Osteoporosis is a major clinical problem in chronic inflammatory diseases such as rheumatoid arthritis. The mechanism of bone loss in this condition remains unclear, but previous studies have indicated that depressed bone formation plays a causal role. Since cytokine-induced nitric oxide (NO) production has been shown to inhibit osteoblast growth and differentiation in vitro, this study was undertaken to investigate the role of the inducible NO synthase (iNOS) pathway in the pathogenesis of inflammation-mediated osteoporosis (IMO) by studying mice with targeted inactivation of the iNOS gene (iNOS knockout [iNOS KO] mice).
IMO was induced in wild-type (WT) and iNOS KO mice by subcutaneous injections of magnesium silicate. The skeletal response was assessed at the tibial metaphysis by measurements of bone mineral density (BMD), using peripheral quantitative computed tomography, by bone histomorphometry, and by measurements of bone cell apoptosis.
NO production increased 2.5-fold (P < 0.005) in WT mice with IMO, but did not change significantly in iNOS KO mice. Total BMD values decreased by a mean ± SEM of 14.4 ± 2.0% in WT mice with IMO, compared with a decrease of 8.6 ± 1.2% in iNOS KO mice with IMO (P < 0.01). Histomorphometric analysis confirmed that trabecular bone volume was lower in WT mice with IMO compared with iNOS KO mice with IMO (16.2 ± 1.5% versus 23.4 ± 2.6%; P < 0.05) and showed that IMO was associated with reduced bone formation and a 320% increase in osteoblast apoptosis (P < 0.005) in WT mice. In contrast, iNOS KO mice with IMO showed less inhibition of bone formation than WT mice and showed no significant increase in osteoblast apoptosis.
Inducible NOS–mediated osteoblast apoptosis and depressed bone formation play important roles in the pathogenesis of IMO.
Chronic inflammatory diseases such as rheumatoid arthritis (RA), ankylosing spondylitis, and inflammatory bowel disease are associated with generalized osteoporosis and an increased risk of fractures (1). The pathogenesis of osteoporosis in these conditions is unclear, but there is evidence to suggest that inhibition of bone formation plays a role in association with systemic overproduction of bone-resorbing cytokines, nitric oxide (NO), and prostaglandins (2).
The evidence implicating NO as a mediator of inflammation-induced bone loss stems from the fact that NO production is increased in inflammatory disorders (3). This appears to be due to activation of the inducible NO synthase (iNOS) pathway by cytokines, such as interleukin-1 and tumor necrosis factor α (4), which are expressed at sites of inflammation. Cytokine-induced iNOS activation has also been shown to inhibit osteoblast function in vitro and to stimulate osteoblast apoptosis (5). While the role of iNOS-induced osteoblast apoptosis in the pathogenesis of inflammation-induced bone loss remains unclear, we previously reported that the NOS inhibitor NG-monomethyl-L-arginine inhibited bone loss in a rat model of inflammation-induced osteoporosis (IMO) (6). In those studies, however, we were unable to clearly define the mechanisms involved, because of the poor specificity of the pharmacologic inhibitors then available to probe NOS function.
In the present study, in order to resolve these issues and clarify the role of the iNOS pathway in inflammation-induced bone loss in vivo, we have examined the effects of systemic inflammation on bone in mice lacking a functional iNOS gene.
MATERIALS AND METHODS
The murine iNOS gene was disrupted by introducing a targeted mutation into embryonic stem cells derived from the 129/Sv mouse strain, as previously described (7). The homozygous, heterozygous, and wild-type (WT) mouse littermates thus generated were back-crossed onto MF1 mice for 3 generations, to generate a colony on a mixed 129 × MF1 background. The phenotype of these mice has previously been extensively characterized, and Western blotting has shown that peritoneal macrophages from these mice do not produce iNOS following cytokine stimulation. Low levels of nitrite have been detected in peritoneal macrophages stimulated for up to 72 hours with lipopolysaccharide and interferon-γ, however, which appears to be attributable either to cytokine-induced activation or to induction of constitutive NOS isoforms (7). The mice were housed in a designated animal facility, routinely maintained on a 12 hours:12 hours light:dark cycle, and given ad libitum access to food and water. The rodent diet used (Diet CRM; Special Diets Services, Essex, UK) contained negligible amounts of nitrite/nitrate. All experiments were undertaken in accordance with UK Home Office regulations.
Induction of IMO
IMO was induced in 14-week-old female iNOS knockout (iNOS KO) and WT mice using an adaptation of the protocol described for rats (8). Systemic inflammation was induced by subcutaneous injections of magnesium silicate or talc (MgSiO4; 16 mg/gm body weight) in 600 μl saline at 5 sites on the upper back on day 0 of the study. Controls received an identical volume of saline at the same sites. The mice received intraperitoneal injections of Calcein (40 μg/gm body weight) on days 10 and 17 to allow determination of the mineral apposition rate, and the experiment was terminated on day 21.
Bone mineral density (BMD)
Using an XCT Research M bone densitometer (Stratec Medizintechnik, Pforzheim, Germany) with a voxel size of 70 μm and analysis software version 5.1.4, BMD at the left proximal tibial metaphysis was measured in mice which had been anesthetized with 0.2% Rompun (Bayer, Bury St. Edmonds, UK) and 10 mg/ml Vetalar TMV (Pharmacia & Upjohn, Crawley, UK). Mice were scanned at the start of the IMO study and again 3 weeks later at the end of the experiment. Quality assurance measurements were performed daily on the peripheral quantitative computed tomography (pQCT) densitometer according to the manufacturer's instructions. Mice were killed at the end of the experiment, and the spleen and long bones were harvested for subsequent analyses. Bone length was measured using Vernier calipers (PAV 0-25; Vaduz, Liechtenstein).
Urinary nitrate and creatinine analyses
NO production was assessed by measuring the stable metabolites of NO, nitrite, and nitrate in 18-hour urine samples which were collected on days 0 and 21 of the experiment by placing mice in metabolic cages. Urinary nitrate/nitrite concentration was determined by the Griess reaction and expressed as a molar ratio to urinary creatinine (which was measured by an autoanalyzer).
Histomorphometric studies were performed on the left tibiae. The bones were dissected free of soft tissues, fixed in 4% buffered formalin/saline (pH 7.4), and embedded in methyl methacrylate. Longitudinal sections (4 μm) were then prepared on silanized slides and stained with von Kossa's stain and Paragon. Histomorphometric measurements were made on 5–6 microscopic fields of sections of the proximal metaphysis distal to the epiphyseal growth plate at 200× magnification with an Axioskop (Carl Zeiss, Welwyn Garden City, UK) coupled to an image analysis system using customized software developed with Aphelion ActiveX Objects (Adcis, Hérouville-Saint-Clair, France). Dynamic histomorphometric measurements were made on the left femorae. Calcein double-labeling was visualized by fluorescence microscopy, and the mineral apposition rates were determined using image analysis. Bone histomorphometric variables were expressed according to the guidelines of the American Society of Bone and Mineral Research Nomenclature Committee (9).
Detection of apoptosis
Methyl methacrylate–embedded sections of tibiae were carefully deplasticized using at least three 3-minute washes with xylene, then rehydrated through a series of alcohol washes and a final wash in phosphate buffered saline (PBS). A nick translation assay was then carried out on the rehydrated sections, following the protocol described by Rogers et al (10). Negative control sections were incubated with nick translation buffer lacking in DNA polymerase, while positive control sections were pretreated with DNase I (Promega, Southampton, UK). Following the nick translation assay, the sections were washed in PBS and counterstained with 1 μg/ml 4′,6-diamidino-2-phenylindole (DAPI) for 10 minutes prior to analysis by fluorescence microscopy. An average of 12 microscopic fields of the proximal tibial metaphysis from each section were analyzed at 400× magnification.
Statistical analyses were performed using SPSS version 9 software (SPSS, Chicago, IL). Significant differences between groups were determined by analysis of variance, followed by post-hoc testing using Fisher's pair-wise comparisons. To take into account the potentially confounding effects of body weight on BMD, body weight was used as a covariable in the statistical analyses of the densitometry data. All data are presented as the mean ± SEM unless stated otherwise. P values less than 0.05 were considered significant.
BMD. Total BMD was significantly reduced in animals that had been injected with MgSiO4 compared with controls, due to a fall in both trabecular and cortical components of BMD (Figures 1A–C). The reduction in BMD was significantly attenuated in the iNOS KO mice compared with the WT mice, and this was statistically significant for total, trabecular, and cortical BMD.
Bone histomorphometry, body weight, and spleen weight. Consistent with the BMD results, trabecular bone volume, trabecular thickness, and cortical thickness were significantly reduced in mice with IMO compared with controls (Table 1). The reductions in trabecular bone volume and thickness which resulted from IMO were smaller in the iNOS KO mice than in the WT mice. The difference in the skeletal responses to inflammation was obvious on examination of low-power photomicrographs of the long bones (Figures 2A–D). At a cellular level, IMO was associated with a significant decrease in several parameters of bone formation, including osteoblast surface, osteoblast number, and mineral apposition rate. These changes were significantly attenuated in iNOS KO mice compared with WT mice. In contrast, IMO had no significant effect on parameters of bone resorption, which were unchanged from baseline in both iNOS KO and WT mice (Table 1).
Table 1. Changes in proximal tibial bone histomorphometric parameters and in body and spleen weights in response to inflammation-mediated osteoporosis (IMO) in wild-type (WT) mice and mice with targeted inactivation (knockout) of the gene for inducible nitric oxide synthase (iNOS KO mice)*
iNOS KO mice
Values are the mean ± SEM of 7 mice per group. BV/TV = trabecular bone volume/total tissue volume; TrabTh = trabecular bone thickness; CtTh = cortical bone thickness; OcS = active resorption surface; ES = total resorption surface; NOc = osteoclast number; ObS = osteoblast surface; NOb = osteoblast number.
Spleen weight, which was used as an indicator of generalized inflammation, was significantly greater in the IMO groups than in the saline control groups, but did not differ between genotypes (Table 1). Mean (±SEM) initial body weights were 30.5 ± 1.2 gm in the WT mice and 27.8 ± 0.8 gm in the iNOS KO mice. Body weights increased significantly in both the WT and iNOS KO IMO groups compared with controls (Table 1), although this was largely attributable to the weight of the injected MgSiO4 and to localized edema surrounding the sites of MgSiO4 injection. Tibial bone length was unaffected by IMO.
Apoptosis. Apoptosis was detected in 14.0 ± 1.4% and 11.5 ± 1.3% of osteoblasts in WT and iNOS KO mice, respectively (see Figures 1D, 2E, and 2F). In the WT IMO group, the level of osteoblast apoptosis was increased significantly, to 44.8 ± 3.4%, while only a modest increase, to 14.2 ± 3.0%, was noted in the iNOS KO IMO group. The difference between the two IMO groups was highly significant (P < 0.001). Osteoclast apoptosis was not observed in any of the sections. While DAPI staining clearly revealed the presence of osteocytes within bone lacunae, no osteocyte apoptosis was observed.
NO production. Baseline urinary nitrate levels were significantly lower in the iNOS KO mice than in the corresponding WT animals. NO production increased significantly in WT mice injected with MgSiO4, but did not change significantly in iNOS KO mice (Figure 1E). Urinary nitrate levels at day 21 were unchanged from the baseline values on day 0 in the saline-treated mice.
This report is the first to describe the use of mice as a model of IMO caused by the subcutaneous injection of MgSiO4. We found that subcutaneous injections of MgSiO4 caused a generalized inflammatory process associated with splenomegaly and significant bone loss. In WT mice, IMO was accompanied by a marked increase in systemic production of NO, as reflected by urinary nitrate/nitrite excretion, while nitrate/nitrite excretion did not change in iNOS KO mice. This confirms that the previously noted increase in NO production in IMO (6) is due to iNOS activation rather than activation of other NOS isoforms.
The inflammatory process in animals injected with MgSiO4 was associated with significant bone loss, as detected by pQCT, which affected both trabecular and cortical compartments of the bone. In accordance with the pQCT data, histomorphometric analysis showed reductions of trabecular bone volume, trabecular thickness, and cortical thickness in animals with IMO. Inhibition of several parameters of bone formation, including osteoblast number, osteoblast surface, and mineral apposition rate, was also observed. In contrast, indices of bone resorption, including osteoclast numbers and resorption surfaces, were not significantly changed from baseline values. These data suggest that the systemic osteoporosis which develops in response to this inflammation model results primarily from decreased bone formation rather than increased osteoclastic bone resorption, which is consistent with the observations in rats with IMO (8) and in humans with chronic inflammatory disease (11).
The abnormalities of bone turnover in IMO, which are observed at sites distant from inflammatory lesions, contrast with the periarticular increases in osteoclast activity that are characteristic of human RA and adjuvant-induced arthritis in experimental animals (12). Periarticular bone loss in adjuvant arthritis has recently been shown to be due mainly to osteoclast activation mediated by up-regulation of receptor activator of nuclear factor κB ligand (RANKL) expression on activated T cells which infiltrate the inflamed joint (12). The lack of an increase in systemic osteoclastic bone resorption in humans with RA (11) and in the IMO model studied here (8) may result from the fact that activated T cells, which promote osteoclastic bone resorption via RANKL expression, tend to accumulate within inflammatory lesions. An additional explanation could be that the stimulatory effects of cytokines and prostaglandins on systemic bone resorption were counterbalanced by suppressive effects of high NO levels on osteoclast activity within the bone microenvironment (13).
A novel observation to emerge from this study was that the reduced bone formation in WT mice with IMO was associated with a significant increase in osteoblast apoptosis. The increase in osteoblast apoptosis was not observed in iNOS KO mice, which is consistent with the fact that the reduction in BMD that resulted from IMO was significantly less in iNOS KO mice than in WT mice, even though the severity of inflammation—as reflected by spleen weight—was similar in the two genotype groups. It is important to emphasize that iNOS KO mice still exhibited a significant reduction in BMD compared with control animals, indicating that factors other than iNOS activation and osteoblast apoptosis contribute to reduced bone formation and bone loss in this model. Further work will be required to investigate the mechanisms involved, but possibilities include direct inhibitory effects of cytokines on osteoblast growth and differentiation and suppressive effects of increased corticosteroid levels on bone formation (14).
Osteoblast and osteocyte apoptosis have previously been shown to contribute to bone loss in corticosteroid-induced osteoporosis (14), and this study demonstrates that a similar mechanism may be at work in inflammation-induced bone loss. Unlike glucocorticoid-induced osteoporosis, we observed no evidence of an increase in osteocyte apoptosis in this study, nor did we observe an increase in osteoclast apoptosis. This suggests that the deleterious effects of iNOS activation and inflammation on bone may be relatively specific for mature osteoblasts.
These observations are consistent with the results of in vitro studies which have shown clearly that high levels of cytokine-induced NO inhibit osteoblast growth and differentiation as well as stimulating osteoblast apoptosis (5, 13). The mechanisms of NO-dependent apoptosis are becoming increasingly well understood, and apoptosis of osteoblasts is known to involve death receptor– and mitochondria-activated pathways, with caspases as the downstream effectors (15, 16). Further studies will be required to define the precise mechanism of osteoblast apoptosis during chronic inflammation.
In conclusion, we have shown that the osteoporosis resulting from MgSiO4-induced inflammation is primarily mediated by reduced bone formation and that this is accompanied by a dramatic increase in osteoblast apoptosis, mediated by activation of the iNOS pathway. When combined with the results of our previous investigations (6), these data demonstrate the importance of the iNOS pathway as a regulator of bone turnover in vivo and suggest that iNOS inhibitors could be of value in the treatment of bone loss in inflammatory conditions such as RA.
The authors thank Claire Clarkin, Lynne Doverty, and Alistair McKinnon for technical assistance with the histologic sample preparation, and Lorna Smith for assistance with the urinary nitrate measurements.