Mitochondria-associated myosin 19 processively transports mitochondria on actin tracks in living cells

Mitochondria are fundamentally important in cell function, and their malfunction can cause the development of cancer, cardiovascular disease, and neuronal disorders. Myosin 19 (Myo19) shows discrete localization with mitochondria and is thought to play an important role in mitochondrial dynamics and function; however, the function of Myo19 in mitochondrial dynamics at the cellular and molecular levels is poorly understood. Critical missing information is whether Myo19 is a processive motor that is suitable for transportation of mitochondria. Here, we show for the first time that single Myo19 molecules processively move on actin filaments and can transport mitochondria in cells. We demonstrate that Myo19 dimers having a leucine zipper processively moved on cellular actin tracks in demembraned cells with a velocity of 50 to 60 nm/s and a run length of ∼0.4 μm, similar to the movement of isolated mitochondria from Myo19 dimer-transfected cells on actin tracks, suggesting that the Myo19 dimer can transport mitochondria. Furthermore, we show single molecules of Myo19 dimers processively moved on single actin filaments with a large step size of ∼34 nm. Importantly, WT Myo19 single molecules without the leucine zipper processively move in filopodia in living cells similar to Myo19 dimers, whereas deletion of the tail domain abolished such active movement. These results suggest that Myo19 can processively move on actin filaments when two Myo19 monomers form a dimer, presumably as a result of tail–tail association. In conclusion, Myo19 molecules can directly transport mitochondria on actin tracks within living cells.

Mitochondria are fundamentally important in cell function, and their malfunction can cause the development of cancer, cardiovascular disease, and neuronal disorders. Myosin 19 (Myo19) shows discrete localization with mitochondria and is thought to play an important role in mitochondrial dynamics and function; however, the function of Myo19 in mitochondrial dynamics at the cellular and molecular levels is poorly understood. Critical missing information is whether Myo19 is a processive motor that is suitable for transportation of mitochondria. Here, we show for the first time that single Myo19 molecules processively move on actin filaments and can transport mitochondria in cells. We demonstrate that Myo19 dimers having a leucine zipper processively moved on cellular actin tracks in demembraned cells with a velocity of 50 to 60 nm/s and a run length of 0.4 μm, similar to the movement of isolated mitochondria from Myo19 dimer-transfected cells on actin tracks, suggesting that the Myo19 dimer can transport mitochondria. Furthermore, we show single molecules of Myo19 dimers processively moved on single actin filaments with a large step size of 34 nm. Importantly, WT Myo19 single molecules without the leucine zipper processively move in filopodia in living cells similar to Myo19 dimers, whereas deletion of the tail domain abolished such active movement. These results suggest that Myo19 can processively move on actin filaments when two Myo19 monomers form a dimer, presumably as a result of tailtail association. In conclusion, Myo19 molecules can directly transport mitochondria on actin tracks within living cells.
Myosins are motor proteins that play critical roles in various cell functions, such as force production, cell motility, morphosis, cytokinesis, vesicle, or macromolecule transportation, and organelle localization through association with actin filaments (1). A number of types of myosins have been discovered, and based on the primary structure of their motor domain, it is thought that >35 classes of myosins are constituted of the superfamily (2). All myosins contain a conserved motor domain and neck domain, followed by a tail domain specific to the myosin class. The motor domain is responsible for actin binding and converting the chemical energy of ATP to mechanical work. Following short neck domain containing IQ motifs interacts with calmodulin (CaM) and/or CaM-like light chain(s) and constitutes a lever arm when myosin moves on actin filaments. In addition, this domain also functions as a regulatory domain of various myosins. A C-terminal tail domain is structurally unique and is often critical for the classspecific functions, such as regulation of the motile activity, association with unique binding partners, and targeting the molecule to the specific intracellular structures. Several myosins in the myosin superfamily members have a coiled-coil domain after the neck region to facilitate its dimer formation, whereas others do not have, and it is thought that the dimer formation of myosin is important for the function and regulation of myosin molecules (3)(4)(5)(6).
Recent studies have revealed that several types of myosin can continuously move on actin filament, which are thought to be involved in cargo transportation in cells (7). These myosins are often called processive myosin, and it is thought that these myosins spend a majority of the actin-activated ATP hydrolysis cycle time in the "strong actin-binding" state called high duty cycle myosin (8). It is thought that two-headed myosins having a high duty cycle can processively walk on actin filaments with hand-over-hand mechanism (9).
Myo19 (also called myosin 19 or class XIX myosin) is an unconventional myosin that is expressed in vertebrates. From the primary structure, it is predicted that Myo19 consists of a motor domain, a neck domain containing three IQ motifs, and a unique tail domain, but missing a predicted coiled-coil domain (2,10). It has been reported that Myo19 is colocalized with mitochondria (11) and has been considered to play a role in actin-based mitochondria dynamics in cells (12)(13)(14)(15)(16)(17). The C-terminal tail domain of Myo19 is necessary and sufficient for mitochondria targeting (11,13,14). It is demonstrated that the C-terminal tail domain binds to outer mitochondrial rho GTPase protein (Miro1 and/or Miro2) (17), thus protecting Myo19 from degradation (16). Miro proteins also associate with KIF5 (also called kinesin 1) and dyneindynactin complex, microtubule-associated motor proteins (18)(19)(20)(21). Therefore, it is anticipated that Miro proteins coordinate both microtubule-based mitochondria movement by KIF5/dynein and actin-based movement by Myo19 in cells. Recent studies have also revealed that Myo19 migrates to the tip of filopodia in cells by stimulations such as glucose starvation and reactive oxygen species (13,15). This suggests that Myo19 may be regulated by such stimulations.
Biochemical studies have revealed that Myo19 has actinactivated ATPase activity like most other myosins (22)(23)(24) with a plus (barbed)-end-directed (23) high duty ratio motor (23,24). The IQ domain of Myo19 binds to CaM (a common light chain of unconventional myosins) as well as myosin regulatory light chains (23).
A critical unanswered question is whether Myo19 is a motor protein that is suitable for cargo transportation. As Myo19 exclusively colocalizes with mitochondria, it is anticipated that Myo19 can transport mitochondria if Myo19 has a suitable motor characteristic for cargo transportation (15,24). Since Myo19 does not have a coiled coil, it has been thought that Myo19 alone is a single-headed motor (2). It is puzzling whether a single-headed Myo19 can move on actin filaments with well-known hand-over-hand mechanism (9). The critical missing information is whether Myo19 can move processively in a way that is suitable for transportation of mitochondria, since accumulation of Myo19 at filopodial tips does not necessarily assure the processive movement of Myo19.
The aim of this study is to clarify whether human Myo19 (HM19) can move processively on actin filaments and transport mitochondria in cells. To clarify the movement of Myo19 and its role in mitochondrial movement, we utilized singlemolecule analysis with a highly sensitive total internal reflection fluorescence (TIRF) microscope by highly inclined and laminated optical sheet (HILO) illumination (25)(26)(27).
We characterized the movement of Myo19 at a singlemolecule level using demembraned and living cell systems. The movement of single Myo19 molecules was further determined in in vitro single-molecule motility assay using single actin filaments by TIRF microscopy. Based upon our findings, we propose that Myo19 is a weak gaiting processive motor, and monomer-dimer transition of Myo19 via its tail domain is critical not only for the movement of single Myo19 molecules on actin filaments but also for actin-based mitochondrial movement in cells.

HM19 constructs used in this study
To test the single-molecule movement of Myo19, we made various HM19 constructs (Fig. 1). To visualize HM19 molecules in living cells at a single-molecule level, we employed HaloTag technique. Advantage of this technique is that we can control the number of fluorescently labeled molecules within the cells by manipulating the concentration of exogenously added HaloTag ligand. HaloTag was added to the N-terminal end of HM19 to avoid possible interference of the HaloTag moiety on the interaction of HM19 and its binding partners such as mitochondria or possible influence on interheavy chain dimer formation (Fig. 1). As shown in Figure 1, there is no obvious predicted coiled-coil domain in HM19. It has been shown that mammalian myosin 6, myosin 7a, and myosin 10 do not readily form a dimer, even though they contain a predicted coiled-coil domain (4,28,29), and it is thought that dimer formation of these myosins is facilitated by the binding of target molecules (4,5,30,31), and that dimer formation is critical for the continuous movement of these myosins (32)(33)(34). To address the importance of possible dimer formation of HM19 on its movement and cargo transporter activity, we made the forced dimer constructs of HM19-leucine zipper (LZ) (Fig. 1).
It has been reported that the tail domain of myosin can function as an intramolecular inhibitor domain (35)(36)(37). Since the tail domain is involved in the regulation of dimer formation of various myosin family members (29,30,38), it is plausible that the tail domain of Myo19 may play a role in dimer formation. In order to study the role of the tail domain in the Myo19 motor function, we also produced a tailtruncated mutant of HM19 construct (Fig. 1).
Although the HaloTag technique assures a single fluorophore in a single HM19 heavy chain, fluorophore-unbound HM19 can be present. In addition to the HaloTag-HM19, therefore, we also produced enhanced GFP (EGFP)-HM19 constructs to observe the single-molecule movement of HM19 ( Fig. 1). In order to study the single-molecule stepping of HM19 movement, C-terminal FLAG tag was introduced in aid of protein isolation, and a c-Myc tag was introduced to label a quantum dot (Qdot) at the C-terminal end of the molecule via anti-Myc antibodies conjugated with Qdot ( Fig. 1). All constructs used in this study were in good agreement in terms of the molecular masses and the response to antibodies (Fig. S1).
Continuous movement of mitochondria containing HM19 on cellular actin structure It has been reported that mitochondrial transport can be driven by kinesin-dynein motors (39). On the other hand, it is unclear whether an actin-based motor, myosin such as HM19, can drive mitochondrial movement or transport. We first examined the mitochondrial movements on actin tracks to elucidate the physiological relevance of HM19 in mitochondrial movement in cells.
To address this question, heavy mitochondrial fraction (HMF) was prepared from HM19-untransfected and HM19transfected human embryonic kidney 293T (HEK293T) cells according to Shneyer et al. (13), since HMF contains HM19 (13). The isolated mitochondria vesicles were introduced into the demembraned U2OS cells, and the movement was observed. We first tested the localization of HMF in the absence of ATP to see whether HM19 is associated with cellular actin. After addition of MitoTracker-stained HMF obtained from untransfected HEK293T cells, the demembraned cells were observed (Fig. S2, A-C). The result clearly showed that the HMF colocalized with actin on demembraned cells, suggesting that mitochondrial membrane vesicles are associated with cellular actin tracks, presumably through HM19 that is known to be associated with mitochondria (11,13). We also prepared mitochondria vesicles from HEK293 cells expressing Halo-HM19-forced dimer (HM19 Full LZ) and examined the association of Halo-HM19 Full LZ with mitochondria and actin tracks. Halo-HM19 Full LZ was mostly colocalized with mitochondria vesicles (Fig. S2, D-F). Moreover, Halo-HM19 Full LZ showed notable colocalization with actin tracks (Fig. S2, G-I), suggesting the association of Halo-HM19 Full LZ on mitochondria with actin tracks. Consistently, EGFP-HM19 Full WT and EGFP-HM19 Full LZ showed marked colocalization with mitochondria. Moreover, EGFP-HM19 Full LZ showed notable localization at filopodia (Figs. S3 and S4).
Next, we examined the movement of mitochondria on demembraned cells using HMF prepared from Halo-HM19 Full LZ-overexpressed HEK293T cells (see the "Experimental procedures" section). Interestingly, HMF prepared from the Halo-HM19 Full LZ-expressing cells moved on the cellular actin tracks in the presence of ATP ( Fig. 2A and Movie S1). The mean run length and the velocity were 0.37 ± 0.09 μm, and 55.7 ± 5.9 nm/s (mean ± SEM, n = 48), respectively (Fig. 2, B, C and Table 1). The histogram of the velocity showed a wide distribution with the combination of two different peaks at 30 and 60 nm/s. On the other hand, we did not observe the movement of HMF obtained from HM19-untransfected cells in the presence of ATP. These results support that the HMF movement is driven by HM19.

Processive movement of a forced dimer of HM19 on demembraned cells
We next asked whether HM19 can move at a singlemolecule level on demembraned cells' actin tracks. For this experiment, we purified EGFP-HM19 Full LZ from EGFP-HM19 Full LZ-transfected HEK293T cells by anti-FLAG affinity chromatography, and the movement was observed with a TIRF microscope (see the "Experimental procedures" section). As shown in Figure 3A, the fluorescent spots of EGFP signals typically showed two-step photobleaching. This indicates that the fluorescent spots contain two EGFP molecules, which are presumably derived from a single dimer molecule of EGFP-HM19 Full LZ containing an LZ module. Single-molecule movement of human myosin 19 in the presence of 1 mM ATP (Movie S2). The single-molecule EGFP-HM19 Full LZ continuously moved on demembraned cellular actin tracks, and the typical time-lapse images of the movement are shown in Figure 3B. The histogram of the run length is shown in Figure 3C. The obtained result was best fit to a single exponential equation to yield an average run length of 0.43 ± 0.08 μm (SEM, n = 55; Table 1). The result indicates that EGFP-HM19 Full LZ moved processively on demembranedcellular actin in the presence of physiological ATP concentration. Figure 3D shows the histogram of the velocity of EGFP-HM19 Full LZ in 1 mM ATP. The velocity was widely distributed with the peaks at 40 nm/s and above 70 nm/s (Table 1). These velocity peaks were similar to those obtained for HMF containing Halo-HM19 Full LZ (Fig. 2C) although the population of molecules with faster velocity was larger.

Processive movement of HM19 in living cells
Colocalization of HM19 constructs with actin or mitochondria was tested by confocal microscopy (Figs. S3-S5). EGFP-HM19 WT (HM19 Full WT, without LZ) localized at mitochondria but not notably localized at filopodial tips (Fig. S4B). On the other hand, EGFP-HM19 Full LZ localized at both mitochondria and the tip of filopodia (Fig. S4C). EGFPtail-less HM19 (HM19 IQ3 ) did not show localization either at mitochondria or at the tip of filopodia (Fig. S4D). These results suggest that the tail domain is required for the mitochondrial targeting that is consistent with the previous reports (11-13, 16, 17). These results also suggest that the dimer formation is important for the movement of HM19 to the filopodial tips. It should be noted that there was no essential difference between EGFP-HM19 constructs and Halo-HM19 constructs in their localizations, consistent with the similar localization of both constructs in living cells and fixed cells (Figs. S3-S5).
Next, we asked whether HM19 can processively move in live cells at a single-molecule level. Halo-tagged HM19 constructs were expressed in HeLa cells, stained with rhodamine 110 (R110) and observed with TIRF microscope by HILO illumination (Figs. 4-6 and S7-S11, and see also "Experimental procedures" section). The HILO illumination has great advantages that not only include being able to easily find the cell structures expressing the labeled protein but also being able to photobleach a wide range of cells for single-molecule observation. We studied the movement on filopodial actin tracks that are unidirectional with barbed ends at the tips of filopodia. We first examined the movement of forced dimer constructs, Halo-HM19 Full LZ and Halo-HM19 IQ3 LZ in living cells showing notable accumulation of fluorescent signals at the filopodial tips at a single-molecule level, consistent with the confocal microscope observation of these constructs (Fig. S6).
We found many moving fluorescence spots in filopodia. Fig. S7 shows a typical two-step photobleaching of R110-Halo-HM19 Full LZ and R110-Halo-HM19 IQ3 LZ. This suggests that the fluorescent spots of the LZ constructs contain two Halo moieties. We also measured the fluorescent intensity of each particles, and the distribution of intensities was analyzed (Fig. S8, B and C). There were two peaks in which the higher intensity was approximately double of the lower intensity, suggesting that the former contains two fluorophores and the latter contains one fluorophore, respectively. The result also suggests that R110-Halo-HM19 Full LZ and R110-Halo-HM19 IQ3 LZ contain two Halo moieties. Fig. S9A shows representative successive snapshot images of the movement of R110-Halo-HM19 IQ3 LZ in filopodia. The kymograph analysis revealed the processive movement of R110-Halo-HM19 IQ3 LZ in filopodia (Fig. S9, B and C). The velocity and run length for R110-Halo-HM19 IQ3 LZ were 0.219 ± 0.008 μm/s and 1.4 ± Halo-HM19 Full LZ was expressed in HEK293T cells and stained with R110 direct reagent. The HMF prepared as described in the "Experimental procedures" section was added to demembraned U2OS cells, and the movement was observed with a TIRF microscope in the presence of 1 mM ATP (Movie S1). The bar represents 1 μm. B, run length of HM19 Full LZassociated mitochondria vesicles. The position of the moving light spot was tracked, and the run length was calculated. The solid line shows the best fit to a single exponential equation, R 0 e −r/λ , where R 0 is the initial frequency extrapolated to zero run length, r is the run length, and λ is the average run length. The average run length was 0.37 ± 0.09 μm (SEM, n = 48). C, the velocity of HM19 Full LZ-associated mitochondria vesicles. The distribution was fitted to two Gaussian equations with the mean velocity of 27.8 ± 1.0 and 61.9 ± 6.4 nm/s (SEM, n = 48), respectively. HEK293T, human embryonic kidney 293T cell line; HM19, human Myo19; HMF, heavy mitochondrial fraction; LZ, leucine zipper; R110, rhodamine 110; TIRF, total internal reflection fluorescence.
Single-molecule movement of human myosin 19 0.2 μm (mean ± SEM, n = 128), respectively (Fig. 4, C and F and Table 2). For R110-Halo-HM19 Full LZ, the mean velocity and the run length were 0.209 ± 0.008 μm/s and 0.95 ± 0.17 μm (mean ± SEM, n = 140), respectively (Fig. 4, B and E and Table 2). The velocity of R110-Halo-HM19 Full LZ in living cells was approximately three times higher than that of in vitro experiment of EGFP-HM19 Full LZ using demembraned-cell system ( show apparent filopodial tip localization in contrast to R110-Halo-HM19 Full LZ and R110-Halo-HM19 IQ3 LZ that showed notable localization at the filopodial tips (Figs. S3-S6). Figure 5A shows a typical photobleaching of R110-Halo-HM19 Full WT. This suggests that the fluorescent spots of the HM19 WT construct contain predominantly one Halo moiety.
The fluorescence intensity distribution of Halo-HM19 Full WT showed that the monomeric form of Halo-HM19 Full WT construct is dominant in contrast to those of LZ constructs (Fig. S8A). Quite interestingly, we were able to observe the movement of Halo-HM19 Full WT on filopodia (Fig. 5B, and see also Movie S3) although this construct did not show notable localization at the tip of filopodia (Fig. 5C, and see also Fig. S5).
Representative kymograph of the movement of Halo-HM19 Full WT is shown in Figure 5D. In this kymograph, the molecule moves toward filopodial tips with occasional fast movement. The mean velocity and run length for R110-Halo-HM19 Full WT were 0.263 ± 0.012 μm/s and 0.91 ± 0.08 μm (mean ± SEM, n = 127) (Fig. 4, A and D and Table 2), respectively. The velocity and run length were comparable to those of R110-Halo-HM19 Full LZ, indicating that HM19 Full WT can continuously move on actin track in filopodia without forced dimerization module. To further study the effect of the dimerization in the continuous movement of HM19, we next tested the tail-less HM19 construct without artificial   To further analyze the movement of R110-Halo-HM19 Full WT, we captured the images at a faster frame rate, 100 ms interval (10 fps). As shown in Figure 6A, R110-Halo-HM19 Full WT shows a wide range of velocity distribution although the majority of velocities were 200 to 300 nm/s. The result suggests that R110-Halo-HM19 Full WT movement includes the random motion on filopodial actin filaments (Movie S4). A similar distribution was also observed with the leucine-zippered tail-less-HM19 construct, Halo-HM19 IQ3 LZ (Fig. S10B). On the other hand, the tail and LZ-lacking HM19 construct, Halo-HM19 IQ3 , showed a wider velocity distribution without major velocity distribution at 200 to 300 nm/s range (Fig. S10A). This suggests that the movement of R110-Halo-HM19 IQ3 is based upon random diffusion rather than active movement, whereas the movement of R110-Halo-HM19 Full WT contains active movement that is similar to R110-Halo-HM19 IQ3 LZ with run length of 1 μm (Fig. 6B) in addition to random diffusion. These results suggest that the full-length HM19 construct in part forms a dimer without the LZ moiety in living cells. To address this notion, we tested another myosin class, myosin 10, which is known to make a dimer via the native coiled coil (Fig. S10, C and D). We found that myosin 10 without coiled coil, that is, monomeric myosin 10, also showed the movement similar to R110-Halo-HM19 IQ3 , whereas myosin 10 with native coiled coil showed predominantly directional movement with 1 μm/s, which is consistent with the velocity of myosin 10 in cells (40,41). The results support our idea that the movement of R110-Halo-HM19 IQ3 is random diffusion on actin filaments, whereas R110-Halo-HM19 Full WT actively moves. The inset of Figure 6A shows a representative kymograph of the movement of R110-Halo-HM19 Full WT. R110-Halo-HM19 Full WT moved continuously toward the filopodial tips, and the movement demonstrated jumps at multiple points (represented by arrowheads), suggesting the presence of diffusive motion during the directional movement. Figure 6C shows the mean square displacement (MSD) analysis of R110-Halo-HM19 Full WT. The plot does not fit random motion: f(t) = 4Dt. When the plot was fitted with f(t) = 4Dt + v 2 t 2 , where v = velocity, t = time, and D = diffusion coefficient, the velocity and the diffusion coefficient were 0.33 ± 0.07 μm/s and 0.050 ± 0.024 μm 2 /s (SEM, n = 9), respectively. Although the apparent velocity of R110-Halo-HM19 Full WT was similar to R110-Halo-HM19 IQ3 LZ (v = 0.31 μm/s, Fig. S11B), the diffusion coefficient of R110-Halo-HM19 Full WT was three to four times larger than that of R110-Halo-HM19 IQ3 LZ (0.014 ± 0.010 μm 2 /s), which was approximately five times less than that of R110-Halo-HM19 IQ3 (0.24 ± 0.01 μm 2 /s at 0.5-2.0 s, Fig. S11A). Figure 6D shows log-log plot of MSD over time. When the plot is fitted with f(t) = ct α , where c = constant, t = time, and α = slope, the α value for R110-Halo-HM19 Full WT is calculated to be 1.47 ± 0.08 (SEM), which is between R110-Halo-HM19 IQ3 (α = 1.17 ± 0.01, Fig. S11C) and R110-Halo-HM19 IQ3 -LZ (α = 1.74 ± 0.01, Fig. S11D). This suggests that the movement of R110-Halo-HM19 Full WT is composed of slow directional movement with the velocity similar to the forced dimer construct and fast diffusion component. These analyses further support an idea that HM19 Full WT without LZ can continuously move on actin in living cells (see Discussion section).
A potential experimental limitation of the use of HaloTag technology is that it is not fully assured whether HaloTag ligands bind all the expressed Halo-tagged protein molecules in cells. This implies that the moving single fluorescence spots may not always be single molecules of HM19s. To address this issue, we also examined the cellular movements of HM19 using EGFP-HM19 Full WT and EGFP-HM19 Full LZ in living HeLa cells (Figs. S12 and S13). While the number of filopodial tips having EGFP-HM19 Full WT was fewer than those of EGFP-HM19 Full LZ (Fig. S3), we found that EGFP-HM19 Full WT also localized at the tip of the filopodia (Fig. S12).  EGFP-HM19 Full WT moved on filopodial actin tracks at a single-molecule level. The velocity and run length were 0.216 ± 0.007 μm/s (mean ± SEM, n = 110) and 0.85 ± 0.14 μm (mean ± SEM, n = 110) for EGFP-HM19 Full WT and 0.185 ± 0.005 μm/s (mean ± SEM, n = 103) and 0.94 ± 0.09 μm (mean ± SEM, n = 222) for EGFP-HM19 Full LZ, respectively ( Fig. S13 and Table 3). These results support that the HM19 Full WT single molecules can move processively on cellular actin tracks in living cells.

Single-molecule movement and stepping of HM19 in vitro
While aforementioned results indicate the continuous movement of HM19 at single-molecule level in cells, the best approach to obtain conclusive evidence of processive movement of HM19 is in vitro single-molecule motility analysis (Fig. 7). To observe the HM19 movement, we made flow chambers in which single actin filaments were immobilized on the glass surface and monitored the movement of isolated EGFP-HM19 Full LZ molecules in physiological ATP concentration. The mean velocity in the presence of 1 mM ATP was 95.1 ± 7.4 nm/s (mean ± SEM, n = 47) (Fig. 7A), which was 1.7 times faster than the velocity at 5 μM ATP in a similar condition ( Table 3). The average run length at 1 mM ATP was 0.18 ± 0.01 μm (n = 47) (Fig. 7B).
To determine the HM19 Full LZ stepping, Qdot was attached to the C-terminal end of an isolated HM19 Full LZ through antic-Myc antibodies (Fig. 8A). We observed the successive HM19 Full LZ-Qdot movement on single actin filaments in the presence of 2 μM ATP. By tracking the center position of HM19 Full LZ-Qdots using FIONA technique (9), we determined the step sizes of HM19 Full LZ ( Fig. 8B and Movie S5). Figure 8C shows the step size distribution of HM19 Full LZ-Qdot. The distribution was symmetrical, and the mean step size of HM19 Full LZ-Qdot was 33.8 ± 1.3 nm (mean ± SEM, N steps/Qdots = 91/26) for forward step and -24 ± 3.3 nm (SEM, N steps/Qdots = 10/6) for backward step. The mean forward step was slightly shorter than a half pitch of F-actin helix (36 nm). In most cases, the back step was only single steps, and we did not observe successive back steps. Figure 8D Table 3). These results indicate that HM19 Full LZ-Qdot moves four to five steps on average at the physiological concentration of ATP on single actin filaments and takes 2.8 steps per second, that is, 2.8 s −1 at 22 C. This value is consistent with the V max of the actin-activated ATPase activity for monomeric truncated HM19 (23,24).

Discussion
Although it is well known that kinesin-dynein motor proteins are associated with mitochondrial movement (39), several myosins such as nonmuscle myosin 2, myosin 5a, myosin 6, and Myo19 (19), are found adjacent to mitochondria (42). However, the role of this myosin is still unknown. Since Myo19 is highly colocalized with mitochondria, it has been thought that it plays a role in mitochondrial dynamics and movement (11). Myo19 can be found at the filopodial tips (13,15). However, it is unclear whether Myo19 can actively and processively move to the filopodial tips or if it localizes at the filopodial tips by other means such as diffusion or movement driven by other motor proteins. Our objective of the present study is to clarify whether HM19 is a processive motor that can actively transport mitochondria. In the present study, we showed for the first time that HM19 can processively move within filopodia to the tips and transport mitochondria through a processive manner. This processive movement of HM19 agrees with the high duty ratio of tail-truncated HM19 by kinetic analyses (23,24), as high duty ratio is one of the critical features to account for processive movement of motor proteins (1,43).
We found that the isolated mitochondria membrane fraction is localized at actin tracks on demembraned cells (Fig. S2). This suggests that endogenous HM19 on mitochondria membranes attaches to actin filaments. This view is supported by the result that the mitochondria membranes were prepared from the cells expressing Halo-HM19 Full LZ; the forced dimer of HM19 not only localized at the actin tracks but also moved on the cellular actin tracks in the presence of ATP (Fig. 3). However, the mitochondrial membrane fraction with endogenous HM19 did not move on the cellular actin tracks in the presence of ATP in in vitro system. It is assumed that endogenous HM19 is present on mitochondria from untransfected cells (13). Therefore, a question is why the endogenous HM19 did not apparently support the isolated mitochondrial movement. It is plausible that the majority of endogenous HM19 molecules on isolated mitochondrial membrane in vitro are "inhibited," presumably because of its monomeric state.
To directly evaluate the transporter activity of HM19, we studied the movement of EGFP-HM19-forced dimer at singlemolecule level and found that it processively moves on cellular actin tracks in demembraned cells (Fig. 3). This implies that HM19-forced dimer can be a processive motor suitable for mitochondria transport. The velocity of EGFP-HM19-forced dimer on demembraned cells showed two velocity peaks, which were similar to those of mitochondria membranes obtained from the cells expressing HM19-forced dimer. This result suggests that HM19-forced dimer supports the movement of isolated mitochondrial membrane. We think that the   (Fig. S5). On the other hand, the tailtruncated Halo-HM19-forced dimer localizes at filopodial tips, but not with mitochondria, which is consistent with the previous reports that the tail domain of HM19 is responsible for mitochondrial binding (11,13,14). A critical question is whether HM19 actually moves toward filopodial tips by its own motor activity. In order to address this question, we studied intrafilopodial movement of HM19 at a singlemolecule level in living cells by using TIRF microscopy.
Halo-HM19-forced dimers continuously moved within filopodia in living cells. Kymograph analysis revealed that the movement is unidirectional, that is, toward filopodial tips.
Since the tip of filopodia is plus end of actin bundles (46), the result is consistent with the previous finding that showed HM19 is a plus-directed myosin (23). Halo-ligand provides more intense fluorescence, which is advantageous for tracking the movement of HM19, whereas a single fluorophore containing moving spots may contain HM19 molecules without bound Halo ligand, thus possibility of having multiple HM19 molecules. To circumvent this problem, we employed EGFP-HM19 Full LZ (Fig. 3). Single fluorescent EGFP spot of EGFP-HM19 Full LZ processively moved on actin track. The result further supports processive movement of a single-molecule HM19 dimer. Since HM19 does not have a coiled-coil domain (2,10), it is thought that HM19 is a monomeric myosin. However, it is known that the two heads of myosin can concertedly interact with actin to produce processive movement upon actin through a hand-over-hand mechanism (9). Quite interestingly, we observed a number of fluorescent spots of Halo-HM19 Full WT molecules moving in filopodia (Figs. 4 and 5), although Halo-HM19 Full WT did not show apparent localization at filopodial tips under confocal microscope (Fig. S5). For Halo-HM19-forced dimers, the fluorescence intensity at filopodial tips was 6 to 10 times brighter compared with Halo-HM19 Full WT (Fig. S6). This indicates that Halo-HM19-forced dimers can more efficiently move to filopodial tips. Note that the velocity of HM19-forced dimer toward the tip of filopodia in living cells (200 nm/s) is much faster than that of isolated EGFP-HM19-forced dimer (60 nm/s) in vitro. The difference in the velocity is consistent with the temperature difference of ATPase activity between in vitro and in vivo conditions (Q10 of HM19 = 2.5 calculated from temperature difference [22 versus 37 C]; data not shown). On the other hand, the kymograph analysis for Halo-HM19 Full WT shows processive movement in filopodial actin tracks in living cells (Figs. 4 and 5), and the velocity of Halo-HM19 Full WT was essentially the same as that of Halo-HM19-forced dimers (Fig. 4). This indicates that Halo-HM19 Full WT moves with its own motor activity. The run lengths of both Halo-HM19-forced dimers and Halo-HM19 Full WT in living cells were 2.5 times longer than that of isolated EGFP-HM19-forced dimer in vitro (1 versus 0.4 μm). This is likely because of the difference in cells and in vitro environments, such as higher effective concentration of Myo19 in cells. A critical question is why HM19 Full WT is not accumulated at the filopodial tips despite the fact that we found processive movement of HM19 Full WT in living cells. Taking into account these findings, we propose the following hypotheses. HM19 Full WT is predominantly inactive within cells, presumably in its monomeric form. However, a certain population of HM19 Full WT is activated, presumably because of the conversion from monomer to dimer by unknown mechanism. Supporting this view, it was found that a part of fluorescent spots of Halo-HM19 Full WT showed two-step photobleaching, suggesting the presence of two fluorophores, although the fluorescent spots of Halo-HM19 Full WT contain mostly one fluorophore (Figs. 5A and S8). Moreover, we found that the run length of HM19 Full WT is similar to HM19 Full LZ, suggesting that the activated form of HM19 Full WT, which is presumably its dimer form, moves similar to the forced dimer of HM19. While these findings suggest that HM19 Full WT can form a dimer in cells, it is likely that the produced dimer is not highly stable in the same way as forced-dimer constructs of HM19. Supporting this view, we found that the accumulation of HM19 Full WT at filopodial tips was much lower than that of HM19-forced dimer (Figs. S6 and S12). We think that the HM19 Full WT can form a dimer while it moves on actin filaments as shown previously for the tailtruncated myosin 10 that the dimer formation can be facilitated via myosin head-actin interaction (33,47). However, once it dissociates from actin track, the dimer of the HM19 Full WT is disassembled into monomers and diffused away from the actin track. On the other hand, the stable forced dimer of HM19 Full , that is, HM19 Full LZ, can rebind to the actin track and move further toward filopodial tips, thus accumulating at the tips. In addition, since the population of HM19 Full WT dimer is likely to be much less than the HM19-forced dimer, it is anticipated that the number of HM19 Full WT molecules processively moving on actin track is much fewer than that of forced dimer constructs. These can explain much lower accumulation of HM19 Full WT at the filopodial tips.
Analysis of MSD data further supports this view (Figs. 6 and S11). In MSD analysis with log-log axes, the slope (α) = 1 indicates that the movement is purely by diffusion and the (α) = 2 indicates the movement with a constant velocity without diffusion. In the case of myosin 5a HMM movement on single actin filaments in vitro, the value is estimated as α = 1.9 (48). This indicates that myosin 5a moves at a constant velocity without diffusion in vitro. On the other hand, the α values were 1.2, 1.7, and 1.5, for HM19 IQ3 , HM19 IQ3 LZ, and HM19 Full WT, respectively (Figs. 6D and S11). This result suggests that the movement of HM19 IQ3 is dominantly diffuse. The movement of HM19 Full WT is more directional than that of HM19 IQ3 , but more diffusive than HM19 IQ3 LZ, which shows constant directional movement. This is in good agreement with the result of the calculated velocity = 330 nm/s of Single-molecule movement of human myosin 19 HM19 Full WT from the MSD analysis, which contains significant nondirectional diffusive movement. Interestingly, time versus MSD of HM19 IQ3 showed a nonlinear relationship, especially in a short period (Fig. S11A, inset). This suggests that HM19 IQ3 single molecules can exhibit directional movement in certain short periods although HM19 IQ3 moves predominantly with a simple diffusion. This may indicate that the tail-truncated single-headed HM19 has the ability to move on filopodial actin filaments, similar to the single-headed myosin 9 (49). It was reported previously that KIF1A, one of the kinesin family members, shows simple diffusive movement with some biased diffusion (50). Therefore, the movement of HM19 IQ3 may also contain a biased diffusion component.
EGFP-HM19 IQ3 shows virtually no filopodial tip accumulation (data not shown), whereas HM19 Full WT showed some filopodial tip accumulation in live cells (Fig. S12). This is consistent with the MSD analysis described above (Fig. 6). The result also suggests that the tail domain contributes to a dimer formation. There is no predictable coiled-coil domain in the tail domain of HM19, and a question is how HM19 can form a dimer in cells. One possibility is that HM19 may form a dimer through the binding of its partner molecules. The reason for this possibility is that one of the potential binding proteins is Miro1/2, which is a protein that is known to recruit and stabilize HM19 with mitochondria. However, the direct binding of HM19 to Miro1/2 protein remains to be fully clarified (16). Alternatively, C-terminal tail domain of HM19 may directly dimerize through the binding of some small molecules as it is reported that phosphatidylinositol 3,4,5-trisphosphate binding at the tail of myosin 10 facilitates dimer formation (29) and phosphatidylinositol 4,5-bisphosphate binding at the C-terminal tail region of myosin 6 promotes dimer formation (30). Further studies are required to clarify the regulation mechanism of dimer formation of HM19.
In order to further clarify the characteristics of processive movement of HM19 at the molecular level, the movement and stepping of HM19 was studied with in vitro single-molecule motility assay. Our result clearly indicates that HM19 dimer processively moves on single actin filaments (Fig. 7). To further analyze the processive movement of HM19, we studied the successive stepping of HM19 using Qdot-attached HM19 dimer (Fig. 8A). The HM19 Full LZ single molecule moved on single actin filaments for multiple successive steps (Fig. 8B). These results are consistent with the results obtained with EGFP-HM19 Full LZ in cells and demembraned cells and support that an HM19 dimer can processively move on actin filaments. The average step size of HM19 Full LZ was 34 nm. It has been thought that the length of the IQ domain of myosin is critical for the lever-arm length, hence its step size (51). HM19 has three IQ motifs that are calculated to be 11 to 12 nm of possible lever arm (52). In the case of myosin 10 having three IQ motifs, an SAH domain after the IQ domain serves as an extension of lever arm to explain its step size of 30 to 36 nm (33). There is no SAH sequence after the IQ domain of HM19, and it is obscure how HM19 shows large steps. It might be possible that the tail domain in addition to the IQ domain contributes to the step size of HM19 Full LZ construct since the forced dimerization module is located at the C-terminal end of the molecule. However, since HM19 IQ3 LZ can also processively move toward filopodial tips, it is less likely that the tail domain is required for processive movement of HM19 in cells. Further study into the long lever arm of HM19 is required.
The run length of HM19 IQ3 LZ on single actin filaments in vitro was 180 nm (Fig. 7B). This is significantly shorter comparing with other processive myosins such as myosin 5a and myosin 10 (47, 53). A possible explanation is weak gating effect of HM19. Based upon the run length of HM19 IQ3 LZ, the duty ratio of HM19 IQ3 LZ is estimated to be 0.6 with the method previously used for myosin 5a (54). This finding suggests that single-molecule HM19 is not a great motor to transport cargos, such as mitochondria over a long distance. This conclusion is consistent with the studies by Usaj and Henn (24) and Shneyer et al. (15). Conversely, since it is highly likely that a large cargo such as mitochondria associates with a number of HM19 molecules, the assumption is that HM19 molecules can concertedly support successive movement of mitochondria in cells. If a single molecule of HM19 can move mitochondria for a long distance, many molecules associated with a single mitochondrion may interfere with each other for the movement. Our results indicate that HM19 Full LZ takes the step size of 34 nm and moves on single actin filaments at 95 nm/s at 1 mM ATP. This indicates the single stepping rate of 2.8 s −1 at 22 C. This rate is the same as 2.8 s −1 of the reported ATPase rate with HM19 IQ3 , a monomeric form at 25 C (24). This further supports the idea that HM19 has a weak gating effect, in which the two heads of HM19 dimer may not influence each other to decrease the cycling rate.
In conclusion, we observed single-molecule movement of HM19 for the first time in in vitro and in cells and found that HM19 dimer moved processively on actin structure in vitro and in cells. We also demonstrated that the WT of HM19 without a forced dimerization module is capable of moving on filopodial actin tracks, presumably because of dimerization of HM19 in cells. Finally, we demonstrated that the C-terminal tail region of HM19 is critical for such movement of HM19 in cells.
Recombinant full-length HM19 (HM19 Full WT) (including 1-970 of Myo19 amino acids) was subcloned into pEGFP-C1 (Clontech) to make EGFP-HM19 Full WT. The stop codon of EGFP-HM19 Full WT was removed by PCR site-directed mutagenesis (59) using Pfu Ultra II (Agilent Technologies), and an oligo DNA containing two sets of GCN4 LZ sequence (60) was then inserted to make EGFP-HM19 Full LZ. The sequence after the end of HM19 Full WT was SGTTH SYTSMKQLEDKVEELLSKNYHLENEVARLKKLVGERTSMK QLEDKVEELLSKNYHLENEVARLKKLVGERTSGSEL.
For in vitro single-molecule motility assay of HM19, the stop codon of EGFP-HM19 Full LZ was removed, and c-Myc/FLAG sequences (amino acid: EQKLISEEDL/DYKDDDDK) were introduced at the 3 0 end. The N-terminal EGFP sequence was converted to HaloTag sequence (61)  Nakarai Tesque] containing 100 μM 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, 80 nM aprotinin, 1.5 μM E-64, 2 μM leupeptin, 5 μM bestatin, and 1 μM pepstatin A), followed by gentle sonication for 10 s. The cells were centrifuged at 1000g (TOMY AR015-24 rotor) for 1 min to pellet the cell debris. The supernatant was centrifuged again at 10,000g for 20 min to precipitate HMF, and the HMF pellet was resuspended in 100 μl of homogenizing buffer. For double staining of R110-Halo-HM19 with mitochondria vesicles, 10 μM MitoTracker Red FM (Thermo Fisher Scientific) was mixed with the HMF prepared from R110-Halo-HM19 Full LZ transfected HEK293T cells. The HM19 Full LZ-HMF was washed thrice with a homogenizing buffer by centrifugation at 10,000g for 20 min to remove unbound MitoTracker and resuspended in the same buffer. The aliquots of the HMF were flash-frozen by liquid nitrogen and stored at −80 C.

Preparation of purified HM19 protein
A nonmuscle-type myosin light chain (MRLC12A) cDNA was obtained from a human kidney cDNA library and transferred to pEB Multi vector (Fujifilm Wako Chemicals). HEK293T cells (Invitrogen) were transfected with the vector DNA, and neomycin-resistant MRLC12A-expressing HEK293T cells were created. EGFP-HM19 Full LZ was then transiently expressed in MRLC12A-expressing HEK293T cells by a transfection of recombinant Myo19 heavy chain cDNAs using PEI Max (Polysciences, Inc) and then purified using anti-FLAG antibody-agarose resins (SIGMA). In short, 6 × 10 7 cells were harvested at 24 h after transfection and homogenized in a lysis buffer (0.25 M KCl, 50 mM Hepes-KOH [pH 7.5], 5 mM EGTA, 15 mM β-mercaptoethanol, 1 mM ATP, 2% CHAPS, and protease inhibitor cocktail [described previously]). The suspension was mixed with anti-FLAG antibody-agarose resins, and the tube containing the suspension was rotated at 4 C for 1 h. After two washes with a wash buffer (0.12 M KCl, 12.5% trehalose, 1 mM EGTA, 15 mM β-mercaptoethanol, 2% CHAPS, and protease inhibitor cocktail), the protein was eluted with an elution buffer (the wash buffer containing 0.1 mg/ml FLAG peptide). The eluted fractions were flash-frozen by liquid nitrogen and stored at −80 C.

Labeling of Halo-HM19s with HaloTag ligands and mitochondria staining in living cells
HeLa or HEK293T cells were cultured with Dulbecco's modified Eagle's medium in 3.5 cm glass bottom dish (Matsunami Glass; D35-14-1-U), and cells were transfected with Halo-Myo19s using Lipofectamine 3000 (Invitrogen) according to the manufacturer's protocol. After 16 to 24 hours, the medium was removed and the dish was washed with Leibovitz-15 medium (Gibco) plus 10% fetal bovine serum, followed by the addition of 0.1 μM R110 Direct HaloTag ligands at 37 C for 15 min. Following thrice washes with Leibovitz-15 medium, the dish was incubated at 37 C for 30 min in Leibovitz-15 medium. After thrice washes with FluoroBrite (Gibco), the cells were incubated at 37 C in a CO 2 incubator before observation. For mitochondria staining, the cells were incubated in a medium containing 0.2 μM MitoTracker for 30 min at 37 C in a CO 2 incubator.
Qdot labeling of HM19 and fluorescent labeling of single actin filaments For the step size and dwell time determinations, Qdot 525 conjugated with goat F(ab') 2 antimouse immunoglobulin G (H + L) were mixed with anti-c-Myc monoclonal antibody (Clontech) as described previously (33,34). F-actin was labeled by Alexa Fluor 568 Phalloidin at a 1:1 ratio overnight and stored at 4 C. For single-molecule experiments with Qdotlabeled HM19, 7.5 μM labeled and unlabeled actins were mixed at the ratio of 1:10, and the mixture was diluted to 1 μM just before the experiments.

Single-molecule motility assay with TIRF microscope
The movement of R110-Halo-HM19s in living cells was observed with a hand-built TIRF/HILO setup (62) using an inverted Nikon Ti2 Eclipse microscope equipped with an objective lens (100× APO TIRF/numerical aperture [NA] = 1.49), a 488 nm Argon-ion laser system (Melles Griot), a backilluminated electron multiplier charge-coupled device camera Single-molecule movement of human myosin 19 (iXON DU-897; Andor Technology), and temperature control system (model INU-KRi; TOKAIHIT). The fluorescent light of cultured cells expressing labeled HM19 proteins was monitored using Nikon NIS Elements software, and the movements of HM19 were recorded after proper photobleaching. The experiments were conducted at 37 C. The movement of isolated EGFP-HM19 Full LZ on demembraned cells and single actin filaments was observed with a hand-built TIRF setup using an Olympus IX71 microscope, an objective lens (Pla-nApo-N ×60, NA = 1.49, oil), a 488 nm argon laser (FITEL HPU50101; Furukawa Electric), and an EMCCD camera (iXON+ DU-897; Andor Technology). We used a 4f imaging system composed of two achromatic lenses whose focal positions were 70 and 200 mm, respectively, as a relay optics. The actual magnification of the microscope containing a relay optics was 170×. Experiments were carried out in a solution containing 25 mM KCl, 20 mM Hepes (pH 7.5), 1 mM MgCl 2 , 0.2 mM CaCl 2 , 143 mM 2-mercaptoethanol, 12 μM CaM, and 1 mM Trolox in the presence of indicated concentration of ATP. For the assay with single actin filaments, flow chambers were prepared by using no. 1.5 glass coverslips and glass slides (Matsunami Glass). Alpha-actinin (catalog no.: A9776; SIGMA) was used to immobilize F-actin, and casein from bovine milk (07319-82; Nakarai Tesque) was used for glasssurface blocking. The movement of fluorescent HM19 single molecules was monitored and recorded using Aodor SOLIS 32 bit software (version 4.28.30050.0). The experiments were conducted at 22 C.

Data analysis
The video images of single molecules captured with a TIRF microscope were analyzed by using in-house 2D Gaussian fitting and tracking software (63). A function for the recorded stage drift was calculated and subtracted from the raw data of each runs for each movie. Run length and velocity of HM19 were concurrently determined by measuring the length and time between on and off the tracks. Time-lapse images, kymographs, and trajectory of the HM19 movements were analyzed using ImageJ software (National Institutes of Health). Numeric calculation was done with Excel software (Microsoft), and the data plotting and statistical analysis were done with Prism software (GraphPad Software, Inc). For step size analysis, an in-house step-fitting software based on an algorithm described in the study by Kerssemakers et al. (64) was used as described previously (33,34).

3D deconvolution microscopy
The colocalization of mitochondria membrane vesicles with cellular actin was observed with DeltaVision OMX (GE Healthcare). U2OS cells were cultured on round cover glass (CG15NH, Ø12 mm; Thorlabs), treated with 100 ng/ml nocodazole for 8 h, and demembraned as shown previously in the presence of Alexa Fluor 488 phalloidin, followed by the addition of MitoTracker-stained HMF. After three washes with PBS, the cover glass was mounted to a slide glass using Fluoro gel with DABCO (EM Science) and sealed with nail polish. The fluorescence image was captured with conventional light path setting using DeltaVision OMX software, and the 3D deconvolution and image registration were then applied using SoftWoRx 6.5.2 software (GE). The maximum intensity of a series of z-axis images was projected to one x-y image, and the images were processed using ImageJ software.

Confocal light microscopy
For the colocalization of HM19 and actin, Halo-HM19transfected HeLa cells cultured on 3.5 cm glass bottom dishes were fixed with 4% paraformaldehyde for 15 min at 37 C. After being washed with PBS twice, the cells were treated with 0.1% Triton X-100 in PBS for 3 min. The cells were washed twice with PBS and quenched with 0.1 M glycine in PBS for 15 min at room temperature. Actin and HaloTag proteins were stained for 1 h with Alexa Fluor 647 Phalloidin (0.4 U/ml) and 1 μM HaloTag tetramethylrhodamine ligand, respectively. Unbound fluorophores were removed using five washes with PBS. For the colocalization of HM19 and mitochondria, HeLa cells were transfected with EGFP-HM19s and stained with MitoTracker as described previously. The cells were observed by using an Olympus FV1000 laser scanning confocal microscope system with an IX81 microscope, an objective lens (PLANAPO N 60×, 1.40 NA, oil), 473 nm, 559 nm, and 635 nm lasers (Olympus). The fluorescence signal was captured in a range of 490 to 540 (or 590) nm for EGFP and R110, 575 to 620 nm for tetramethylrhodamine, 575 to 675 nm for MitoTracker, and 655 to 755 nm for Alexa Fluor 647. FV10-ASW Viewer, version 3.1, software was used to control the equipment and data analysis.

Data availability
All data are contained within the article.