The Bone Resorption Inhibitors Odanacatib and Alendronate Affect Post-Osteoclastic Events Differently in Ovariectomized Rabbits

The present study is a follow-up of the recent study reported by Pennypacker et al. [7]. In brief, 6-month-old female rabbits were randomized into four groups: (1) sham-operated (n = 11), (2) OVX and treated with vehicle (n = 12), (3) OVX and treated subcutaneously with ALN (300 μg/kg twice/week, n = 11), and (4) OVX and treated orally with ODN given in a control diet (Teklad global high fiber 2031C; Harlan, Indianapolis, IN) at a concentration of 0.0016 % (w/w) corresponding to a steady-state exposure of AUC = 4 μM/24 h of ODN (n = 12). Since the plasma protein binding is lower in the rabbit (87.3 %) than in humans (97.5 %), the free fraction of ODN exposure in this study was 0.51 μM/24 h, or approximately threefold of the human daily free fraction calculated from the 50-mg once-weekly clinical dose (6–7 μM/24 h). Treatment of the rabbits was started on the third day after ovariectomy and continued for 27 weeks, after which the animals were killed. For histomorphometry, lumbar vertebra L4 was removed, immersed in 70 % ethanol, and embedded undecalcified in 80 % methyl methacrylate–20 % dibutyl phthalate. For immunohistochemistry, bone specimens (L2) were obtained from a new study including three rabbits in each of the four groups. The four groups were treated as described above except for the ODN group, which was treated orally in the control diet at a concentration of 0.0043 % (w/w) corresponding to a steady-state exposure of AUC = 7.5 μM/24 h of ODN, and the ALN group, which was treated once weekly at a dose of 300 μg/kg. Bone specimens were fixed in 4 % formaldehyde, decalcified in 0.2 M EDTA-Na₂ containing 0.05 % formaldehyde, and embedded in paraffin.

Paraffin sections (3.5 μm thick) from L2 were processed for toluidine blue (1 %), tartrate-resistant acid phosphatase (TRAP) activity, and immunohistochemical stains. Toluidine blue staining was performed as previously described [6]. TRAP activity staining was done as described by van de Wijngaert and Burger [15], sections were counterstained with Mayer’s hematoxylin and mounted. For immunohistochemistry, the procedure included deparaffinization and rehydration of the sections, followed by epitope retrieval in Tris–EDTA buffer (pH 9.0) overnight at 60 °C. Immunostainings for cathepsin K were performed using a polyclonal goat IgG antibody (sc-6507; Santa Cruz Biotechnology, Santa Cruz, CA, USA) as primary antibody and a mouse anti-goat IgG (Jackson ImmunoResearch Laboratories, West Grove, PA, USA) as secondary antibody. The mouse anti-goat IgGs were detected with a polymeric alkaline phosphatase–conjugated system (PowerVision; ImmunoVision Technologies, Hillsborough, CA, USA) and visualized by liquid permanent red (DakoDenmark, Glostrup, Denmark). Sections were counterstained with Mayer’s hematoxylin and mounted.

Histomorphometric parameters were assessed in the cancellous bone of Masson-Goldner trichrome–stained sections (6 μm) prepared from the plastic-embedded L4 as described [7]. The parameters included the trabecular bone volume per tissue volume (BV/TV), trabecular thickness (Tb.Th), trabecular number (Tb.N), trabecular spacing (Tb.Sp), as well as the fraction of trabecular bone surface covered by eroded (ES/BS), osteoclast (Oc.S/BS), reversal (Rv.S/BS), osteoid (OS/BS), and osteoblast surface (Ob.S/BS). Reversal surfaces were defined as eroded surfaces (identified through broken lamellae) without osteoclasts. For every single hit on reversal perimeters the presence of osteoid in the vicinity was recorded. Vicinity was defined as being within the same 2D remodeling unit as the reversal surface itself. Furthermore, discrimination was made between cuboidal (C.Ob.S) and flat osteoblasts based on the cell morphology. Cuboidal osteoblasts were defined as cells with a cuboidal or plumb shape lining osteoid, and flat osteoblasts were defined as cells with a flat nucleus and thin cytoplasm lining osteoid. The trabecular bone volume, trabecular thickness, trabecular number, and trabecular spacing were measured at a total magnification of 100× using the Osteomeasure Analysis System (OsteoMetrics, Decatur, GA, USA), whereas all other histomorphometric analyses were done in a systematic random way at a total magnification of 200× using a light microscope (DM2500; Leica, Herlev, Denmark) equipped with an eyepiece containing a Merz graticule with parallel curved lines [16].

Osteoclasts were characterized in more detail in Masson-Goldner trichrome– and toluidine blue–stained sections. The total number of osteoclast profiles in the cancellous area of the vertebral body was assessed in Masson-Goldner trichrome–stained sections from each rabbit. Discrimination was made between profiles of osteoclasts attached to the bone matrix and profiles of osteoclasts detached from the bone matrix. Profiles of detached osteoclasts were categorized as either osteoclast profiles separated from the bone surface by a layer of mononucleated cells or osteoclast profiles found farther away from the bone surface, i.e., more than one cell layer from the bone surface or above bone forming osteoblasts. Regarding the attached osteoclast profiles, discrimination was made between profiles found in resorption cavities more than two lamellae deep and profiles found on noneroded or shallow cavities, i.e., less than or equal to a depth of two lamellae. This analysis was done at a total magnification of 200× using a light microscope (Leica DMRXAZ), where broken lamellae were visualized by polarized light. The presence of demineralized collagen at the osteoclast–bone interface and the presence of intracellular vesicles in osteoclasts were analyzed in Masson-Goldner trichrome– and toluidine blue–stained sections, respectively, at a total magnification of 400×.

Cell densities of reversal cells and osteoblasts were measured at a total magnification of 400× using the Osteomeasure Analysis System (OsteoMetrics). Here, all cell profiles connected to the bone surface were counted regardless of the presence of a visible nucleus. Samples for electron microscopic analysis were prepared as previous described [17] and analyzed with a transmission electron microscope (TEM 208; Philips, Eindhoven, The Netherlands) operated at 80 kV.

Data for each group were tested for normality using the D’Agostino-Pearson omnibus test. When the data allowed for parametric statistics, the unpaired t-test was used, applying the Welch correction in the presence of unequal variances. When nonparametric statistics were needed, the Mann–Whitney test was used. The correlation between two variables was analyzed using the Spearman rank correlation test. Best-fitted lines were made by linear regression including a comparison of the slopes of the regression lines. p < 0.05 was considered statistically significant. All statistical analyses were performed with GraphPad Prism 5 (GraphPad Software, La Jolla, CA,USA).

In agreement with other reports [18, 19], OVX significantly reduced bone volume and trabecular numbers, significantly increased trabecular space, and did not affect trabecular thickness, thus supporting estrogen deficiency–induced bone loss (Table 1). According to the present measurements, ALN or ODN treatment of OVX rabbits induced thicker trabeculae but did not prevent either the reduction in trabecular number or the increase in trabecular space, as was also reported in OVX monkeys treated with ODN [20].

Since both ODN and ALN target the osteoclast, we performed a systematic analysis of the extent of bone surface covered by eroded surface and by osteoclasts. Vehicle-treated OVX rabbits (OVX + Veh) did not demonstrate any significant changes in the extent of eroded surface or osteoclast surface compared to the sham group (Fig. 1), which is consistent with other long-term OVX studies [19]. Treatment of OVX rabbits with ODN resulted in a twofold increase in the extent of eroded surface and a threefold increase in osteoclast surface, whereas ALN did not induce any significant changes in these parameters. These observations are consistent with earlier reports on ODN effects in monkeys [6, 21] and on ALN effects in multiple species [22‐24].

Because osteoclasts responded differently to ODN and ALN, a more thorough examination of the treatment-related effects on osteoclasts was undertaken. Osteoclast profiles were generally found in contact with the bone surface in the OVX + Veh group (Fig. 2a). Although this was also the case for the majority of the osteoclast profiles in the ALN- and ODN-treated OVX rabbits, a substantial proportion of the profiles appeared detached from the bone surface (Fig. 2a, b). Upon quantification, the proportion of detached osteoclast profiles proved to be increased sixfold in ALN- and threefold in ODN-treated OVX rabbits compared to vehicle-treated OVX rabbits (Fig. 3a). The proportion of detached osteoclast profiles was thus two times higher in the group receiving ALN than in the group receiving ODN (Fig. 3a). Further analysis on the detached osteoclast profiles revealed that more than half of these profiles in sham, vehicle-, and ODN-treated OVX rabbits were separated from the bone surface by a single layer of mononucleated cells, i.e., reversal cells, which were found directly interposed between the osteoclast itself and the bottom of the resorption cavity (Figs. 2b, 3b). In contrast, the fraction of osteoclast profiles showing this type of contact with reversal cells was two times lower in the ALN versus ODN treatment. Instead, in the presence of ALN, the detached osteoclast profiles were more frequently found at ≥2 cell layers away from the bone surface or even above bone forming osteoblasts (Figs. 2a, 3b), thus suggesting that ALN treatment changes the adhesion and function of the osteoclasts more dramatically than ODN.

Another unique effect of ODN concerns the bone in contact with osteoclast profiles. In ODN-treated OVX rabbits, osteoclasts appeared about twice as frequently on noneroded surfaces or shallow cavities (≤2 lamellae deep) compared to the situation in the other groups, including ALN treatment (Figs. 2c, 3c). Hence, even though ODN treatment leads to an increased extent of osteoclast and eroded surfaces, erosion is not deep and the total amount of resorbed bone matrix should not necessarily be as significant as suggested by the extent of eroded surfaces.

Staining of undegraded, demineralized collagen frequently appeared at the osteoclast–bone interface in ODN-treated OVX rabbits (Figs. 2d, 3d), as was previously demonstrated in mouse and rabbit bones treated with a broad-spectrum cysteine proteinase inhibitor [12, 13]. Furthermore, ODN treatment induced an accumulation of toluidine blue–stained intracellular vesicles in osteoclasts (Figs. 2e, 3e), as reported earlier in ODN-treated monkeys and interpreted as accumulation of collagen fragments [6]. No stained vesicles were observed in sham, vehicle-, or ALN-treated OVX rabbits (Fig. 3e). These observations thus indicate compromised extracellular and intracellular degradation of demineralized collagen in the presence of ODN, in line with earlier reports on the effect of cathepsin K inhibition [8, 12, 25].

Interestingly, the immunoreactivity of cathepsin K appeared to be increased in ODN-treated OVX rabbits compared to sham, vehicle-, and ALN-treated OVX rabbits (Fig. 2f), which suggests that increased expression/intracellular accumulation of cathepsin K is induced by ODN treatment. This observation is consistent with earlier observations obtained in various experimental systems [8, 26, 27].

The reversal surfaces are defined as eroded surfaces vacated by osteoclasts [11]. They are the surfaces that physically connect bone resorption to formation, meaning that they may play a critical role in the initiation of bone formation during bone remodeling. Because ODN and ALN affect osteoclasts differently, it is important to analyze whether this difference has an impact on the reversal phase. We first analyzed the extent of reversal surfaces in the four groups of rabbits (Fig. 4a). Ovariectomy of the rabbits (OVX + Veh) had no significant effect on the extent of reversal surfaces compared to sham. Compared to OVX + Veh rabbits, ODN treatment increased the extent of reversal surface more than twice, whereas ALN treatment induced a widespread distribution of the values, with no significant effect being found (Fig. 4a). Furthermore, when relating the extent of reversal surface to the extent of osteoclast surface measured in each biopsy, a clear positive correlation was observed in all groups, meaning that the more bone surface is covered by osteoclasts, the larger is the extent of reversal surfaces (Fig. 4b). Interestingly, however, as indicated by the slope of the regression lines, the increases in reversal surface were significantly smaller in the ODN group compared with all of the other groups, suggesting that osteoid deposition starts sooner. In line with this, reversal surfaces were more than twice as frequently close to osteoid surfaces in the presence of ODN than in the presence of ALN (Fig. 4c). Furthermore, compared to all of the other groups, cell densities on the reversal surfaces tended to increase in ODN-treated OVX rabbits, which was the only group showing a significantly increased reversal cell density compared with sham (Fig. 4d). Altogether these observations indicated that ODN and ALN affected the reversal phase differently and thereby also the connection of resorption and formation. Thus, compared to ALN, ODN appeared to allow higher bone surface cell densities and a faster initiation of the bone formation step during bone remodeling.

In the previous report on the comparative effects of ODN and ALN in OVX rabbits [7], mineralizing surfaces and bone formation rates were shown to increase in response to OVX and to be higher upon treatment of the OVX rabbits with ODN compared with ALN. Here, we analyzed additional bone formation parameters. Figure 5(a, b) shows that the extent of osteoid and osteoblast surfaces also increased in response to OVX but that they were not changed upon further treatment with ODN or ALN. In this respect, one should note that the measurements in the ALN group were widely distributed, with values ranging from approximately 2–87 %, thus revealing variable effects of ALN on the extent of osteoid and osteoblast surfaces.

Interestingly, however, a more detailed analysis demonstrated clearly distinct effects of ODN and ALN on osteoblasts. Compared with ALN, ODN led to a threefold higher prevalence of cuboidal osteoblasts (Fig. 5c), a morphology reflecting a higher cell density. In line with this observation, ODN also led to significantly higher osteoblast densities compared with both ALN-treated and sham rabbits (Fig. 5d). Thus, overall, the higher prevalence of cuboidal osteoblasts and the higher osteoblast density may contribute to the increased mineralizing surfaces and bone formation rates observed upon ODN treatment compared with ALN treatment [7].