Osteoclasts (OCs) are terminally differentiated, multinucleated cells responsible for the physiological and pathological destruction of bone.1, 2 Increased numbers and/or activation of OCs underscore several debilitating osteolytic diseases, including osteoporosis, osteoarthritis, Paget's disease and tumor-mediated bone loss. Formed by the fusion of mononuclear myeloid-lineage cells under the aegis of RANKL and M-CSF, these specialized polykaryons reside on bone surfaces and promote skeletal remodeling and turnover throughout life. Upon contact with bone, OCs rapidly reorganize their cytoskeleton and membrane architecture to reflect their functionally polarized status. This rearrangement is accompanied by the segregation of the OC plasmalemma into several functionally distinct membrane domains, of which the bone-apposed ruffled border (RB) membrane is most characteristic and represents the “resorptive apparatus” of the cell. The RB, circumscribed by a tight ring of filamentous actin (ie, the sealing zone, SZ), serves as an exocytotic release site for protons and osteolytic enzymes (largely cathepsin K) required to degrade both the mineral and organic phases of bone.3–5 In addition, the RB facilitates the synchronous ingestion and removal of degraded bone matrix from the underlying resorptive lacunae by endocytosis, whereby it is expelled into the extracellular milieu at the basolateral functional secretory domain.6, 7 To maintain its exquisite structural and functional organization, the RB must therefore be furnished with high levels of vesicular trafficking, with each transport pathway being tightly coupled to the underlying cytoskeleton to ensure a delicate balance in membrane flow and directionality during the resorptive process.8, 9 Whereas recent inroads have been made toward establishing the nature of these vesiculotubular transport routes and their associated regulatory machinery,10–20 comparatively few molecular insights have been provided to explain how these transport vesicles are intimately linked to the osteoclastic cytoskeleton during bone resorption.
In other polarized cell types (eg, epithelial cells), transport of intracellular cargo along the microtubule and actin cytoskeleton is powered by mechanochemical motor proteins of the AAA+ (ATPases associated with diverse cellular activities) superfamily, which utilize energy liberated from the hydrolysis of ATP. Local actin-based movements (occurring in the cortical regions of cells) are driven by myosin family members, whereas long-range microtubule-mediated motility is regulated by the “tug-of-war” activities between kinesins (plus-end) and cytoplasmic dynein (minus-end).21, 22 Recently, we have shown that two molecular adaptors of the dynein motor complex, Tctex-1 and LIS1, regulate vesicular trafficking during osteoclastic bone resorption.13, 14 Targeted suppression of Tctex-1 disrupts Rab3D-mediated secretory vesicle trafficking in OCs and impairs bone resorption in vitro.13 Similarly, depletion of LIS1 coincides with a loss of microtubule stability, reduced cathepsin K secretion, and greatly diminished bone resorptive capacity.13 Whereas these studies imply an important role for cytoplasmic dynein in bone resorptive function, the precise involvement of this macromolecular motor complex in OC formation and function has yet to be addressed. To further dissect the role of the dynein-dynactin complex in OCs, we have systemically profiled the expression, localization, and function of osteoclastic dynein-dynactin complex during RANKL-driven OC formation and bone resorption. Herein, we show, for the first time, that the integrity of the dynein-dynactin complex is not only critical to the timely formation of OCs but also their functional activity, serving to maintain the spatial and temporal control of key intracellular organelles, including the Golgi and cathepsin K–containing lysosomes, each essential to the bone resorptive process.
Materials and Methods
Antibodies and reagents
Rabbit polyclonal antibodies against Dynein HC (R-325), Kinesin LC1, and Dynein LC (H-60) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). Mouse monoclonal antibodies against GM130, p150Glued and dynactin p50 were purchased from BD Biosciences (Erembodegem, Belgium). Mouse and rabbit monoclonal antibodies against phospho-histone H3 (Ser 10), IκBα, phospho-ERK1/2, phospho-p38, total p38, phospho-p65, and total p65 were purchased from Cell Signaling, Technology, Inc., Danvers, MA, USA. Other commercial antibody sources include mouse monoclonal anti-dynein intermediate chain (74.1) (Chemicon, Boronia, VIC, Australia), mouse anti-chicken α-tubulin (Sigma, St. Louis, MO, USA), mouse monoclonal anti–cathepsin K (Millipore, Billerica, MA, USA), mouse monoclonal ERK1/2 (V114a) (Promega, Madison, WI, USA), and mouse monoclonal β-actin (JLA20) (Calbiochem, La Jolla, CA, USA). Mouse monoclonal anti-c-Src antibody was a gift from Dr W Langdon (University of Western Australia, Australia), rabbit polyclonal V-GLUT1 was kindly provided by Dr R Jahn (Max Planck Institute for Biophysical Chemistry, Germany), and a rabbit polyclonal antibody against CLIP-17023 was kindly provided by N. Galjart (Erasmus MC, The Netherlands).
Alexa Fluor 488-, 546-, 647- and horseradish peroxidase (HRP)-conjugated goat anti-mouse immunoglobulin G (IgG), and goat anti-rabbit IgG were all purchased from Invitrogen, Carlsbad, CA, USA. Rhodamine- and Alexa Fluor 647-conjugated phalloidin were obtained from Invitrogen. LysoTracker Red DND-99 and Hoechst 33258 (bis-benzimide) were purchased from Molecular Probes (Eugene, OR, USA). A cathepsin K Detection Kit was purchased from Calbiochem. The complete cDNA encoding human p50 (dynamitin)-EGFP construct was a kind gift from RB Vallee (University of Massachusetts Graduate School of Biomedical Sciences, USA). GST-RANKL160–318 (rRANKL) recombinant proteins were expressed and purified in our laboratory as previously described.24 Cell culture media was purchased from Gibco (Carlsbad, CA, USA), whereas fetal bovine serum and penicillin-streptomycin were purchased from Invitrogen.
Macrophage and OC culture
Bone marrow macrophages (BMMs) were isolated as previously described.14 Briefly, total bone marrow was extracted from tibias and femora of 6 to 8-week-old C57BL/6 mice (purchased from the Animal Resource Centre, Murdoch, WA, Australia) in accordance to the UWA Institutional Animal Ethics Committee Guidelines. Mouse monocytic RAW264.7 cells were grown as per American Type Culture Collection guidelines (ATCC, Manassas, VA, USA) in α-modified MEM (α-MEM) supplemented with 10% fetal calf serum (FCS). All cell cultures were maintained in 5% CO2 at 37°C. OCs were generated from either RAW264.7 or BMMs cells differentiated with RANKL (100 ng/mL) alone or in combination with M-CSF (25 ng/mL) in vitro as previously described.20 Mature primary human OCs were sourced from giant cell tumor of bone (GCT) tissue according to previous protocols.14 GCT was collected fresh from patients postoperatively (Sir Charles Gairdner Hospital, Nedlands, WA, Australia). All patients consented and experiments were approved by a Human Ethics Committee at UWA. Cells were harvested and processed for total RNA extraction, velocity density gradient centrifugation, immunoblot analysis, or seeded onto either glass coverslips, Osteologic Discs (BD Biosciences), devitalized cortical bovine bone, or whale dentine surfaces before being fixed and processed for immunofluorescence microscopy and scanning electron microscopy. Only cells that stained positive for TRAP activity and contained three or more nuclei were scored as mature OCs.
Reverse transcription (RT)-PCR
Total RNA was isolated from cultured cells using RNeasy Mini Kit (QIAGEN, Dusseldorf, Germany) in accordance with the manufacturers' protocol. For RT-PCR, cDNA was prepared from 2 µg of total RNA, using reverse transcriptase with an oligo-dT primer. All PCR was carried out using 2 µl of each cDNA, using cycling parameters 94°C, 45 seconds; 54°C, 45 seconds; and 72°C, 45 seconds for 30 cycles (except the calcitonin receptor (CTR) whose annealing temperature was 60°C), with primers designed against the following mouse sequences: DHC (forward: 5′-GGTGGGAGTGCATTACGAGT-3′; reverse: 5′- TGTTGCTGAATGGCAGAGTC-3′), DIC (forward: 5′-CGGAGGAAGAAAAACAGCAG-3′; reverse: 5′-CCCGATGCTTAGACCAATGT-3′), DLIC (forward: 5′- TCTCACATCCGCAAGTTCTG-3′; reverse: 5′-CTGGCAACATTGGATGACAC-3′), DLC (forward: 5′-ATGTGCGACCGGAAGG-3; reverse: 5′- TTAACCAGATTTGAACAGAAGAATG-3′, p150Glued (forward: 5′-AGATGGTGGAGATGCTGACC-3′; reverse: 5′-GAGCCTTGGTCTCAGCAAAC-3′), p50 dynamitin (forward: 5′-GGACGCAGACACTCAGAACA-3′; reverse: 5′-CTGCTGGGTGGTATCCAAGT-3′), CTR (forward: 5′-TGGTTGAGGTTGTGCCCA-3′; reverse: 5′-CTCGTGGGTTTGCCTCATC-3′), and acidic ribosomal phosphoprotein (36B4) (forward: 5′-TCATTGTGGGAGCAGACA-3′; reverse: 5′-TCCTCCGACTCTTCCTTT-3′), which served as an internal loading control. PCR samples were analyzed by agarose gel electrophoresis and quantified using ImageJ software (NIH). For the detection of DIC splice variants, primers designed against the following mouse sequences were used: Dyn 2 Ex1a (forward: 5′-GGCCGGTGTATCTGTTTCAAC-3′; reverse: 5′-GAGGGACCCAGTACTCAG-3′) and Dyn2 Ex1b (forward: 5′-CAGTTGGAGAGGGACGTTC-3′; reverse: 5′-CTCGAGGGGGAAAGTCAAC-3′).
Cultured cells were washed with ice-cold PBS and lysed in 1× RIPA buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 1% sodium deoxycholate) supplemented with protease inhibitors: 1 mM phenylmethylsufonyl fluoride (PMSF) and Complete Mini EDTA-free protease inhibitor cocktail (Roche, Indianapolis, IN, USA). After incubation on ice (30 minutes), cell lysates were cleared by centrifugation at 15,000 rpm for 15 minutes at 4°C. For immunoblotting, 10 to 20 µg of protein were resolved by SDS-PAGE (10% to 15% polyacrylamide gels) and then transferred overnight onto nitrocellulose membranes (Hybond ECL; Amersham Life Science, Arlington Heights, IL, USA). Membranes were blocked with 5% skim milk in Tris-buffered saline–Tween 20 (TBS-T) for 60 minutes, and then probed with primary antibodies diluted in 1% (w/v) skim milk powder in TBS-T overnight at 4°C. Membranes were washed and then incubated with HRP-conjugated secondary antibodies and developed using the Western Lightening Ultra Detection Kit (Perkin Elmer, Melbourne, VIC, Australia), in combination with the FujiFilm LAS-3000 Gel Documentation System (FujiFilm, GE Healthcare, Rydalmere, NSW, Australia).
Velocity density gradient centrifugation
Velocity density gradient centrifugation was performed as previously described.14 Briefly, ∼107 osteoclastic cells were washed with ice-cold PBS, and scraped with 500 µl of lysis buffer (50 mM HEPES-KOH, pH 7.2, 150 mM NaCl, and 1 mM MgCl2) supplemented with protease inhibitors: 1 mM phenylmethylsufonyl fluoride (PMSF), 1 mg/mL aprotinin, 1 mg/mL leupeptin, and 0.7 mg/mL pepstatin (all from Sigma), and then homogenized by passing cells through a 25-gauge syringe (at least 10 strokes). Proteins were then cleared by centrifugation at 3000 rpm for 15 minutes at 4°C. Five hundred microliters of the clarified supernatant was layered on top of 12 mL of 5% to 20% linear sucrose density gradient prepared in the lysis buffer. After centrifugation at 150,000 g for 18 h in a SW40 rotor (Beckman Coulter, Gladesville, NSW, Australia), 1 mL fractions were collected and equal volumes analyzed by immunoblotting using antibodies to DHC, p150Glued, IC74, DLC, KLC1, α-tubulin, p50, and CLIP-170. Immuno-intensities were quantified by densitometry using ImageJ (National Institutes of Health, Bethesda, MD, USA).
Retrovirus mediated transduction of BMMs
Retroviral vectors, pMX-EGFP and pMX-p50-EGFP were constructed by subcloning EGFP or p50-EGFP cDNAs into pMX-puro vector (kindly provided by Dr. Kitamura, University of Tokyo, Tokyo, Japan). Retrovirus packaging was performed by transfection of the pMX vectors into the packaging cell line Platinum-E (Plat-E).25 Preparation of viral stocks and viral transduction was performed according to methods previously described.26 After transduction and puromycin selection, BMMs were differentiated into OCs in the presence of RANKL (100 ng/mL) and M-CSF (25 ng/mL).
Immunofluorescence confocal microscopy
Immunofluorescence was performed as previous described.14 In brief, cells grown on glass coverslips, or devitalized bovine bone discs were fixed with either 4% paraformaldehyde (PFA) in PBS for 15 minutes at room temperature (RT) and permeabilized for 5 minutes with 0.1% Triton X-100 in PBS or with methanol (−20°C) for 10 minutes. In some instances cells were preincubated with microtubule stabilizing buffer (MTSB) (80 mM PIPES pH 6.8, 1 mM MgCl2, 4 mM EGTA) supplemented with 0.5% Triton X-100 for 30 seconds at RT before methanol fixation to release free cytosolic dynein. Nonspecific binding of antibodies was blocked by 3% FCS in PBS for 30 minutes, after which cells were incubated with primary antibody diluted in 0.2% BSA-PBS for 2 h at RT. After extensive washing, bound primary antibodies were visualized with Alexa Fluor secondary antibody-conjugates (Invitrogen, Grand Island, NY, USA). F-actin was stained with Rhodamine- or Alexa Fluor 488/647-conjugated phalloidin and nuclei visualized by Hoechst 33258 dye (Molecular Probes). Samples were mounted with ProLong Gold anti-fade mounting medium (Invitrogen) and imaging performed using a NIKON A1Si confocal microscope (Nikon, Melville, NY, USA) equipped with 20× (dry) and 60× (oil) immersion lenses. Images were collected using the systems NIS-C Elements software (Nikon). Quantitation of co-localization correlation coefficients (Pearson's, Rr = 0–1) was performed using the Intensity Correlation Analysis ImageJ macro (NIH). This plug-in generates an image, scatterplot, and correlation coefficient of co-localized pixels between two 8-bit images (or stacks). The plugin merges selected 8-bit channels (eg, red and green) to a pseudocolored image which highlights the co-localized pixels. Pixels are considered co-localized if their intensities are higher than the threshold of their channels (set at 50 by default), and if the ratio of their intensity is higher than the ratio setting value (set at 50% by default). Positive correlation was considered when Rr > 0.8. Quantitation was performed on at least 10 to 20 cells from two independent experiments. Quantitation of p50-GFP expression and cathepsin K immuno-intensity profiles were performed using a custom-written Linescan Intensity Profiler (MATLAB, The Mathworks, Inc., Natick, MA, USA) as previously described.27 p50-GFP expression levels in OCs were defined as either low (ie, relative GFP intensity = 100 to 150 AU), moderate (150 to 200 AU), or high (>200 AU), respectively. Golgi ribbon integrity was assessed using Particle Analysis macros in ImageJ (NIH). Cathepsin K (CTSK) secretion was quantified by measuring the number of OCs containing CTSK within F-actin rings, compared with the total number of active (F-actin ring positive) OCs. At least 50 to 100 OCs were analyzed per bone slice in each group (n = 6).
Time-lapse confocal microscopy
For the live cell imaging, OCs expressing either GFP or p50-GFP were grown on glass-bottom culture dishes (35 mm petri dish, 10 mm microwell with no. 1.5 coverglass) (MatTek Corporation, Ashland, MA, USA) and incubated with LysoTracker Red DND-99 (100 nm, Invitrogen) or cathepsin K fluorogenic substrate (Calbiochem) according to manufacturers' instructions to visualize late endosomes and cathepsin K containing lysosomes, respectively. Cells were imaged by time-lapse fluorescence confocal microscopy using an inverted 40× (water) objective lens under controlled atmospheric conditions (37°C and 5% CO2/Air) in a Tokai Hit Stage Top incubator (INUG2E-TIZ, Fujinomiya-shi, Shizuoka-ken, Japan). Frames were captured every 5 seconds for 5 minutes using NIS-C Elements software (Nikon). Image stacks and cumulative vesicle trajectories were analyzed using the ImageJ MTrackJ plug-in (NIH).
Cell proliferation assay
The Promega CellTiter 96 AQueous MTS cell proliferation assay (cat # G5421) was used according to manufacturers' protocol (Promega). Briefly, 20 µl of MTS-PMS reagent was added to cells expressing p50-GFP or GFP alone (100 µl of media) and incubated for 4 h at 37°C. Absorbance at 490 nm was quantified by spectrophotometry.
Bone resorption assays and medium CTx measurement
BMM-derived OCs transduced with either GFP (control) or p50-GFP were seeded directly onto devitalized dentine discs to assess resorptive activity. After 48 h, cells were fixed with 4% PFA and either stained with rhodamine-conjugated phalloidin to visualize F-actin rings or stained for TRAP activity (387-A, Sigma). The number of TRAP + ve multinucleated OCs were scored before assessment of resorptive activity. After the removal of OCs using a soft brush, resorption pits were visualized by scanning electron microscopy and reflective confocal microscopy. The total resorbed area/dentine slice and depth of resorptive pits were quantified using ImageJ (NIH) software. Medium crosslink peptide sequence of collagen type 1 (CTx-1) levels were determined using CrossLaps for Culture ELISA kit (Immunodiagnostic Systems, Scottsdale, AZ, USA) according to the manufacturer's instruction.
Statistics and data presentation
Results were statistically analyzed using a two-tailed Student's t test using Microsoft Excel (Microsoft Corp., Redmond, WA, USA). All data shown are representative of at least three independent experiments and expressed as mean ± SD.
Expression and subcellular localization of the dynein-dynactin complex during RANKL-induced OC differentiation
Cytoplasmic dynein is a large multisubunit complex (∼1.6 MDa) composed of two heavy chains (DHC, 530 kDa) that contain the motor and ATP hydrolytic sites, several copies of an intermediate chain (IC, 74 kDa), light intermediate chains (LICs, 50 to 60 kDa), and several light chains (LCs, ∼22 kDa)28 (Supplemental Fig. 1). Attachment of the dynein motor to specific cargo is facilitated by its multifunctional adaptor dynactin, a multimeric complex consisting of a rod shaped domain (Arp-1 filament and actin capping proteins) and a projecting arm (p150Glued), which are linked by an association with the p50 subunit dynamitin.29, 30 To investigate the role of this macromolecular machine in OCs, we first examined the mRNA expression profiles of selected subunits of the dynein-dynactin complex during M-CSF/RANKL-driven OC differentiation. For this purpose, mouse BMMs were cultured under pro-osteoclastogenic conditions (25 ng/mL M-CSF, 100 ng/mL RANKL) for up to 7 days, upon which multinucleated cells formed and expressed the OC maturation marker calcitonin receptor (CTR) by RT-PCR (Fig. 1A). RT-PCR analyses employing primer sets specific for the indicated conventional mouse dynein and dynactin subunits confirmed that all major dynein (DHC, DLIC, and DLC) and dynactin (dynamitin, p150Glued) subunits were expressed in OCs. Whereas in some instances we observed variability in expression levels during distinct stages of RANKL-mediated differentiation (eg, DHC), these differences were not statistically significant when comparing mRNA expression levels between unstimulated BMMs (day 0) and fully mature OCs (day 7). To complement these analyses we also screened for OC-specific isoforms of the dynein IC, which has recently been shown to determine the structural and functional organization of a dynein complex for a given cell type.31 As shown in Fig. 1B, whereas the IC2A is the primary splice variant of the brain dynein complex (positive control), IC2C is the major, if not sole, splice variant of the OC dynein complex (Fig. 1B), indicating that the organization of the dynein-dynactin complex in OCs is distinct from that in brain.
Next, to gain insights into the molecular associations that exist between dynein and dynactin in OCs we compared the co-sedimentation profiles of dynein-dynactin subunits with microtubules (α-tubulin) and microtubule plus-end associated proteins, namely the light chain of the kinesin motor (kinesin light chain, KLC), and the CAP-Gly domain-containing microtubule plus-end associated protein (CLIP-170) by biochemical subcellular fractionation assays (Fig. 1C–D). Immunoblot analysis showed peak expression and co-migration of dynein (HC, IC, LC) and dynactin (p50/dynamitin, p150Glued) subunits in sucrose fraction(s) (FR) 10 to 12, which overlapped with a fraction of microtubules (α-tubulin; FR 9to 12) implying that they all form a common intracellular complex in OCs. In contrast, the peak migration of kinesin (KLC) and CLIP-170, as well as the bulk of α-tubulin, were detected in lower density fractions (ie, KLC FR6-8; CLIP-170 FR4-6; α-tubulin FR4 to 8) consistent with their role as plus-end microtubule proteins. Together, these data are consistent with the position that the minus-end dynein-dynactin motor exists as a distinct molecular complex from the plus-end motor kinesin on microtubules in OCs.
OC formation requires proliferation and differentiation of committed monocytes into OC-progenitor cells before the onset of cell fusion.1 To gauge the potential function of dynein-dynactin in osteoclastogenesis, we next monitored the subcellular distribution(s) of the dynein-dynactin complex during distinct phases of RANKL-induced OC differentiation (Fig. 1E). In this instance, we employed RANKL-stimulated RAW 264.7 cells as an OC model to circumvent the need for M-CSF, which might mask potential RANKL-specific changes in dynein-dynactin localization during OC differentiation. As shown in Fig. 1E, dynein (DLC, green) and dynactin (p150Glued, red) co-localized (yellow) to a discrete perinuclear spot (likely the centromere, arrow) in unstimulated OC progenitor cells (day 0), consistent with its reported localization patterns in other cells types.32 By comparison, dynein relocalized to mitotic spindles and kinetochores (as revealed by costaining for α-tubulin, arrows) in proliferating OC precursor cells (day-1 post-RANKL treatment), hinting that this dynein might participate in mitosis during the early phases of OC differentiation. The subcellular distribution of dynein and dynactin also changed markedly during the onset of cell fusion (day 3). In this instance, dynein and dynactin co-localized on small vesicular-like punctae within nanotubules of precursor cells that were in the process of fusing with larger multinucleated pre-OCs (inset, arrows), probably reflecting minus-end mediated delivery of cargo between fusing mononuclear cells with larger syncytia. After the maturation phase (day-5), dynein and dynactin accumulated within the perinuclear region of OCs and lined filamentous structures reminiscent of microtubules. Taken together, these data demonstrate that the OC dynein-dynactin complex is recruited to discrete intracellular sites during the proliferation, differentiation, and fusion of OC progenitors into mature multinucleated cells, implying specific functions for this motor complex at defined stages of OC formation.
The RB is a plus-end microtubule domain on which the dynein-dynactin complex coexists during bone resorption
Next, we examined the interplay between dynein-dynactin and the OC cytoskeleton during the onset of polarization and bone resorption. For this, we used primary human osteoclastic cells derived from giant cell tumor (GCT), as they offer unique advantages over rodent cells when defining dynein localization patterns on bone ie: (1) they rapidly and uniformly resorb bone; (2) they exhibit greater affinity to dynein antibodies raised against human peptides; (3) they are sufficiently large to enable fine subcellular structures to be readily resolved by confocal microscopy; and (5) they directly reflect bona fide human cells.14 Human OCs were cultured under both non-polarizing (glass coverslips) and highly polarizing/bone-resorbing conditions (bovine bone discs) for up to 48 h, after which the cells were fixed and stained with specific antibodies to dynein (DLC) or dynactin (p150Glued) in combination with either α-tubulin or phalloidin to illuminate microtubules or actin-filaments, respectively. Consistent with its conventional role as a microtubule-based motor, dynein tightly co-localized with the OC microtubule network on both glass and bone surfaces, faithfully following the reorganization of the microtubule architecture during polarization/bone resorption. In contrast, minor overlap between dynein and actin filaments were observed in OC, irrespective of their polarized status, with partial co-association detected in proximity of the podosomal belt and within the sealing zone/RB region. We also confirmed co-localization between dynein and dynactin subunits in highly polarized human OCs actively resorbing bone (Fig. 2A–B). Cross-correlation analyses (Pearson's, Rr = 0–1) further supported these co-localizations, with dynein showing strong correlation with both dynactin (p150Glued) (Rr = 0.831 ± 0.05) and α-tubulin (Rr = 0.889 ± 0.06), but comparatively modest correlation with F-actin (Rr = 0.494 ± 0.09) (Fig. 2C).
Because the exact polarity of microtubules (plus-ends vs minus-ends) in OCs remain ill-defined, we also immunostained OCs with CLIP-170, a marker of dynamically growing microtubule plus-ends, to orientate OC microtubule plus-end pole(s). To this end, OCs cultured on either glass or bone were stained with CLIP-170 and its distribution compared with F-actin, microtubules, and dynein-dynactin. As shown in Fig. 3, CLIP-170 (green) typically stained the peripheral tiplike projections of microtubules (red) that emanated away from the nuclear region (magenta) of OCs cultured on both glass (Fig. 3A, panel iv) and bone (Fig. 3B, panels v–viii). Interestingly, in bone-resorbing OCs, we noted localized densities of CLIP-170 positive-ends that projected into the RB/SZ region (Fig. 3B, panels i–iv, white arrows) as defined by the actin-ring (blue), thereby orientating that the RB membrane as a “plus-end” microtubule domain. Furthermore, we also detected strong co-localization between CLIP-170 and p150Glued, both on glass and (Fig. 3A, panels v–viii) and at the RB level on bone (Fig. 3B, panels ix–xii). These co-localization patterns were also observed when OCs where cultured on bone-mimicking Osteologic discs (Supplemental Fig. 2). Quantitative analysis of co-localization coefficients (Rr) further confirmed this position, with strong association observed between CLIP-170 and p150Glued (Rr = 0.956 ± 0.04); microtubules (Rr = 0.841 ± 0.07); and F-actin (Rr = 0.830 ± 0.06), respectively (Fig. 3B and Supplemental Fig. 3). Together, these data position the RB as a plus-end microtubule domain and indicate that a percentage of dynein-dynactin coexists with CLIP-170 in OCs at plus-end microtubule poles.
Disruption of the dynein-dynactin complex impairs osteoclastogenesis
To investigate the function of the dynein-dynactin complex in OCs, we utilized a retroviral-based transduction system (pMX) to overexpress a GFP-fusion chimera of p50/dynamitin (Fig. 4A), which has been shown to competitively inhibit endogenous dynein activity.33 Exogenous expression of dynamitin/p50 destabilizes dynein-dynactin interaction(s) by displacing its endogenous p50 counterpart and thus is routinely employed as a tool for monitoring dynein function.34 The transduction efficiency and expression levels of, either p50-GFP or GFP alone (control) into mouse BMM-derived OCs was confirmed by fluorescence microscopy and immunoblotting (Fig. 4B–C).
Having validated successful delivery and expression of p50-GFP into OCs, we next examined the potential effect of dynein-dynactin disruption on osteoclastogenesis. To this end, BMMs were transduced with retroviruses encoding either p50-GFP or GFP alone and then stimulated with M-CSF/RANKL for up to 7 days. At specific time points, OCs were fixed and stained for TRACP and the morphology and number of TRACP-positive multinucleated cells (>3 nuclei) between p50-GFP and GFP control groups were compared. Whereas the majority of OCs formed at day 7 by the p50-GFP group were indistinguishable from those derived from the GFP control cells (Fig. 4D), with each displaying typical microtubule organization and podosomal belts (Supplemental Fig. 4), we did note aberrations in the morphology and spreading of p50-GFP transduced OCs, however, only upon pronounced overexpression of the fusion chimera (eg, in Fig. 5A). Moreover, quantitative analyses of OC numbers yielded significant and reproducible differences in the total numbers of OCs formed between p50-GFP and GFP control groups that were most evident in the early phases of OC formation (ie, days 3 to 5), indicating that disruption of the dynein-dynactin complex delayed the onset of osteoclastogenesis (Fig. 4E). The delay in OC formation rates was further reflected by reduced expression of RANKL-induced OC marker proteins, including cathepsin K, c-Src and the vesicular glutamate transport V-GLUT135 in p50-GFP-expressing cells compared with those expressing GFP alone (Fig. 4F).
To investigate the potential mechanisms underlying this perturbation in OC differentiation, we first checked for disturbances in key RANKL-signaling cascades, including nuclear factor of kappa B (NF-κB) and mitogen-activated protein kinase (MAPK). NF-κB activity was monitored by the degradation and phosphorylation signatures of IκBα and p65 (p-p65). As illustrated in Fig. 4G, the RANKL-induced degradation/phosphorylation profiles of IκBα and p65 (p-p65) were comparable between p50-GFP- and GFP-expressing OC progenitor cells. Little differences were also detected in the activation levels of MAPK signaling pathways, including p-ERK and p-p38 when examined within short (Fig. 4G; within 1 h) or prolonged periods (Fig. 4F; 1 to 5 days) of RANKL-stimulation, indicating that these signaling pathways alone were not sufficient to account for the delays in OC formation rates upon dynein-dynactin disruption.
Next, we checked for differences in the proliferative status of the OC progenitor cells. Proliferation of OC progenitors expressing GFP or p50-GFP was monitored using MTS assays after exposure to M-CSF (50 ng/mL) for 12 h (Fig. 4I). Consistent with previous reports,32 we found that disruption of dynein activity by p50 overexpression significantly delayed the proliferation rates of OC progenitors. Because dynein (DLC) was found to localize to mitotic kinetochores of OC progenitor cells, 1-day post RANKL (Fig. 1E, white arrows), we reasoned that the delay in cell proliferation might reflect disturbances in mitotic division. To this end, we monitored the number of OC progenitors exiting mitosis (1 day post-RANKL stimulation) by fluorescence microscopy after staining with the mitosis-specific anti-phospho-histone 3 marker.36 In keeping with this position, we found that progenitor cells overexpressing p50-GFP rapidly accumulated in a prometaphase-like state (Fig. 4J, white arrows), displaying >200% the mitotic index of GFP-expressing control cells (Fig. 4K). Taken together, these data indicate that the disruptions in osteoclastogenesis after dynein-dynactin uncoupling are attributable, at least in part, to mitotic delay of OC progenitor cells.
Uncoupling of the dynein-dynactin complex disrupts Golgi ribbon integrity and spatial organization of acidified late endosomes/lysosomes in OCs
The dynein-dynactin motor complex is principally responsible for the unilateral transport of a variety of intracellular cargo toward microtubule minus-ends.37, 38 Therefore, we next assessed the effects of p50-mediated dynein-dynactin disruption on the spatiotemporal distribution and motility of key intracellular organelles and cargo in OCs, namely the Golgi and acidified lysosomal compartments, the latter of which is known to house cathepsin K, an enzyme that is essential to the bone resorptive process.39 Golgi architecture and stability was first assessed by immunofluorescence confocal microscopy using the specific cis-Golgi marker GM130. As shown in Fig. 5A, whereas GFP-expressing control OCs displayed characteristic Golgi ribbons that tightly circumscribed their concurrent nuclei, OCs expressing p50-GFP displayed drastic alternations in Golgi ribbon morphology consistent with a disruption in dynein activity. These ribbon phenotypes ranged from mildly fragmented (broken ribbons), compacted (collapsed ribbons) to dispersed/fragmented (scattered ribbons) (Fig. 5B), an effect that correlated with the level of p50-GFP overexpression (as defined by GFP-intensity profiles shown in Fig. 5C), with complete ribbon dispersal most evident in OCs highly expressing exogenous p50-GFP, as evidenced by quantitative analysis (Fig. 5D). Similarly, we observed clear redistribution of acidified late-endosomes (LysoTracker Red) and cathepsin K–containing lysosomes (cathepsin K fluorogenic substrate) from a mainly perinuclear location in GFP-expressing control OCs toward the cell periphery upon the overexpression of p50-GFP as monitored by live cell confocal microscopy (white arrows, Fig. 6A). Interestingly, despite this dramatic redistribution, analysis of vesicle trajectories revealed that late endosomes/lysosomes were still capable of bidirectional movement in both centrifugal and centripetal directions between the center (nuclei) and the periphery (plasma membrane, PM) in OCs expressing p50-GFP (Fig. 6B–C). In GFP-expressing control cells, most lysosomes were typically concentrated in the perinuclear region displaying highly motile properties, with many moving rapidly toward or away from the cell center and occasionally reversing, characteristic of dynein function (Supplemental Video 1). By comparison, p50-GFP-expressing cells exhibited a more peripheral dispersion of lysosomes (yellow arrows, Fig. 6A), some of which displayed short-range and stationary movement, while others alternated between this behavior with very rapid local-range plus-end movement that appeared to fuse with the plasma membrane (PM) (Supplemental Video 2). Whereas we did not observe a complete loss of minus-end events in p50-GFP expressing OCs, the peripheral dispersion of late endosomes/lysosomes together with the increased number/length of plus-end (kinesin-driven) movements observed are both consistent with a net reduction in dynein function in vivo.
Dynein-dynactin dysfunction diminishes osteoclastic bone resorption as a result of impaired cathepsin K secretion
Finally, we examined the impact of dynein-dynactin disruption on the OCs functional capacity to resorb bone. For this purpose, BMM-derived OCs expressing either GFP or p50-GFP were cultured on dentine discs for 48 h, after which cells were fixed and either stained with rhodamine-conjugated phalloidin to assess OC polarization (F-actin ring formation) or removed to visualize the number and morphology of the underlying resorptive pits by scanning electron microscopy (SEM). Assessment of OC polarization showed little differences in the ability of OCs to form F-actin rings (Fig. 7A). Quantitative analysis of OC resorptive parameters revealed a slight, albeit modest change, in the total resorbed area per dentine disc (Fig. 7B, D), however, this difference failed to reach statistical significance. This may be due, in part, to the heterogeneous expression levels of p50-GFP observed in OCs and residual endogenous dynein activity that was inherent in our approach. Nonetheless, closer inspection of individual resorptive pits revealed clear distinctions in the pit depth and morphology between GFP transduced and p50-GFP transduced cells, the latter of which was deemed morphologically shallow as confirmed by confocal microscopy and quantitative analysis (Fig. 7C, E).
Considering the drastic redistribution of the Golgi and lysosomes after dynein disruption, we reasoned that the reduced pit depth may reflect disturbances in the OCs ability to deliver and/or secrete cathepsin K (CTSK) at the RB. Consistent with this position, we observed obvious impairments in the ability of OCs to deliver CTSK to the RB during bone resorption when p50-GFP was overexpressed, with considerably less CTSK reaching the RB exit site, but instead accumulating within the nuclear (n) region (Fig. 7F–G, arrows). This reduction in CTSK delivery to the RB was verified by immuno-intensity profiles (Fig. 7H) and quantitative analyses of CTSK secretion relative to F-actin rings/SZs (Fig. 7I). Furthermore, the defect in cathepsin K delivery was also confirmed biochemically by analysis of medium CTx-1 concentration, which revealed a significant decrease in CTx-1 release in OCs overexpressing p50-GFP compared with GFP control cells (Fig. 7J). Together, these data indicate that the structural and functional integrity of the dynein-dynein complex is an important requirement for efficient bone resorption, serving to regulate the delivery and/or release of cathepsin K during the bone resorptive process.
The dynein-dynactin motor complex is an elegant mechanochemical machine that orchestrates the coupling of intracellular cargo to microtubules, thereby empowering their directionality toward minus-ends. Despite the critical importance of dynein-dynactin being well-established in a multitude of cell systems, its involvement in bone-resorbing OCs has, up until now, remained obscure. Here, by exploiting the ability of exogenous p50/dynamitin to disrupt dynein activity in vivo, we show for the first time that the dynein-dynactin motor complex: (1) regulates the proliferation and mitotic progression of OC progenitors during RANKL-induced differentiation; (2) is essential to the integrity and stability of key intracellular organelles (ie, the Golgi and lysosomes) in OCs; and (3) is required for the efficient delivery and/or secretion of osteolytic cargo (namely cathepsin K), and thus constitutes an integral component of the osteoclastic bone resorptive machinery. Collectively, this study represents the first comprehensive analysis of a microtubule-based motor of any kind in OCs and unveils remarkable versatility for dynein-dynactin in OC formation and function.
By combining RT-PCR with subcellular fractionation assays and immunofluorescence microscopy we show that most, if not all, of the conventional dynein-dynactin subunits are endogenously expressed in OCs. Whereas the majority of these subunits appear to be constitutively expressed throughout osteoclastogenesis (DHC, DLIC, and p50/dynamitin), others exists as distinct subunits and/or are differentially localized during RANKL-mediated differentiation, indicating that they may be recruited for specific phases of OC formation and activation. For instance, unlike the brain dynein complex, which primarily houses the intermediate chain subunit IC2A, the OC dynein complex contains a unique IC splice variant, namely IC2C. Similarly, the dynein LC identified here corresponds to LC8, which is modestly downregulated during RANKL-driven OC differentiation, implying that other variable light chains must exist in OCs to facilitate specific motor-cargo interactions. Indeed, we have recently reported the existence of another light chain protein, ie, Tctex-1 in OCs, which functions as a molecular adaptor to link Rab3D-bearing secretory vesicles to dynein and microtubules.14 Additionally, we have further shown that LIS1, a multifunctional regulator of the dynein-dynactin complex, serves to couple cathepsin K–containing lysosomes to the dynein-dynactin complex during bone-resorption.13 Thus, it is conceivable that dynein-dynactin adopts unique structural configurations in OCs by selectively substituting individual subunits, with each variant conferring a specialized function during OC formation and bone resorption.
Mutations in the cytoplasmic dynein complex and its regulators underscore a plethora of neurological diseases that are conserved from mice to humans.40 Deletion of dynein-dynactin subunits in flies and mice correspond with embryonic lethality, thereby exemplifying the fundamental importance of this motor complex during development.38, 41, 42 Our findings that a loss of dynein-dynactin integrity disrupts OC formation and bone resorption, further extends the critical importance of this motor complex to bone. Although uncoupling the dynein-dynactin complex did not prevent the final progression and fusion of mononuclear cells into mature OCs at day 7, it significantly retarded the onset of osteoclastogenesis particularly at the early phases of OC differentiation. This disruption in OC formation was attributed, at least in part, to significant delays in the proliferation and mitotic exit rates of precursor cells rather than major impairments in prototypical RANKL-signaling cascades including NF-κB and MAPK. These findings are in keeping with previous studies in non-neuronal cells, which indicate that whereas dynein activity is essential for mitotic spindle pole organization and cell proliferation,32, 43 it is largely dispensable for the activation of NF-κB and MAPK signaling pathways.44–46
Perhaps the most striking morphological aberrations observed after p50-overexpression was the collapse and dispersal of the Golgi ribbons and loss of the spatial positioning of acidified late-endosomes/lysosomes. These findings were not completely unexpected and they reaffirm the wealth of data reported in mononuclear cell types documenting the crucial role for dynein-dynactin in maintaining the peri-nuclear positioning of the Golgi and endosomes-lysosomes in fibroblastic cells.33, 38 Reflecting this drastic redistribution of intracellular organelles, disruption of dynein activity led to impaired osteoclastic bone resorption function. Whereas dynein dysfunction neither altered OC intracellular acidification (Supplemental Fig. 5), nor its capacity to resorb the mineral phase of bone and/or hydroxyapitite substrates (Supplemental Fig. 5)—implying that the bulk of the acidification machinery (ie, V-ATPases) remained “in-check,”—removal of the organic component of bone was significantly impaired as evidenced by the reduced pit depth and decreased cathepsin K delivery/secretion at the RB membrane, a phenotype reminiscent of that observed in cathepsin K– and β3-integrin-deficient mice.39, 47 Moreover, the magnitude of the reduced pit depth and impaired cathepsin K secretion phenocopies that were observed after the depletion of dynein-light chain (eg, Tctex-1), and dynein regulatory molecules (eg, LIS1, Rab3D) in OCs inferring that it may be a general phenomenon upon disruption of dynein-mediated transport.13, 14, 20
How might dynein-dynactin regulate the delivery of cathepsin K during bone resorption? In the absence of further experiments, we can only speculate as to the precise role of dynein-dynactin in the bone resorptive process. Nonetheless, an attractive model posits dynein-dynactin as a molecular anchor, serving to tether cathepsin k–containing secretory lysosomes to microtubules plus-ends at the RB in preparation for their release and membrane fusion during the resorptive process (Supplemental Fig. 6). Indeed, despite its canonical role as a minus-end directed motor, dynein often accumulates at the plus-ends of microtubules, where it functions along with dynactin and CLIP-170 to modulate the recruitment and attachment of diverse intracellular cargos.30 The observed co-localization between the dynactin subunit p150Glued and CLIP-170 at the RB level of bone-resorbing OCs would be in keeping with this position. This scenario is further strengthened by recent optical trapping and total internal reflection microscopy (TIRF) experiments, which showed that dynein functions to tether and stabilizes dynamic microtubule plus-ends at the cell cortex via an intimate association with dynactin.48 An alternative but less favored possibility is that dynein-dynactin may serve to ensure the stability of the Golgi apparatus during the trafficking of cathepsin K from the Golgi to secretory lysosomes, with the observed reduction in cathepsin K secretion being simply a consequence of compromised Golgi integrity after a net loss of dynein activity. Irrespective of the exact mechanism, our data clearly highlight an important role for dynein-dynactin in the transport of cathepsin K, placing this motor complex among the growing list of molecular participants known to regulate vesicular trafficking during osteoclastic bone resorption.8
In summary, our findings unveil a previously unanticipated but important role for the dynein-dynactin complex in OC formation and function. Additionally, we establish for the first time, the RB membrane as a plus-end microtubule domain upon which dynein-dynactin intimately associates to facilitate the microtubule-coupled delivery of acidified intracellular cargo (eg, cathepsin k). Future studies will focus on the generation and characterization of conditional knockout mice to OC-specific dynein subunits like IC2C and/or its regulators (ie, Tctex-1 and LIS1), which will undoubtedly shed important new light on the precise role of this enigmatic microtubule motor in bone physiology.
All authors state that they have no conflicts of interest.
We are grateful to Drs RB Vallee, N Galjart, W Langdon, and R Jahn for providing cDNA constructs and antibodies. Special thanks to Drs Tamara Abel and Paul Rigby for their technical assistance with the live cell imaging. All microscopy was carried out using facilities at the Centre for Microscopy, Characterization and Analysis, the University of Western Australia. PYN is supported by an International Postgraduate Research Scholarship (IPRS) and University Postgraduate Award (International Student) (UPAIS). This work was further supported by the National Health and Medical Research Council of Australia (NHMRC) (ID1029292), which went to NJP, JX, and MHZ.
Authors' roles: PYN, TSC, HZ, MHZ, and NJP designed the experiments and analyzed the data; PYN, TC, SY, HZ, ESMA, ECK, and HTF performed the experiments; NJP, HZ, MHZ, and JX supervised the experiments NJP; and PYN wrote the manuscript. NJP, MHZ, JX, and HZ revised the manuscript. All authors approved the final version of the manuscript.