Myosin X Regulates Sealing Zone Patterning in Osteoclasts through Linkage of Podosomes and Microtubules

Abstract

Osteoclasts use actin-rich attachment structures in place of focal adhesions for adherence to bone and non-bone substrates. On glass, osteoclasts generate podosomes, foot-like processes containing a core of F-actin and regulatory proteins that undergo high turnover. To facilitate bone resorption, osteoclasts generate an actin-rich sealing zone composed of densely packed podosome-like units. Patterning of both podosomes and sealing zones is dependent upon an intact microtubule system. A role for unconventional myosin X (Myo10), which can bind actin, microtubules, and integrins, was examined in osteoclasts. Immunolocalization showed Myo10 to be associated with the outer edges of immature podosome rings and sealing zones, suggesting a possible role in podosome and sealing zone positioning. Further, complexes containing both Myo10 and β-tubulin were readily precipitated from osteoclasts lysates. RNAi-mediated suppression of Myo10 led to decreased cell and sealing zone perimeter, along with decreased motility and resorptive capacity. Further, siRNA-treated cells could not properly position podosomes following microtubule disruption. Osteoclasts overexpressing dominant negative Myo10 microtubule binding domains (MyTH4) showed a similar phenotype. Conversely, overexpression of full-length Myo10 led to increased formation of podosome belts along with larger sealing zones and enhanced bone resorptive capacity. These studies suggest that Myo10 plays a role in osteoclast attachment and podosome positioning by direct linkage of actin to the microtubule network.

Introduction

Osteoclastic bone resorption is dependent on tight attachment of the cell to the bone substrate. When on bone, osteoclasts form the sealing zone, a ring-shaped structure that surrounds the ruffled border, through which protons and proteases are secreted to effect resorption. Sealing zones are composed of structural units with certain similarities to podosomes that osteoclasts form on non-bone substrates (1). Podosomes, termed for their foot-like appearance, are typified by short F-actin bundles orientated perpendicular to the substrate (2, 3). Podosomes form a cylindrical core of F-actin with a “cloud” of microfilaments radiating from the base of the core (1). Actin-binding proteins, integrins, and regulatory proteins such as kinases associate specifically with the actin core or the radiating cloud of F-actin that surrounds the podosomes and with specific regions of the sealing zone. Microtubules also are critical for the integrity of these actin-rich structures, as several groups have demonstrated that podosome and sealing zone formation and function in macrophages and osteoclasts are dependent on an intact microtubule system (4–8).

Myosins are actin-based molecular motors involved in many cellular processes including muscle contraction, cell migration, cell polarization, and mitosis. Myosins consist of a head domain that hydrolyzes ATP and binds F-actin, IQ domains that bind calmodulin or calmodulin-like proteins, and tail domains that allow the myosins to attach to cargo. Categorized by the motor head domain, myosins are currently grouped into over 20 classes (9). We recently showed that regulated proteolysis of the “classical” myosin IIA isoform stimulates the onset of cell fusion during osteoclastogenesis (10). Myosin X (Myo10), which is expressed in most vertebrate tissues at low levels, is the sole member of the class X myosins (11). Myo10 possesses a long tail that has been predicted to contain a coiled-coil domain adjacent to its IQ motifs, allowing dimerization as seen with classical myosins (11); however, empirical studies suggest that this region may instead form a single α-helix that lengthens the head domain (12). Beyond this region, each Myo10 tail also possesses three pleckstrin homology domains (PH), one myosin tail homology 4 domain (MyTH4),² and a FERM domain (11). PH domains have been shown to be involved in protein-protein interactions along with mediating signaling transduction through binding of phosphoinositides (13). FERM domains link cell membrane proteins to the cytoskeleton (14), while the less-studied MyTH4 domains have been shown to be involved in microtubule binding (15). These various tail domains allow Myo10 to bind and potentially transport multiple proteins. Myo10, through immunoprecipitation and hybrid assays, has been shown so far to bind actin, microtubules, VASP, products of PI3 kinase, and beta integrins (16–19). The importance of microtubule binding to Myo10 recently has been underscored by several studies demonstrating its role in assembly and orientation of meiotic and mitotic spindles via its MyTH4-FERM domains (18, 20, 21).

In this study, we have examined the role of Myo10 in osteoclast podosome patterning and sealing zone formation. We hypothesize Myo10 is a potentially important contributor to osteoclast attachment, resorption, and migration due to its capacity to bind integrins and as an effector of PI3 kinase, a signaling molecule known to play a crucial role in osteoclast spreading and motility (22–25). Here we demonstrate that Myo10 plays a role in osteoclast attachment and subsequent cell spreading and migration by linking podosomes/sealing zones to the microtubule network. This work is the first to define a role for Myo10 in podosome-based adhesion, as well as demonstrating its role as a linker between the two cytoskeletal systems in osteoclasts.

EXPERIMENTAL PROCEDURES

Reagents

Affinity-purified chicken antibody against mouse Myo10 was previously described (26). The specificity of this antibody in immunocytochemistry is demonstrated in supplemental Fig. S1, showing lack of reactivity by preimmune serum as well as loss of specific labeling in Myo10 siRNA-treated cells. In addition, supplemental Fig. S1 shows a Western blot of lysate from mature osteoclasts probed with the affinity-purified antibody. This Western demonstrates that murine osteoclasts express full-length Myo10 (at ∼250 kDa), consistent with results seen in macrophages (16). Minor bands represent breakdown products that are routinely seen in lysates from osteoclasts, which are highly proteolytically active. A “motor-less” 165-kDa isoform of Myo10 found in brain (26) is not detected in osteoclasts. Loading control antibodies to GAPDH and β-actin, both mouse monoclonal antibodies, were purchased from Abcam (Cambridge, MA). A mouse monoclonal anti-β-tubulin antibody was purchased from Invitrogen (Carlsbad, CA).

Cell Culture, Immunocytochemistry, and Microscopy

Osteoclasts were generated either from RAW264.7 macrophages (American Type Culture Collection, Manassas, VA) or murine bone marrow as previously described (10, 27–29) in the presence of a soluble form of RANKL (30). For immunocytochemistry, osteoclasts were cultured either on glass coverslips or thinly cut ivory slices and fixed as previously described (10, 27–29). For visualization of microtubules, glutaraldehyde was added at a final concentration of 0.75% to each fixative solution. Primary antibodies were added in a standard PEG blocking buffer, and were detected using Alexa-labeled secondary antibodies (Invitrogen). F-actin was labeled using Alexa-coupled phalloidin, also from Invitrogen. Nuclei were labeled by incubation with bisbenzimide for 10 min following the secondary staining. Cells were visualized using either a Nikon Eclipse 80i epifluorescent microscope with SPOT software (Diagnostic Instruments, Sterling Heights, MI) or a Zeiss 510 META laser scanning confocal microscope at the Campus Microscopy and Imaging Facility, The Ohio State University. Cell perimeter was measured using SigmaScan Pro 5.0 software (SPSS Science, Chicago, IL).

Competitive RT-PCR of Myo10 mRNA

To determine Myo10 mRNA expression levels by competitive RT-PCR, primers were created that corresponded to murine sequences. The sense primer was of the sequence 5′-AACAATGGACAGCTTCTTTCCCG-3′, while the antisense primer was of the sequence 5′-GCGATAGCATTCGTTGGCAATGG-3′. For an internal standard, a cDNA was created that corresponded to the expected PCR product using the primers above, but contained an internal deletion of 21%, a T7 promoter element, and a tail of 15 adenosines, as previously described (28, 31). This product was transcribed in vitro using the MAXIscript system (Ambion, Austin, TX), and 1 pg of the resulting RNA (the internal standard) was added to 1 μg of osteoclast total cellular RNA prior to reverse transcription and PCR. These reactions were performed using the Superscript First-strand synthesis System and TaqDNA Polymerase, both from Invitrogen. The resulting RT-PCR products were run in a 2% agarose gel and stained with ethidium bromide to visualize the relative intensities of the bands, which were measured using Quantity One software (Bio-Rad).

Immunoprecipitation

Immunoprecipitation was performed essentially as previously described (10, 32). The cells were solubilized in M-PER supplemented with protease inhibitors, and the resulting lysates were centrifuged for 10 min at 20,000 × g to remove insoluble material. Precleared lysates were incubated with Myo10 or tubulin antibody for 12–16 h at 4 °C and then with anti-chicken IgY-agarose (Gallus Immunotech, Cary, NC) or protein A-Sepharose for 30 min at 4 °C. The complexes were washed with NET-GEL buffer (50 mm Tris-Cl, pH 7.4, 150 mm NaCl, 1 mm EDTA, 5 mm sodium azide, 0.1% Nonidet P-40, and 0.25% gelatin), run in SDS-PAGE, transferred to Hybond membrane and probed by Western analysis. [³⁵S]Methionine/cysteine pulse-labeling of cells for analysis of Myo10 stability was performed as previously described (10).

Western Analysis

For Western analysis of whole cell lysates, osteoclasts were harvested with M-PER reagent (Pierce Biotechnology), run in pre-cast PAGE 4–20% gradient gels (Bio-Rad) and transferred to Hybond membrane (GE HealthCare Biosciences, Piscataway, NJ). Primary antibodies were allowed to bind to the membranes using standard methodology, and were detected using horseradish peroxidase-labeled secondary antibodies coupled with SuperSignal West Pico Chemiluminescent reagents (Pierce Biotechnology).

RNAi-mediated Knockdown of Myo10

To suppress murine Myo10 expression, siRNAs were designed and synthesized by Ambion (Austin, TX). siRNA 73578, which was used for all relevant experiments, was found to have optimal activity at 75 nm, whereas siRNA 73762, which was used to confirm many of the results, was optimally active at 50 nm. For all experiments, a non-targeting dsRNA from Ambion (Austin, TX) was used as a negative control, and siRNAs homologous to siRNAs 73578 and 73762 but containing three point mutations in the middle of the sequence also were used as negative controls (co73578, co73762). RAW264.7 cells were plated and stimulated with GST-RANKL to form osteoclasts. On day 4 of differentiation, targeting siRNAs or an equivalent concentration of a negative control siRNA was added to Lipofectamine 2000 (Invitrogen) in plain Dulbecco's modified Eagle's medium and added to the cells. After 3 h, Dulbecco's modified Eagle's medium containing 20% fetal bovine serum, l-glutamine, and 100 ng/ml RANKL were added, and the Lipofectamine complex was allowed to remain on the cells for 16 additional hours. Using this method, transfection efficiencies of >95% were achieved, as previously reported (28). For immunocytochemical analysis, the cells were scraped and replated on ivory slices or glass coverslips immediately following the transfection. For RNA analysis, total cellular RNA was harvested 2 days post-transfection with RNA-Bee (Tel-test, Inc., Friendswood, TX). For protein analysis, whole cell lysates were harvested 1–4 days post transfection with M-PER.

Bone marrow-derived osteoclasts were transfected via electroporation as previously described (10, 28, 29). On day 4 of differentiation in culture, after osteoclasts had formed, cells were scraped, pelleted, and resuspended in siPORT buffer (Ambion, Austin, TX). The cells were electroporated at 250 V/50 μF with the siRNA solution (75 nm for siRNA 73578, 50 nm for siRNA 73762) or an equal concentration of negative control siRNA, then plated in standard differentiation medium on glass or ivory for immunocytochemistry, or plastic for RNA and protein analysis. As for RAW264.7-derived osteoclasts, transfection efficiencies were >95%.

Osteoclast Functional Assays

For measurements of bone resorptive capacity, 3 days post-transfection, osteoclasts were plated on BD BioCoat Osteologic Discs (BD Biosciences, San Jose, CA). These discs are composed of calcium phosphate mineral, which is capable of stimulating sealing zone formation in the absence of protein matrix (33). Control and siRNA-treated cells were kept on the discs for 3 days. The cells were removed by the addition of bleach for 5 min and several washes with water. Resorbed areas (clearings) were assayed by photographing the resulting discs under low magnification, and quantifying these areas with SigmaScan Pro 5.0 software (SPSS Science, Chicago, IL) as previously described (10, 28, 29). Equal numbers of images were compared among test groups. Groups were assayed for number of clearings, area per clearing, and total resorption.

Motility was measured by the use of 8.0-μm pore Transwell migration chambers (Corning Life Sciences, Acton, MA). The bottom side of the membrane was coated with collagen (3 mg/ml diluted 1:2 with 100% ethanol) and dried overnight. Immediately following the transfection, the cells were scraped and replated at a low density on the upper side of the chamber. After 4 days (when RNAi treatment caused Myo10 suppression), the cells were stimulated to migrate overnight by the addition of 40 μg/ml osteopontin peptide to the bottom of the well. Cells on the upper side of the membrane were removed with a cotton swab, and the remaining cells were fixed and stained for tartrate resistant acid phosphatase using a Leukocyte Acid Phosphatase kit (Sigma).

Microtubule Depolymerization

Osteoclasts 4 days post-transfection were placed at 4 °C for 3 h. Subsequently, the chilled medium was replaced with prewarmed (37 °C) medium for times indicated. Actin podosome belts (defined as a row of podosomes at the cell periphery) versus podosome clusters and podosome rings were counted in equal number of frames from three separate experiments.

Overexpression of Full-length and Myo10 Tail Fragments

Full-length bovine Myo10 (amino acid residues 1–2052) or various tail fragments were PCR-amplified from the plasmid GFP-M10 (34) and subcloned into the pEF6/V5-His expression vector (Invitrogen). The MyTH4 domain fragment consisted of amino acid residues 1541–1712, while the MyTH4-FERM domain fragment consisted of residues 1541–2052. Each were overexpressed in RAW264.7 cells and selected by 3 μg/ml of blasticidin. For each construct at least four individual clones were assessed and found to produce similar results.

RESULTS

Myo10 Is Absent from Mature Podosome Belts and Sealing Zones

Myo10 distribution was visualized by immunocytochemistry in RAW264.7- and mouse bone marrow-derived osteoclasts plated on glass coverslips or thinly cut ivory. Initially, we observed Myo10 distribution in osteoclasts that possessed mature adhesion structures. On glass, these are defined as peripheral belts of podosomes, a hallmark of fully mature osteoclasts (4). Myo10 demonstrated a generally diffuse distribution throughout the cytoplasm without specifically overlapping podosomes, but was somewhat enriched immediately adjacent to peripheral podosome belts, although this enrichment was not present in all cells. Fig. 1A illustrates this distribution in marrow-derived osteoclasts, while Fig. 1B shows a similar result in RAW264.7-derived cells. Arrows indicate the enrichment of Myo10 adjacent to podosome belts. In resorbing osteoclasts on ivory, Myo10 was present throughout the cell yet was absent from the mature sealing zone (Fig. 1C, single sections and z-stacks). The general exclusion of Myo10 from peripheral podosome belts and sealing zones suggest that it is not a direct component of mature osteoclast adhesion structures. However, given Myo10 ability to bind integrins as well as to mediate effects of PI3 kinase (a signaling molecule with known effects on osteoclast spreading and migration), we hypothesized that it might play a role in formation or positioning of immature podosome structures and sealing zones. Thus, its role in osteoclasts was explored further.

FIGURE 1.

Distribution of Myo10 in mature osteoclasts. A, mouse bone marrow-derived osteoclasts were plated on glass and examined by confocal microscopy for the localization of Myo10 (green) and F-actin (red). Arrows indicate regions of Myo10 enrichment near the podosome belt. Scale bars, 10 μm. B, images of RAW264.7-derived osteoclasts were prepared as in A. C, osteoclasts were plated on thinly cut ivory and examined by confocal microscopy to view Myo10 (green) in relation to the actin ring/sealing zone (F-actin, red). Myo10 is completely absent in the sealing zone in both RAW264.7 and MBM derived osteoclasts. Z-stack images of an MBM-derived osteoclast are shown to illustrate Myo10 distribution throughout the cell. Scale bars, 20 μm.

Myo10 Is Associated with Maturing Adhesion Structures

Peripheral belts of podosomes and circular sealing zones are not generated de novo; their formation occurs through ongoing processes of podosome or sealing zone patterning (4, 33). Fig. 2A illustrates schematically how podosome belts and sealing zones are formed from distinct precursor structures. Maturing osteoclasts, when cultured on a non-bone substrate such as glass, initially form clusters of podosomes that are dispersed through the cell interior. As the cells mature, these clusters become rings that are internal to the cell periphery, until finally the cell forms a peripheral podosome belt (4). The latter step from ring to belt is microtubule-dependent, because disruption of microtubules at this phase inhibits its progression. The formation of sealing zones in bone-attached cells is less well-defined; however, the nascent sealing zone appears to begin as an actin-rich patch, which then expands into a ring-like structure (33). However, sealing zone integrity also is dependent on microtubules, because their disruption causes sealing zone collapse (7, 8).

FIGURE 2.

Myo10 is positioned between microtubules and actin of forming podosome belts and sealing zone. A, schematic of podosome belt and sealing zone formation in cultured osteoclasts. B, a marrow-derived osteoclast on glass with an immature podosome ring was labeled for Myo10 (green), F-actin (red), and microtubules (blue). Myo10 localizes between F-actin and microtubules, as shown in the merged image. Scale bars, 10 μm. C, a RAW264.7-derived osteoclast on bone containing a nascent sealing zone (actin patch) was labeled for Myo10 (green) and F-actin (red). Myo10 is present in a ring at the perimeter of the actin patch. Scale bar, 10 μm. D, Myo10 or β-tubulin was immunoprecipitated from maturing RAW264.7 osteoclasts and the resulting precipitates were Western blotted for the same proteins. Under all conditions, co-precipitation of β-tubulin and Myo10 was noted. In the last lane, immunoprecipitations were performed with MOPC-21, an irrelevant mouse IgG control antibody.

We examined the distribution of endogenous Myo10 in relationship to microtubules and/or F-actin in cells that had yet to form mature podosome belts or sealing zones. Fig. 2B illustrates a cell cultured on glass, in the transition stage between a podosome ring and the mature belt. The individual and merged images demonstrate a band of Myo10 overlapping the outer edges of the podosome ring, with microtubules positioned outside but terminating near the Myo10 (arrows). This arrangement, in which Myo10 is sandwiched between podosomes and microtubules, was noted in all podosome rings in the process of transitioning to the belt phase. However, this arrangement of Myo10 as an intermediary between microfilaments and microtubules was not seen in mature osteoclasts containing podosome belts, suggesting that the positioning of Myo10 between podosomes and microtubules may be a transient event that occurs during podosome patterning and is no longer needed once stable podosome belts are formed. Additionally, we searched osteoclasts cultured on ivory to find nascent sealing zones in the actin patch stage. When cells were labeled for Myo10 and F-actin, we found that Myo10 surrounds the actin patch in a ringed distribution (Fig. 2C). Given Myo10 distribution in encircling podosome rings and nascent sealing zones, we hypothesize that this myosin may play a role in patterning of osteoclast adhesion structures.

Myo10 contains a MyTH4 domain and this domain has been shown to bind to microtubules (18). Our immunocytochemistry suggests that Myo10 may interact with microtubules in maturing osteoclasts. To examine this by biochemical means, we immunoprecipitated either Myo10 or β-tubulin and Western blotted these precipitates for their potential binding partner. In some cases, immunoprecipitations were performed in the presence of taxol or phalloidin to stabilize the microtubules and microfilaments, respectively. Fig. 2D demonstrates that under all conditions tested, Myo10 and β-tubulin were able to co-precipitate. These results, coupled with the immunochemistry shown in Fig. 2B, indicate that Myo10 may act as a linker between the microfilament and microtubule arms of the cytoskeleton in osteoclasts.

RNAi-mediated Myo10 Suppression Causes Decreased Sealing Zone Perimeter, Resorption, Cell Spreading, and Motility

To examine the function of Myo10 in osteoclasts, RNA interference was used to suppress Myo10 expression in both RAW264.7- and mouse bone marrow-derived osteoclasts. In Fig. 3A, targeting and negative control siRNAs were transfected on day 4 of osteoclast differentiation and assayed for Myo10 mRNA levels 2 days post-transfection. Three negative control double-stranded oligonucleotides failed to decrease the levels of Myo10 mRNA in RAW264.7 osteoclasts while both siRNA 73578 and 73762 decreased the mRNA expression by ∼90% (n = 3) when measured by competitive RT-PCR (Fig. 3A, top panel). Four days post-transfection in RAW264.7 osteoclasts, Myo10 protein levels were reduced to 62.0 + 4.5% (siRNA 73578) or 44.2 + 5.6% (siRNA 73782) of control levels (n = 3; Fig. 3A, middle panel). Similar efficiencies of knockdown were achieved in mouse marrow-derived osteoclasts (Fig. 3B). To further explore the timing of Myo10 suppression, control- or siRNA-treated cells were examined over a 4-day period. Fig. 3C shows an example of a Western blot in which the timing of suppression was assessed, whereas Fig. 3D presents a graphical compilation of multiple such experiments. As shown, modest levels of protein knockdown were achieved early post-transfection, but were substantial at day 4. This lag in suppression was suggestive of slow turnover kinetics of the Myo10 protein. However, previous reports have indicated a rapid turnover (2–2.5 h) for transfected Myo10 in HeLa cells (35, 36). To determine whether Myo10 demonstrates a greater stability in our cultures, pulse-chase analysis was performed on wild-type osteoclasts. Cells were pulse-labeled with [³⁵S]methionine/cysteine and Myo10 was immunoprecipitated from cell lysates after varying times in medium lacking radiolabel. As shown in Fig. 3E, Myo10 demonstrated great stability in osteoclasts, as even after 48 h of chase, 80% of the originally labeled Myo10 protein remained. Extrapolation from these data predicted a half-life of 4.7 days. This result is generally consistent with our finding that siRNA treatments resulted in ∼45–65% knockdown at 4 days post-transfection. Therefore, for all subsequent experiments with siRNA-treated osteoclasts, cells were examined on day 4 post-transfection when Myo10 protein levels were at their lowest.

FIGURE 3.

Suppression of Myo10 in osteoclasts. A, competitive RT-PCR and Western analysis of RAW264.7-derived osteoclasts. siRNAs 73762 and 73578 decreased Myo10 mRNA expression, whereas three negative controls (control, a generic non-targeting double-stranded RNA oligonucleotide; mut73762 and mut73578, mutant versions of 73762 and 73578) did not affect Myo10 expression. mRNA was measured 2 days post-transfection. Protein levels measured by Western analysis 4 days post transfection also demonstrate siRNA-specific loss of Myo10 while exhibiting no change in any of the controls. β-Actin levels did not change. B, competitive RT-PCR and Western analysis of marrow-derived osteoclasts. SiRNAs 73762 and 73578 elicited specific knockdown of Myo10, similar to that seen in RAW264.7-derived cells. C, Western analysis shows time course of Myo10 protein expression following transfection of siRNA 73578 into marrow-derived osteoclasts. The control used was a non-targeting RNA oligonucleotide. D, multiple experiments like those in panel C were quantified and graphed to show mean expression + S.D. relative to control transfectants (n = 3). E, wild-type osteoclasts were pulse labeled with [³⁵S]methionine/cysteine for 2 h, followed by chases of 0, 1, 2, 24, and 48 h. Myo10 was immunoprecipitated from samples adjusted to equivalent numbers of cpm. Results shown are the averages of 4 experiments + S.D.

siRNA-treated osteoclasts on ivory were examined by labeling with fluorescent phalloidin to determine changes in the sealing zone. In both RAW264.7 and mouse bone marrow-derived osteoclasts, the major change seen was a decrease in the perimeter of the sealing zone compared with controls. Fig. 4A, left, shows representative sealing zones from control- and siRNA-treated cells, each containing 5 nuclei and photographed at the same magnification. Sealing zone perimeters were measured in multiple cells and found to decrease by forty percent while no overall change in nuclear number was seen (Fig. 4A, graphs). Thus, suppression of Myo10 causes decreased sealing zone perimeter without affecting cell fusion. Although only one control is shown in this figure, subsequent experiments were performed comparing all three siRNA controls, which gave identical results (data not shown).

FIGURE 4.

Myo10 suppression leads to a decrease in cell and sealing zone spreading. A, 4 days post transfection, osteoclasts treated with a control non-targeting oligonucleotide or siRNA 73578 were fixed and stained with phalloidin. Confocal images demonstrate the decrease in sealing zone perimeter resulting from Myo10 suppression. Both control and siRNA-treated cells depicted have 5 nuclei. Scale bars, 10 μm. Graph represents mean relative sealing zone perimeter + S.D. or nuclear number + S.D.; n = 62 for RAW264.7 cells and n = 45 for mouse bone marrow (MBM) cells from three separate experiments. B, confocal images demonstrate the decrease in cell perimeter resulting from Myo10 suppression. Scale bar, 50 μm. siRNA treatment of RAW264.7 and mouse bone marrow-derived osteoclasts resulted in decreased perimeter 4 days post-transfection. Data points indicate mean relative perimeter + S.D. of at least 50 cells. C, osteoclast perimeter and nuclear number were quantified for control- and siRNA 73578-treated cells 4 days post-transfection. Bars represent mean relative perimeter or nuclear number + S.D.; n = 42 for RAW264.7 cells and n = 32 for MBM cells. D, resorptive capacity of control- versus siRNA 73578-treated cells was compared using synthetic bone substrate. Myo10 suppression decreased the number and size of clearings leading to a decrease in total resorption. Bars represent mean resorption + S.D. for three separate experiments. E, osteopontin-directed motility was measured in both RAW264.7 and MBM osteoclasts by Transwell migration assays. Myo10 suppression by siRNA 73578 resulted in loss of motility in both cell types. Bars represent mean relative motility + S.D. for three separate experiments in which at least 390 cells were assayed for each sample.

siRNA-treated osteoclasts on glass were labeled with fluorescent phalloidin to determine podosome structure and cell spreading. Whereas the siRNA-treated cells were capable of generating podosomes, they exhibited a readily noticeable difference in cell size when compared with controls. As shown in Fig. 4B, siRNA-treated cells possessed a cell perimeter similar to that of control cells over the first 3 days post-transfection, but were noticeably diminished in perimeter, to about 60% of controls, on day 4 when Myo10 protein expression was at its lowest (refer to Fig. 3D). These results show a direct correlation between levels of Myo10 protein expression and osteoclast perimeter. When average cell perimeter was measured on day 4, the resulting perimeters for both RAW264.7- and marrow-derived osteoclasts treated with siRNA 73578 were about sixty percent of the control (Fig. 4C). Although only one control is shown in this figure, subsequent experiments were performed comparing all three siRNA controls, which gave identical results (data not shown). Again, no differences in nuclear number were caused by Myo10 knockdown. This indicates that the changes in cell perimeter and sealing zone perimeter are not due to changes in the level of cell fusion. siRNA 73762 gave qualitatively similar results, as demonstrated in supplemental Fig. S2, A and B.

Because Myo10 knockdown caused changes in both sealing zone and cell spreading, we assessed functional changes by resorption and migration assays. First, osteoclasts treated with control or Myo10 siRNAs were plated on synthetic bone substrate for 3 days to allow for resorption. The clearing size and number of clearings along with total resorbed area was measured. We found that Myo10 siRNA-treated osteoclasts resorbed less in all measurements (Fig. 4D). In addition to defects in resorption, the motility of cells treated with siRNA 73578 was diminished when examined in Transwell migration assays. When stimulated to undergo directional migration by addition of osteopontin to the underside of Transwell filters, siRNA-treated osteoclasts (either RAW264.7- or marrow-derived) demonstrated levels of migration only 40% of that of controls (Fig. 4E). Similar results were achieved with siRNA 73762, as shown in supplemental Fig. S2C.

RNAi-mediated Myo10 Suppression Inhibits Microtubule-dependent Podosome Belt Formation

As described above, podosome belt positioning at the cell periphery in macrophages and osteoclasts is dependent on intact microtubules (4–6). When microtubules of mature osteoclasts are disrupted, the podosome belt collapses, but recovers by transitioning through the podosome cluster and ring stages before reforming the belt. Thus, disruption of microtubules recapitulates the formation of podosome belts seen in maturing osteoclasts. Because Myo10 interacts with microtubules in osteoclasts, we transiently depolymerized microtubules in these cells and assessed how Myo10 suppression affected podosome repositioning. Control- and siRNA-treated cells were placed at 4 °C for 3 h, which caused complete disruption of the microtubule network and a collapse of the podosome belt. The initial 3-h cold treatment caused all cells to be devoid of a microtubule network or a podosome belt in both the control- and siRNA-treated cells without detachment of the cells (Fig. 5A, top panels). After 24 h of recovery, microtubule networks were regenerated. Whereas control cells re-formed mature podosome belts, siRNA-treated cells organized their podosomes in internal clusters or small rings (Fig. 5A). This can be seen in a low magnification view of both control- and siRNA-treated cells after 24 h recovery (Fig. 5B). Almost 100% of the control cells recovered their peripheral podosome belts after 24 h, while less than 3% of siRNA-treated cells did so (Fig. 5B, graph). Thus, the defect in Myo10-suppressed cells appears to be in podosome patterning and not formation, as Myo10 siRNA-treated cells were capable of forming podosomes, but did not localize them correctly. Because siRNA-treated cells were capable of re-forming microtubules following their disruption but could not position podosomes in a microtubule-dependent manner, these results further suggest that Myo10 might act as a link between podosomes and microtubules. However, previous studies by others have shown that the accumulation of stable microtubules in mature osteoclasts is dependent on the small GTPase Rho (37). To determine whether Rho activity might be altered in Myo10-suppressed cells, we assayed levels of active Rho in control- or siRNA-treated cells. Rho activity was unaltered, as was activity of the related GTPase Rac (data not shown).

FIGURE 5.

Suppression of Myo10 leads to an inability to regenerate podosome belts after microtubule depolymerization. A, osteoclasts at 4 days post-transfection with non-targeting control or targeting siRNA 73578 were cold-treated to depolymerize microtubules, and allowed to recover for 0 or 24 h at 37 °C. Cells plated on glass were viewed by confocal microscopy after fixation and labeling for microtubules (red) and F-actin (green). Control cells recovered peripheral podosome belts 24 h following microtubule disruption, but siRNA-treated cells did not. Scale bars, 20 μm. B, low magnification confocal photomicrographs of cells recovered for 24 h demonstrate the differences in podosome distribution between control- and siRNA-treated cells. Scale bars, 30 μm. Graphical representation of the proportion of cells exhibiting a podosomal belt versus podosome clusters/rings. Bars represent the mean + S.D. of three experiments, in which at least 350 cells were assayed for each experiment.

Overexpression of MyTH4-containing Tail Fragments Inhibits Podosome Belt and Sealing Zone Formation

To delineate regions of Myo10 potentially involved in podosome patterning, and to support the emerging hypothesis that Myo10 links podosomes to microtubules during patterning of adhesion structures, we created plasmid constructs that contained various Myo10 tail domains alone or in combination in the pEF6/V5-His expression vector (Fig. 6A). These constructs, in addition to an empty vector for control, were transfected in RAW264.7 cells to generate stably transfected clones that overexpressed each domain. The cells were then differentiated into osteoclasts and labeled with phalloidin to visualize the F-actin podosome structures and general cell morphology. At least four clones for each construct were examined; all produced similar results. Previous studies have used overexpression of truncated Myo10 tail domains for dominant negative inhibition of Myo10 function (16, 18, 38). Because MyTH4 is the minimal microtubule binding domain of Myo10 and the combination of MyTH4-FERM is more efficient at binding microtubules than the MyTH4 domain alone, we examined the podosome phenotypes of control and MyTH4 or MyTH4-FERM overexpression cells on glass (15, 18). MyTH4- and MyTH4-FERM-overexpressing osteoclasts were unable to form podosome belts, but instead formed internal rings and clusters (Fig. 6B, top row). This effect was particularly notable in cells expressing MyTH4-FERM domains. To be sure this was not a differentiation-dependent patterning effect, microtubules in mature osteoclasts overexpressing MyTH4 or MyTH4-FERM were depolymerized by cold treatment, and cells were allowed to recover overnight. MyTH4-expressing osteoclasts, like siRNA-treated cells, could not recover podosome belts and were inhibited from transitioning out of the internal ring stage. MyTH4-FERM-expressing osteoclasts were even more affected and could generate only very small podosome rings (Fig. 6B, middle row). Further, these cells made poor sealing zones when cultured on ivory. While MyTH4-expressing cells were capable of making small sealing zones, MyTH4-FERM-expressing cells primarily formed actin patches, reminiscent of those preceding sealing zone formation (Fig. 6B, bottom row).

FIGURE 6.

Overexpression of the MyTH4 domain suppresses podosome belt and sealing zone formation. A, diagram of Myo10 tail domain clones used in this figure. Each was stably overexpressed in multiple RAW264.7 macrophage clones and differentiated into osteoclasts. B, osteoclasts overexpressing the MyTH4 and MyTH4-FERM domains were labeled with fluorescent phalloidin to visualize F-actin, revealing that dominant negative overexpression of Myo10 tail domains caused the cells to be unable to form podosome belts under normal growth conditions (top row) or after microtubule depolymerization (middle row). Further, these cells formed only small sealing zones or actin patches on ivory (bottom row). Scale bars, 20 μm.

Overexpression of Full-length Myo10 Leads to Increased Podosome Belt and Sealing Zone Formation

To reaffirm the role of Myo10 in podosome positioning and formation of the sealing zone, full-length Myo10 was cloned into the pEF6/V5-His expression vector and stably overexpressed in RAW264.7 cells. The empty vector was used as a control for these studies. We noted that RAW264.7 cells could not readily tolerate long-term overexpression of the full-length myosin. Of three experiments where 12 individual clones were chosen for expansion each time, only two clones (FLX1 and FLX2) survived to produce enough cells for further study. By Western, the Myo10 band intensities of FLX1 and FLX2 increased to 140 + 1.3% and 199 + 3.5% of control cells, respectively (data not shown). The pEF/V5-His vector uses the human elongation factor-1α promoter to keep overexpression levels low compared with more commonly used viral promoters, but even under these conditions, RAW264.7 cells apparently tolerate long-term overexpression of Myo10 poorly.

The resulting clones were plated on glass and the F-actin phenotype was examined by labeling with fluorescent phalloidin. Initially, undifferentiated RAW264.7 macrophages were examined (Fig. 7A, top panels). Whereas control transfectants exhibited normal F-actin distribution, clones FLX1 and FLX2 generated distinct podosome belts, a phenotype normally present only in osteoclasts. When the clones were differentiated into osteoclasts, control cells generated typical podosome belts, but the podosomes of FLX1 and FLX2 were arranged in unusual conformations (Fig. 7A, bottom panels). Myo10-overexpressing cells produced podosomes in multiple ring and belt structures. Higher magnification of one of these cells demonstrates the appearance of intensely labeled belts with multiple internal rows of podosomes (Fig. 7B). Perimeter and nuclear number was measured for these cells and both were found increase when Myo10 was overexpressed (Fig. 7B, graph). When cells overexpressing Myo10 were cultured on ivory and stained with fluorescent phalloidin to visualize the sealing zone, sealing zone perimeter was increased (Fig. 7C; note differences in scale of photos and associated graph), leading to cells with greater resorptive capacity (Fig. 7D). Thus, while suppression of Myo10 function by either RNA interference or dominant negative inhibition by the MyTH4 domain inhibited podosome belt formation and sealing zone spreading, overexpression of the full-length Myo10 forced an increase in peripheral and internal podosome patterning and sealing zone spreading. These results and those previously described clearly indicate a role for Myo10 in regulating adhesion structure patterning through a microtubule-dependent process.

FIGURE 7.

Overexpression of full-length Myo10 promotes formation of adhesion structures. A, full-length Myo10 was stably overexpressed in RAW264.7 macrophages. FLX1 and FLX2 refer to individual clonal lines transfected with the full-length Myo10 while control cells were transfected with empty vector. Overexpression of full-length Myo10 promoted podosome belt formation in macrophages as indicated by fluorescent phalloidin labeling (top panels). In osteoclasts differentiated from the transfected clonal lines, overexpression of Myo10 induced excessive formation of podosome belts and rings (bottom panels). Scale bars, 50 μm. B, (left) a high magnification view of a single mature osteoclast overexpressing Myo10 illustrates excessive podosome belt and ring formation (left panel). Scale bar, 20 μm (right). Cells overexpressing Myo10 show greater cell size and nuclear number than controls. Bars represent mean relative cell perimeter and nuclear number + S.D. C, (left) Myo10 overexpressing cells were plated on thinly cut ivory, and F-actin was visualized to demonstrate the relative sizes of sealing zones. Scale bars, 10 μm, (right) quantification of sealing zone perimeter represents mean relative + S.D. D, Myo10-overexpressing cells were plated on synthetic bone substrate for 3 days, and resorptive capacity was measured. Quantification represents mean of three experiments + S.D.

DISCUSSION

The relatively recently described class X myosin was first characterized as the founding member of a group of myosins containing MyTH4, FERM, and multiple pleckstrin homology domains (11). In other cell types, Myo10 has been shown to be involved in producing protrusions at the cell membrane as it plays a role in filopodia extensions (35, 38) and axonal guidance (39) and is found to localize in lamellipodia (11) and phagocytic cups (16). Podosomes and sealing zones are actin-based structures that interact with, and are dependent on, intact microtubules for formation and function. Whereas a few proteins have been proposed to link the podosome and microtubule networks, Myo10 is the first potential linker known to directly bind both actin and microtubules. The data presented here suggest a model in which a ring of Myo10 encircles immature podosome rings or nascent sealing zones, and through interaction with microtubules, provides force to drive expansion of these adhesion structures into peripheral podosome belts and mature (i.e. fully expanded) circular sealing zones. As demonstrated above, loss of Myo10 function through siRNA-mediated knockdown resulted in smaller sealing zones and poor cellular spreading; these defects not surprisingly negatively influenced cell capacities for motility and bone resorption. Further, overexpression of dominant negative tail fragments carrying the MyTH4 domain also inhibited podosome belt and sealing zone formation. This finding, along with the failure to properly pattern podosomes after microtubule depolymerization, suggests that Myo10 does indeed act as a linker between podosomes and microtubules. We also demonstrated that endogenous Myo10 has very slow turnover kinetics in osteoclasts, with a predicted half-life in excess of 4 days. This is in contrast to findings in HeLa cells, in which transfected GFP-Myo10 was shown to possess a half-life of 2–2.5 h (36). This finding illustrates that regulation of Myo10 stability can be markedly different in different situations. It further suggests that whereas Myo10 may play transitory roles in some cell types, its ongoing presence is required in osteoclasts for continuous formation and rearrangement of the bone resorptive apparatus.

Previous studies have suggested roles for other proteins in mediating linkage between microtubules and podosomes. Linder et al. (6) suggested a role for the actin-binding protein WASP in mediating podosome to microtubule interaction through a second intermediate, CIP4 (CDC42-interacting protein 4). The WASP polyproline domain that binds CIP4, a microtubule binding protein, was found to block podosome regeneration without affecting the microtubule network when injected into macrophages (6). Perhaps relevant to these findings, Myo10 has been proposed to interact with and transport VASP, a WASP related protein that shares similar structural motifs such as the polyproline domain, to the tip of filopodia (17). A second protein, kinesin KIF1C, also has been proposed as the microtubule-podosome link (41). Again, it was suggested that KIF1C binds microtubules and a second intermediate binds actin. Immunoprecipitation of KIF1C from HUVECs resulted in co-precipitation of nonmuscle myosin IIA, which was purported to directly interact with KIF1C, although conditions by which these experiments were performed may have resulted in precipitation of indirectly bound actin-associated proteins. Although Kopp et al. (10) found that pharmacological inhibition of nonmuscle myosin IIA in macrophages via blebbistatin (which also inhibits macrophage myosin IIB) decreases podosome formation, our own studies in osteoclasts demonstrated that specific inhibition of myoIIA via RNA interference did not affect podosome formation. Nonetheless, multiple linker proteins may be responsible for microtubule regulation of podosome/sealing zone function, and the results here indicate that Myo10 may mediate this connection via its microtubule-binding MyTH4 tail domain specifically in transitions to mature podosome belts and sealing zones. Unfortunately, we were unable to utilize video live cell imaging to follow Myo10 trafficking in podosome positioning/sealing zone formation. As described in Fig. 7, osteoclasts tolerate overexpression of Myo10 very poorly, and the obligatory low levels of fluorescently tagged myosin that we were able to achieve did not allow for proper visualization.

Myo10 may be playing additional roles in osteoclasts beyond podosome/sealing zone patterning. Although recent evidence suggests many PH domains are utilized for protein-protein interaction, the Myo10 PH also domains have been shown to bind PI3 kinase products (42). GFP-tagged Myo10 PH domains bind the plasma membrane of mouse myoblasts (43), and Myo10 is recruited to the phagocytic cups of macrophages in a PI3-kinase-dependent manner (16). Preliminary experiments using Myo10 PH domains as a dominant negative suggest a role in podosome formation (not shown). Myo10 also has been shown to bind beta integrins via the FERM domain (19). Both PI3-kinase signal transduction and integrin-mediated signaling and attachment play roles in proper osteoclast function (22, 44, 45). Studies are on-going to better examine potential relationships between these activities and Myo10 function. Additionally, a recent study demonstrated that the bone morphogenetic protein BMP6 induces Myo10 expression in endothelial cells and Myo10 is required to guide migration of these cells toward BMP6 gradients (46). Although the effects of BMPs on osteoclast activity is poorly understood, a number of studies have suggested a role for these proteins in promoting osteoclast recruitment and activity (40, 47–50). It will be of interest to determine if BMPs play a role similar to that in endothelial cells by regulating osteoclastic Myo10 expression, motility, and potentially resorption.

In summary, these studies reveal several novel aspects of Myo10 function in mammalian cells. While previous reports demonstrated that this motor protein linked actin- and microtubule networks during spindle formation (18, 20, 21), this study illustrates that this linkage is of import to mammalian cells in regulating adhesion and motility. The presence of MyTH4 domains in actin-based motor proteins is particularly gratifying in that it suggests that these myosins act as direct linkers between the cytoskeletal networks without a requirement for other protein intermediates. Given the presence of microtubule binding domains in Myo10, it is tempting to speculate that this motor protein may serve to link actin and microtubules for a number of cellular processes. The current study demonstrates a central role for Myo10 in positioning of adhesion complexes, at least in those cells like osteoclasts, where adhesion is mediated by podosome-based structures. Of continued interest will be the functions of the FERM and PH domains of Myo10 in osteoclast activity, as well as Myo10 overall role in potential regulation of the bone resorptive process through PI3K, integrin, BMP, and binding partners yet to be determined.