In vitro model of bone to facilitate measurement of adhesion forces and super-resolution imaging of osteoclasts

Abstract

To elucidate processes in the osteoclastic bone resorption, visualise resorption and related actin reorganisation, a combination of imaging technologies and an applicable in vitro model is needed. Nanosized bone powder from matching species is deposited on any biocompatible surface in order to form a thin, translucent, smooth and elastic representation of injured bone. Osteoclasts cultured on the layer expressed matching morphology to ones cultured on sawed cortical bone slices. Resorption pits were easily identified by reflectance microscopy. The coating allowed actin structures on the bone interface to be visualised with super-resolution microscopy along with a detailed interlinked actin networks and actin branching in conjunction with V-ATPase, dynamin and Arp2/3 at actin patches. Furthermore, we measured the timescale of an adaptive osteoclast adhesion to bone by force spectroscopy experiments on live osteoclasts with bone-coated AFM cantilevers. Utilising the in vitro model and the advanced imaging technologies we localised immunofluorescence signals in respect to bone with high precision and detected resorption at its early stages. Put together, our data supports a cyclic model for resorption in human osteoclasts.

Figure 1

Thin, smooth and optically accessible 3D-culture model for bone biology research made from bone particles. AFM image of human bone particles (a) used in bone coating dried on a coverslip in low density. Appearances of bone particle coated glass coverslips (b) with 6 different coating buffer compositions: 1) 10 mM PO₄, 0.05% BSA and 0.025% Triton X-100, 2) ethanol, 3) ion exchanged water, 4) 10 mM PO₄ and 0.1% milk, 5) 10 mM PO₄, 0.1% milk and 0.1% Triton X-100, and 6) 10 mM PO₄ and 150 mM NaCl. A cross section image of bone particle coating acquired with laser scanning confocal microscope demonstrating the uniform thickness of the layer (c). Data was captured by setting a detector to capture reflected light, with excitation wavelength 476 nm. An AFM image captured in acoustic oscillation mode illustrates smoothness of the densely coated bone layer (d). Electron micrographs of bone coating at different magnifications, showing smooth surface (e) and type-I collagen fibril (f). The presence D-bands on the collagen fibrils (arrowheads in f) suggest that also the organic components of the bone are present and folded in native conformation. Scale bar 1 μm in (a), scale bar 10 μm in (c,d), 400 nm in (e,f).

Figure 2. OC morphology on different substrata.

Peripheral blood-derived OC morphology on (a) glass, (b) vitronectin coated glass, (c–e) sawed human bone slice and (i–k) human bone particle coating. On the glass-based substrates (a,b) the OCs are typically larger with more OCPs fused together and actin exhibits predominantly PD-structures, organised in rafts and rings. On the bone-derived substrates (c–n) OCs are induced to resorb bone that can be visualised from the presence of RPs. The average size of OCs is smaller and actin exhibits more complex patterns: wide actin belts represent SZs, actin patches (APs) with high but uneven phalloidin staining were aligned with RPs and small circular structures <5 μm in diameter (SAR). Bone marrow derived CD34+ OCPs on sawed human bone slice (f–h) and human bone particle coating (l–n) exhibit high resorption activity as demonstrated by a number of RP on the substrata. For both of the features PDs and SZs with both blood- and bone marrow derived OCs grown on the different substrates, the representative features were selected from N > 20 examples. Scale bar 20 μm.

Figure 3. Super-resolution examination of phalloidin-stained actin structures in human OCs exhibiting SZs or APs.

Panels (a–f) show SZ-structure at different magnifications captured with wide field (a,b), confocal (c,d) and STED (e,f) microscopes. STED+ shows deconvolved data. SZ exhibits a uniform dense actin staining with closely-knit structure of long parallel and crossing filaments. Whereas, AP (g–m) has a more network-like appearance with occasional brighter puncta. Here length of straight filaments is shorter and the network has more holes than in SZ. In high magnification of the deconvolved STED images from AP (j–m), on most of the bright puncta visualised we noticed a small hole that may correspond to an endocytosis event where a vesicle is budding from RB membrane by branched actin polymerisation. The representative SZ and AP was selected from N > 20 observed cases. Scale bar: 50 μm in (a), 10 μm in (b,c), 5 μm in (d,e), 1 μm in (f), 2 μm in (g,h), 0.5 μm in (i), 0.2 μm in (j–m) and 2 μm in (n–o).

Figure 4. There are fewer nuclei in resorbing cells grown on bone coating than in multinuclear cells grown on glass.

We counted the nuclei on three separate bone coated coverslips or clean coverslips, and found that on bone coating the OC precursors are less likely to fuse into large cells with many nuclei. In OCs associated with resorption pits, we typically found one nucleus tightly coupled within the actin structure, indicating a more active role in the transcription required in the resorption process. Resorbing peripheral blood-derived OCs on human bone slice (a,d) and human bone particle coating (b,e) with actin (green) and nucleus (red). Actin in (a) shows SZs with and in (b) AP both with a nucleus (arrows) tightly coupled with the resorptive zones. On glass (c,f) the nuclei are located high above the plane where PDs are seen. There is a significant difference in the number of nuclei in resorbing cells cultured on bone coating and on glass (g). For both cases N > 100, to be counted on glass the cell needed to have more than one nucleus and on bone coating it needed to express resorption specific features of be located on top of a RP. Scale bar 20 μm.

Figure 5. V-ATPase localisation in human OCs exhibiting APs, SZs and PDs.

F-actin is displayed in green, V-ATPase in magenta and reflection from bone in cyan. (a–d) a 2.5 μm maximum projection of OCs above RPs with ongoing active resorption show an accumulation of V-ATPase at the bone interface. Actively resorbing cells are devoid of V-ATPase filled vacuoles, but have smaller puncta representing acidified endosomes and lysosomes. A 2.5 μm maximum projection (e–h) of an OC displaying a SZ structure, and a medium intensity of V-ATPase in the middle of the SZ. Further away from the bone interface are seen vacuoles filled with V-ATPase, along with endosomes and lysosomes. In a 5 μm maximum projection of OCs grown on glass (i–k) no accumulation V-ATPase was observed at the glass interface, in abundant endosomes and lysosomes and V-ATPase filled vacuoles were a typically seen stored inside the cell (bottom center). Boxes in (b,f,j) outline the field of view in the merge images (d,h,k). Arrowheads point to V-ATPase stained vacuoles. We selected the representative images from 18 V-ATPase stained SZs and N > 20 APs and PDs. Scale bar in (a,e,i)is 10 μm and in (d,h,l)5 μm.

Figure 6. Dynamin localisation in human OCs exhibiting APs, SZs and PDs.

(a–d) RP (bone surface in cyan) coincided with AP structures exhibiting a network of filamentous actin (green) and dynamin (magenta) staining. (a) actin at AP-structures was organised in form of intensely stained discs of networked actin with a number of bright puncta. A reflection signal image from the bone coating (cyan) where dark patches signify indentations i.e. RPs. (b) dynamin (mainly dynamin 2) showed staining throughout the plasma membrane and an intense staining at AP-structures, indicating specific accumulation at these zones and vesicular transport. (d) an overlay image of dynamin and actin shows dynamin localising at the puncta in actin network. An OC grown on bone coating exhibiting a SZ (e–h), does not show accumulation of dynamin at the SZ or at its center. An OC grown on (i–k) glass shows PDs and end an even low-intensity dynamin signal throughout the cell. We selected the representative images from 10 dynamin stained SZs and N > 20 APs and PDs. The scale bars are: 5 μm in (d), 20 μm in other images.

Figure 7. OC adhesion energy measurements with bone coated cantilevers in an AFM force spectroscopy setup.

Electron micrographs of BSA coated (a) and human bone particle coated (b) tipless AFM-cantilevers. (c) shows a schematic diagram of an adhesion measurement that illustrates the steps of a force distance cycle. Adhesion force curves of peripheral blood-derived OCs on glass with BSA coated (d) and human bone particle coated (e) cantilevers. The shaded areas in the curves represent adhesion energies and the lowest point of the curve represents the maximum adhesion. (f) Comparison and statistical analysis of adhesion energy between BSA and bone coating at 3 time points, 30 s, 120 s, 300 s. Values are means ± SE (n = 5 with BSA and n = 6 with bone coated cantilevers, assessed by Mann-Whitney test). Scale bar 10 μm in (a,b).

Figure 8. Schematic illustration of the proposed mechanism of resorption cycle and vesicular transport during the cycle.

Actin encapsulated vacuoles filled with V-ATPase and proteases are transported to plasma membrane forming the RB area. Filamentous and dynamic branched actin forms the observed AP structures at RB during this process. The vacuoles are likely to be acidified and when fusing with plasma membrane they initiate a rapid decrease of pH in the microenvironment. The V-ATPases remain on the membrane and may maintain the acidic microenvironment for a while. During this process begins endocytosis and subsequent removal of the resorption products. This process needs both dynamin and Arp2/3, and hence, the forces driving the process seem to be generated by branched actin polymerisation. After the active phase stops V-ATPase and both low level of actin remain on the plasma membrane and the resorption may be re-initiated at the same location by fusion of more V-ATPase filled acidified vesicles. The mechanism we propose describes resorption as a dynamic process that relies bursts of acidified vesicles and explains the emergence of clusters of small RPs.