Trabecular bone organoids: a micron-scale 'humanised' prototype designed to study the effects of microgravity and degeneration

Abstract

Bone is a highly responsive organ, which continuously adapts to the environment it is subjected to in order to withstand metabolic demands. These events are difficult to study in this particular tissue in vivo, due to its rigid, mineralised structure and inaccessibility of the cellular component located within. This manuscript presents the development of a micron-scale bone organoid prototype, a concept that can allow the study of bone processes at the cell-tissue interface. The model is constructed with a combination of primary female osteoblastic and osteoclastic cells, seeded onto femoral head micro-trabeculae, where they recapitulate relevant phenotypes and functions. Subsequently, constructs are inserted into a simulated microgravity bioreactor (NASA-Synthecon) to model a pathological state of reduced mechanical stimulation. In these constructs, we detected osteoclastic bone resorption sites, which were different in morphology in the simulated microgravity group compared to static controls. Once encapsulated in human fibrin and exposed to analogue microgravity for 5 days, masses of bone can be observed being lost from the initial structure, allowing to simulate the bone loss process further. Constructs can function as multicellular, organotypic units. Large osteocytic projections and tubular structures develop from the initial construct into the matrix at the millimetre scale. Micron-level fragments from the initial bone structure are detected travelling along these tubules and carried to sites distant from the native structure, where new matrix formation is initiated. We believe this model allows the study of fine-level physiological processes, which can shed light into pathological bone loss and imbalances in bone remodelling.

Fig. 1. The development of an analogue bone system with equivalent anatomical, biophysical and biochemical characteristics.

a Bone tissue (exemplified using a C. elaphus sp. antler) is composed of a network of trabecular rods, packed together to become compact bone (left). This tissue is heavily mineralised, composed of Calcium Phosphate (left), with little protein content (right), where sulphated content is detected in marrow cavities. b The femoral head is adapted to withstand great mechanical loads such as the upper body weight (BW), and is therefore composed of large amounts of trabecular bone (black box). Each trabecula is adapted at the micron level to withstand forces acting from multiple directions. c To ensure an anatomically relevant structural base for a bone organoid, thermally treated micro-trabecular particles from femoral heads were used (500–1000 μm). These particles present the lamellar structure (d, box) of bone and surface topography (e) that is essential for normal cell sensing and attachment. They also have a relevant biochemical composition, comprised entirely of a calcium-phosphate phase (Micro-XRF) (f). Traces of sulphur indicative of organic matter can be detected, however, at the limit of the detection threshold. X-ray diffraction analysis (g) confirmed these are composed of a biologically derived phase of hydroxyapatite (red bars), the mature bone mineral. h The micro-trabeculae are naturally highly electrostatic (left), which was exploited for insertion into liquid cell suspension droplets (right) to generate miniaturised bone avatars. A hanging-drop culture system was used to suspend trabeculae and primary female bone effector cells and to direct attachment onto the trabecular surface through gravitational sedimentation (h, i). These were cultured for 48 h to maximise surface colonisation and self-organisation of the cells. In the absence of bone scaffolds, osteoblasts come into direct contact with each other to form a spheroid (j, left), whereas in the presence of a trabecula they successfully populate the surface and display osteogenic phenotypes (right). Primary female osteoclast precursors are also able to attach individually and co-cultured with osteoblasts, generating a complete remodelling unit (k). Their presence can be differentiated on trabeculae as osteoblastic nuclei are larger than osteoclastic ones. Scale bars as indicated.

Fig. 2. Maturation of the human osteoclastic constructs and simulation of the specialised bone resorption activity encountered in vivo.

Human osteoclast precursor cells of monocytic origin (a) are incubated in the presence of pro-osteoclastic factors RANKL and M-CSF to generate a population of multi-nucleated osteoclasts, which display the ruffled morphology indicative of their active phenotype (c, black arrow). The fusion of monocytes into polykaryons can be observed in constructs as well after 8 days of culture (d left and zoomed, right), where they are seen agglomerating on the surface of trabeculae. Some of the formed structures are delimited (dotted lines, arrows) to allow a better visualisation. At the interface with the scaffold, these cell aggregates form belts (e, top and bottom) that are rich in tubulin and actin (white arrows and dotted line). When cultured for 2 weeks, these cells can create perforative resorption patterns resembling pores and lacunae (f, black box and g, red arrows). Scale bars as indicated.

Fig. 3. Maturation of the human osteoblastic constructs and simulation of the osteoid producing events encountered in vivo.

Human osteoblastic cells cultured in non-osteogenic (control) medium for 14 days (a) display an unchanged osteoblastic phenotype, whereas cells cultured in osteogenic medium for this period display a highly dendritic phenotype within an interconnected network, indicative of an osteocytic phenotype (b). Constructs seeded with cells and cultured on non-adherent surfaces for 7 days are able to recapitulate in vivo events such as osteoid production. New matrix is produced, which can be morphologically distinguished from the compact bone structure (b, left, white arrows). Cavities within trabeculae are used as a sites where new matrix formation takes place (b, right), and dendritic cells are seen embedded within this matrix (c, 1). The surface of constructs is also a site of osteoblastic-osteocytic activity, with large projections extending across the trabeculae (c, 2, 3). In addition, communication and aggregates (red arrow) can be observed at the surface, which are rich in tubulin and actin (d). e In contrast, constructs that were cultured in the absence of trabecula, including 7 days inside a perfused rotating vessel, develop a lobular embryoid-type morphology. Scale bars as indicated.

Fig. 4. Simulating bone loss in constructs and the effects on local mechanics.

Resorption of the bone surface was investigated on trabeculae from different cell-seeded groups after 8 days of culture to detect the manifestations in each type of condition. Cell-free constructs and those that were cultured with osteoblasts show only signs of endogenous resorption, originating from the native bone tissue (a-1,2 and 3,4, respectively). Osteoclast-cultured constructs show a significant pattern of resorption (a-5, 6), which was observed in mixed constructs as well (a-7, 8), developed with both osteoblastic and osteoclastic cells. Dark boxes = areas of no resorption; red boxes = areas of resorption. The mechanical properties of these groups were investigated by subjecting the cultured constructs to compressive forces and observing the fracture patterns, related to their trabecular distribution (b), and changes in maximum force required to collapse the structure, a feature related to their integrity (c). The values selected for comparison, for which representative examples are presented in b with orange arrows, were chosen as the maximum compressive force that can generate the ultimate structural failure. The least-resistant organoids were those that contained mixed osteoblasts-osteoclasts and osteoclasts alone (**p = 0.01 and *p = 0.03, respectively, n = 3), which required significantly less force to reach terminal failure. Data is presented as mean ± SD. The resorptive patterns encountered in mixed constructs simulating a whole remodelling unit (d) were reticulate in appearance, similar to those encountered in human trabecular bone from femoral heads. Figure e is taken from Gentzsch et al. by permission from Springer Nature. Scale bars as indicated.

Fig. 5. Changes in cell resorption observed with exposure to simulated microgravity.

a To simulate a reduction in the force of gravity that the femoral head trabeculae are normally adapted to resist, particularly when upright (left), organoids were placed either in static, typical culture conditions, where they experience the normal gravitational pull, or in dynamic, rotational conditions, inside a NASA-synthecon reactor vessel, where a laminar movement of the culture medium suspends constructs in a state of orbital buoyancy (right). During this culture model, the forces acting on the trabecular organoids, resulting from the gravitational force (F_g), centrifugal forces (F_c) and hydrodynamic drag forces (F_d) (b) balance each other during each orbital revolution (c), thus preventing sedimentation and simulating weightlessness. Following 6 days of culture under these conditions, resorption induced by cells can be observed in osteoclastic and osteoblastic-osteoclastic constructs (d, red arrows and box), which appears longitudinally extended in morphology (e). This resorption pattern resembles a second class of lacunae encountered in trabecular bone from human femoral heads (f) and is different structurally from the type encountered in constructs cultured under static conditions (g, h). Extracts from the incubation medium of constructs subjected to simulated microgravity were taken as pellets (i), or immuno-purified from liquid phase (j). In both cases, large protein complexes were detected between 60 and 70 kDa, containing an association of the target protein, native cell and medium proteins (j right, black rectangles). These regions were probed with antibodies during western blotting (i), or the protein was immuno-selected using magnetic isolation (j, left). The protein Sclerostin (i) was detected in osteoblastic samples, but additionally in osteoclast and mixed samples/medium. This pattern was also observed in concentrated medium samples (j, left—immuno-selected samples and right—all samples including medium). RANKL (i) was not expressed in osteoblast constructs, in the absence of osteoclasts. RANKL was also detected in concentrated medium samples, in all but osteoblast samples (j). The endocrine protein PTHR1 and integrin CX43 (i) were only detected in osteoclastic constructs. Figure f is taken from Gentzsch et al. by permission from Springer Nature. Scale bars as indicated.

Fig. 6. Encapsulation of trabecular organoids into biochemical microenvironments that can generate bone physiological or pathological prototypes.

a and b present two examples each of constructs that were encapsulated into a physiological (human fibrin) and tumorigenic environment (Matrigel^®). Constructs that were encapsulated in fibrin (day 19) create extensive projections (orange arrows) into the surrounding environment (a-top), which stretch from multiple points on the surface (middle). These constructs are heavily populated with cells (bottom). Constructs embedded in Matrigel^® (day 16) also display highly proliferative cells, however, the pattern and morphology of these cells is more representative of a tumouroid, as they are densely packed into atypical phenotypes (pink arrows). These lobular proliferative regions have been delimited with a dotted white line for easier visualisation. c The chemical composition of fibrin-embedded constructs and nature of projections was probed using X-ray fluorescence mapping to allow for spatial detection of new features, with different elements coding for distinct anatomical structures. The projections emerging from the construct surface (grey arrows and black arrows) contain a large amount of Phosphorus (P). Calcium (Ca) is also present in the newly forming structures. Potassium (K), an intracellular ion present in cells is seen within these dendritic structures (pink arrows) and co-localised with the Phosphorus deposits, however, discrete deposits of P are found on their own surrounding the K structures (green arrows). Areas rich in Sulphur (S) are also seen localised to the regions where projections and matrix are forming (white arrows). Organoids were scanned with Micro-CT as a fused mass (d) to detect compositional gradients based on density. Owing to the dense nature of the trabecula, organoids behave like bone tissue when subjected to X-Ray radiation and as such, display selective attenuation at the surface (the beam hardening effect). Therefore, beam artefacts were removed and a function was applied to remove the lower density ranges of data (e). White arrows show discrete pockets of a denser (blue) material in different constructs in the near vicinity or emerging from the very dense (red) surface, and trapped within the lowest density (green), surrounding fibrin material. Scale bars as indicated.

Fig. 7. From bone loss to bone remodelling and fracture repair.

Mixed-cell organoids were investigated following 23 days of total culture, including 5 days inside a simulated microgravity vessel under two rotating conditions (simulated Micro G and Normal G Dynamic). a When imaged using brightfield microscopy, areas of significant bone loss (red squares) can be observed at various locations adjacent to the organoid surface in both conditions. Large tubular projections (black arrows) can be observed emerging in both conditions from the original organoid and into the surrounding fibrin matrix. Some dense mineral-tubular structures (T) are seen forming from the surface (top panel, white dotted lines). These projections carry mineral (red arrows) to sites away from the main structure. These structures can reach ~1 millimetre, as observed in this simulated microgravity-cultured construct (b). These large projections were chemically mapped at high resolution using Micro-X-Ray Fluorescence and were shown to contain a mixture of protein and mineral content (c, d), where the elements Sulphur, Phosphorus and Calcium were co-localised in both conditions along the length of these tubule formations (d). e Whole maps of constructs revealed intricate networks of mineralised matrix projections also containing cells in both conditions (coloured arrows) and differences could be assessed morphologically. In these particular examples selected, a construct from the normal gravity condition (top) contains a higher number of projections compared to the simulated microgravity construct (bottom), which contains a main tubular structure forming into the matrix, but which displays many potentially emerging structures, based on the distribution of potassium (K), in networks surrounding the initial structure. Molecularly, a non-quantitative protein assessment of constructs revealed that the Sclerostin and RANKL presence was pronounced within each condition compared to that of the endocrine Parathyroid Hormone Receptor 1 and to the shear-responsive Connexin 43 channels (f). The presence of PTHR1 in the modelled microgravity group was not clearly detected, however, it was present in the normal gravity condition, which may indicate differences in the regulatory pathway controlling the bone remodelling process between the two contexts.

Fig. 8. Evolution of constructs and potential applications of the model.

a Osteoblast constructs that were previously cultured under low-shear conditions for 7 days and subsequently seeded into fibrin matrices, are able to form micro-trabecular shaped soft structures during 3 weeks of culture and to generate nucleation points at sites far away from the initial organoid structure (red squares). This new matrix shows a mixture of protein and inorganic content and a phosphorus-led mineralisation (b). c It is possible to quantitatively assess the degree of new bone formation by measuring the ratios of elements from the initial construct (yellow shape) compared to the overall organoid (green shape) to assess the degree of change. d In these examples from osteoblastic constructs that were either cultured in a static condition (top) or simulated microgravity (bottom) for 7 days and subsequently developed in fibrin for 3 weeks, a rate of remodelling (e) could be calculated from the acquired maps by normalising the average elemental content of the organoid to that of the initial trabecular construct. Alternatively, in the example provided in f (left) from a previous organoid, regions of interest can be compared within or across samples, provided that the surface area selected is constant (f, right). As shown, the differences between Area 1, containing a visibly larger amount of new matrix compared to Area 2, can be successfully translated numerically based on the elemental content. Owing to the small diameter of organoids, constructs can be used with more automated systems to determine the effects of various agents (g). Change can be assessed by recovering the organoids and analysing them volumetrically (individually or as a mass) with technologies such as micro-CT, which facilitate a density-based selection (as exemplified here using a function selecting the highest density mineral, i.e., the initial bone). Further tests can be applied, such as illumination with polarised light, which will vary based on the orientation of crystallographic axes within the construct, linked to the degree of new bone formation. Additional colorimetric and fluorescence-based assays specific to bone tissue can be used in conjunction. Scale bars as indicated.