The authors state that they have no conflicts of interest.
Osteocyte apoptosis is spatially and temporally linked to bone fatigue-induced microdamage and to subsequent intracortical remodeling. Specifically, osteocytes surrounding fatigue microcracks in bone undergo apoptosis, and those regions containing apoptotic osteocytes co-localize exactly with areas subsequently resorbed by osteoclasts. Here we tested the hypothesis that osteocyte apoptosis is a key controlling step in the activation and/or targeting of osteoclastic resorption after bone fatigue. We carried out in vivo fatigue loading of ulna from 4- to 5-mo-old Sprague-Dawley rats treated with an apoptosis inhibitor (the pan-caspase inhibitor Q-VD-OPh) or with vehicle. Intracortical bone remodeling and osteocyte apoptosis were quantitatively assessed by standard histomorphometric techniques on day 14 after fatigue. Continuous exposure to Q-VD-OPh completely blocked both fatigue-induced apoptosis and the activation of osteoclastic resorption, whereas short-term caspase inhibition during only the first 2 days after fatigue resulted in >50% reductions in both osteocyte apoptosis and bone resorption. These results (1) show that osteocyte apoptosis is necessary to initiate intracortical bone remodeling in response to fatigue microdamage, (2) indicate a possible dose-response relationship between the two processes, and (3) suggest that early apoptotic events after fatigue-induced microdamage may play a substantial role in determining the subsequent course of tissue remodeling.
Turnover of cells and matrix occurs in all organs and tissues and is essential to maintain their health and integrity. In bone, turnover of cells and matrix occurs simultaneously through remodeling, wherein osteoclastic resorption removes-and osteoblastic infilling replaces-microscopic regions of bone that have reached the end of their functional life.(1–4) Perhaps the best-characterized example of a microscopic region of bone reaching the end of its life occurs when it sustains microdamage caused by fatigue. Left undetected and unrepaired, microdamage in bone leads to compromised mechanical properties and fragility. Fatigue microdamage in living rodent, canine, and human bone has been shown to activate localized bone remodeling in the region around microcracks, so that damaged areas of bone are removed and replaced.(1,5–7) This remodeling of bone in a nonrandom, lesion-specific manner has been termed “targeted remodeling” by Parfitt and others.(8–10) The cellular mechanisms responsible for targeting this remodeling process to damaged areas of bone remain obscure.
Verborgt et al.(11) showed that osteocytes surrounding fatigue microcracks in vivo undergo apoptosis, which both preceded the onset of osteoclastic resorption and co-localized exactly with the areas of bone that were subsequently resorbed. Apoptosis is a complex, highly regulated cell death process that can result from a range of developmental, genetic, and environmental cues,(12–15) including injury, mild toxic stimuli, loss of cell attachment to matrices, and loss of trophic signals needed to maintain cell viability.(16–20) Also of importance, the removal of apoptotic cells and their debris occurs through a noninflammatory process, carried out by specialized phagocytic cells (e.g., macrophages) that recognize newly exposed markers such as phosphatidylserine.(21–23) It is hypothesized that osteoclasts, which arise from the macrophage lineage, fulfill this phagocytic role in bone with regard to the disposal of apoptotic osteocytes.(2,24,25)
The tight spatial and temporal coupling of osteocyte apoptosis to both bone microdamage and intracortical remodeling has led to the hypothesis that osteocyte apoptosis is a key controlling step in the activation and/or targeting of osteoclastic resorption in response to bone fatigue.(5,11) In these studies, we tested this hypothesis in an in vivo rat ulnar fatigue model. We sought to determine whether suppression of osteocyte apoptosis using a broad-spectrum caspase inhibitor would alter the onset of fatigue-induced intracortical bone remodeling. Oligopeptide inhibitors of the caspases necessary for apoptosis(13,18,26–28) have been successfully used in animal models to attenuate cell death and minimize tissue damage after ischemic injury to heart, liver, gut, and brain, as well as septicemia.(26,29–35) Consequently these inhibitors provide a unique opportunity to pharmacologically modulate osteocyte apoptosis in situ.
MATERIALS AND METHODS
In vivo fatigue
The left ulna of adult female Sprague-Dawley rats (16 wk old) were cyclically loaded in vivo following the procedure described in detail elsewhere.(5,11) This loading protocol induces localized fatigue-induced bone matrix microdamage and subsequent intracortical remodeling of the damaged region in a temporally and spatially coordinated manner. A key advantage to this model is a virtually complete absence of baseline intracortical remodeling in the rat ulnar cortex. Consequently, any observed remodeling responses result from osteoclast activation produced by the experimental manipulation.
Ulna were cyclically loaded (maximal load, 16 N, 2 Hz) to a predetermined decrease in whole bone stiffness using a small servohydraulic loading system (Model 8841; Instron, Canton, MA, USA). This load level produces initial peak strains of ∼3800 ± 500 μstrain,(5) a level greater than normally achieved during habitual loading but corresponding to peak strains reported in military recruits and racehorses undergoing high level physical activities.(36–41) As bone, like other composite materials, loses stiffness as microcracks form during fatigue loading,(42–46) changes in whole ulna stiffness were monitored throughout the experiment using the system LVDT, and loading was stopped after a 23% stiffness decrease was attained. This fatigue level was shown in previous studies to induce bone microdamage and osteocyte apoptosis and to activate intracortical remodeling in rat ulna without causing a stress fracture response.(5,11) Loading was conducted under isoflurane anesthesia, and animals were permitted unrestricted cage activity and ad libitum access to food and water before and after loading. All procedures were conducted under Institutional Animal Care and Use Committee approval.
In vivo apoptosis inhibition
The pan-caspase inhibitor, Q-VD-OPh (quinolyl-valyl-O-methylaspartyl-[-2,6-difluorophenoxy]-methylketone; MP Biomedicals, Livermore, CA, USA), was used in these studies, because it is effective at lower concentrations than related compounds(28) and also lacks the potential renal toxicity of caspase inhibitors containing fluoromethylketone-leaving groups (e.g., zVAD-FMK).(28,47,48) Adult rats were assigned randomly to one of four experimental groups (N = 5/group). Two groups were subjected to fatigue loading to initiate intracortical resorption as described above, whereas the remaining two groups were not loaded. Beginning 2 h before fatigue loading, one group of animals (FAT + Ap Inh) received the pan-caspase inhibitor Q-VD-OPh (20 mg/kg/d split into two doses 12 h apart), whereas a second group (FAT + Veh) received a corresponding dose of DMSO vehicle. Two groups of independent nonfatigued control animals (NonFAT + Ap Inh, NonFAT + Veh) received similar treatments. Independent, controls were used rather than the nonloaded contralateral limbs, because localized bone injury has been reported in the literature to result in systemic skeletal responses.(49–53)
Two inhibition experiments were carried out. In the first experiment, animals were treated daily with Q-VD-OPh or vehicle for 14 days to test whether continuous suppression of osteocyte apoptosis would alter the activation of intracortical bone resorption. In the second study, animals were treated only for the first 2 days after loading to test whether partial suppression of osteocyte apoptosis would produce a correspondingly limited change in bone resorption. All animals in both experiments were killed 14 days after loading, at which time bone resorption is fully activated after the experimental fatigue challenge.(5,11,54) Both left and right ulna were harvested, fixed in neutral buffered formalin, and processed for immunohistochemistry and histomorphometric evaluation. Ulna were decalcified in 10% EDTA and dehydrated in ethylene glycol monoethyl ether, cleared in methyl salicylate, and embedded in paraffin. Cross-sections were cut at 5 μm thickness through the 5-mm-long segment of the ulnar mid-diaphysis where osteocyte apoptosis and bone remodeling occur in this fatigue model.(5,11)
Osteocyte apoptosis was measured by counting cells immunohistochemically stained for biological markers of apoptosis or exhibiting a pyknotic or atypical nucleus. Staining for the activated (cleaved) form of caspase-3 (c-cas-3), an effector caspase required for regulated cell death, allowed assessment of apoptotic cell numbers in control animals and also of caspase inhibitor efficacy in treated animals. In addition, cells were stained for phosphorylated histone, H2AX, which is associated with double-stranded DNA damage in apoptosis.(55,56) Glass-mounted sections were deparaffinized, rehydrated, and treated with 3% hydrogen peroxide for 20 min to inhibit endogenous peroxidase activity. A methanol-sodium hydroxide-based antigen retrieval system (DeCal; Biogenex, San Ramon, CA, USA) was applied for 30 min at room temperature, after which samples were blocked (Dako Cytomation, Carpinteria, CA, USA) for a further 30 min. Sections were incubated with rabbit anti-rat antibody to cleaved caspase-3 (9661; Cell Signaling Technologies, Carpinteria, CA, USA) or a rabbit-anti mouse H2AX antibody (AB3369; Chemicon, Temecula, CA, USA) at dilutions of 1:50 and 1:800, respectively. Sections were incubated in a humidified chamber overnight at 4°C. Detection was performed using a goat anti-rabbit secondary antibody with a streptavidin-biotin-conjugated system and developed with a DAB substrate chromogen system (Dako), after which the sections were counterstained using a 5% light green solution, dehydrated, and coverslipped using nonfluorescing mounting medium. Rat tibial growth plates processed in the identical manner as the bone samples were used as positive staining controls, because hypertrophic chondrocytes express the apoptosis marker proteins used in these experiments.(57–60) Thymus tissue was also used as a nonskeletal positive control tissue, because thymocytes undergo extensive apoptosis during T-cell development and clonal selection.(61–63) Species-appropriate antibody controls were also examined in each experiment. For each apoptosis antibody, the numbers of stained and unstained osteocytes were counted throughout the entire ulnar cross-section under bright-field illumination using a ×40 magnification objective. In addition, both normal and morphologically atypical (retracted and pyknotic) osteocytes were counted in parallel sections using a ×63 magnification objective following the method described by Bentolila et al.,(5) to provide independent morphological confirmation of cell death. For all assessment methods, apoptotic osteocytes are expressed as a percentage of total osteocytes.
Intracortical resorption space numbers and sizes were measured from ulnar cross-sections as an index of remodeling activity. Data were collected for the entire cross-section using a ×40 magnification objective and are expressed per cortical bone area.
Effects of Q-VD on osteoclast formation in vitro
Osteoclast formation assays to evaluate potential direct effects of the caspase inhibitor were carried out using total rat bone marrow following a method adapted from Takahashi et al.(64) Marrow cells were cultured in αMEM/10% FBS containing 10 nM 1,25(OH)2D3 (Sigma, St Louis, MO, USA). One day after plating, cells were treated for 6 days with Q-VD-OPh at concentrations between 0 and 50 μM, which bracketed the estimated peak serum levels for inhibitor. On day 7 after plating, cells were stained for TRACP using a detection kit (86C; Sigma), and the number of TRACP+ (osteoclast-like) cells containing three or more nuclei was counted in 10 high-power fields from each of four replicate wells. Experiments were repeated twice with comparable results. In addition, we evaluated the effects of the Q-VD and apoptosis inhibition in vivo on rat bone marrow osteoclastogenic capacity. Rats (fatigued and nonfatigued, N = 2–3/group) were treated with Q-VD-OPh or with DMSO vehicle as described earlier. After treatment, the rats were killed, marrow was isolated from both femora, and total marrow cultured to induce osteoclastogenesis; no Q-VD-OPh or vehicle was added to the cultures. Cultures were assayed for TRACP+ multinucleated cells as described above.
Differences among groups in numbers of resorption spaces and apoptotic osteocytes in vivo, and of TRACP+ cells in vitro, were evaluated using ANOVA with posthoc analysis of between-group differences using the Dunn procedure (GraphPad Instat for Macintosh; GraphPad Software, San Diego, CA, USA).
Continuous caspase inhibition after fatigue
Osteocyte apoptosis was effectively absent from control bones (<2%), but increased >600% in the ulnar cortices of fatigue-loaded, vehicle-treated (FAT + Veh) animals (Fig. 1; p < 0.001 versus control). Treatment with Q-VD-OPh throughout the 14-day period after fatigue loading completely suppressed osteocyte apoptosis (Fig. 1). All methods of assessing apoptosis (staining for cleaved caspase-3 and H2AX, enumeration of pyknotic osteocytes) gave similar results (table in Fig. 1).
Intracortical resorption was activated by fatigue loading in vehicle-treated animals (Fig. 2, FAT + Veh), consistent with numerous studies using this model. Rs.N/B.Ar in this group was 1.8 ± 0.5 /mm2 versus a complete absence of resorption in controls (Fig. 2; p < 0.0001). Also in agreement with previous studies, fatigue-induced osteoclastic activity occurred only in regions of osteocyte apoptosis. In contrast, animals receiving Q-VD-OPh showed no activation of intracortical remodeling in fatigue-loaded ulna (Fig. 2, FAT+ Ap Inh).
Short-duration caspase inhibition after fatigue
Treatment with Q-VD-OPh for only 2 days immediately after ulnar loading reduced the number of fatigue-induced apoptotic osteocytes measured on day 14 postloading by nearly 50% compared with vehicle-treated animals (p < 0.03) and also suppressed the increase in resorption by nearly 75% (Fig. 3; p < 0.01). Resorption space areas did not differ significantly among groups (FAT + Veh: 0.0178 ± 0.0125 mm2; FAT + Ap Inh: 0.0158 ± 0.0111 mm2; p = 0.37).
Osteoclast differentiation in vitro
Formation of TRACP+ multinucleated cells in total rat bone marrow cultures was unaffected by Q-VD-OPh, indicating that the ability of the inhibitor to prevent bone resorption in vivo was unlikely to be the result of direct inhibition of osteoclastogenesis (Fig. 4). Similarly, treatment with Q-VD-OPh in vivo did not affect the ability of rat marrow cells to form osteoclasts in culture. Total marrow cultures established from rats treated with the Q-VD-OPh apoptosis inhibitor or vehicle in vivo showed no suppression of osteoclastogenic potential. All in vivo treatments groups yielded similar numbers of TRACP+ cells in culture as follows: FAT + Ap Inh, 7 ± 4 TRACP+ cells/well; FAT + Veh, 6 ± 5 TRACP+ cells/well; NonFAT + Ap Inh, 7 ± 5 TRACP+ cells/well; NonFAT + Veh, 5 ± 4 TRACP+ cells/well, indicating that neither fatigue nor the apoptosis inhibitor caused a systemic suppression of the osteoclast-forming potential of the bone marrow.
Apoptotic cell death and the subsequent removal of apoptotic cell debris by specialized phagocytic cells occur in numerous tissues during development(12–15) and in response to injury.(16–20) This highly orchestrated physiological process clearly differs from nonregulated (necrotic) cell death in several respects including its initiation by specific ligand-receptor interactions (e.g., Fas-Fas ligand), its pathways of cellular degeneration (often dependent on caspases) that produce distinctive forms of cell debris, and the mechanism for removal of that debris by phagocytic cells without a major inflammatory response.(13,14,24–26,30,65–70) The close spatial and temporal correlations among fatigue-induced microdamage in bone, osteocyte apoptosis, and the subsequent resorption and replacement of the damaged areas suggested strongly that damage-induced bone remodeling is a skeletal manifestation of that common physiological process and led us to hypothesize(5,11,71) that osteocyte apoptosis plays a controlling role in the initiation and/or targeting of damage-induced bone remodeling. The results of this study provide direct evidence that apoptosis is essential if fatigue-induced remodeling is to occur. Moreover, our findings suggest that osteocytes, and not other cell types, are the cells whose apoptotic death likely triggers the activation and targeting of bone resorption in this circumstance. This concept received further support by recent work by Tatsumi et al.,(72) who showed that selectively induced osteocyte death by gene-targeted ablation caused rapid bone loss.
We chose a pan-caspase inhibitor for these studies because caspase action is essential for most apoptotic cell death, and therefore these enzymes are logical targets for pharmacological modulation of apoptosis.(30,73–75) However, despite widespread success in minimizing tissue damage after injury,(13,18,26–35) the use of an oligopeptide caspase inhibitor that lacks cell type specificity raises obvious questions about whether its ability to inhibit resorption resulted from actions on cells other than osteocytes, in particular by direct effects on the differentiation or activity of osteoclasts. Miura et al.(76) reported that a specific caspase-3 inhibitor impaired osteoblast recruitment from the bone marrow stromal cells, suggesting that some osteoblast apoptosis is needed to maintain recruitment from the precursor pool. In contrast, Halasy-Nagy et al.(30) established that caspase inhibitors do not directly inhibit the recruitment, differentiation, or ongoing activity of osteoclasts. We confirmed these observations in vitro, finding that formation of TRACP+ multinucleated cells by total rat marrow (and in separate experiments by RAW 264.7 cells; data not shown) was unaffected by Q-VD-OPh at concentrations as high as 50 μM, a level that exceeds the estimated serum concentration of the inhibitor. Bone marrow from animals treated in vivo with Q-VD-OPh also did not show any reduction of in vitro osteoclastogenic potential, confirming that there was no systemic suppression of the osteoclast-forming potential of the bone marrow from the apoptosis inhibitor. The absence of background intracortical remodeling in the adult rat ulnar cortex(5,11,27,71) rules out direct effects on both extant osteoclasts and osteoblasts. Finally, the mid-diaphysis has effectively no marrow cavity (Fig. 3). Therefore, osteocytes comprise the overwhelming majority of cells in the adult rat ulnar mid-diaphysis and by far the most likely site of Q-VD-OPh action in this system.
Spatial and temporal correlations between bone microdamage, osteocyte apoptosis, and remodeling led us previously to hypothesize(5,11,71) that osteocyte apoptosis is a key controlling step in the activation and/or targeting of osteoclastic resorption after bone fatigue. The current studies, showing that continuous pharmacological suppression of osteocyte apoptosis using a pan-caspase inhibitor completely prevented fatigue-induced activation of osteoclastic resorption within the cortex, argues strongly that osteocyte apoptosis is in fact needed to activate bone resorption at microdamage sites.
Our finding that caspase inhibition only during the initial 48 h after fatigue loading suppressed both osteocyte apoptosis and remodeling activation to similar extents not only argues that the amount of osteoclastic resorption in this system is directly related to the amount of osteocyte apoptosis but also indicates that key signaling events stemming from osteocyte death and necessary to activate bone resorption, occur early after acute microinjury. This scenario parallels the events described after focal ischemia perfusion-reperfusion injury in brain, myocardium, and transplants. In those circumstances, acute local injury also leads to a characteristic series of tissue responses, including (1) rapid, local release of cytotoxic molecules such as NO, oxygen radicals, and calcium, (2) induction of localized apoptosis, (3) “walling in” of the apoptotic region by surrounding cells, and (4) production and localized release of pro-inflammatory cytokines and prostaglandins by cells surrounding the area of focal injury. These factors attract the phagocytic cells that ultimately remove the damaged area and allow subsequent ingrowth of granulation tissue and the start of tissue renewal.(19,77–81) In these other focal injury systems, inhibiting the initial apoptosis burst after acute injury was shown to be sufficient to attenuate all of the ensuing tissue responses.(31,82) These parallels with focal injury and repair responses in nonskeletal tissues may provide key insights into the signaling pathways that link osteocyte death to osteoclast recruitment.
The triggers for localized osteocyte apoptosis around microdamage are not well understood but are undoubtedly complex. The few osteocytes directly transected by microcracks probably die immediately as a result of trauma that is large in scale relative to the size of the osteocytes. The physiological implication of this core injury is poorly understood not only in bone but also in other systems where similar focal injury occurs (e.g., focal infarcts). There are important parallels, however, between damage-induced osteocyte death and localized ischemia-reperfusion injuries in other tissues. Tami et al.,(83) in our laboratory, examined the changes in fluid flow through the osteocyte lacunar-canalicular system in the area surrounding microcracks (which corresponds to the region where we observed osteocyte apoptosis) and found that these areas suffered from disrupted fluid and solute transport. The hypoxia and reperfusion resulting from such disruptions can generate multiple potential stressors capable of inducing localized apoptotic cell death.
Whereas our results showed a requirement for localized osteocyte apoptosis to activate resorption, neither the molecular mechanisms responsible for generating osteoclastogenic and chemoattractant signals nor the cellular source of those signals has been clearly established. However, intriguing recent data indicate that osteoclastogenic signaling in this system could arise from several sources. Loss of antiresorptive signals as osteocytes die (i.e., local loss of inhibition) has been widely considered as a mechanism(84–86); alternatively, “cell-mediated activation” hypotheses posit that dying osteocytes or their neighbors produce pro-osteoclastogenic signals. Kurata et al.(25) showed that focal wounding of MLO-Y4 cells in vitro triggered release of RANKL and macrophage-colony stimulating factor (M-CSF), although whether these key signaling molecules come from dying cells or the non-apoptotic surviving cells was not examined. Verborgt et al.(71) in our laboratory showed that osteocytes juxtaposed to apoptotic osteocytes at microcracks actively mount an anti-apoptotic response and proposed that these surviving cells could be a major source of signals promoting the subsequent osteoclastic response. In a key recent paper, Kogianni et al.(87) showed that apoptotic debris from osteocytes can stimulate osteoclasts. However, it is not known whether osteocyte apoptotic debris in vivo is sufficiently “diffusible” to migrate from the damage site to the bone surface and initiate signaling to osteoclasts or whether it remains fixed in the tissue because of constraints imposed by the mineralized bone matrix and their relatively large size compared with the very limited permeability of the small pericellular spaces of the lacunar-canalicular system. In the latter case, these debris particles could serve as a contact-based homing signal encountered by osteoclasts tunneling through bone.
In summary, we showed that osteocyte apoptosis plays a fundamental role in the initiation of bone remodeling in response to localized tissue damage. The features of this skeletal damage response parallel the highly orchestrated, apoptosis-based processes observed in physiological circumstances ranging from tissue morphogenesis to ischemic injury repair. Based on these findings, it seems reasonable to conclude that removal of apoptotic debris is the “raison d'être” for targeted remodeling of damaged bone and the basis for its underlying mechanisms. Moreover, it also seems reasonable to speculate based on these findings that osteocyte apoptosis may assume a similarly critical role in the initiation of bone remodeling triggered by other stimuli such as estrogen depletion and immobilization.
This work was supported by NIH Grant AR 41210 and the National Space Biomedical Research Institute through NASA NCC 9-58.