which myosin V selectively competes with dynein, effectively removing organelles from the MT tracks during retrograde runs and displacing them onto actin filaments (Gross et al., 2002). Similarly in mouse melanocytes, myosin V captures melanosomes moving rapidly on MTs and offloads them onto actin tracks. Local movement on actin eventually allows them to reach distal ends of melanocytes and this is a prerequisite for skin pigmentation (Wu et al., 1998). Our data suggest that anterograde movements of axonal mitochondria can be strongly disrupted by myosin V, while retrograde movements are modestly disrupted by both myosins V and VI. This is also consistent with previous analysis of myo Va null mice (Bridgman, 1999) in which axonal vesicles showed increased anterograde movement. However, our data show not just increased duty cycle and run length but also increased velocity after myosin V and VI knock-down. This suggests that myosins exert a force on mitochondria that competes with MT-based motors, and may thus facilitate their off-loading from MT tracks and/or docking on actin. It is possible that some or all of the “stationary” periods we observe are actually comprised of very short bidirectional movements on actin that do not complement mitochondrial motility, consistent with the view of myosin V and VI as “dynamic tethers” (Woolner and Bement, 2009). Apart from their roles in organelle transport, a parallel body of evidence does indicate a role for myosin V and VI in anchoring cytoplasmic organelles. Myosin V has been implicated in tethering vesicles to actin during spermatogenesis (Mermall et al., 2005). As for Myosin VI, its monomeric form, low processivity (Lister et al., 2004) and long dwell time (Altman et al., 2004; Noguchi et al., 2006) are generally inconsistent with organelle movement, thus it is perhaps more likely to be a candidate for anchoring than for moving mitochondria on actin.
What physiological role could be played by myosins opposing protracted mitochondrial movements in the axon? Mitochondria often do not have a uniform or constant distribution in axons; against the backdrop of mass net movement in the anterograde and retrograde directions they are also selectively localized (Hollenbeck and Saxton, 2005) and activated (Verburg and Hollenbeck, 2008) at specific sites. These include a variety of regions with high demand for mitochondrial function (Hollenbeck and Saxton, 2005). Achieving and subsequently reorganizing such a non-uniform distribution requires that mitochondria be not only transported to these regions, but also retained there. The primary mechanism proposed for mitochondrial docking is linkage to the cytoskeleton and evidence exists for neurofilaments (NFs), MTs and actin filaments as docking sites. Associations of mitochondria with NFs have been suggested by ultrastructural observation (e.g., (Hirokawa, 1982)) and in vitro binding (Wagner et al., 2003) but Drosophila neurons display complex saltatory mitochondrial movements despite lacking intermediate filaments (Goldstein and Gunawardena, 2000). Ultrastructural studies have also shown links between mitochondria and MTs (Linden et al., 1989a; Linden et al., 1989b; Leterrier et al., 1994), and recent observations by Kang and coworkers (Kang et al., 2008) have implicated the brain specific protein syntaphilin in mitochondrial docking to MTs in a mouse model; however no syntaphilin homologue has been found in invertebrates (Lloyd et al., 2000). Recent evidence indicates that the mitochondrial proteins Milton and Miro GTPase can regulate MT-based movement in the axon. Miro binds the kinesin-1 HC directly (Macaskill et al., 2009b) and also as part of a complex with the protein Milton (Glater et al., 2006; MacAskill et al., 2009a) in a manner that depends on Ca levels (Saotome et al., 2008; Macaskill et al., 2009b; Wang and Schwarz, 2009) and probably controls kinesin function. While both Miro and Milton are required for transport of mitochondria to synapses (Stowers et al., 2002; Guo et al., 2005), their role in generating a non-uniform distribution of mitochondria in the neuron is not yet clear.
Another body of evidence favors actin filaments as a mitochondrial docking substrate. Mitochondria-rich zones in neurons are often actin-rich but MT- and NF-poor. In addition, in the absence of actin, axonal mitochondrial transport exhibits greater velocities and excursion lengths (Morris and Hollenbeck, 1995; Ligon and Steward, 2000) indicating that some interaction with actin actually impedes MT based transport. Most relevant here, mitochondria can be attracted to and retained at axonal sites of experimentally-induced receptor-based signaling (Chada and Hollenbeck, 2003; Verburg and Hollenbeck, 2008) only in the presence of intact actin filaments (Chada and Hollenbeck, 2004). Along with the data presented here, this argues for a transport system in which MTs act as tracks for most normal mitochondrial traffic in the axon, while myosin motors interact with actin to oppose movement transiently, but in a fashion that becomes stable in the presence of specific signals and, perhaps, additional docking proteins. The ability of myosins to produce short, slow movements of axonal mitochondria (Morris and Hollenbeck, 1995; Ligon and Steward, 2000) would thus be incidental to their main function there. It seems likely that mitochondrial transport requires several levels of regulation, including the regulation of motor activity, motor-organelle binding, interruption or re-initiation of movement, and docking.
Drosophila neurons differ from most vertebrate neurons in that they lack neurofilaments, and in culture their axons have a very small diameter. Neurofilaments are not only abundantly present in vertebrate axons but are also known to interact with mitochondria (Wagner et al., 2003). Their possible influences on the relative roles of actin, myosins and MT motors in vertebrate neurons cannot be addressed here. Thus, some of the transport phenotypes we observed upon myosin depletion could be different in vertebrate neurons. .
The effects of myosin V depletion on both mitochondrial traffic and length could be independent functions of the motor protein (Mermall et al., 1998). However it seems likely that mitochondria require stationary phases for proper assembly and function of fission machinery. Hence a perturbation that leads to increased transport would interfere with their ability to undergo normal fission, shifting the balance of fission and fusion and producing longer mitochondria. Consistent with this, Saotome et al (Saotome et al., 2009) observed that EF-hand mutations in the Miro protein increased mitochondrial motility by suppressing Ca-dependent mitochondrial arrest and also lead to a >20% increase in mitochondrial length. Conversely, a dMiro null mutation decreased both the motility and length of axonal mitochondria (Russo et al., 2009). In addition, De Vos et al (De Vos et al., 2005) have shown that the induction of mitochondrial fission using ATP synthase and electron-transport inhibitors in neurons was accompanied by increased occupancy of mitochondria by the fission protein DRP1. However, if F-actin was disrupted, then inhibitor treatment did not increase DRP1 co-localization and fission did not increase. This indicates that F-actin is required for successful recruitment of fission proteins to mitochondria, and our results imply that this requirement includes myosin-actin interactions that halt mitochondrial movement.
We thank Haiqiong Li and Mandana Amiri for myosin V tail co localization with mitochondria data. We thank Daniel Suter and Henry Chang (Purdue University) for helpful comments during the study, Dan Kiehart (Duke University) for Anti-myosin II, James Sellars (NHLBI, NIH) and Don Ready (Purdue University) for anti-myosin V and Kathryn Miller (Washington University St Louis) for anti-myosin VI. PJH thanks Norbert Perrimon (Harvard Medical School) for hosting a sabbatical visit. This work was supported by a grant from the National Institute of Neurological Disorder and Stroke (NS 27073) to PJH, and from the Purdue Research Foundation to DP and PJH.