Introduction
Bone diseases, such as osteoporosis and most cancer metastases to bone, are characterized by loss of bone mass and increased fracture risk. This is largely due to excessive bone resorption compared to bone formation during bone remodeling. These skeletal diseases have triggered the development of a series of inhibitors of osteoclasts, the cells responsible for bone resorption. These inhibitors comprise estrogen, selective estrogen receptor modulators (SERMs), calcitonin, bisphosphonates such as alendronate (ALN), as well as the newly developed denosumab [
1]. Although these antiresorptive drugs are demonstrated to improve bone mineral density and to reduce fracture risks in patients, they are also known to reduce bone formation as a consequence of the coupling between resorption and formation during bone remodeling. Recent efforts have aimed at finding alternative treatments based on stimulation of osteoblasts, the bone forming cells. These have brought intermittent parathyroid hormone (PTH) therapy onto the market and anti-sclerostin into clinical trials [
2]. However, despite their clinical efficiency, the bone formation induced by these anabolic agents is not strictly linked to the sites of bone resorption, and they may not reestablish the site-specific bone resorption–bone formation balance that is lost in pathological bone remodeling [
3,
4].
Odanacatib (ODN) is a newly developed inhibitor of osteoclastic bone resorption presently in a clinical phase III trial. It is unique compared with the standard antiresorptives because of its ability to inhibit bone resorption while keeping bone formation generally unaffected or even stimulated at specific bone sites, as supported by observations in clinical and preclinical studies with ovariectomized (OVX) monkeys and rabbits [
5‐
7]. ODN is therefore considered to represent a new generation of bone resorption inhibitors that may bring alternative benefits to osteoporotic patients compared to the current standard of care. The mechanism whereby ODN inhibits bone resorption is well understood [
8,
9]; however, how ODN preserves bone formation requires further investigations.
Osteoclasts remove bone by secreting protons and cathepsin K, a potent collagen-degrading proteinase [
10]. The protons solubilize the mineral, thereby denuding the collagen fibers, which make up 90 % of the organic bone matrix. Once denuded, collagen becomes available to cathepsin K. ODN is a specific inhibitor of cathepsin K and inhibits thus the resorptive activity of the osteoclast by inhibiting collagen degradation [
8,
9]. This means that ODN inhibits the very last step of the resorption process, while other anti-osteoclastic drugs, like bisphosphonates and anti-RANKL antibody, act more upstream. We speculated that by uniquely inhibiting this very last step of the resorption process, ODN could also uniquely modify the signals left in the resorption lacunae vacated by the osteoclasts and therefore also modify the cellular events which follow immediately after osteoclastic resorption, such as osteoblast recruitment. These events are obviously important in connecting bone resorption and formation. We thus hypothesized that ODN might affect bone formation—in contrast to classical anti-osteoclastic drugs—by acting on this connecting step, known as the “reversal phase” [
11]. Interestingly, inhibition of cathepsin K by a broad-spectrum cysteine proteinase inhibitor induced unusual features at the reversal phase in a mouse model [
12], and similar observations were made in a rabbit model [
13] and in pycnodysostosis, a disease caused by a mutation in the gene encoding cathepsin K [
12]. It is worth noting that a drug affecting the reversal phase may be of special relevance to osteoporosis because this disease is precisely characterized by a lack of coordination between resorption and formation, and prolonged or arrested reversal phase has been reported in estrogen deficiency– and age-related bone loss [
11,
14]. Although these reports point to a failure at the level of the reversal phase, a phase linking resorption and formation has remained largely undercharacterized [
11,
14].
The present study aimed at identifying what differentiates ODN from classical anti-osteoclastic drugs, such as ALN, in the OVX rabbit model [
7]. We histomorphometrically compared their effects on the target cell, the osteoclast, and on the post–osteoclastic events of the reversal phase. It is a follow-up of a previous histomorphometric study performed on OVX rabbits, where ODN was shown to differ from ALN in its ability to preserve bone formation [
7]. The recent report on this earlier study strictly concerned the extent of mineralization surfaces and bone formation rates and did not show data either on the resorption or on the reversal phase, which we here hypothesize to determine the subsequent bone formation levels.
Discussion
It is well known that ODN and ALN have in common to be inhibitors of osteoclastic bone resorption, but that ODN differs from ALN and other traditional antiresorptives by its ability to preserve bone formation. The present study emphasizes a series of additional distinctive properties of ODN. These relate to bone resorption itself, the reversal phase, and osteoblast recruitment. Interestingly, the effects of ODN on the latter two shed insight into how ODN may favor bone formation.
Regarding resorption, our data show that treatment with ODN clearly changes the geometry of the resorption cavities compared to controls and a classical antiresorptive, ALN. Upon ODN treatment, a larger surface is eroded but erosion is less deep. The latter characteristic is in accordance with the repeated demonstration of shallower excavations in the pit assay in response to lowered levels of cathepsin K activity [
28‐
31]. It is of interest that ODN thereby mimics the effect of estrogen, which reduces both cathepsin K expression [
32,
33] and the depth of the pits [
30]. As discussed by the latter authors, a shallower geometry may help in preventing perforations of trabeculae; and it was also shown to favor bone stiffness [
34]. Thus, a unique property of ODN is to allow bone remodeling while simply reducing its rate and changing the geometry of the package of matrix, which is renewed with less risk of fragilization. This effect on geometry is probably a result of ODN slowing down collagen degradation without directly affecting demineralization. The rate of collagen degradation relative to the rate of demineralization is a major determinant of the shape of the resorption lacunae, as discussed by Soe et al. [
35]. It should be noted that others also reported that ALN has no effect on erosion depth [
23].
The reversal phase has been proposed to prepare the eroded bone surfaces for bone formation [
12], and reversal phase arrest prevents subsequent bone formation [
11,
12,
14]. The reversal phase is thus a key for successful remodeling. As explained below, ODN shows unique effects on the reversal phase compared to ALN and control: (1) it shortens it, thus shortening the duration between the end of resorption and the initiation of bone deposition, and (2) it favors osteoblast recruitment.
In earlier studies, the duration of the reversal phase was estimated by relating the extent of the reversal surface to the extent of the osteoclast surface, and it was shown that reversal surfaces relative to osteoclast surfaces were smaller in situations where resorption is known to be ideally coupled with formation, like in primary hyperparathyroidism [
11,
14]. In contrast, it was shown to be larger in situations of less efficient coupling, like in osteoporosis [
11,
14]. In the present study, the ODN-induced shortening of the reversal surface relative to the osteoclast surface thus points to an efficient coupling and faster initiation of bone formation in the presence of ODN compared to all other conditions investigated here. This interpretation is supported by the fact that bone deposition is more frequently seen in the neighborhood of reversal surfaces upon ODN treatment than in all the other conditions.
A major event occurring during the reversal phase is recruitment of osteoblasts to the bone surface. It should be noted that the result of this process is not well reflected by the extent of osteoblast surface since osteoblast-lineage cells may spread to a variable extent over the bone surface, where they may adopt either a flat or a cuboidal morphology. The result of osteoblast recruitment is thus better appreciated through cell density. Here, we show that ODN treatment favors higher osteoblast densities and more cuboidal osteoblasts compared to ALN treatment. It makes sense to believe that this higher cell density would result in higher bone formation rates and mineralizing surfaces, as previously reported by Pennypacker et al. [
7]. Interestingly, cell densities appeared already higher at the level of the reversal phase, which is in accordance with the view that reversal cells may differentiate into mature osteoblasts.
What is the mechanism whereby ODN treatment affects the reversal phase? Here, two issues should especially be emphasized. First, the signals left by the osteoclast in the resorption lacuna are probably also the signals seen by the cells colonizing these lacunae after the departure of the osteoclast. Since ODN acts on the very last step of the resorption process, ODN treatment is likely to affect these signals. Obvious changes due to the presence of ODN are less efficient degradation of organic matrix molecules, thus resulting in the accumulation of demineralized collagen; of other bone matrix molecules [
25]; as well as of matrix-associated growth factors [
36]. Collagen and growth factors, like transforming growth factor-β, have been shown to be chemoattractants for osteoblast-lineage cells [
37,
38]. Furthermore, electron microscopic studies of mouse and rabbit bone cultures performed in the presence of cathepsin K inhibitors showed numerous osteoblast-lineage cells in close contact with osteoclasts and enwrapping the demineralized collagen fibers that the osteoclasts themselves were unable to degrade [
12,
13]. The same osteoblast-lineage cells were also shown to deposit cement lines and collagen fibers in the vacated pits [
12,
39]. Here, we have extended these studies in several ways: we used a selective cathepsin K inhibitor, ODN; we quantified the interactions of the osteoclasts with the periosteoclastic cells; and we showed that ODN favors these interactions compared to ALN. Taking these observations together, a likely scenario contributing to increased osteoblast recruitment is that collagen leftovers and undegraded growth factors stimulate the recruitment and the activity of the cells colonizing the vacated resorption lacunae, which are the cells that subsequently differentiate into mature bone forming osteoblasts.
In relation with the mechanism whereby ODN affects the reversal phase and bone formation, one should also emphasize the impressive increase in osteoclast surface induced by ODN. Several other situations of increased osteoclast surface have been reported, which include impairment of c-Src and chloride channel 7 (ClC-7) activity [
40,
41]. Interestingly, all these situations have been associated with increased bone formation [
41,
42]. This effect is poorly understood and has led to the proposal that osteoclastic anabolic factors might be involved [
43]. One may also note that all these situations, including the ODN-induced increase in osteoclast prevalence, lead to an increase in the interface between osteoclasts and all cells surrounding the osteoclasts. Among the latter cells are not only reversal cells, as discussed above, but also (1) bone remodeling compartment canopy cells, which are osteoblast-lineage cells and likely contribute to the formation of mature bone forming osteoblasts [
16,
44], and (2) endothelial cells, which appear also to be involved in bone formation during remodeling of cancellous bone [
17]. The contributions of canopy and endothelial cells should be assessed in future investigations.
It is interesting that in the present model the positive effects of ODN and ALN on bone formation result in increased trabecular thickness but not in prevention of OVX-induced loss of trabeculae, as also reported elsewhere [
20]. Therefore, the observations reported here may concern surviving trabeculae, which have been found to correspond with the trabeculae submitted to strain [
45]. One should thus be aware that the ODN effects reported in the present study might be conditioned by mechanical loading.
In conclusion, so far treatments of bone diseases have focused mainly on bone resorption itself or bone formation itself. However, bone diseases may not simply be due to a direct failure of resorption or formation but result from a lack of coordination between resorption and formation. Here, we propose that ODN is a bone resorption inhibitor which acts positively on the mechanism coordinating resorption with formation, through recruitment of osteoblast progenitors.