In this study, we aimed at developing a biomineralized collagen membrane that allows the differentiation of osteoclasts from human peripheral blood monocytes and osteoclastic resorption by depositing a CoCaP coating on the mineralized collagen fibrils of the membrane.
1. CoCaP coating increases stiffness of biomineralized collagen membrane
Biomineralized collagen membranes (BiominCol in figures) were prepared using a modified PILP method as described previously [14, 15]. The membranes had a fibrillar surface structure, as observed by SEM (Fig. 1A). Incubation of the membranes in a CaP solution containing Co2+ did not lead to apparent changes of the surface structure, and no mineral deposits were observed (Fig. 1B). This result was unexpected, as previous studies have shown that CaP nodules (spherulitic mineral aggregates) were deposited on tissue culture plastic upon immersion in CaP solutions, with or without Co2+, similar to the one used here.[17, 18] Moreover, it has been shown that CaP mineral deposits formed on collagen scaffolds upon incubation into concentrated SBF solutions [19–21]. TGA results showed that the mass remaining after combustion of the organic phase of the membranes increased from 67% (w/w) for the uncoated collagen membrane to 71% (w/w) for the CoCaP-coated one (Fig. 1C), indicating that the CoCaP coating slightly increased the mineral content of the membrane. A few aspects related to the properties of the biomineralized collagen membrane and the CaP coating method used here may explain why this extra mineral deposition, originating from the coating, was not observed on SEM images. First, the surface area of the mineralized collagen membranes was much higher than that of cell culture plates used in the earlier studies,[17, 18] which means that the same amount of mineral that would normally deposit on a flat surface could now form on many more available nucleation/growth sites, resulting in the formation of smaller mineral particles. Second, the CoCaP coating step provided less total mineral than, for example, the PILP method used for intrafibrillar collagen mineralization; the CoCaP coating solution had about ten times lower ion content than the PILP solution, due to the lower reaction volume used, as described in the methods section. This may explain why there was no noticeable fiber diameter increase after the coating step (uncoated BiominCol at 188 ± 45 nm and CoCaP-coated BiominCol at 156 ± 30, n = 20), in contrast to the observed fiber diameter increase following the intrafibrillar mineralization.[14, 15] Finally, and perhaps most importantly, the biomineralized collagen membranes used in our study already contained intrafibrillar CaP mineral before immersion into the CoCaP coating solution, which may have interfered with nucleation and growth of CaP deposits on the fibrous collagen surface. We propose that CoCaP coating preferentially crystallized on the pre-existing intrafibrillar CaP. This seems plausible based on earlier studies that compared the rate of hydroxyapatite (HA) crystal growth on collagen, elastin and HA surfaces. These studies showed that the growth rate of HA on the HA surface was several orders of magnitude higher than on the polymeric substrates.[22–24] It should be noted that we tried to incorporate Co2+ intrafibrillarly, by adding the ion to the PILP solution. This was, however, unsuccessful as no intrafibrillar mineral was formed. This result is in accordance with previous reports in which inhibitory effects of cations such as Cu2+ [10] and Sr2+ [11] on the PILP-induced biomineralization were observed.
Taken together, based on the SEM characterization of the structural properties of mineralized collagen membranes before and after immersion in the CoCaP coating solution, there was no evidence for the deposition of the mineral on the surface; however, it is suggested that the new mineral deposited on the pre-existing intrafibrillar mineral, which could not be observed using SEM imaging. This hypothesis was supported by the TGA data, presented above, as well by the results of the ICP-MS analysis of the Co2+ content of coated membranes, measured for different initial concentrations of Co2+ in the solution and presented as the number of Co2+ per 106 Ca2+ (Fig. 1D). The ICP-MS results showed that increasing concentrations of Co2+ in the CaP coating solution, caused corresponding increases in the amount of Co2+ detected after dissolution of the coated membranes in acid, which confirmed that Co-containing mineral was deposited in the membrane. For a starting concentration of 1 µM Co2+ there were 20 Co2+ per million Ca2+. This formulation was chosen for all other experiments in this study. CoCaP-coated collagen membranes showed a continuous release of Co2+ in water over a period of 16 days (Fig. 1D), with an average release of approximately 400 ppt per refreshment period (2–3 days), which is equivalent to a molarity of around 7 nM.
Nanoindentation measurements further showed that the stiffness of the CoCaP-coated collagen membrane (about 10 MPa) was about one order of magnitude higher relative to the uncoated membrane (Fig. 1D), which is likely a result of the presence of additional mineral. It should be noted that both the uncoated and CoCaP-coated collagen membranes prepared here were significantly less stiff than the biomineralized collagen membrane reported in an earlier study (177 MPa).[15] Although a similar biomineralization method was used in both studies, differences in fiber density and density of crosslinks between the materials obtained in the two studies may explain different mechanical properties.
The TGA, ICP-MS, and nanoindentation data showed that the CoCaP coating process changed the physicochemical properties of the substrate, by incorporation of Co in its composition and by increasing its stiffness, without significantly altering the surface morphology of the biomineralized collagen membrane.
2. Osteoclasts differentiate from monocytes and show actin rings on CoCaP-coated mineralized collagen membranes, but not on uncoated membranes
Monocytes derived from human peripheral blood were cultured on uncoated and CoCaP-coated biomineralized collagen membranes, and stimulated to differentiate into osteoclasts by the addition of 40 ng/mL RANK-L to basic cell culture medium. After 7 days in differentiation medium, a few large and multinucleated cells were observed on the uncoated membranes (Fig. 2A), but not on CoCaP-coated membranes (Fig. 2B). TRAP-positive cells were not observed on either of the membranes after 7 days, but appeared on both membranes after 14 days, showing concentrated actin structures (Fig. 2C-D). These actin structures were dispersed through the cell and did not form clear actin rings. After 21 days of culture in differentiation conditions, a few osteoclasts on the CoCaP-coated membranes exhibited circular, high-intensity actin structures that resemble actin rings (Fig. 2F). It should however be noted that the number of osteoclast with clearly defined actin rings was limited. These structures were not detected on the uncoated biomineralized collagen membrane (Fig. 2E).
In an earlier study, we have shown that biomineralized collagen membranes supported osteoclast formation. Nevertheless, the formed osteoclasts were unable to form stable actin rings or sealing zones, and were therefore incapable of resorption. Here, we observed for the first time osteoclasts with actin rings on the CoCaP-coated biomineralized collagen membranes modified. As was described in the previous section, the coated membranes contained additional CaP with Co2+ and had increased stiffness relative to the uncoated ones. Both modifications may have contributed to the difference in osteoclast phenotype observed, namely the capacity to form actin rings, but it is likely that the increased stiffness is the main factor. Few studies investigated the effects of Co2+ on formation and resorptive activity of osteoclasts in the similar concentration range as used here. In a study by Patntirapong et al., surface of tissue culture well plates was coated with thin CaP layers containing Co2+ using solutions with a Co2+ concentration of 0.1, 1, and 5 µM, respectively. These coatings, which were shown to release about 1, 10, and 50 ng of Co2+ in 500 µL cell culture medium, respectively over a period of 3 days, supported the formation of a higher number of murine osteoclasts (1 µM condition in particular), and larger resorbed areas (all three conditions).[9] For comparison, after 2-day incubation in water of the CoCaP-coated collagen membrane used in our study, the Co2+ concentration was around 0.4 µg/L (equivalent to around 7 nM), which is lower than the Co2+ released in the 0.1 µM condition of the above discussed study (2 ug/L). Another study showed that 10 nM of dissolved Co2+ had either no effect on the resorptive activity of forming human osteoclasts, or caused a slight drop in the resorptive activity of mature osteoclasts.[25] Taken together, the CoCaP-coated collagen membranes used in our study contained a lower amount of Co2+ than the studies in which a stimulatory effect on differentiation and resorptive activity of osteoclasts was observed. Nevertheless, differences in experimental set up (e.g. cell type) do not allow a direct comparison and therefore, an effect of Co2+ ions on osteoclast formation observed in our study cannot be fully excluded.
While low doses of Co2+ seem to be able to stimulate osteoclast resorptive activity, there is no information on whether they play a role in actin ring formation. Substrate stiffness, on the other hand, has been suggested to have an influence on actin ring formation. In a study where the formation of podosomes by osteoclasts cultured on polyacrylamide and polydimethylsiloxane surfaces ranging in stiffness from 30 kPa to around 1800 kPa was investigated, it was shown that podosome belt formation was possible in this range of stiffness, as long as integrins were activated (e.g., by coating the substrate surfaces with either collagen or vitronectin).[26] These findings suggested that actin ring formation can occur, to a certain extent, independently of substrate stiffness, if all other conditions (e.g. integrin activation) are met. In another study, it was shown that osteoclasts were also able to form actin rings on collagen-coated coverslips, but not on the surface of collagen gels.[27] This study suggested that the mechanical properties and/or the porous, mesh-like surface of the gel impeded actin ring formation.
In our study, we have shown that biomineralized collagen, with stiffness of around 1000 kPa, did not allow actin ring formation on its surface. Since biomineralized collagen contains ligands for β1 integrin activation (but no ligands for the vitronectin receptor), and stiffness within the range that allows actin ring formation, it is suggested that the porous structure was the inhibiting factor for actin ring formation. However, when stiffness was increased up to 10 MPa by coating the membrane with CoCaP, actin rings were sporadically observed, which suggests that the inhibiting effect of the fibrillary surface on actin ring formation can be at least partially overcome by increasing the substrate stiffness.
4. Presence of Co and increased stiffness stimulate limited resorption of the coated biomineralized collagen membrane
Resorption of uncoated and CoCaP-coated biomineralized collagen membranes was evaluated by measuring the Ca2+ concentration in the cell culture supernatant, up to 21 days. This method allows for indirect quantification of resorptive activity, and is suitable for materials for which quantification of resorbed area or volume by conventional imaging methods is difficult, for example because of the high roughness or porosity of the substrate as in the case of collagen membranes. Nevertheless, a limitation of this method is a relatively low sensitivity; as the Ca2+ concentration in medium varies with many processes associated with cell culture, only major differences in resorptive behavior can be quantified. For example, when biomineralized collagen is immersed in cell culture medium in the absence of cells, there is a depletion of Ca2+ from the medium (data not shown). Nevertheless, the substrate-based effects are accounted for in the results as a comparison is only made between cells cultured in basic versus differentiation medium, on the same substrate. Therefore, any difference in Ca2+ concentration should be a result of cell differentiation and resorptive activity.
On the uncoated membranes, no significant differences in Ca2+ concentration were observed between cultures in basic and differentiation medium, except at the last time point of 21 days, when there was a significantly higher Ca2+ content in the basic medium condition. This is an unexpected finding, as no Ca2+ release is expected to occur from the substrate in the absence of cells, or from activity of macrophages. On the CoCaP-coated membrane, there were significant differences in Ca2+ concentration between basic and differentiation medium at 2, 9 and 16 days, with differentiation medium having a higher Ca2+ content. While it is unlikely that resorption occurred as early as on day 2, as osteoclasts were not observed on the coated membrane even after 7 days (Fig. 2), it is plausible that the higher Ca2+ content in differentiation medium at 9 and 16 days was due to resorption of the substrate. However, these results must be interpreted with caution, due to the fact that only a few resorbing osteoclasts were observed and that this technique is not optimal for determining small differences in osteoclastic resorption.
To further investigate whether the differences in Ca2+ concentration were due to osteoclast resorption, the same uncoated and CoCaP-coated membranes for which the ICP-MS analysis of the medium was performed, were observed by SEM, after removing the cells (Fig. 4). On uncoated biomineralized collagen membranes, no obvious signs of resorption were observed. There were areas on the membrane where the fibrillar structure appeared to be disturbed, which probably correspond to places of cell attachment. On the CoCaP-coated membranes, the same areas with disrupted fibrillar structure were also present, but interestingly, some of these areas also exhibited resorption lacunae-like features, i.e., pockets indented in the surface of the biomaterial. These were of the same size as the actin rings observed by confocal laser microscopy. The observation of actin rings by fluorescence microscopy, together with SEM observations of resorption lacunae, and increased Ca2+ in cell culture medium after 9 and 16 days all point towards events of osteoclast resorption taking place on the CoCaP-coated biomineralized membrane, although to a limited extent. Another interesting observation was that all resorption lacunae observed using SEM had a pit-like structure, while no trench-like resorption lacunae were observed.
In the previous section, we discussed the influence of substrate stiffness on actin ring formation by osteoclasts on different types of substrate. To our knowledge, no studies exist that investigated the effect of stiffness on osteoclastic resorption, especially in the ranges of kPa to MPa, which would be useful to compare to our own data. From various resorption studies on bone, dentine, or CaP ceramics, it is evident that in-vitro resorption on surfaces with higher stiffness is common.[28–30] We have also performed differentiation and resorption experiments on porous β-TCP ceramic disks to confirm that the differentiated osteoclasts were able to resorb this substrate (Supplementary Fig. 1). The stiffness of all these substrates is in the GPa range, thus orders of magnitude higher than the biomineralized collage substrates used in our study. Although resorption studies on polymer-CaP composites exist,[31–33] and these materials can have lower stiffness, the resorption results in these studies are commonly not discussed in the context of the mechanical properties of the substrate.
Taken together, the results of this study showed that coating a biomineralized collagen membrane with a layer of Co-containing CaP had a twofold effect on the membrane properties, i.e. the stiffness of the membrane was increased by an order of magnitude, to 10 MPa, and Co was added to the substrate (Fig. 1D, 1F). Additional CaP (Fig. 1C) that was added by the coating process was involved in both effects by, on the one hand stiffening the membrane and on the other acting as a carrier of Co. Interestingly, no obvious effects of the additional coating was observed on the fibril thickness and hence the porosity of the membrane (Fig. 1A, 1B). Osteoclasts, differentiated from human peripheral blood monocytes, were able to form on the coated membranes, and were also capable of resorbing it, albeit to a limited extent. This is supported by the observation of actin ring formation using fluorescence microscopy (Fig. 2F), as well as by the observation of pit-like formations on the membranes, using SEM after cell removal (Fig. 4B, 4D). The higher concentration of Ca2+ in cell culture medium of differentiated cells at days 9 and 16 also support the few, and sporadic resorption events observed (Fig. 3B). Although the effect of Co2+ ions, which were present at a low concentration (Fig. 1D), cannot be excluded, we hypothesize that the main contributor to the ability of osteoclasts to resorb the coated membrane was the increase in membrane stiffness (though still significantly lower than that of bone or dentin slices), that allowed sealing of a compartment for resorption, partly overcoming the resorption-inhibiting effect of the large pores.
The main aim of this study was to modify the biomineralized collagen membrane in such a way that it can be resorbed by osteoclasts by adding Co2+ ions to the material. As discussed above, since the method we had to use to achieve this changed more than only the chemical properties of the membrane, the limitation of this study is that we cannot conclusively provide evidence for which of these properties is responsible for the observed biological response. To this end, additional experiments separating the effects of stiffness and chemistry are required.