Summary
Endocytosis and the subsequent intracellular trafficking of the endocytosed material are important determinants of cellular function. Osteoclasts, cells of the monocyte/macrophage family, are specialized for the internalization and processing of bone matrix. Transcytosis of endocytosed material has been observed in osteoclasts but the precise mechanism controlling this process is unclear. Here, we investigate the regulation of these trafficking events. To establish the directionality and kinetics of trafficking events in resorbing osteoclasts, we devised a system using fluorescent low-molecular-weight markers as probes to follow the route taken by the digested bone matrix. We demonstrate that this route is largely distinct from the pathway followed by proteins taken up by receptor-mediated endocytosis at the basolateral plasma membrane. Endocytosis and transcytosis from the ruffled border are fast processes, with a half-life of the endocytosed material inside the cells of 22 minutes. We demonstrate the crucial role of the microtubule network in transport from the ruffled-border area and provide evidence for a role of the cytoskeleton in the overall efficacy of trafficking. Moreover, we analyse the effect of the V-ATPase inhibitor bafilomycin A1 on endocytic uptake, which gives insight into the pH-dependent regulation of membrane trafficking and resorption in osteoclasts.
Introduction
Rapid exocytosis and transcytosis of endocytosed material are not limited to specialized cell types, such as neurons, but are present in a range of cells, thus expanding the cellular models for intracellular polarized trafficking greatly. Osteoclasts are terminally differentiated, multinucleate cells specialized for bone-matrix degradation, a process that involves extracellular-matrix digestion, endocytosis and subsequent transcytosis (Nesbitt and Horton, 1997; Salo et al., 1997). Therefore, osteoclasts are dynamic cells with a high membrane turnover and are in close contact with the external environment through the continuous process of exocytosis of vesicles and the parallel endocytosis of digested material.
At the beginning of resorption, the osteoclast cytoskeleton rearranges so that actin forms a dense belt-like structure close to the bone surface, the `actin ring' (Lakkakorpi and Väänänen, 1996). Polarization of the cell is detectable and, owing to fusion of exocytotic vesicles containing vesicular ATPase (V-ATPase), which supplies the large amounts of protons needed to solubilize the mineral component of the bone, the part of the plasma membrane enclosed by the actin enlarges into the highly convoluted ruffled border (for a review, see Stenbeck, 2002). A possible uncoupling of vesicle fusion from vesicle recycling might be responsible for this drastic expansion of the plasma membrane. Depending on their size, osteoclasts are capable of excavating pits several micrometres deep and tens of micrometres long during overnight cultures (Chambers et al., 1984). An average excavation speed of around 1 μm per hour can be estimated (S. Nesbitt, personal communication). Thus, if all digested material passes through the cell, an estimated 1.3×10–16 moles of collagen fragments per hour would cross an osteoclast. In MDCK cells, approximately 2.4×10–19 moles of dextran per hour cross the cell (von Bonsdorff et al., 1985). Taking into account the fact that osteoclasts are about ten times larger than MDCK cells, a very efficient transport system within the osteoclasts is predicted to accommodate the large amount of material that is passing through the cell. This inwardly polarized traffic of degraded bone is accompanied by the apposition of a corresponding amount of ruffled-border membrane, which needs to be recycled for homeostasis to occur. This dynamic equilibrium is supported by the finding that basolaterally endocytosed transferrin is found in the area of the ruffled border after 30 minutes of chase (Palokangas et al., 1997). It was also recently suggested that the ruffled border is divided into a peripheral secretory subdomain and a centrally located uptake zone (Mulari et al., 2003). Studies with differentially targeted virus particles, in conjunction with the localization of lysosome-associated membrane proteins (LAMPs) and the small GTPase Rab7 to this plasma-membrane domain, also indicate a high similarity of the ruffled border to the late-endosome/lysosome compartment (Palokangas et al., 1997; Salo et al., 1996).
The cytoskeleton plays an important role during osteoclast bone resorption, because both interfering with the microtubule network and disruption of the actin cytoskeleton inhibit resorption (Shimizu and Sasaki, 1991; Zaidi, 1990). This could reflect the altered adhesion properties of the osteoclasts after drug treatment. However, in other cell types, the cytoskeleton was shown to play a vital part in vesicular trafficking and transcytosis (Apodaca, 2001). Vesicular carriers travel along microtubules towards their destination and are translocated to the actin cytoskeleton at the cell periphery before fusion to the plasma membrane (Goode et al., 2000). Disruption of microtubules with nocodazole or colchicine blocks transport from the early to the late endosome and the basolateral to apical transcytosis of several proteins (Apodaca, 2001). In contrast to receptor-mediated endocytosis, which is microtubule independent, adsorptive and fluid-phase endocytosis are slowed by administration of colchicine (Piasek and Thyberg, 1980). Actin filaments have been linked to endocytosis from the apical plasma membrane in several polarized cell types (Apodaca, 2001) and the actin-disrupting drug cytochalasin has been shown to slow basolateral to apical transcytosis significantly (Maples et al., 1997). In osteoclasts, both microtubules and F-actin are found in the central part of the ruffled border, which is surrounded by the actin ring of the sealing zone (Mulari et al., 2003). The rearrangement of the actin cytoskeleton at the beginning of the resorption cycle is stimulated by engagement of the major osteoclast integrin αvβ3, which activates a heteromultimeric signalling complex that includes actin-binding proteins, the tyrosine kinase pp(60c-src) and the lipid phosphatidylinositol-3-kinase (PI-3-kinase) (Hruska et al., 1995). Consequently, inhibitors of tyrosine phosphorylation or PI-3-kinase block osteoclastic bone resorption. However, PI-3-kinases also have a role beyond the organization of the actin cytoskeleton in intracellular trafficking (Backer, 2000).
The large amount of membrane that moves through osteoclasts during the resorption cycle could provide the cell with a monitoring parameter for its resorption activity. To establish the directionality and kinetics of trafficking events in resorbing osteoclasts, the requirements for endocytosis from the ruffled border and the pathway taken by the endocytosed material, we used soluble fluorescent marker molecules as probes. This strategy made the labelling of the bone and its associated problems (e.g. surface restriction of the label or the necessity of whole animal experiments with tetracycline) redundant. Here, we provide evidence for the crucial role that the microtubule network plays in transport from the ruffled border area. Moreover, we analyse the effect of the V-ATPase inhibitor bafilomycin A1 on endocytic uptake, which gives insight into the pH-dependent regulation of membrane trafficking and resorption in osteoclasts. Additionally, we demonstrate that the route followed by the endocytosed material is largely distinct from the pathway taken by molecules endocytosed through receptor-mediated endocytosis from the basolateral domain of the plasma membrane.
Materials and Methods
Materials
Fluorescent dextrans with different molecular weights (3000 kDa and 70,000 kDa), rhodamine-conjugated phalloidin and Alexa-488-labelled goat anti-mouse secondary antibodies, DiI-LDL and TRITC-labelled transferrin were from Molecular Probes (Leiden, The Netherlands), fluorescein-labelled bovine serum albumin (FITC-BSA) (∼70,000 kDa) and fluorescein-labelled phalloidin were from Sigma (Poole, UK), Cy5-labelled goat anti-mouse secondary antibodies were from Kirkegaard & Perry Laboratories (Gaithersburg, MD, USA). Tissue-culture materials were obtained from Invitrogen (Paisley, UK). Bafilomycin A was a kind gift from J. Mattsson (AstraZeneca, Mölndal, Sweden). Nocodazole, cytochalasin D, latrunculin B and calcitonin were from Calbiochem (Nottingham, UK). Stock solutions of the inhibitors were prepared in dimethyl sulfoxide (DMSO) and, in every relevant experiment, an equal volume of DMSO was added to control samples. Anti-αvβ3-integrin antibodies (clone 23C6) have been described previously (Davies et al., 1989).
Labelling of dentine disks with 5-(and-6)-carboxy-SNAFL-2
For labelling with the seminaphtofluorescein dye 5-(and-6)-carboxy-SNAFL®-2, 50 dentine disks were sonicated, polished on Whatman 3MM paper and suspended in 0.1 M sodium bicarbonate pH 8.3 before the addition of 300 μg 5-(and-6)-carboxy-SNAFL-2 (Molecular Probes, Leiden, The Netherlands) for 1 hour at room temperature. The disks were then washed with PBS and incubated with 0.15 M hydroxylamine, pH 8.5, for 1 hour at room temperature, washed several times with water, sterilized in 70% ethanol and air dried.
Isolation and culture of osteoclasts
The procedure used to culture rabbit osteoclasts was modified from the original method developed by Tezuka et al. (Tezuka et al., 1992). Briefly, osteoclasts were mechanically disaggregated from long bones of 5-day-old rabbits in alpha-MEM containing 2 mM L-glutamine, 100 IU ml–1 penicillin and 100 μg ml–1 streptomycin. The cells were then pelleted, resuspended in alpha-MEM containing 2 mM L-glutamine, 100 IU ml–1 penicillin, 100 μg ml–1 streptomycin and 10% foetal bovine serum (FBS), and allowed to attach to sonicated dentine slices (100-150 μm thick, surface area 0.3 cm2) at 37°C in 5% CO2/95% air for 60 minutes. Non-adherent cells were then washed away and the remaining cells were cultured for 20 hours in α-MEM containing 2 mM L-glutamine, 100 IU ml–1 penicillin, 100 μg ml–1 streptomycin and 10% FBS.
Uptake of fluorescently labelled markers
For incubation with the fluorescent test substances, the cells on dentine slices were transferred into MEM-BSA [α-MEM buffered with 20 mM HEPES (pH 7.0) containing 2 mM L-glutamine, 100 IU ml–1 penicillin, 100 μg ml–1 streptomycin and 0.1% BSA] and incubated for 1-20 minutes in MEM-BSA containing 30-260 μM fluorescent markers. The incubations were stopped by washing the cells on dentine slices twice with PBS for 20 seconds each, followed by the immediate addition of 3% paraformaldehyde, 2% sucrose in PBS for 10 minutes at room temperature. Experiments were repeated at least three times and representative images of a minimum of 50 different osteoclasts are shown.
In some experiments, the cells on dentine slices were preincubated for 30 minutes at 37°C in MEM-BSA containing the appropriate drug (25 nM bafilomycin A, 10 μM cytochalasin D, 1 μM latrunculin, 50 nM paclitaxel or 10 μM nocodazole or vehicle) before exposure to the fluorescent markers.
Immunodetection
Resorbing osteoclasts were identified by staining with rhodamine- or FITC-conjugated phalloidin and/or immunostaining with anti-αvβ3 antibodies. After fixation, the cells were incubated for 30 minutes in 5% newborn bovine serum in PBS containing 0.02% sodium azide (wash) and then processed for immunodetection. Primary antibodies were diluted in wash and the cells were incubated for 30 minutes at room temperature. Bound antibodies were detected with Alexa-488- or Cy5-labelled goat anti-mouse secondary antibodies.
Confocal microscopy and three-dimensional reconstruction
Fluorescent marker and antibody distribution were monitored with a Leica TCS-NT confocal laser-scanning microscope (CSLM) (Leica, Heidelberg, Germany) using standard filter settings and sequential scanning to avoid overlap of emission from the fluorophores. The thickness of the optical section was calculated with the help of the Leica TCS-NT software and was set to 1.013 μm for the x-y sections and 1.008 μm for the z-x sections. In some images, the x-y sections were added together with the extended focus option of the Leica TCS-NT software to allow the visualization of the area just above the bone surface to the bottom of the pit. Experiments were repeated at least three times for each marker. For dextran uptake and trafficking kinetics, the amount of fluorescence inside at least 15 osteoclasts was determined after removing the bottom sections containing the pit area from the stack using the Leica TCS-NT software package, the remaining sections were then added together and the fluorescence intensity determined as above and expressed as arbitrary fluorescence units. For three-dimensional (3D) reconstruction, the optical thickness of the x-y sections was set to 0.5 μm and sections were collected from the top of the cell to the bottom of the pit. The sections were then added together with Imaris software (Bitplane, Zürich, Switzerland) and displayed as z-x sections of the whole cell.
Statistical analysis
All experiments were repeated at least three times and, in each experiment, the number of αvβ3-antibody-positive, multinucleate cells was determined and examined for TRITC-dextran endocytosis. In experiments with inhibitors, at least 300 osteoclasts were observed to determine the number of endocytosis-positive cells. In the kinetics experiments, the fluorescence intensities of at least 15 osteoclasts were measured. Results obtained for each experiment were expressed in arbitrary fluorescence units (± standard error) or as the proportion of endocytosis-active osteoclasts in total osteoclasts (± standard error) and compared with the vehicle control with the StatView SE package (Abacus Concepts, Berkeley, CA, USA). Results from independent experiments were pooled before plotting. Differences were analysed using Student's t-test and considered significant if P<0.05.
Results
Directionality of endocytosis in resorbing osteoclasts
Osteoclasts plated on glass are non-resorbing and highly motile, and form peripheral plasma-membrane-associated filamentous actin structures. By contrast, osteoclasts plated on bone or dentine are less motile, excavate pits and form the typical actin-rich sealing zone, where actin is largely condensed (Fig. 1F,H, arrows); these structures are readily distinguishable from the cortical actin (Fig. 1F,H, triangles). When we observed the uptake of inert fluorescent molecular markers (FITC-labelled dextran-3000 and FITC-labelled dextran-70,000) in osteoclasts plated on glass and dentine, there was a marked difference between the cells plated on glass (Fig. 1A-D) and the cells plated on dentine (Fig. 1E-H). The non-resorbing osteoclasts plated on glass endocytosed both markers to a similar extent (Fig. 1A,C), whereas resorbing osteoclasts plated on dentine only endocytosed the low-molecular-weight dextran-3000 (Fig. 1E). Extending the incubation time with the markers from 5 minutes to 20 minutes did not change the result (data not shown). We have previously shown that the sealing zone allows the passage of low-molecular-weight markers like dextran-3000 (Stenbeck and Horton, 2000). In osteoclasts plated on glass, sealing might not be as effective as on dentine, because few acidify the subcellular space (Zimolo et al., 1995), thus allowing passage of high-molecular-weight markers, or endocytosis takes place at random sites because matrix-derived cues are missing.
Substrate dependence of dextran endocytosis. Rabbit osteoclasts cultured overnight on glass (A-D) or dentine (E-H) were incubated for 5 minutes with either FITC-labelled dextran-3000 (FD3) (A,E) or FITC-labelled dextran-70,000 (FD70) (C,G). Cells plated on glass are non-resorbing and endocytose both markers to a similar extent (A,C), whereas resorbing osteoclast plated on dentine only endocytose low-molecular-weight dextran 3000 (E,G). F-actin was visualized with rhodamine-conjugated phalloidin (B,D,F,H). The typical ring-like actin structure of resorbing osteoclasts is only visible in cells plated on dentine (F,H, arrow). The outline of the cortical actin observed in B,D,F,H (triangle) is indicated in A,C,E,F with a white dashed line. Images shown are single CSLM sections taken close to the substrate surface, which is visible in E and G because of the binding of the negatively charged FITC-labelled dextran. Scale bar, 10 μm.
We hypothesized that, during resorption, the endocytic activities of osteoclasts shift towards the ruffled border. To verify our hypothesis, we incubated osteoclasts plated on dentine with cytochalasin D, a drug that disrupts actin filaments. Incubating resorbing osteoclasts with cytochalasin D interferes with the sealing zone and allows high-molecular-weight molecules like FITC-BSA, which are normally excluded, access to the resorption area (Stenbeck and Horton, 2000). We used this observation to load the resorption area with FITC-BSA and to follow its endocytosis after removal of cytochalasin D. Osteoclasts on dentine were incubated for 10 minutes with cytochalasin D or vehicle control before the addition of FITC-BSA for 5 minutes in the presence (Fig. 2B) or absence (Fig. 2A) of the drug. The cells were then washed and chased in medium without inhibitor and FITC-BSA for 30 minutes (Fig. 2D); the control sample without cytochalasin D is shown in Fig. 2C. As shown in Fig. 2B,D, cytochalasin D addition caused the appearance of FITC-BSA staining in the resorption area. Remarkably, subsequent uptake of FITC-BSA was only observed in osteoclasts in which the labelled marker had entered the resorption area (Fig. 2D), indicating that most of the endocytic activities was taking place at the ruffled border.
Endocytic activities of osteoclasts shift towards the ruffled border during resorption. Rabbit osteoclasts cultured overnight on dentine were incubated with vehicle control (A,C) or 10 μM cytochalasin D for 10 minutes (B,D) before the addition of FITC-BSA for 5 minutes and then either fixed directly (A,B) or washed and chased in medium without inhibitor and FITC-BSA for 30 minutes (C,D). FITC-BSA accumulates in the pit immediately below the cell only upon treatment with cytochalasin D (B, arrowheads). After the chase, endocytosis is only observed in cells that had been exposed to cytochalasin D (fluorescent structures inside of the cell are indicated with arrows in D), whereas vehicle-treated cells remained refractory to the label (C). The outline of the cells as delineated by αvβ3 antibodies is depicted by a dashed white line. Images shown are representative of 50 osteoclasts and lateral views of the cells taken with CSLM with a thickness of 1.008 μm. Scale bar, 10 μm.
Pathways of fluid-phase endocytosis and bone-matrix resorption converge
To investigate whether molecules added to the outside of the resorption area enter the same vesicle pathway as the digested bone matrix, we covalently labelled the surface of the dentine with the green-fluorescent dye 5-(and-6)-carboxy-SNAFL-2. Osteoclasts were plated on the labelled dentine and incubated for 4 hours before the addition of TRITC-labelled dextran-3000 for 30 minutes. After fixation, the distribution of labelled bone matrix and fluid-phase marker was observed with CSLM (Fig. 3A,B, respectively). The experiment was performed early in the resorption culture to ensure that pits would not be very deep, thus allowing large amounts of fluorescent surface marker to be available to monitor intracellular trafficking of bone matrix. Additionally, the cells were exposed to the fluid-phase tracer for longer periods to label at equilibrium most parts of the endocytic pathway. Most of the labelled vesicular structures contain both labels (yellow in Fig. 3C); the few vesicular structures containing only the fluid-phase marker are located towards the central part of the cell (Fig. 3C, arrowheads). This result indicates that the endocytic pathways of digested bone matrix and low-molecular-weight dextran largely converge.
Pathways of endocytosed fluid-phase marker and resorbed bone matrix converge. Osteoclasts were plated on fluorescently labelled dentine and incubated for 4 hours before the addition of TRITC-labelled dextran 3000 for 30 minutes. After fixation, the distribution of labelled bone matrix (A,C, green) and low-molecular-weight dextran (B,C, red) was analysed with CSLM. Most of the visible vesicular structures contain both labels (C, yellow); the few vesicular structures containing only dextran label are located towards the centre of the cell (C, arrowheads). The outline of the cells as delineated by αvβ3 integrin antibodies is depicted as a dashed white line. Scale bar, 10 μm.
Co-localization of dextran with endocytic markers
To establish the route that the internalized material is taking and to confirm that the colocalization observed in Fig. 3 was not due to extensive exchange between endocytic compartments after endocytosis, pulse-chase experiments with fluorescently labelled low-density lipoprotein (LDL) and transferrin were performed. Transferrin is commonly used to label the recycling endosome and LDL is a marker for endocytic traffic directed towards the lysosomal compartment (Gruenberg and Maxfield, 1995). After endocytosis, transferrin bound to its receptor reaches the early endosomal compartment, from where it is delivered via recycling endosomes back to the plasma membrane (Gruenberg and Maxfield, 1995). Depending on the cell type and labelling conditions, a proportion (up to 25%) of the transferrin has a half-life of only 5-7 minutes inside the cell, whereas a larger proportion accumulates in the recycling endosome and it can take more than 60 minutes before these molecules are recycled back to the plasma membrane (Hopkins et al., 1994). In osteoclasts, it has been shown that endocytosed transferrin reaches the ruffled border after 30 minutes of incubation (Mulari et al., 2003; Palokangas et al., 1997), establishing that there is cross-talk between the recycling and late endosome/ruffled border. However, the amount of transferrin that reaches the ruffled border has not been established. To label all subcompartments of the recycling pathway with transferrin, osteoclasts were incubated for 1 hour with TRITC-transferrin before the addition of FITC-labelled dextran-3000 for 10 minutes (Fig. 4A,B). Similar to previous observations (Mulari et al., 2003), there was limited colocalization between the endocytosed markers, which were found in the area close to the bone surface and at the cell periphery (Fig. 4A). The lateral view of the cell underscores this observation, because colocalization is seen only in close proximity to the pit (Fig. 4B, arrowhead). These results indicate that the endocytosed dextran spends limited time in the transferrin-positive compartment. This conclusion was reinforced by examining cells that were labelled for only a short period with both markers. In this condition, the spatial separation between ruffled border and transferrin receptor became apparent, transferrin being found mainly at the leading edge of the cell, whereas endocytosed dextran localized in the middle of the osteoclasts (Fig. 4C, red and green, respectively).
Distribution of dextran in endocytic compartments. Osteoclasts cultured on dentine disks were incubated for 1 hour with TRITC-transferrin (A,B, red) before the addition of FITC-labelled dextran-3000 for 10 minutes (A,B, green). The cells were then fixed and processed for immunostaining with antibodies against the αvβ3 integrin (A-H, blue). Limited colocalization of both markers can be observed in the area of the ruffled border (A, yellow). The lateral view of the cell taken at the position indicated by the dotted white line in (A) confirms this observation because colocalization is seen only close to the pit (B, arrowhead). When the cells were labelled for 10 minutes with both markers together, the spatial separation between ruffled border and transferrin receptor became apparent. Transferrin is found mainly at the leading edge of the cell, whereas endocytosed dextran localized to the middle section of the osteoclasts (C,D, red and green, respectively). A similar picture was also observed when osteoclasts were prelabelled with DiI-LDL (E,F, red) for 30 minutes before the addition of FITC-labelled dextran for 10 minutes (E,F, green). LDL is found distributed throughout the osteoclasts and reaches the area close to the ruffled border, where colocalization with endocytosed dextran can be observed (E,F, yellow and arrowheads). When the cells were labelled for 5 minutes with both markers together and then chased for 25 minutes, the amount of colocalization was limited and localized to the cell periphery (G, yellow and arrowheads), whereas the area close to the ruffled border was devoid of colocalization because of the absence of dextran staining, which had moved towards the cell interior (H). Osteoclasts containing only endocytosed dextran but none of the other markers are indicated with a star in (A) and (E). The direction of movement of the cells is indicated with an open arrow in (B,D,F,H). The asterisks in (E) indicate the pits excavated by the osteoclasts and hence are devoid of surface label. Scale bar, 10 μm.
Similarly, the pathway taken by the lysosomal maker DiI-LDL has been well described in other cell types. After endocytosis, DiI-LDL first reaches the early sorting endosome, where colocalization with transferrin can be observed. The half-life of the DiI-LDL inside this organelle is 8 minutes, after which it is sorted away towards the late-endosome/lysosome pathway (Ghosh et al., 1994), where degradation occurs. Endocytosed marker molecules that are indigestible can recycle back from these late endosomal compartments to the plasma membrane with a half-life of 180 minutes. When osteoclasts were prelabelled with DiI-LDL for 30 minutes before the addition of FITC-labelled dextran-3000 for 10 minutes, DiI-LDL is found distributed throughout the osteoclasts and, like transferrin, it reaches the area close to the ruffled border, where colocalization with endocytosed dextran is observed (Fig. 4E,F, red and green, respectively). However, when the cells were labelled for 5 minutes with both DiI-LDL and FITC-labelled dextran-3000, and then chased for 25 minutes, the amount of colocalization was limited and restricted to the cell periphery (Fig. 4G, arrowheads). By contrast, the area close to the ruffled border was devoid of dextran staining, which had moved towards the cell centre. These data indicate that only a small proportion of the dextran is delivered to DiI-LDL-containing late endosomes. Interestingly, under all incubation conditions, osteoclasts containing only endocytosed dextran but none of the other markers were observed (Fig. 4A,C, star).
Kinetics of fluid-phase-marker endocytosis and trafficking
To assess the rate of trafficking of endocytosed material in osteoclasts, we quantified the uptake and transport of TRITC-labelled dextran-3000 by measuring the amount of fluorescence inside cells in time-course and pulse-chase experiments. We also used computer-assisted 3D reconstruction of x-y cross-sections taken at a 0.5 μm step size through the cells to visualize the trafficking of the dextran through osteoclasts (Fig. 5A,B,D). Whereas, early in the labelling period, the dextran is exclusively found close to the bone surface and the pit (Fig. 5A,B), after 60 minutes of chase, the label is distributed throughout the cell and can be observed close to the basolateral membrane surface (Fig. 5D). For the kinetic experiments, osteoclasts were incubated for 1 minute, 5 minutes, 10 minutes and 30 minutes with the endocytic marker before fixation and immunodecoration with antibodies directed against the αvβ3 integrin to visualize the outline of the cells. The amount of fluorescence inside cells was determined using the Leica TCS-NT software package and expressed as arbitrary fluorescence units (Fig. 5C). A very rapid accumulation of dextran in the pit occurs over the first 10 minutes of incubation (Fig. 5A,B), indicating that the pit area provides a sink for the negatively charged label. This observation is in line with our previous data, which demonstrated that actively resorbing osteoclasts accumulate the negatively charged low-molecular-weight dye Lucifer Yellow in the area surrounded by filamentous actin underneath the osteoclasts (Stenbeck and Horton, 2000). Furthermore, it provides an explanation for the steep increase in fluorescence detected inside the osteoclasts between the 1 minute and 5 minute time points (Fig. 5C). Interestingly, the kinetic analysis of TRITC-labelled dextran-3000 internalization shows that the process is exponential, starting to level off after about 30 minutes of labelling (Fig. 5C). This might reflect the fact that the fluorescence signal is becoming saturated and, for this reason, we measured the clearance of the endocytosed material in pulse-chase experiments. When the clearance of endocytosed marker was followed by incubating the cells first for 10 minutes with the TRITC-labelled dextran-3000 before a chase period of 15 minutes, 30 minutes or 60 minutes in medium without label, the half-life of the tracer inside of the osteoclasts was determined to be 22 minutes (Fig. 5E), demonstrating the fast transcytotic activities inside these cells.
Kinetics of dextran endocytosis. For uptake kinetics, osteoclasts plated on dentine were incubated for 1 minute, 5 minutes, 10 minutes and 30 minutes with TRITC-labelled dextran-3000 before fixation and immunodecoration with antibodies against the αvβ3 integrin to identify the cells (A-C). To follow the trafficking of the label through the osteoclast, cells plated on dentine were labelled for 10 minutes with TRITC-labelled dextran-3000 before a chase period of 15 minutes, 30 minutes or 60 minutes in medium without label (D,E). Computer-assisted 3D reconstruction of x-y sections taken at a 0.5 μm step size was used to visualize the trafficking of the dextran through the osteoclasts (A,B,D, red indicates TRITC-labelled dextran and blue indicates αvβ3 integrin). At early time points, the dextran is exclusively found close to the bone surface and the pit (A,B) but, after 60 minutes of chase, the label is distributed throughout the cell and can be observed close to the basolateral membrane surface and outside the cell (D, arrowhead). The amount of fluorescence inside of the cells was determined using the Leica TCS-NT software package and expressed as arbitrary fluorescence units. Columns show mean fluorescence and s.e.m. (C,E).
Cytoskeletal requirements of dextran trafficking
In order to address the influence of the cytoskeleton on the uptake and trafficking of dextran in resorbing osteoclasts, we incubated cells settled overnight on dentine disks for 30 minutes with cytochalasin D, nocodazole or paclitaxel, before adding TRITC-labelled dextran-3000 for 5 minutes or 20 minutes in the presence of the inhibitors (Fig. 6). Cytochalasin D is a cell-permeable toxin that binds the plus ends of actin filaments, preventing their elongation and leading to their disassembly. Nocodazole binds to tubulin dimers, precluding their addition to growing microtubules; this lowers the effective tubulin dimer concentration below the critical concentration for tubule assembly and so leads to depolymerization. Paclitaxel, by contrast, binds to the microtubule filament and stabilizes it, thus inhibiting its disassembly.
Cytoskeletal requirements of dextran endocytosis and intracellular trafficking in osteoclasts. Osteoclasts settled overnight on dentine disks were incubated with vehicle (A), 10 μM cytochalasin D (B), 10 μM nocodazole (C) or 10 μM paclitaxel (D) for 30 minutes before the addition of TRITC-labelled dextran-3000 in the presence of the inhibitors. Representative lateral images of osteoclasts after 20 minutes of labelling are shown after immunodecoration with antibodies directed against the αvβ3 integrin (A-D, red indicates TRITC-labelled dextran and green indicates αvβ3 integrin); arrows indicate the position of the bone surface relative to the label. Scale bar, 10 μm. The number of osteoclasts with dextran endocytosis was counted and expressed as a percentage. Columns show mean percentage of osteoclasts with dextran endocytosis and s.e.m.
The number of osteoclasts with dextran endocytosis was counted after fixation and immunodecoration with antibodies directed against the αvβ3 integrin. None of the inhibitors reduced the number of osteoclasts with dextran endocytosis efficiently (5 minute incubations; cytochalasin D, 16.9±11.05% inhibition, n=332; nocodazole, 17±8.7% inhibition, n=348; paclitaxel, 13±4.5% inhibition, n=349) (Fig. 6E). After 20 minutes of labelling, a larger inhibitory effect was observed for the drugs affecting microtubules; this was similar for both microtubule-stabilizing agrents (34.2±4.4% inhibition, n=523) and microtubule-destabilizing agents (33.2±5.7% inhibition, n=296), indicating that endocytosis is dependent on microtubule turnover (Fig. 6E). Despite the supporting evidence for a role for F-actin in endocytosis in other biological systems, no significant reduction in the proportion of endocytosis-positive osteoclasts was observed in the cytochalasin-D-treated samples (12.4±10.9% inhibition, n=423) (Fig. 6E). Similar results were obtained when the experiments were carried out in the presence of 1 μM latrunculin B (data not shown), which causes actin depolymerization by a completely different mechanism from cytochalasin D. However, despite the limited effect that cytoskeletal inhibitors had on the number of osteoclasts undergoing endocytosis, the net amount of endocytosed material was reduced under all experimental conditions. When the fluorescence intensities inside treated cells were analysed and compared with the vehicle control cells, a 60% reduction in fluorescence intensity was observed for the cytochalasin-D-, nococdazole- and paclitaxel-treated cells, indicating that the efficiency of endocytosis and trafficking depends on intact microtubule and F-actin networks (data not shown).
The most striking observation was, however, made when the distribution of the label inside of the osteoclasts was monitored. In nocodazole-treated samples, no progression of the label from the area of resorption towards the cell centre was detected, even after 20 minutes of incubation (Fig. 6C). When the distance that the dextran had travelled from the resorption area after 20 minutes of incubation was measured, a 41% reduction was observed compared with the vehicle-treated cells (data not shown). By contrast, osteoclasts incubated with paclitaxel show a label distribution that is similar to that observed in the vehicle-treated control, indicating that an intact microtubule network, but not microtubule turnover, is necessary for the removal of endocytosed material from the area of the ruffled border (Fig. 6D). Upon cytochalasin-D treatment, the distribution of label was similar to that observed in control samples, demonstrating a limited influence of the actin cytoskeleton on intracellular dextran trafficking in osteoclasts (Fig. 6B).
pH dependence of dextran trafficking
The V-ATPase inhibitor bafilomycin A1 strongly inhibits bone resorption during short-term (30 minutes) and chronic (overnight) treatments (Hall et al., 1995; Sasaki et al., 1994; Stenbeck and Horton, 2000). Furthermore, bafilomycin affects endocytic trafficking in a range of cell types, and has been shown to interfere with recycling and transport to lysosomes and the trans-Golgi network, as well as fluid-phase endocytosis (Drose and Altendorf, 1997). To test whether the inhibitory effect of bafilomycin on bone resorption is due solely to inhibiting the acidification of the resorption area or whether there is an additional effect on endocytic trafficking, osteoclasts plated on dentine disks were pre-incubated for 30 minutes with either 25 nM bafilomycin A1 or vehicle control. TRITC-labelled dextran-3000 was then added for 5 minutes before fixation and immunodecoration with antibodies directed against the αvβ3 integrin. Representative images are shown in Fig. 7A,B. When the number of osteoclasts with dextran endocytosis was counted, a modest inhibitory effect was observed. Bafilomycin A1 at 25 nM decreased the number of endocytosis positive osteoclasts by 34.7±11.3% (n=752) (Fig. 7, table). Higher concentrations of bafilomycin (100 nM), which are commonly used to alter the pH of internal organelles, had a stronger inhibitory effect (data not shown).
The pH dependence of dextran trafficking. Osteoclasts were plated on dentine disks and incubated for 16 hours before a 30 minute pre-incubation with 25 nM bafilomycin A1 or vehicle control. TRITC-labelled dextran-3000 was then added for 5 minutes before fixation and immunodecoration with antibodies against the αvβ3 integrin. Representative single CSLM sections for cells treated with the vehicle control (A) or bafilomycin A1 (B) are shown; the outline of the cells as delineated by αvβ3 antibodies is depicted by a dashed white line. Scale bar, 10 μm. The number of osteoclasts presenting endocytosed dextran was counted and expressed as the percentage of endocytosis inhibition ± s.e.m. (Table).
Discussion
In this study, we devised a system to examine in detail the intracellular trafficking of endocytosed markers in bone-resorbing osteoclasts. During the non-resorptive phase, osteoclasts take up large amounts of external fluid, as can be seen in cells plated on glass that are incubated with inert fluid-phase markers. Once resorption has started, however, the uptake is restricted to the area of resorption. As a consequence, externally added fluid-phase markers are only endocytosed if they have access to this area. Fluid-phase markers taken up at the ruffled border follow a similar intracellular route as the digested bone matrix, as demonstrated by the colocalization of endocytosed TRITC-labelled dextran with labelled dentine. By contrast, only limited colocalization of these markers was observed with ligands taken up via receptor-mediated endocytosis at the remaining plasma membrane, indicating that the cross-talk between the different endocytic compartments in osteoclasts is low. However, the ruffled border receives material from both the recycling and the late-endosome/lysosome pathways, because both transferrin and LDL reached the ruffled border. We did not observe the secretion of transferrin or LDL at the ruffled border, although this might be a result of technical limitations. In other cell types, it has been shown that the leading edge of the cell is the preferred exit site for recycling molecules; in osteoclasts, it is likely that areas of high membrane turnover, like the ruffled border, are preferred exit sites for material endocytosed from the basolateral part of the cell, whereas material endocytosed from the ruffled border exits the cell at the side opposite the bone surface. Even though we cannot estimate the amount of dextran that is directly recycled back to the plasma membrane in osteoclasts, a tracer half-life of 22 minutes does not fit with the kinetics observed for molecules entering the recycling pathway, nor with molecules that are targeted back to the plasma membrane from the lysosome. This further indicates that dextran takes an alternative route through the osteoclast. Furthermore, we could not detect colocalization of dextran with the early-endosome marker EEA1 (data not shown), indicating that the amount of dextran in this compartment at any given time is low. This might reflect the fact that maturation of the endocytic compartment is a rapid event in osteoclasts.
The detailed analysis of the transcytotic pathway in the time-course experiments showed that dextran is first detectable close to the resorption area and, with longer incubation times, is then distributed throughout the osteoclast. The 3D reconstruction of dextran localization shows that, after 60 minutes of chase, there is still label in the pit but that most of the staining inside of the osteoclasts has moved from the resorption site towards the upper part of the cell. This observation fits well with the proposed release of the endocytosed material at a specialized exit site at the plasma membrane opposite the bone surface (Salo et al., 1996). The half-life of the label inside of the osteoclasts was estimated to be 22 minutes, which would sustain the speed requirements of the large membrane turnover during bone resorption.
Rapid transport of organelles and vesicles is mediated by the microtubule network, whereas the slower transport occurs via actin filaments (Goode et al., 2000). Our data demonstrating the strong inhibitory effect of the microtubule-disrupting drug nocodazole on dextran trafficking underlines the crucial role of the microtubule network in transport from the ruffled border and substantiates the hypothesis of Väänanen's group that matrix uptake is mediated by microtubules (Mulari et al., 2003). Surprisingly, all cytoskeleton-disrupting drugs had limited influence on the number of osteoclasts showing dextran endocytosis, and their inhibitory effect only became apparent when the amount of endocytosed dextran was measured. Experiments measuring fluid-phase endocytosis in macrophages treated with colchicine, a drug that (like nocodazole) disrupts microtubules, showed a similar extent of uptake reduction (Piasek and Thyberg, 1980), suggesting that microtubules exert kinetic control on endocytosis.
Inhibition of bone resorption by bafilomycin A1 requires ten times lower concentrations of the drug than are needed to inhibit intracellular organelle acidification in other systems (Palokangas et al., 1997). This might be explained by the fact that the V-ATPase inserted into the ruffled border plasma membrane is more accessible to the drug or might reflect a hypersensitivity of the osteoclast-specific isoform of the V-ATPase to bafilomycin A1 (Drose and Altendorf, 1997; Li et al., 1999). Our data show that bafilomycin A1 at low concentrations affects not only the acidification of the extracellular space but also endocytic trafficking from the ruffled border. A recent study of the effect of bafilomycin A1 on transferrin uptake and dextran fluid-phase endocytosis in macrophages and osteoclasts plated on glass indicated that low concentrations of the drug (5-10 nM) can inhibit dextran fluid-phase endocytosis, whereas receptor-mediated endocytosis of transferrin was stimulated (Xu et al., 2003). Contrary to this study, we observed a much milder inhibitory effect on dextran endocytosis in osteoclasts plated on dentine and only higher concentrations of the drug, known to inhibit intracellular organelle acidification in other systems, showed a similarly strong inhibitory effect. These differences highlight the absolute substrate dependency of osteoclast behaviour and stress the sensitivity of osteoclast function to the maintenance of adequate levels of organelle acidification, indicating a possible dual mechanism for the inhibitory action of bafilomycin on bone resorption.
An important question for our understanding of the process of bone resorption is the nature of the receptor responsible for bone-matrix uptake. There is the attractive possibility that the αvβ3 integrin functions as the endocytic matrix receptor, because of its high expression levels and its localization to the ruffled border. Some integrins, like the αvβ3 integrin, are endocytosed and recycled back to the plasma membrane, making them ideal candidates for this function (Bretscher, 1996; Lawson and Maxfield, 1995). In osteoclasts, the p85 subunit of Class I PI-3-kinase has been shown to be part of the integrin signalling complex and has been localized to the ruffled-border membrane (Hruska et al., 1995; Nakamura et al., 1997). Interestingly, when we used the PI-3-kinase inhibitors wortmannin and LY290042, a strong inhibitory effect on endocytosis was observed (65% inhibition with 100 nM wortmannin, 55% with 10 μM LY290042; data not shown). There is, however, evidence that wortmannin treatment of osteoclasts also interferes with the progression of Golgi-derived vesicles towards the plasma membrane and that it abrogates resorption without disrupting the adhesion structures of the actin ring (Nakamura et al., 1997; Stenbeck and Horton, 2000). Thus, other wortmannin-sensitive PI kinases that are not part of the integrin signalling complex, such as Class III enzymes, might interfere with endocytic uptake and the regulation of osteoclast bone resorption via an integrin-independent mechanism.
A large proportion of integrin internalization is via clathrin-dependent endocytosis and there is evidence that the necessary machinery is localized to specific regions of the ruffled border (Altankov and Grinnell, 1995; Memmo and McKeown-Longo, 1998; Mulari et al., 2003). Our finding that low-molecular-weight fluid-phase markers can efficiently enter the pathway taken by digested bone-matrix proteins does not necessarily contradict this hypothesis. However, dextran is first observed in the area surrounded by filamentous actin, an area described as devoid of clathrin (Mulari et al., 2003). Additionally, we could not detect any colocalization of dextran with clathrin (data not shown) and our inhibitor studies support a pinocytic uptake mechanism. A non-clathrin-dependent mechanism of bone-matrix endocytosis would have the advantage that the size of the formed vesicle is not restricted, thus allowing the rapid internalization and recycling of large portions of ruffled border. A detailed analysis of this process has to await the possibility of performing live cell imaging of integrin turnover in bone-resorbing osteoclasts.
Acknowledgements
We thank G. Schiavo and B. Leitinger for critical reading of the manuscript, and members of the Horton laboratory for helpful discussions. G.S. is an ARC postdoctoral fellow and M.A.H. is a recipient of a Wellcome Trust program grant.
- Accepted October 9, 2003.
- © The Company of Biologists Limited 2004
References
Article navigation
Related articles
Cited by...
More in this TOC section
Similar articles
Other journals from The Company of Biologists
Cell scientist to watch – Mahak Sharma
“It's important to keep a focus in the initial few years and to show that you're productive.”
Mahak Sharma and her lab at the Indian Institute of Science, Education and Research investigate vesicular trafficking machinery and its appropriation by intracellular pathogens. She shares her journey from India to the USA and back again to set up her own lab, and her thoughts on why doing your homework pays off at conferences.
Have you seen First Person, our new interview series with early-career researchers? In this issue, we catch up Olivia Susanto from the Beaton Institute in Glasgow, UK.
Cell science at a glance – Actin assembly mechanisms at a glance
In this new poster and accompanying article, Eugen Kerkhoff and colleagues summarise and discuss the current knowledge on the actin cytoskeleton and associated motor proteins and their modes of action in actin nucleation, elongation and force generation.
Hypothesis – A rim-and-spoke hypothesis to explain the biomechanical roles for cytoplasmic intermediate filament networks
Rudolf E. Leube and colleagues hypothesise that cytoplasmic intermediate keratin filaments comprise a rim-and-spoke arrangement supporting the plasma membrane and connecting to the nucleus, to deliver their mechanosensory functions.
Articles of interest in other journals
From Development:
DNA damage induces a kinetochore-based ATM/ATR-independent SAC arrest unique to the first meiotic division in mouse oocytes
Simon I. R. Lane, Stephanie L. Morgan, Tianyu Wu, Josie K. Collins, Julie A. Merriman, Elias ElInati, James M. Turner, Keith T. Jones
Development 2017 144: 3475-3486; doi: 10.1242/dev.153965
Afadin orients cell division to position the tubule lumen in developing renal tubules
Lei Gao, Zhufeng Yang, Chitkale Hiremath, Susan E. Zimmerman, Blake Long, Paul R. Brakeman, Keith E. Mostov, David M. Bryant, Katherine Luby-Phelps, Denise K. Marciano
Development 2017 144: 3511-3520; doi: 10.1242/dev.148908
Transfer to Biology Open
If your submission to Journal of Cell Science is unsuccessful, did you know you can transfer your paper and any reviews directly to our sister journal, Biology Open? The majority of papers transferred with reviews are accepted for publication. Find out how here.