Portions presented at the 45th Annual Meeting of the Orthopaedic Research Society in Anaheim, California, U.S.A., February 1–4, 1998.
As a result of fatigue, bone sustains microdamage, which is then repaired by bone-remodeling processes. How osteoclastic activity is targeted at the removal of microdamaged regions of bone matrix is unknown. In the current studies, we tested the hypothesis that changes in osteocyte integrity, through the initiation of regulated cell death (apoptosis), are associated with fatigue-related microdamage and bone resorption. Ulnae of adult rats were fatigue-loaded to produce a known degree of matrix damage. Osteocyte integrity was then assessed histomorphometrically from terminal deoxynucleotidyl transferase–mediated deoxyuridine triphosphate–nick end labeling (TUNEL)–stained sections to detect cells undergoing DNA fragmentation associated with apoptosis; toluidine blue–stained sections were used for secondary morphological confirmation. Ten days after loading, large numbers of TUNEL-positive osteocytes were found in bone surrounding microcracks and in bone surrounding intracortical resorption spaces (∼300% increases over controls, p < 0.005). TUNEL labeling in loaded ulnae at sites distant from microcracks or resorption foci did not differ from that in control bone. Osteocytes in toluidine blue–stained sections showed equivalent trends to TUNEL-stained sections, with significant increases in pyknotic nuclei and empty lacunae associated with microcracks and intracortical resorption spaces. TUNEL-positive osteocytes were observed around bone microdamage by 1 day after loading (p < 0.01 relative to baseline), and their number remained elevated throughout the entire experimental period. Increases in empty lacunae and decreases in normal osteocyte numbers were observed over time as well. These studies show that (1) osteocyte apoptosis is induced by bone fatigue, (2) this apoptosis is localized to regions of bone that contain microcracks, and (3) osteoclastic resorption after fatigue also coincides with regions of osteocyte apoptosis. The strong associations between microdamage, osteocyte apoptosis, and subsequent bone remodeling support the hypothesis that osteocyte apoptosis provides a key part of the activation or signaling mechanisms by which osteoclasts target bone for removal after fatigue-induced matrix injury.
Bone sustains microdamage as a result of normal wear and tear processes (i.e., fatigue).(1–5) This damage will be repaired by focal bone remodeling.(5,7–9) This process has been described as targeted remodeling, wherein osteoclasts specifically remove regions of bone matrix that have sustained matrix microdamage or reached the end of their functional life span.(10) How such foci of damaged bone might be targeted by osteoclasts is unknown.
Osteocytes have been hypothesized to play a role in this targeted remodeling process.(3,5,6,9) Numerous studies have suggested that bone microdamage could injure osteocytes, thereby signaling the remodeling of a region of bone.(5,6,11) However, only recently have data directly supported this assertion. Changes in osteocyte integrity were found to localize with regions undergoing bone remodeling in fatigued loaded rat ulnae.(9) Qiu et al.(12) have shown that osteocyte apoptosis occurs in association with surgically induced matrix damage in bone, providing evidence to suggest that osteocyte apoptosis may be a critical step in the process by which damaged bone and osteocytes are signaled and targeted for removal by osteoclastic resorption. Several recent studies posit a role for osteocyte apoptosis in local control of bone resorption. Studies in growing bone report that osteocyte apoptosis is associated with osteoclastic resorption.(13,14) There is an overall increase in osteocyte apoptosis after estrogen withdrawal in humans and rats.(15,16) However, a spatial relationship between osteocyte apoptosis and the new bone-remodeling foci initiated by estrogen withdrawal has not been shown. The hypothesis that bone fatigue could induce osteocyte apoptosis, and that such focal changes in osteocyte integrity at sites of matrix microdamage could identify areas for subsequent remodeling, provides for an intriguing targeting mechanism. In the current studies, we tested the hypothesis that changes in osteocyte integrity, through the initiation of regulated cell death (apoptosis), are associated with fatigue microdamage and consequent bone resorption.
MATERIALS AND METHODS
Right ulnae of adult female Sprague–Dawley rats (5–6 months old) were subjected to fatigue loading in vivo through a modification of end-load bending developed by Bentolila et al.(9) We have shown that this model will activate intracortical remodeling in fatigued ulnar cortices in rats. With rats under inhaled isoflurane–induced anesthesia (0.5–3%), fatigue loading of ulnae was performed under load control using a 20-N maximum load range, as in our previous studies; the axial displacement range was measured using the LVDT (linear variable differential transducer) system (Sensotec, Cleveland, OH, U.S.A.). Loading was conducted at 4 Hz. Ulnae were fatigued to a single stopping point based on loss of bone stiffness, which reflects the formation of microdamage in bone.(1,2,4,17,18) Fatigue-induced losses of ulnar structural stiffness were determined from increases in ulnar compliance.(9) Before and after loading, animals were allowed unrestricted cage activity and ad libitum access to food and water. Procedures were conducted with approval from the Institutional Animal Care and Use Committee of the Henry Ford Health Sciences Center.
In the first series of experiments, right ulnae of 10 rats were fatigue-loaded to a prefailure stopping point of a 30% decrease in ulna whole-bone stiffness, which represents a point in the fatigue life of bone at which matrix damage has occurred but well in advance of fatigue fracture.(1,2,4,16,17) Left ulnae were not loaded and served as paired internal controls. Spatial relationships between changes in osteocyte integrity and microcracks or between changes in osteocyte integrity and intracortical osteoclastic activity were studied 10 days after loading, when intracortical resorption has been shown to occur in this model.(9)
In the second series of experiments, ulnae from 40 rats were loaded as described for experiment 1. Changes in osteocyte integrity over time after fatigue loading were then studied at baseline (0 days) and 1, 3, 7, and 10 days after loading.
At necropsy, ulnae were manually dissected free of soft tissues, fixed in formalin, decalcified in ethylenediaminetetraacetic acid, and embedded in paraffin for histological sectioning. Sections of the ulnar diaphysis were cut at 5 μm thickness and adhered to silane-coated slides. Sections were stained using the end-labeling approach to distinguish osteocytes undergoing DNA fragmentation characteristically associated with apoptosis. Terminal deoxynucleotidyl transferase (TdT)–mediated deoxyuridine triphosphate (dUTP)–nick end labeling (TUNEL) was used to identify the DNA fragments generated by endonuclease activity in cells.(19,20) The Apotag system (Oncogene Research Products, Cambridge, MA, U.S.A.) was used for the TUNEL studies. Sections were placed in phosphate-buffered saline equilibration buffer for 5 minutes and incubated for 60 minutes with TdT/digoxygenin-labeled dUTP at 37°C. The reaction was stopped by immersing the sections with stop/wash buffer at 37°C for 30 minutes. Peroxidase-labeled antidigoxygenin with diaminobenzidine/cobalt/nickel staining was used to localize TUNEL-stained cells; fast green was used as a counterstain.(21,22) Rat thymus served as a positive control for apoptotic cells. Deoxyribonuclease-treated control bone was used as a positive control for TUNEL. TdT was not added to negative controls. High levels of background staining with TUNEL staining have been reported from studies using this method to demonstrate apoptosis in tissue sections.(23–25) Accordingly, we incorporated a secondary morphologically based assessment of osteocyte integrity (toluidine blue–stained sections) to ensure that comparable changes in osteocyte integrity were observed in association with microcracks and resorption spaces using two different methods of study to cross check trends observed from TUNEL studies.(21–25)
To assess the whether changes in osteocyte integrity are associated with microdamage or intracortical resorption foci, specimens were examined from bones at 10 days after fatigue loading. Osteocyte number densities (per square millimeter) were determined (1) in bone surrounding (±100μm) microcracks (Mdx) or resorption spaces (Rs, with and without visible microdamage); (2) in bone regions distant from microcracks or resorption spaces in fatigued ulnae; and (3) in nonloaded, control ulnae. On the basis of specific staining, data were collected as follows. From TUNEL-stained sections, the osteocyte number densities in lacunae with TUNEL-negative (−) staining, with TUNEL-positive (+) staining, and in empty lacunae (E.lac.) were determined (Fig. 1). From toluidine blue–stained sections, number densities of osteocytes with normal-appearing nuclei (N.Ot.) and injured osteocytes (defined as osteocytes with pyknotic nuclei [P.Ot.] and empty osteocyte lacunae) were determined (Fig. 1) by following methods we have described previously.(9) To assess time-dependent changes in osteocyte integrity, osteocyte densities were determined in bone immediately around microcracks, which could be identified at all time periods. Measurements were performed using point count stereological methods, using a 10 × 10-mm eyepiece grid reticule at 250× magnification; tissues were sampled for the entire middle one-third of the bone length, where microdamage and remodeling occur in this model.(9) Because osteocyte numbers were not normally distributed, nonparametric statistics were used for analyses. For the first experiment (i.e., changes at 10 days), the Wilcoxon signed-rank test for matched pairs was used to determine whether changes in injured/apoptotic osteocytes, normal osteocytes and empty lacunar number densities in fatigued limbs were significant relative to nonloaded control bones. For the second experiment, the Kruskal–Wallis analysis of variance was used assess apoptotic osteocytes, normal osteocytes, and empty lacunae between groups over time, with post hoc comparison performed against baseline (day-0) values using the Mann–Whitney U test. Statistical analyses were performed using the StatView software package (version 4.5; Abacus Concepts, Berkeley, CA, U.S.A.). Data are reported as means ± SD.
Figure 2 summarizes the data for TUNEL-positive and TUNEL-negative osteocytes and empty lacunae in relation to microdamage and bone resorption. In bone immediately surrounding microcracks, the numbers of TUNEL-positive osteocytes was increased almost 400% over control levels (p < 0.005). In contrast, in bone regions of fatigue-loaded ulnae distant from microcracks, the number of TUNEL-positive osteocytes was similar to that in nonloaded control bones (15 and 13% of total cells, respectively), indicating osteocyte apoptosis in association with bone microdamage. Empty lacunae near microcracks were markedly increased as well (∼8-fold over controls, p < 0.005). In addition, TUNEL-positive osteocytes and empty lacunae were increased significantly in bone immediately surrounding intracortical resorption spaces (p < 0.005 relative to control levels), indicating a strong association between regions of osteocyte apoptosis and bone resorption foci.
Morphologically based studies of osteocytes from toluidine blue–stained sections are shown in Fig. 3. These studies showed equivalent trends to TUNEL-stained sections, with dramatic and significant increases in the number of osteocytes with pyknotic nuclei and empty lacunae in specific association with microcracks and intracortical resorption spaces (5- and 10-fold increases, respectively; p < 0.005). Osteocytes with pyknotic nuclei and empty lacunae were rarely observed in normal, control ulnar cortex (< 5% of all osteocytes). In fatigued bone, away from either microcracks or resorption spaces, the numbers of injured osteocytes were similar to control tissues. Together these data indicate that after fatigue loading in vivo, there is a significant, focal loss of osteocyte integrity in association with fatigue microdamage and with intracortical resorption foci, and that these changes occur through the regulated death of osteocytes.
Because TUNEL and toluidine blue–stained sections revealed effectively identical trends with regard to osteocyte changes associated with microcracks, data from only one study method (TUNEL staining) are reported to demonstrate time-dependent changes in osteocytes after fatigue loading (Fig. 4). At baseline, (day 0 after fatigue loading), the numbers of TUNEL-positive (apoptotic) stained osteocytes and empty lacunae immediately were unchanged from control levels. On 1 day after fatigue loading, the number of TUNEL-positive osteocytes was increased nearly 4-fold over baseline values (p < 0.01). TUNEL-positive osteocyte number remained constant at this elevated level for all subsequent time periods examined. In contrast, the number of normal (TUNEL-negative) osteocytes decreased continually 1, 3, and 7 days after loading (p < 0.01, relative to baseline at each time period), with no subsequent change observed at day 10. The number of empty lacunae showed an inverse pattern to that observed for TUNEL-negative osteocytes, with significant increases from baseline 3, 7, and 10 days after loading.
The current studies show that (1) osteocyte apoptosis is induced by bone fatigue, (2) this apoptotic response is localized to the regions containing bone microcracks, and (3) subsequent osteoclastic removal of bone after fatigue also coincides with regions of osteocyte apoptosis. Apoptosis, or regulated cell death,(26) is implicated in the control or signaling of remodeling and replacement of skeletal tissues, such as growth cartilage and fracture callus.(21–23,27–29) Apoptosis is associated with local remodeling responses to physical injury in other tissues, as well.(30–32) It has been hypothesized that bone microdamage could cause osteocyte injury, thereby signaling remodeling of a region of bone.(5,6,11) Bentolila et al.(9) in our laboratory reported recently that after bone fatigue, osteocyte integrity was impaired in compact bone adjacent intracortical resorption spaces. Qiu et al.,(12) also from our laboratory, showed that osteocyte apoptosis occurs in response to surgically induced local damage in bone, suggesting that osteocyte apoptosis may be a critical step in the process by which damaged bone and osteocytes are signaled and targeted for removal by osteoclastic resorption. Several additional lines of support have been reported recently, supporting the hypothesis that osteocyte apoptosis is associated with osteoclastic activity in normal animal and human tissues. Noble et al.(14) examined human adult, pediatric, and pathological bone (heterotopic bone and osteoarthritic osteophytes) and found a positive relationship between bone turnover and osteocyte apoptosis; they posited a role for osteocyte apoptosis in local control of bone resorption. Tomkinson et al.(15,16) found that estrogen withdrawal in human and rats result in osteocyte apoptosis. Bronckers et al.(13) reported that osteocyte apoptosis in developing bone was associated with osteoclastic resorption, whereas Noble et al.(33) found that osteocyte apoptosis in developing bone could be modulated by mechanical loading. Data from the current studies support the idea that focal osteocyte apoptosis, induced by local bone fatigue damage, is associated with activation, signaling to or targeting of intracortical resorption processes after bone fatigue.
The current studies show changes in osteocyte integrity, consistent with regulated cell death, in association with microcracks and with later bone resorption induced by modest levels of fatigue in bone. Osteocyte apoptosis in response to severe overload was recently reported by Noble et al.(34) Using an adaptation of the in vivo fatigue/bone-remodeling model, they cyclically loaded ulnae in young rats (∼90 g) at very high strains to the point of “plastic deformation” and found there was a significant increase in the number of apoptotic osteocytes on the whole-bone level. It is difficult to make direct comparisons between those experiments and the current ones because of fundamental differences in animal age and loading intensity in these experiments. Nevertheless, their data are noteworthy because they show that even in young animals, subjected to very different loading conditions, introduction of matrix damage results in osteocyte death through apoptosis.
In the current studies, the TUNEL staining method was used to distinguish osteocytes undergoing DNA fragmentation characteristically associated with apoptosis,(19,20) with large changes in the number of TUNEL-positive stained osteocytes observed in association with microcracks and resorption spaces. A high level of background staining was observed in control tissues, consistent with the presence of high background staining in other studies using the TUNEL method to show apoptosis in tissue sections.(23–25,35) A second morphologically based assessment of osteocyte integrity to cross-check the observed association with microcracks and resorption revealed lower background levels of abnormal-appearing cells in toluidine blue–stained sections (∼5% vs. 7–15% in TUNEL sections). However, both TUNEL– and toluidine blue–stained sections showed essentially identical changes with regard to osteocyte integrity around microcracks and resorption spaces. Together these data show there is a marked, focal loss of osteocyte integrity in association with fatigue microdamage and with intracortical resorption foci, and these changes result from the induction of osteocyte apoptosis by bone fatigue in vivo.
The experiments to assess changes in osteocyte integrity over time after fatigue revealed that osteocyte apoptosis around bone microdamage had begun by 1 day after fatigue loading. The maintenance of high levels of apoptosis in the tissue over relatively long time periods after fatigue (10 days after loading) seems somewhat unexpected, since apoptosis at the cell level in vitro is a very rapid process, progressing over time periods of several hours.(26) Similar long-term maintenance of high apoptosis levels in tissues in vivo have been reported after focal injury in both brain and cardiac infarct model.(35–37) It possible that apoptosis in vivo is a much more protracted process than programmed cell death in vitro. The time course data in the current experiments suggests another possibility as well. The numbers of empty lacunae increased throughout the study period, whereas the number of normal osteocytes around microcracks decreased commensurately. Thus, over time, early apoptotic cells “die,” giving rise to additional empty lacunae. High levels of osteocyte apoptosis appear to be maintained as additional cells from the normal cell compartment undergo apoptosis.
Previous electron-microscopy studies have reported that osteocytes juxtaposed to resorbing osteoclasts undergo degeneration, with subsequent phagocytosis by osteoclasts.(30,31) It has been widely interpreted from these morphological studies that osteoclastic resorption processes cause the involutional changes in osteocytes in bone that they are about to resorb, i.e., that local acidic pH, enzymatic, or cytokine influences from osteoclasts induce changes in osteocytes. However, the results of the current studies indicate the opposite sequence of events—that osteocyte degeneration precedes bone resorption. Osteocyte apoptosis around bone microdamage was initiated very soon (by 1 day) after fatigue loading. In contrast, intracortical remodeling in the rat fatigue model is not evident until 7–10 days, many days after the onset of changes in osteocyte integrity.(9) Thus the early osteocyte apoptosis around bone microdamage may provide an important local signal to remodel an area of bone.
Apoptosis occurs with a range of developmental or genetically regulated cues.(26,32,38) Moreover, it also follows exposure to mild injury, mild toxic stimulus, loss of attachment of cells to their matrices, and loss of cytokines and factors needed for maintenance of cell viability.(38–41) Thus osteocyte apoptosis may result from a number of processes likely to occur during the microdamage process. Osteocytes are widely and extensively distributed throughout the bone matrix, and their canalicular processes infiltrate completely throughout the tissue. Bone matrix disruption could be expected to directly injure osteocytes, leading to cell death. Matrix microdamage is also likely to disrupt canaliculi, altering canalicular flow and impairing the nutrition and metabolic activities of cells; this would also lead to apoptosis. Osteocytes are attached to their surrounding matrix through adhesion molecules and to their neighboring cells through gap junctions.(42,43) Loss of cell attachment and/or cell to cell communication have been shown to be strong apoptosis-inducing stimuli.(44)
The mechanisms by which apoptotic cells might signal or target osteoclastic activity are unknown, but several intriguing possibilities exist. Apoptotic cells have been shown to be preferentially targeted in macrophage phagocytosis, with the membranes of apoptotic cells exposing specific markers that are targeted by macrophages.(45–47) Accordingly, Bronckers et al.(13) have suggested that osteocyte apoptosis in developing bone could provide a targeting signal for resorption via cell fragmentation products; they showed that in developing bone that diffuse TUNEL-positive material (which was not related to nonspecific binding) could be found in the bone matrix, and appeared to concentrated near empty lacunae. This is consistent with the idea that material from apoptotic osteocytes can move through bone canaliculi, and provide a source of potentially transmissible signals. Release of local factors, such as proteases and cytokines, is also associated with apoptosis. A family of proteases (i.e., caspases), related to interleukin converting enzyme, are key elements in the apoptotic pathway.(41,48–50) Several of the interleukins are well-established mediators of bone resorption activity. Hogquist et al.(51) showed that apoptotic lymphocytes release interleukin-1, which is a primary stimulus for macrophage and osteoclast activity.(10,52)
Bone sustains microdamage as a result of normal wear and tear processes (i.e., fatigue). (1–6) Microdamage will be repaired by focal bone-remodeling processes.(6–9) Parfitt(10) described this process as targeted remodeling, wherein the osteoclastic activity is targeted at the removal of regions of bone matrix which have sustained matrix microdamage or reached the end of their functional life span. How such foci of bone might be targeted by osteoclastic activity has been unknown. The results of the current study indicate that microdamaged bone matrix translates directly to osteocyte injury and apoptosis and degeneration. The local mechanisms initiating osteocyte apoptosis are currently unknown, as are the local signals that may produced by apoptotic osteocytes to attract or target osteoclasts. Nevertheless, the strong associations between microdamage and osteocyte apoptosis, and between apoptosis and subsequent bone remodeling support the hypothesis that osteocyte apoptosis provides a key part of the activation or signaling mechanisms by which osteoclasts target bone for removal after fatigue-induced matrix injury.
We thank K. D. Lundin-Cannon for assistance with animal experiments and histological studies. O. Verborgt was supported by a fellowship from the University of Antwerp. This work was supported by grants AR41210 and AR44712 from the National Institute of Arthritis and Musculoskeletal and Skin Diseases.