Introduction
Osteoporosis, a major public health problem for over 200 million people around the world [
1], is a skeletal condition characterized by low bone density and microstructural degradation of bone tissue, which together increase the risk of fracture [
2]. Drugs such as bisphosphonates (BPs), denosumab, vitamin preparations, calcium preparations, parathyroid hormone (PTH) agents, and anti-sclerostin antibodies are used to treat osteoporosis [
3]. Although BPs are the most widely used agents in the treatment of osteoporosis [
4], the suppression of bone resorption by a BP also leads to a subsequent decrease in bone formation. Decreased bone turnover in bone metabolism leads to a decline in bone quality and increases the risk of atypical fractures [
5]. Bisphosphonates were also suggested to be involved in the development of medication-related osteonecrosis of the jaw [
6].
The limited duration of a regimen for bone-forming agents such as PTH and anti-sclerostin antibodies poses several issues, including the decrease in bone density after the treatment period and the constraints imposed by the treatment schedule. There is a strong demand for the development of new medications for osteoporosis that can overcome these limitations.
Osteoclasts resorb bone by producing cathepsin K, a powerful cysteine proteinase that disassembles collagen [
7]. When cathepsin K is inhibited by E-64, a cysteine proteinase inhibitor, bone resorption is suppressed [
8,
9]. In cathepsin K knockout mice, bone resorption is inhibited and bone formation is enhanced due to an increase in sphingosine 1-phosphate (S1P) produced by osteoclasts [
10].
Based on these findings, the cathepsin K inhibitor odanacatib (ODN) was developed for the treatment for osteoporosis. ODN reduced bone turnover in the lumbar vertebrae of surgically menopausal cynomolgus monkeys, suppressing trabecular bone resorption and increasing bone mass [
11]. The addition of ODN to human osteoclasts significantly reduced the release of CTX-I [
12]. Clinical trials of ODN reported increases in bone density, reductions in fracture risk, and the promotion of bone formation [
13]. Although the development of ODN was discontinued due to an increased risk of stroke observed in a phase 3 trial [
14], the findings obtained are valuable for future studies. ODN and BPs are both bone resorption inhibitors, but they have different impacts on subsequent bone formation. Detailed analyses of the effects of these drugs on bone remodeling could enhance our understanding of bone metabolism and thus contribute to the development of novel treatments for osteoporosis.
However, it is challenging to analyze bone remodeling at the cellular level in vivo. Bone remodeling is a process that spans weeks to months, and its analysis requires observations of bone resorption followed by bone formation at the same site. Histological analyses and bone morphometry provide information from a single time point, and it is impossible to analyze the spatiotemporal relationship between bone resorption and formation, as well as cell interactions. Long-term observation is challenging with in vivo imaging, due to the invasiveness of the observation process. Moreover, the limited vertical resolution in the imaging of the bone marrow makes it difficult to accurately evaluate the bone matrix.
To solve these problems, our team developed an in vitro system that reconstructs the bone cell network involving osteoclasts, osteoblasts, and osteocytes within the mineralized nodule, enabling the visualization of bone modeling and remodeling phenomena by 2-photon microscopy [
15]. We used this system in the present study to investigate the effects of E-64 and zoledronic acid (ZOL), a bisphosphonate, on bone remodeling over a long term.
Discussion
We compared the effects of different bone resorption inhibitors on bone remodeling using an experimental system that replicates the network of osteoblasts, osteoclasts, and osteocytes with the extracellular bone matrix in vitro. In the control group, the resorption and refilling of matrix were clearly indicated by the disappearance and reappearance of the SHG signal at the same regions (Figs.
1b, c,
7d). The quantification of SHG volume also demonstrated the balanced decrease and increase in the matrix (Fig.
2b). The increase in the tdTomato-positive region during the formation phase represents the maturation of osteoclasts (Figs.
1b,
2d, e), accompanied with increase in lysosomes [
24] and mitochondria [
25] to degrade intracellular and extracellular material [
24] and aid in the cell’s recovery [
25], respectively. On the other hand, osteoclasts decreased in the formation phase (Figs.
1b,
2d, S4a), probably due to osteoclast apoptosis. However, we have not examined the mode of cell death. In this in vitro system with limited nutrition, it is possible that not only apoptosis but also necrosis due to energy depletion may be involved.
In contrast, in the ZOL group, temporal changes in matrix were not as obvious as those in other groups (Fig.
1b, c). The quantification also revealed that bone resorption and the subsequent formation were suppressed (Fig.
2b, c). Osteoclasts were rarely seen (Figs.
1b,
2d, e), which agreed with previous reports showing that bisphosphonates cause the apoptosis of osteoclasts [
2,
26] and inhibit the differentiation and activity of osteoclasts [
27]. It has been proposed that BPs reduce the bone remodeling rate indirectly by suppressing bone resorption, while a study of 2021 demonstrated that BPs directly affect the osteoblasts [
28]. Although our present data are not sufficient to determine whether BPs affected the osteoblasts directly or indirectly, the suppression of matrix resorption and subsequent formation by a BP was replicated in our system.
In the E-64 group, the resorption pits seemed smaller and shallower than those in the control group, and refilling occurred in an unsynchronized manner (Figs.
1b, c, 8f, g). Although there was no significant difference compared to the control group, the amount of SHG reduction was smaller (Fig.
2b, c). E-64 decreases the resorption amount and resorption area by osteoclasts in vitro [
29]. The treatment of osteoclasts with cysteine protease inhibitors resulted in the formation of abnormal pits filled with demineralized but un-degraded matrix on the surface of ivory slices [
30]. Cysteine proteinase inhibitors were also reported to suppress osteoclastogenesis [
21]. Our present findings also demonstrated the suppression of multinucleation of osteoclasts by E-64 treatment (Fig.
1e, g, Suppl. Fig S3c, d). Interestingly, small osteoclasts tended to persist in the formation phase in the E-64 group (Figs.
1b,
2d, e). Because larger osteoclasts are more prone to undergo apoptosis than smaller ones [
31], suppression of multinucleation may have suppressed the apoptosis of osteoclasts in E-64 group. Another possible explanation is that suppression of bone resorption that consumes large amounts of adenosine triphosphate produced by glycolysis and oxidative phosphorylation [
32] may have suppressed apoptosis. Regarding bone formation, cysteine protease inhibitors have been shown to suppress osteoblast proliferation, differentiation, and functions [
29,
33]. This study also revealed the suppression of the mineralized nodule formation, though not significantly (Suppl. Fig. S3e, f). The E-64 treatment resulted in a matrix volume that was comparable to that of the control group at F3, by reductions of both resorption and formation (Fig.
2b, c).
When we analyzed each parameter in the 16 ROIs per field of view, we observed that resorption and formation were moderately correlated in the control group as previously [
18], indicating the topological and quantitative coordination between resorption and formation (Fig.
3c, d). The osteoclast volume and osteoblast volume were strongly correlated with both resorption and formation, while the osteoclast volume and osteoblast volume were moderately correlated (Figs.
4a,
5a,
7a). In addition, there were correlations between the number of nuclei and volume of osteoclasts, and the average volume of osteoclast and bone resorption capacity in the control group (Fig.
4c,
1h), which agreed with a previous study indicating the correlation of number of nuclei and resorptive activity of osteoclasts [
34].
On the other hand, in the ZOL group, resorption and formation correlated only weakly, and the correlation among osteoclasts, osteoblasts, and matrix was also reduced. Moreover, only negligible correlation was detected between resorption and formation in the E-64 group, suggesting the loss of the topological and quantitative coordination. This result may not be attributable only to the suppression of matrix formation by osteoblasts, because the effect on the nodule formation was greater in the ZOL group than in the E-64 group (Suppl. Fig. S3e, f). In the E-64 group, the correlation between the osteoclast volume and resorption was preserved to some extent, whereas the osteoclast volume and formation seldom correlated (Fig.
4a).
The correlations between matrix resorption/formation and the osteoblast volume at some time points were moderate in the ZOL group, while only weak correlations were observed in the E-64 group (Fig.
5a). The correlations between the osteoclast volume and osteoblast volume were negligible in the ZOL group, although there were some weak correlations in E-64 group (Fig.
7a). These data suggested that the link between osteoblasts and osteoclasts or matrix was disturbed in the E-64 group.
It is noteworthy that the osteoblast sphericity at R2 was moderately correlated with formation (Fig.
6c). We observed many cuboidal osteoblasts in the resorption sites (Fig.
7d, g). The cuboidal osteoblasts, which are rich in rough endoplasmic reticulum and Golgi apparatus, are deemed as mature osteoblasts which produce bone matrix [
35]. We have also reported that cuboidal osteoblasts with high sphericity indicate mature osteoblasts [
17]. However, in this study, considering the time points (R2), the location, and morphology, we hypothesized that cuboidal osteoblasts could potentially be reversal cells, a precursor of osteoblasts that promotes the transition from the bone resorption phase to the bone formation phase [
36]. The reversal cells also secrete matrix metalloproteinases to remove undigested collagen remnants and prepare the bone surface for subsequent bone formation [
36,
37]. Further studies are needed to determine the characteristics of these cells.
Meanwhile, this moderate correlation between formation and osteoblast sphericity almost disappeared in the E-64 group and the emergence of spherical osteoblasts at the resorption site was suppressed, which may be the cause of the loss of the topological and quantitative coordination of matrix resorption and formation (Figs.
6b, c,
7g). Possible explanations for the loss of spherical osteoblasts are as follows. (1) E-64 treatment suppresses the secretion of coupling factors such as Cthrc1 [
38] and C3a [
39] from osteoclasts directly, or indirectly through the inhibition of cathepsin K leading to some negative feedback effects on osteoclasts. (2) Few collagen remnants that required cleaning by reversal cells were formed in the resorption pits because of the suppression of cathepsin K. This leads to the reduction in the reversal cells appearing at the resorption sites. (3) E-64 suppressed the expression and/or function of receptors for coupling factors such as S1P and ephrin B2 in osteoblasts [
40]. A determination of which of these possibilities is correct will contribute to our understanding of the mechanisms of coupling.
A major issue in this study is whether these observations in our system reflect in vivo phenomena. Few studies have evaluated the effect of wide-spectra cysteine proteinase inhibitors in vivo [
41,
42]. For example, the administration of E-64 suppressed lipopolysaccharide-induced bone resorption [
42]. In contrast, the effects of specific inhibition for cathepsin K have been studied in several species of animals. An osteoclast-specific deletion of cathepsin K suppressed bone resorption and increased the formation of cancellous bone [
10]. The cathepsin K inhibitor SB-553484 suppressed the resorption of cancellous bone and promoted the formation of cortical bone [
43].
ODN suppressed the bone turnover of the lumbar spine of Rhesus monkeys while maintaining the osteoclast number, and increased bone mineral density [
11]. ODN treatment reduced the trabecular and intracortical bone formation rate (BFR) and increased the endocortical BFR and periosteal BFR in femurs [
44]. The cathepsin K inhibitor ONO-5334 and alendronate preferentially increased the cortical bone mass and the trabecular bone mass, respectively [
45].
In human studies, ODN reduced both resorption and formation markers [
46]. Histomorphometric analyses revealed that ODN reduced osteoblast parameters and eroded surfaces but increased the osteoclast number later [
47]. Together the above-cited studies suggested that the inhibition of cathepsin K suppresses bone resorption and subsequent formation in the cancellous bones to some extent while promoting bone formation in the cortical bones. In this study, E-64 suppressed both resorption and formation to a lesser extent than ZOL, which does not contradict the in vivo findings. However, few studies have addressed the effect of cathepsin K inhibition on coupling.
Jensen et al. demonstrated that some osteoclasts were detached from the bone surface and shallow resorption pits were increased in ODN-treated animals [
23], as this study (Fig.
1c). However, they also observed an increase in the reversal cells and cuboidal osteoblasts by ODN treatment, whereas we observed a loss of spherical osteoblasts in the resorption pits (Figs. S6a,
6a,
7f, g). The differences in findings may be attributable to the difference in the spectra of E-64 and cathepsin K-specific inhibition, and/or to the animal species used. To confirm whether our system can reproduce the phenomena that occur in vivo, the effects of E-64 on bone remodeling and coupling should be examined in a mouse model.
Another issue is that there is also a difference in oxygen concentration between this in vitro system and the in vivo environment. Mature osteoclast and osteoclast precursors are under an oxygen tension of 5% in vivo, a lower level than that of the atmospheric conditions [
48,
49]. Both osteoclasts and osteoblasts behave differently under oxygen tensions of 5% and 21%, as to proliferation, differentiation, and migration [
50].
In conclusion, our in vitro system demonstrated not only the quantitative and topological correlation between bone resorption and bone formation, suggesting the existence of coupling, but also the correlations between osteoblasts and osteoclasts and between matrix resorption/formation and these cells. We also observed that the inhibition of cysteine proteinase disrupted the coordination between resorption and formation. Although in vivo evidence remains to be obtained, our present findings provide valuable insights for a deeper understanding of coupling mechanisms and the development of innovative therapies targeting osteoporosis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.