The determinants of cancellous bone turnover and trabecular structure are not understood in normal bone or skeletal disease. Bone remodeling is initiated by osteoclastic resorption followed by osteoblastic formation of new bone. Receptor activator of nuclear factor κB ligand (RANKL) is a newly described regulator of osteoclast formation and function, the activity of which appears to be a balance between interaction with its receptor RANK and with an antagonist binding protein osteoprotegerin (OPG). Therefore, we have examined the relationship between the expression of RANKL, RANK, and OPG and indices of bone structure and turnover in human cancellous bone from the proximal femur. Bone samples were obtained from individuals with osteoarthritis (OA) at joint replacement surgery and from autopsy controls. Histomorphometric analysis of these samples showed that eroded surface (ES/BS) and osteoid surface (OS/BS) were positively associated in both control (p < 0.001) and OA (p < 0.02), indicating that the processes of bone resorption and bone formation remain coupled in OA, as they are in controls. RANKL, OPG, and RANK messenger RNA (mRNA) were abundant in human cancellous bone, with significant differences between control and OA individuals. In coplotting the molecular and histomorphometric data, strong associations were found between the ratio of RANKL/OPG mRNA and the indices of bone turnover (RANKL/OPG vs. ES/BS: r = 0.93, p < 0.001; RANKL/OPG vs. OS/BS: r = 0.80, p < 0.001). These relationships were not evident in trabecular bone from severe OA, suggesting that bone turnover may be regulated differently in this disease. We propose that the effective concentration of RANKL is related causally to bone turnover.
OSTEOARTHRITIS (OA) is a common skeletal disease in both men and women, characterized by progressive degenerative damage to the articular joint cartilage and in severe cases, considerable pain and disability. The etiology of OA is not understood and there are no proven treatments to reduce the severity or slow the progression of this disease. For many sufferers of hip and knee OA, symptoms become sufficiently severe to require resolution by joint replacement surgery.
Although the major focus in OA has been on the articular cartilage, the disease also is associated consistently with marked changes in the subchondral bone. Analysis of femoral heads in hip OA has shown that the subchondral cancellous bone typically is sclerotic and trabecular bone volume is increased compared with age-matched controls.(1–3) It has been postulated that such changes are caused by alteration of loading through the joint because of the articular disease(4) or, alternatively, that they are secondary to pathology within the joint. However, similar changes also are found at sites distal to the joint articular surface, in the proximal femur, and in the iliac crest,(2, 5) suggesting that they are not reactive simply to the joint pathology. The different bone structure in hip OA is accompanied by altered mechanical properties, with a more rigid cancellous bone.(6) Because this more rigid bone would have a reduced ability to absorb shock, it has been postulated that the bone changes in OA might exacerbate the disease or could even precede and be causative of the cartilage degeneration.(7, 8)
The cellular and molecular mechanisms that lead to particular trabecular structures in healthy bone or in the pathology of OA are not well understood. However, now there is a large amount of information about the factors that are capable of regulating the differentiation and activity of the cell types that are responsible for the remodeling of bone, the osteoblast, and the osteoclast. It is understood that physiological bone remodeling is achieved by the initial activity of osteoclasts to resorb bone followed by the formation of new bone by osteoblasts and that these processes are strongly interrelated. Signals that promote osteoclast differentiation appear to act via receptors expressed by cells of the osteoblast lineage,(9) suggesting that in situ, localized activation of osteoclastogenesis leads to site-directed remodeling.
Considerable progress has been made toward an understanding of the mechanisms responsible for the formation and activation of osteoclasts. A large number of hormones and cytokines have been identified that can stimulate the development of osteoclasts from their hematopoietic precursors.(9, 10) Recently, a cell-surface member of the tumor necrosis factor (TNF) ligand family, termed receptor activator of nuclear factor κB ligand (RANKL),(11) or variously termed TNF-related activation-induced cytokine osteoprotegerin ligand/osteoclast differentiation factor (TRANCE(12)/OPGL(13)/ODF(14)) was shown to be central in both osteoclast development and activity.(13) Agents that induce bone resorption have been shown to induce the presentation of RANKL on the surface of osteoblastic cells, which are then able to promote osteoclast formation.(14, 15) RANKL binds to a TNF-receptor superfamily member RANK(16) that is expressed on the surface of osteoclasts and their precursors.(17) RANK has been shown to be essential for osteoclast formation, and RANK-deficient mice exhibit profound osteopetrosis.(18) Likewise, a natural RANKL antagonist, a soluble TNF-receptor family member termed OPG, can inhibit osteoclast formation and bone resorption.(19) Overexpression of OPG in mice resulted in an osteopetrotic phenotype,(19) whereas mice in which the OPG gene is deleted develop extensive osteoporosis and increased numbers of osteoclasts.(20) OPG is expressed by a wide variety of cell types, including osteoblasts, in which its expression is down-regulated by many of the same factors that promote bone resorption and RANKL expression.(21) Thus, there is now good evidence that the local amount of RANKL, relative to OPG, is important in osteoclast formation.(22)
We set out to examine the relationship in human cancellous bone between the expression of RANKL, OPG, and RANK and parameters of bone structure and turnover. We found strong associations between the ratio of RANKL/OPG messenger RNA (mRNA) and indices of bone turnover, consistent with the notion that the corresponding RANKL and OPG molecules have a central role in human bone remodeling. In contrast, these relationships were not evident in trabecular bone from severe OA of the hip, suggesting that bone turnover may be regulated differently in this disease.
MATERIALS AND METHODS
Human bone specimens
Proximal femur surgical specimens were obtained from 16 patients (8 from the left side and 8 from the right side) undergoing total hip arthroplasty for primary OA (8 women, 49-84 years old, and 8 men, 50-85 years old; mean ± SD; age, 68.6 ± 11.8 years). The control group (11 left and 2 right) was selected from routine autopsy cases (7 women, 20-83 years old, and 6 men, 24-85 years old; 60.7 ± 21.1 years), not known to have suffered from any disease affecting the skeleton. The mean age of the OA group did not significantly differ from the control group. The intertrochanteric region was chosen for sampling because the cancellous architecture in this region depends on stresses in the proximal femoral shaft while being unaffected by the sclerotic and cystic changes of the osteoarthritic femoral head.(23) The surgical and autopsy femoral heads were graded macroscopically for OA by assessment of the location and degree of fibrillation and degeneration according to the criteria of Collins.(24) At surgery, primary OA femoral heads were either grade III or grade IV, characterized by cartilage loss, eburnation of bone, osteophytes, and remodeling of the articular contour; none of the autopsy femoral heads had worse than grade II OA.
A 10-mm tube saw core biopsy specimen from the intertrochanteric region of the femur was taken for each OA case at surgery. For the control group, a cancellous bone sample was cut with a band saw from the same site. Each biopsy specimen was divided lengthwise for tissue processing and molecular analyses. There was insufficient bone tissue for histology for one OA case and one control case. For histology, the tissue sample was placed in 70% ethanol and the remainder was processed for total RNA extraction. Fixed tissue was processed undecalcified, embedded in K-plast resin, and sectioned on a Jung K microtome (Reichert Jung, Heidelburg, West Germany). Sections, 5 μm thick, were stained by the von Kossa silver method and counterstained with hematoxylin and eosin to distinguish between the mineralized bone, the osteoid, and the cellular components of the marrow. Morphometric analysis was performed using an ocular-mounted 10 × 10 graticule at a magnification of ×100. Estimates were made of the following parameters: percentage of bone tissue volume (BV/TV); surface density of bone in squared millimeters per cubed millimeters (BS/TV); specific surface of bone in squared millimeters per cubed millimeters (BS/BV); trabecular thickness (Tb.Th) in microns; trabecular separation (Tb.Sp) in microns; trabecular number (Tb.N) in number per millimeter; percentage of osteoid volume (OV/TV); percentage of osteoid surface (OS/BS); and percentage of eroded surface (ES/BS).
High-quality total RNA was isolated from fresh surgical cancellous bone (after storage at 4°C up to 24 h in sterile RNAse-free phosphate-buffered saline [PBS]) and control bone (obtained 24-96 h after death) samples, as we have described previously.(25) Briefly, bone fragments were placed in 4 M guanidinium thiocyanate solution and homogenized using an Ultra-Turrax (TP 18-10; Janke & Kunkel, KG, IKA WERK, Staufen i. Breisgau, Germany). The homogenized sample was clarified by centrifugation (1000g for 5 minutes), vortexed in 0.1 vol of 2 M sodium acetate, pH 4.0, and then extracted with 1 vol of phenol and 0.2 vol of chloroform/isoamylalcohol (49:1). Total RNA was precipitated with isopropanol, resuspended in 1 × 10 mM Tris-HCl, 1 mM EDTA (TE) containing 0.1 vol of 3 M sodium acetate, pH 5.2, then reextracted with 0.5 vol phenol, followed by 0.5 vol chloroform/isoamylalcohol. The RNA was then extracted with 3 vol of 4M sodium acetate, pH 7.0, to remove contaminating proteoglycans, and then precipitated at −20°C overnight. Total RNA was recovered by centrifugation, washed with 75% ethanol, air-dried, dissolved in diethylpyrocarbonate (DEPC)-treated water, and stored at −70°C until further use.
Reverse transcription-polymerase chain reaction
First-strand complementary DNA (cDNA) was synthesized from 1.6 μg of total RNA isolated from 13 OA and 12 control bone samples, using a cDNA synthesis kit, as per manufacturer's instructions (cat. no. A3500; Promega Corp., Madison, WI, USA). cDNA was then amplified by polymerase chain reaction (PCR) to generate products corresponding to mRNA encoding human OPG, RANKL, RANK, and the housekeeping gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH), using oligonucleotide primer sequences and reaction conditions, as we have previously described.(26) GAPDH was chosen to ensure that the various mRNA levels were normalized against the total mRNA content in the samples. PCR was performed for 26 cycles for GAPDH and 30 cycles for each of OPG, RANKL, and RANK. In all cases, the primers were mRNA-specific in that the recognition sites of the upstream and downstream primers resided in separate exons or at intron/exon boundaries in the genomic sequence. To allow semiquantitation of the PCR products, preliminary experiments were performed to ensure that the number of PCR cycles were within the exponential phase of the amplification curve. Amplification products were resolved by electrophoresis on a 2% wt/vol agarose gel poststained with SYBR-1 Green (cat. no. S-7565; Molecular Probes, Eugene, OR, USA). The intensity of the PCR products was quantified using FluorImager and ImageQuant software (Molecular Dynamics, Sunnyvale, CA, USA). Amplified products corresponding to OPG, RANKL, and RANK mRNA are represented as a ratio of the respective PCR product/GAPDH PCR product. It was important for this method that all samples in each amplification reaction were electrophoresed in a single gel, and good interassay variability was obtained by adhering to this procedure. To validate the extraction of RNA from autopsy bone, we confirmed that there was no evidence for any significant degradation of the RNA according to ethidium bromide-stained formaldehyde gels or the differential degradation of particular mRNA species, as we have described previously.(25)
The statistical significance of differences between the control and experimental groups were determined by Student's t-test. Regression analysis was used to examine the relationship between PCR products and bone histomorphometric variables (EXCEL; Microsoft Corp., Redmond, WA, USA). The critical value for significance was chosen for p = 0.05.
In this study we sought to examine the relationship between indices of human cancellous bone structure and remodeling and gene expression of the cytokines RANKL, OPG, and RANK, which are potentially involved in regulating bone turnover in the bone microenvironment. We have previously reported on bone histomorphometric parameters in trabecular bone samples taken from a number of sites in the proximal femur and the iliac crest.(1, 5, 27) In this study we have compared the bone architecture and bone remodeling in OA with control individuals, specifically in cancellous bone samples isolated from the intertrochanteric region of the proximal femur.
Comparison of mean bone structural indices and bone remodeling indices between OA and control individuals
As described in the Materials and Methods section, histomorphometry was performed on cancellous bone samples obtained from patients with severe OA undergoing total hip arthroplasty and from controls without evidence of severe OA pathology taken at autopsy. The mean values for the structural parameters of trabecular bone at this site were similar for the OA and control groups (Table 1). To enable comparison of age-matched control and OA groups, analysis also was performed after exclusion of the two younger control individuals (Table 1). In the latter comparison, OA bone had significantly decreased mean Tb.Sp (p < 0.03), consistent with our previous reports.(5) The static indices of bone turnover, OV/TV, OS/BS, and ES/BS showed the same overall relationships (Table 1), whether all samples were included in the analysis or only those from individuals greater than 40 years of age. In both cases, ES/BS was significantly decreased in the OA group (p < 0.01), indicating reduced bone resorption at this bone site in OA. Although the mean values for parameters corresponding to bone formation, OV/TV, and OS/BS were higher in OA than in the controls, these did not reach a difference of statistical significance.
Table Table 1.. Trabecular Bone Structure and Bone Turnover Indices in OA and Control Intertrochanteric Cancellous Bone Samples
Age dependence of bone structural indices and bone remodeling indices in OA and control individuals
When the histomorphometric measures were plotted against age, the following relationships emerged. First, we found a significant decrease in BV/TV with age for the controls (r = −0.70; p < 0.008; Fig. 1), as reported previously.(2) This relationship was not seen for the age-matched control cohort, which excludes the two younger samples, and the OA group (Fig. 1). In the case of ES/BS, there was a significant increase with age (r = 0.65; p < 0.02) for controls, indicating an increased extent of bone resorption with age (Fig. 2A). This relationship was not observed for the age-matched control and OA groups (Fig. 2A). The OA values for ES/BS segregated such that they were, in general, lower than the control values in each corresponding age range, which was consistent with the difference in mean values discussed previously (Table 1). For OS/BS there was a significant increase with age for controls (r = 0.61; p < 0.03; Fig. 2B), indicating an increased extent of bone formation with age. This relationship was maintained even after exclusion of the two younger control individuals. This, together with an age-dependent increase in ES/BS, indicates an increase in the rate of bone turnover with aging in control individuals. In the OA samples, OS/BS was not significantly correlated with age. Despite the lack of correlation with age in OA for either ES/BS or OS/BS, ES/BS significantly correlated with OS/BS for both the OA and the control groups (r = 0.57 and p < 0.02 and r = 0.79 and p < 0.001, respectively; Fig. 3). This indicates that the processes of bone resorption and bone formation remain coupled in OA, as they are in controls. However, the intercepts of the regression lines shown in Fig. 3 are different (p < 0.003), indicating that for a given amount of resorption there is more bone-forming surface in the OA group than for the control group (Fig. 3). These results have important implications for bone remodeling in OA.
Expression of RANKL, OPG, and RANK mRNA in OA and control trabecular bone
Reverse-transcription (RT)-PCR was used to examine the expression of RANKL, OPG, and RANK mRNA in cancellous bone samples contiguous with those used for histomorphometric analysis from OA patients and autopsy control individuals. Because the results of this study rely on the ability to reliably harvest mRNA species from autopsy bone, we were careful to exclude the possibility of either overall RNA degradation or targeted degradation of specific mRNA species in the bone samples. Thus, RNA samples were used only where there was no evidence of degradation on ethidium bromide-stained gels. In addition, extensive testing showed conclusively that none of the mRNA species of interest was selectively degraded in bone samples stored at 4°C and there was no evidence for selective degradation between the time of death and autopsy in non-OA controls.(25)
mRNA corresponding to each of RANKL, OPG, and RANK was abundant in cancellous bone samples analyzed from the proximal femur. This is the first direct demonstration that these molecules are expressed in the human bone microenvironment. Typical PCR products corresponding to mRNA expression for RANKL, OPG, and RANK for the OA and control groups are shown in Fig. 4. The PCR products were authenticated by comparison with the expected product size and by hybridization to an internal oligonucleotide probe. The number of PCR cycles used for each of these transcripts was within the exponential phase of the amplification curve, enabling comparison of expression between samples. Relative levels of RANKL, OPG, and RANK mRNA were determined by normalizing values to the GAPDH mRNA level determined for each sample.
Expression of OPG, RANKL, and RANK mRNA in females and males
The pooled OA and control group data for OPG/GAPDH mRNA showed no difference between females and males (female: 0.27 ± 0.14 [mean ± SD], n = 14; male: 0.25 ± 0.24, n = 11; p = NS). There also was no difference for RANKL/GAPDH (female, 0.66 ± 0.47; male, 0.55 ± 0.54; p = NS) or RANK/GAPDH (female, 0.48 ± 0.30; male, 0.55 ± 0.47; p = NS) mRNA expression. In addition, when the data were separated into the OA and control groups, no significant difference was found between females and males. Based on these comparisons, analyses of the data were made, independent of sex, for the OA and control groups.
RANK mRNA expression is elevated in OA bone
The mean values obtained for RANKL, OPG, and RANK mRNA expression in the OA and control groups are shown in Table 2. In OA, the mean mRNA expression of RANK/GAPDH was significantly elevated in comparison with the control group (p < 0.01). In contrast, there was no significant difference in mean RANKL/GAPDH or OPG/GAPDH mRNA levels between the OA and control groups. There was comparable variance from the mean for RANKL mRNA in OA and the controls.
Table Table 2.. RT-PCR Product/GAPDH Ratio in OA and Control Intertrochanteric Cancellous Bone Samples
Relationships between RANKL versus OPG mRNA, RANK versus OPG mRNA, and RANKL versus RANK mRNA expression in the proximal femur
When the results for RANKL and OPG mRNA, and RANKL and RANK mRNA were coplotted, significant positive associations were observed. First, RANKL/GAPDH mRNA versus OPG/GAPDH mRNA values were positively associated for both control and OA samples (r = 0.55 and p < 0.05 and r = 0.79 and p < 0.001, respectively; Fig. 5A). However, the slope of the regression line for the control samples was greater than for the OA samples (p < 0.003) so that for a given level of RANKL mRNA, the corresponding level of OPG mRNA was greater in OA samples. Per unit of OPG gene expression, the level of RANKL gene expression in OA was about half that in the controls (regression slopes = 1.67 for OA vs. regression slopes = 3.06 for controls). Taken together, these results suggest that there is a relationship between the positive and negative osteoclastic influences in the bone microenvironment and that the inhibitory influence is relatively greater in OA. When the RANKL/GAPDH mRNA levels were plotted versus RANK/GAPDH mRNA, the level of RANKL gene expression increased in parallel as the level of RANK gene expression increased in both OA and controls (r = 0.95 and p < 0.001 and r = 0.66 and p < 0.01, respectively; Fig. 5B). However, the results segregated such that for a given level of RANK mRNA, the level of RANKL mRNA was higher in the control group. Hence, for a given level of OPG or RANK gene expression, there was a lower level of RANKL gene expression in OA compared with controls (Figs. 5A and 5B). The plot of RANK/GAPDH mRNA versus OPG/GAPDH mRNA expression shows the level of RANK gene expression associated positively with OPG gene expression in OA (r = 0.74; p < 0.003) but not in controls (Fig. 5C). These data suggest that in vivo, as has been shown in vitro,(14) a competitive binding between RANK and OPG with RANKL is a significant factor in osteoclastogenesis. The mean ratios of RANKL/OPG mRNA and RANKL/RANK mRNA were significantly lower (p < 0.02 and p < 0.001, respectively), and RANK/OPG mRNA was significantly increased (p < 0.03) in OA (Table 3). As discussed in more detail in the following sections, there is now good evidence in cell culture model systems that it is the effective concentration of RANKL, as reflected by the ratio of RANKL/OPG, that is of central importance in osteoclastogenesis and osteoclast activity.(22)
Table Table 3.. RT-PCR Product Ratios in OA and Control Intertrochanteric Cancellous Bone Samples
Expression of RANKL, OPG, and RANK mRNA as a function of age
Plotting the values corresponding to mRNA levels for RANKL, OPG, and RANK as a function of age showed that the RANKL/GAPDH mRNA ratio increased with age in the control group (r = 0.62; p < 0.02; Fig. 6A), although this relationship was dependent on inclusion of the two younger samples. There was a positive association between RANK/GAPDH mRNA expression and age for the OA samples (r = 0.58; p < 0.04; Fig. 6C) and the level of RANK gene expression tended to be higher than controls at each age. OPG/GAPDH mRNA showed no dependence on age (Fig. 6B) and remained relatively constant with aging. It should be noted that although the mean OPG/GAPDH mRNA expression was similar in both groups (Table 2), the variance in this value in OA was nine times the variance in controls so that the level of OPG gene expression may be more tightly regulated during aging in controls than in OA.
Because, as mentioned previously, it is the ratio of RANKL/OPG that appears to be of central importance in osteoclastogenesis and osteoclast activity, it was of great interest to investigate this relationship during aging. It was found that the ratio of RANKL/OPG mRNA levels increased in parallel as age increased in both OA (r = 0.58; p < 0.03) and the controls (r = 0.60; p < 0.03; Fig. 7A), although this relationship in the controls was dependent on inclusion of the two younger samples. Importantly, the values segregated such that the RANKL/OPG mRNA levels in control bone were higher than for OA across the entire age range examined, consistent with our observation of reduced bone resorption in OA, compared with controls. The RANKL/RANK mRNA ratio was age-dependent in OA (r = 0.58; p < 0.04) but was not dependent on age in controls (Fig. 7B), which was reflected in the 24 times greater variance in this value in controls than OA (Table 3). More importantly, RANKL/RANK mRNA versus age showed a clear segregation of values between controls and OA individuals, such that the values for OA were lower than controls across the age range examined. RANK/OPG mRNA was not age dependent in either group (Fig. 7C).
Bone volume and bone turnover correlate with the ratio of RANKL/OPG mRNA expression in control bone but not in OA
Because it is the ratio of RANKL/OPG that determines the effective level of RANKL, we explored the relationship between the RANKL/OPG mRNA ratio and indices of trabecular bone volume and turnover. There were 11 control and 12 OA cases in which both molecular and histomorphometric data were available for analysis. The plot of BV/TV versus RANKL/OPG showed a significant decrease in controls (r = −0.67; p < 0.05; Fig. 8). However, no relationship was seen in the OA samples or in the age-matched control cohort excluding the two younger samples. A strong association was found between both ES/BS and RANKL/OPG mRNA (r = 0.93; p < 0.001) and OS/BS and RANKL/OPG mRNA (r = 0.80; p < 0.001) for control bone samples. In both cases, there was a strong positive correlation, suggesting that the ratio of RANKL/OPG may well be of central importance in bone remodeling (Figs. 9A and 9C). The relationships between ES/BS and RANKL/OPG mRNA levels and OS/BS and RANKL/OPG mRNA levels were determined by the levels of RANKL mRNA. In neither OA nor controls was there an observable relationship between ES/BS or OS/BS and OPG mRNA or RANK mRNA. However, ES/BS increased in the control samples as the RANKL/GAPDH mRNA ratio increased (r = 0.83; p < 0.001; results not shown) and OS/BS also was found to increase in controls with the RANKL/GAPDH mRNA ratio (r = 0.84; p < 0.001; results not shown).
The relationships between RANKL/OPG mRNA and ES/BS and OS/BS were not evident for the OA samples (Figs. 9B and 9D). We propose that there may be a fundamental difference in the regulation of bone turnover and remodeling in trabecular bone in OA.
Presently, the etiology of OA is unknown. It is possible that the different presentations of OA, with different spectra of joints involved, indicate that OA is actually a cluster of diseases, with the common feature of relatively noninflammatory degeneration of the articular cartilage. Thus, OA of the hand, OA of the hip, and generalized OA appear to have different genetic bases.(28, 29) Despite these differences, a common feature of OA is a set of changes in the subchondral trabecular bone. Characteristically, the subchondral bone is sclerotic, with increased Tb.Th and decreased intertrabecular spacing.(1) The altered trabecular architecture is associated with altered mechanical properties, such that the tissue is rendered more rigid and less able to absorb shock.(6) These changes in the subchondral bone therefore may exacerbate the disease by transferring more of the load to the articular surface. In hip OA, it has been found that the trabecular bone more distal from the affected joint surface also is reproducibly different from bone in the corresponding site in control individuals.(5) We have described similar architectural changes as found in the femoral head in the intertrochanteric region,(2) more than 15 cm from the joint surface, as well as in the iliac crest. These findings in severe OA have led several investigators to hypothesize that the bone changes in OA may precede and even be causative of the degeneration of the articular cartilage, as first proposed by Radin et al.(7) Thus, the trabecular bone changes may precede the joint degeneration of OA or may arise secondarily to the joint pathology or indeed may occur in parallel with the cartilage damage, driven by the same causative agent(s) that led to cartilage disease. Whichever of these is the case, to devise effective treatments for OA, clearly, it is important to develop an understanding of the processes that lead to the bony component of this disease.
Previous studies have described the relative maintenance with age of trabecular bone volume in OA, compared with the well-known decrease in trabecular bone volume with age in non-OA controls.(2) Although we have confirmed the relationship between BV/TV with age for control subjects without progressive OA, comparison with OA subjects will require a greater sample size than the current study. In addition to the conservation of trabecular volume in OA, the femoral cancellous bone adjacent to the intertrochanteric region in OA is reported to have less well-mineralized bone than non-OA bone. However, bone in OA with increased BV/TV is structurally more rigid.(2, 6) Our results suggest that the principal reason for these differences in trabecular bone in OA may be an altered bone turnover balance, where a decrease in eroded surface (ES/BS), relative to bone formation (OS/BS; Fig. 3), could result in maintenance of BV/TV. This suggestion would be consistent with other studies of the hip and the iliac crest, which provided evidence that bone volume in OA is increased or maintained as a result of reduced bone turnover.(30, 31) Because those observations were made at sites distant from the joint, they were thought not to be subject to loading abnormalities or simply reactive to the joint pathology.
Our data suggest that the decreased ES/BS, relative to OS/BS, in OA primarily is caused by reduced resorption surface in OA (Fig. 3). We interpret these data as describing events within the bone remodeling structures known as basic multicellular units (BMUs), in which both osteoclastic resorption and osteoblastic bone formation occur.(32) Our results focus on cytokines that promote osteoclast formation and activity. Almost certainly, signals also control the rate and extent of bone formation by osteoblasts, although the nature of these signals has not been determined. It is artificial to separate, conceptually or functionally, the processes of bone resorption and formation. However, as shown previously,(33) and now by our results, the relative extent of the two processes may alter under particular circumstances and disease states. Therefore, we were interested in exploring the molecular basis for the difference in resorption between the two groups. After the identification of RANKL as a central player in osteoclastic formation and activity, we designed experiments to investigate its role in the human bone microenvironment. Our data provide new insights into the expression of RANKL, as well as its receptor/effector partner RANK, and OPG, the natural inhibitor of RANKL, in cancellous bone. We have shown for the first time that mRNA corresponding to all three molecules is expressed abundantly in human cancellous bone. Although this result might have been expected, based on a growing amount of information derived largely from cell culture systems and from nonhuman species (reviewed by Hofbauer et al.),(22) the way in which the expression of these molecules relates to events in the human bone microenvironment needs to be established.
The data indicate strong positive associations between the expression of RANKL and OPG mRNA, as well as RANKL and RANK mRNA, in the cancellous bone of the proximal femur. This suggests that, as OPG and RANK increase, RANKL also is increased. However, we found that resorption increases as the RANKL/OPG ratio increases. The changes in the magnitude of this ratio can be the result of changes in the magnitude of RANKL and OPG independently or linked. The data show that the effective osteoclastogenic influence, represented by the ratio of RANKL/OPG, is positively associated with age in both control and OA bone and that this is largely driven by the changing RANKL levels. Experiments in vitro, in which bone marrow stromal cells and osteoblast-like cells were treated with proresorptive agents, have shown concomitant up-regulation of RANKL mRNA and down-regulation of OPG mRNA or up-regulation of RANKL with no change or even an increase in OPG, with the end result in each case an increase in the RANKL/OPG mRNA ratio.(15, 34) In interpreting in vivo results, it is important to note that while RANKL is expressed by a relatively restricted number of cell types, including stromal/osteoblastic cells and T cells, OPG is expressed widely by many cell types. The lower ratio of RANKL/OPG mRNA in OA (Table 3) occurs in the context of no statistical difference in the mean RANKL/GAPDH and OPG/GAPDH mRNA expression (Table 2). This is indicative of a robust and multidimensional control process operating to maintain the integrity of the skeleton. The association between RANKL mRNA and RANK mRNA levels (Fig. 5) suggests either that the same stimulus regulates the expression of both molecules or, more likely, that the same stimulus up-regulates RANKL and causes recruitment of RANK-expressing osteoclast precursor cells. Although these overall relationships were observed in both control and OA bone, the data show clear differences. For a given level of RANKL mRNA there was a higher level of OPG mRNA in OA, consistent with a lower net osteoclastogenic influence in OA samples. In fact, we recently have reported that OA bone has lower levels than control of mRNA for interleukin-6 (IL-6) and IL-11,(25) both of which act on cells of the osteoblast lineage to promote osteoclast formation.(10)
Coplots of the molecular measures of RANKL, OPG, and RANK mRNA and indices of bone structure and turnover revealed strong associations in bone samples from control individuals. The bone resorption index ES/BS was strongly positively associated with RANKL/OPG mRNA levels and this was found to be almost entirely because of the RANKL component. The ratio of RANKL/OPG, as already discussed, is likely to represent the actual local osteoclastogenic influence. This relationship strongly supports the concept that our surrogate measures of RANKL and OPG mRNA levels relate directly to levels of expression of the corresponding proteins in the bone tissue and that the effective concentration of RANKL is dominant in controlling the extent of bone resorption in trabecular bone. It also was interesting to find that the index of bone formation OS/BS was also strongly positively associated with the RANKL/OPG mRNA ratio. This result is consistent with the notion that bone resorption and bone formation are tightly coupled in the human bone microenvironment and that bone turnover is initiated by bone resorption, which is in turn dependent on the effective levels of RANKL. It is important to note here that the molecular indices are completely consistent with the histomorphometric parameters, a highly significant finding given the totally different means of assessing these two types of data. Most importantly, the data shown in Figs. 8 and 9 indicate that the relationships between RANKL/OPG and bone turnover were not evident for the OA bone samples. We thus conclude that the molecular mechanisms of bone turnover may be different fundamentally in OA. We further speculate that the trabecular bone structures in OA arise by subversion of the physiological RANKL-controlled mechanisms.
The mechanisms that lead to the particular shape and structure of bony trabeculae in general and in the proximal hip in particular are clearly complex, involving both mechanical and chemical inputs. However, we have obtained strong evidence in this study that RANKL may be of central importance in determining the trabecular volume and turnover at this site. The strikingly different results obtained with bone from individuals with severe OA suggest that our experimental approach of molecular histomorphometry has great potential to point to the mechanisms that lead to the altered bone structures found in this disease. If, as we hypothesize, this altered bone is an early event in OA, causing or exacerbating the condition, an understanding of these mechanisms, together with early recognition of potential OA sufferers, may enable preventative treatment of these individuals in the future.
The authors thank the Orthopedic Surgeons and Sister Lynn Betts in the Royal Adelaide Hospital for their support and cooperation in the collection of femoral specimens and the Mortuary Staff of the Institute of Medical and Veterinary Science for the collection of autopsy specimens. This work was supported in part by the Adelaide University Faculty of Medicine Research Committee, Adelaide Bone and Joint Research Foundation, Royal Adelaide Hospital and Arthritis Foundation of Australia.