The authors state that they have no conflicts of interest.
This study investigates the impact of α-CGRP on bone metabolism after implantation of polyethylene particles. α-CGRP knockout mice showed less osteolysis compared with wildtype mice. The local neurogenic microenvironment might be a crucial factor in particle-induced osteolysis.
Introduction: Periprosthetic osteolysis is the major reason for aseptic loosening in joint arthroplasty. This study aimed to investigate the potential impact of α-calcitonin gene–related peptide (α-CGRP) deficiency on bone metabolism under conditions of polyethylene particle–induced osteolysis.
Materials and Methods: We used the murine calvarial osteolysis model based on polyethylene particles in 14 C57BL 6 mice and 14 α-CGRP–deficient mice divided into four groups of 7 mice each. Groups 1 (C57BL/J 6) and 3 (α-CGRP knockout) received sham surgery, and groups 2 (C57BL/J 6) and 4 (α-CGRP knockout) were treated with polyethylene particles. Qualitative and quantitative 3D analyses were performed using μCT. In addition, bone resorption was measured within the midline suture by histological examination. The number of osteoclasts was determined by counting the TRACP+ cells. Calvarial bone was tested for RANKL expression by RT-PCR and immunocytochemistry.
Results: Bone resorption was significantly reduced in α-CGRP–deficient mice compared with their corresponding wildtype C57BL 6 mice as confirmed by histomorphometric data (p < 0.001) and μCT (p < 0.01). Osteoclast numbers were significantly reduced in group 3 and the particle subgroup compared with group 1 (p < 0.001). We observed a >3-fold increase of basal RANKL mRNA levels within group 1 compared with group 3. Additional low RANKL immunochemistry staining was noted in groups 3 and 4.
Conclusions: In conclusion, α-CGRP knockout mice did not show the expected extended osteolysis compared with wildtype mice expressing α-CGRP. One of the most reasonable explanations for the observed decrease in osteolysis could be linked to the osteoprotegerin (OPG)/RANK/RANKL system in α-CGRP–deficient animals. As a consequence, the fine tuning of osteoclasts mediating resorption in α-CGRP–null mice may be deregulated.
Periprosthetic osteolysis is the major reason for aseptic loosening in joint arthroplasty.(1) Particles, especially ultra-high-molecular-weight polyethylene particles (UHMWPE) generated by shear and frictional forces, affect the surrounding tissues by reaching into the articular cavity and, through phagocytosis, into the newly originated capsule. Depending on the number of particles, this initiates an aseptic inflammatory response, which can be compensated for initially, but subsequently leads to an inflammatory destruction of the bone when the number of particles increases.(2) These mechanisms have been the subject of a great deal of research. Calcitonin gene–related peptide (CGRP)-immunoreactive nerve fibers were found in the interface membrane which, as Ahmed et al.(3) concluded, might reflect a pathophysiological response contributing to the aseptic loosening of hip prostheses. Until now, the influence of the nervous system on bone metabolism, especially the neurotransmitter CGRP and its role in polyethylene particle–induced osteolysis, has not been confirmed and described in detail by other research.
CGRP is generated by alternative splicing from the Calca gene in various cells, especially in neuronal cells.(4) The neuropeptide CGRP has multiple physiological roles. CGRP affects the metabolism of skeletal muscle, liver, and kidney and inhibits glycogen synthesis.(5,6) It has been determined previously that CGRP receptors are expressed not only in brain and peripheral tissue but also in the adrenal and pituitary glands and exocrine pancreas.(7,8) Furthermore, CGRP is a potent vasodilatator and neurotrophic effector(9,10) and can act as a mediator in the neurogenic inflammatory response.
In bone, CGRP immunopositive nerve fibers have been found in the vicinity of the periosteum,(11) bone marrow,(12,13) epiphyseal plate, and in all tissues surrounding joints.(14,15) The nerve fibers in the epiphyseal plate indicate the possibility that there is involvement of CGRP in bone remodeling.(11) Bjurholm et al.(16,17) showed an expression of CGRP receptors in osteoblasts that leads to an increase in cAMP and calcium, which is linked to CGRP. Osteoblasts respond to CGRP by increased growth.(18) The interaction between CGRP and bone cells has been elucidated by several groups who described an inhibition of bone resorption in vitro.(10,19,20) Transgenic mice overexpressing CGRP have increased bone formation activity and increased trabecular bone volume and osteoblast activity.(21) Moreover, Schinke et al.(22) showed that α-CGRP–deficient mice display a decreased bone formation rate, thereby showing that α-CGRP is a physiological activator of bone formation.
Furthermore, CGRP inhibits the differentiation and recruitment of osteoclast precursors.(23) The expression of the CGRP receptor on primary osteoclasts has a modulatory function on osteoclasts by a direct mechanism.(23,24)
Taken together, these findings identify CGRP as an anabolic factor in bone metabolism. Therefore, we hypothesized that the anabolic effects of CGRP may contribute to the maintenance of bone in particle-induced osteolysis. To test this hypothesis, we used a murine polyethylene particle–induced osteolysis model and studied the extent of osteolysis in α-CGRP knockout compared with wildtype mice.
MATERIALS AND METHODS
All animal studies were approved by the university's ethics committee and the local authorities according to official guidelines. We used a calvarial model of UHMWPE particle–induced osteolysis in 14 12-wk-old C57BL/J 6 male mice and 14 12-wk-old male α-CGRP–deficient animals. Genotyping of α-CGRP–deficient and their respective wildtype controls was performed as described.(21,25) To avoid any differences that could be attributed to genetic background variations, the α-CGRP–deficient mice were backcrossed more than seven times with C57BL/J 6 animals.(22) The murine calvarial model was recently introduced by our group(26) and was based on the original model of calvarial osteolysis.(27) The animals were divided into four groups of seven. The animals in groups 1 (wildtype) and 3 (α-CGRP–null) underwent sham surgery only, whereas the animals in groups 2 (wildtype) and 4 (α-CGRP–null) received UHMWPE particles.
Commercially pure UHMWPE polyethylene particles (Ceridust VP 3610) were obtained from Clariant (Gersthofen, Germany). A detailed morphological particle description has already been published by our group.(26) To eliminate endotoxins, the particles were washed twice in 70% ethanol for 24 h (ethyl alcohol, absolute, for Molecular Biology; Aldrich Co., Milwaukee, WI, USA) at room temperature using a rocking device (Thermal Rocker; Laboratory Line Instruments, Melrose Park, IL, USA). Testing for endotoxins using a quantitative Limulus Amebocyte Lysate (LAL) assay (Charles River, Kent, UK) at the detection level of <0.25 EU/ml was negative. This endotoxin test is commercially available and was applied according to the manufacturer's directions. Endotoxin-free solutions and materials were used as recommended by the manufacturer. The particles were washed in PBS (8.5 g NaCl, 1.43 g K2HPO4, 0.25 g KH2PO2, and 1000 ml aqua dest., pH 7.2–7.4) three times and subsequently dried in a dessicator.
The surgical procedure has been described previously.(26) Briefly, the mice were anaesthetized with ketamine and scrubbed, and an incision was made over the calvarian sagittal midline suture. In the sham controls (groups 1 and 3), the incision was closed without any further intervention, whereas groups 2 and 4 received 30 μl of dried polyethylene particles; afterward, the incision was sutured, and the animals were returned to their cages.
After death, the mice were decapitated, and the skulls were analyzed by μCT with a SkyScan 1072 scanner and associated analysis software (SkyScan, Aartselaar, Belgium). By scanning the mouse skulls with a magnification of ×25, using an average resolution of 19 μm, shadow images of the skulls were developed with rotation steps of 0.9° following an angle of 180°. During scanning, the skulls were placed in a tightly fitting rigid plastic tube to keep the samples in position. The X-ray source was set at a voltage of 80 kV and a current of 100 μA, using an exposure time of 4.9 s. In the next step, cross-section images were created corresponding to the absorption values using Cone Beam Reconstruction software (SkyScan) based on a Feldkamp algorithm. Furthermore, the software made it possible to display the osseus fraction of the cross-sections by removing artifacts. A total of 1023 cross-sections of the skulls were created with an average section-to-section distance of 19 μm. To accurately quantify the microarchitecture of mouse skulls we used CT-Analyser (SkyScan), which enabled a 3D analysis and calculation of specific morphometric parameters including BMD (BV/TV) and bone surface/bone volume (BS/BV). To produce a visual representation of the results, images of the mouse skulls were developed using CT-Volume and Analyze software (BIR; Mayo Clinic).
Specimen retrieval and histological processing
The calvaria were removed as an elliptical plate of bone defined by the foramen magnum, auditory canals, and orbits.(28) Skin and brain tissue were eliminated. The undecalcified calvaria were cut into four pieces vertical to the midline suture, oriented on edge, and embedded in polymethylmethacrylate. The embedded tissues were cut into 4-μm sections in the coronal plane using a Reichert-Jung microtome (Model 2065; Reichert-Jung, Heidelberg, Germany). The sections were stained with Goldner dye, TRACP, and anti-RANKL and analyzed by transmission light microscopy. The existence and dimension of osteoid formation, granulomatous foreign body reaction, and bone resorption and the presence of various cell types including macrophages, foreign body giant cells, fibroblasts, and osteoclasts were evaluated. In addition, the number of osteoclasts was determined in TRACP-stained sections.
Using a standard high-quality light microscope, the specimens were photographed with a digital camera (Coolpix 995; Nikon, Düsseldorf, Germany). Each section was digitally photographed at a magnification of ×20 with the midline suture in the middle of the field. Histomorphometric analysis was performed using image analysis software (UTHSCA Image Tool, IT Version 3.0; University of Texas, San Antonio, TX, USA). The area of soft tissue, including any bone resorption pits in the midline suture, was traced in Goldner sections as described previously(28,29) to determine the eroded surface in the midline suture. Briefly, using one microscopic field at a magnification of ×10, the regions of interest (i.e., the nonosseous tissue area adjacent to and in continuity with the midline suture) was circled by the operator, and the software automatically recorded and calculated the determined area. Within this field, the number of osteoclasts per bone perimeter was determined. Osteoclasts were identified as large multinucleated TRACP+ cells located on the bone perimeter within a resorption lacunae. The values of each available section were averaged per animal. These averages per animal for bone resorption area and for the number of osteoclasts were averaged again for each group.
Extraction of mRNA and real-time RT-PCR analysis
The mouse bone tissue samples were frozen in liquid nitrogen, pulverized manually, and dissolved in TRIzol Reagent (Invitrogen, Paisley, Scotland). RNA was isolated according to the “TRIzol-method” and purified by using the RNeasy Mini Kit (Qiagen, Hilden, Germany). The purification process included DNase treatment using the RNase free DNase Set (Qiagen). RNA was analyzed by quantitative real-time RT-PCR (Rotorgene Cycler; Corbett Research, Mortlake, Australia). Master Mix and Reverse Transcriptase were components of the one-step QuantiTect SybrGreen RealTime RT-PCR-Kit (Qiagen). For RANKL-mRNA amplification, the validated QuantiTect Primer Assay (Mm_Tnfsf11_1_SG QuantiTect Primer Assay, accession number: M_011613; Qiagen) was used. β-actin mRNA (primer-sequence: 5′-AATCGTGCGTGACATCAAA-3′; 3′-CAAGAAGGAAGGCTGGAAAA-5′; Carl Roth, Karlsruhe, Germany) and GAPDH mRNA (primer sequence 5′-AAATTCAACGGCACAGTCAA-3′, 3′-TCTCCATGGTGGTGAAGACA-5′) were used as references.
Immunochemistry of RANKL was performed using the peroxide technique with diaminobenzidine (DAB) as the cytochrome. Expression of RANKL was examined in serial sections. The anti-RANKL antibody was obtained by Santa Cruz Biotechnology (Heidelberg, Germany), and sections were stained according to Pettit et al.(30)
RANKL expression was evaluated using a grading score in which the count of <10 cells/microscopic field of view was defined as “low grade,” a count of >10 cells but <50% of the total amount of cells/microscopic field of view was graded as “moderate,” and RANKL expression of >50% of the cells/microscopic field of view was determined to be “high grade.”
The results are expressed as mean ± SE. The data were analyzed using a two-tailed Student's t-test and by two-way ANOVA (Microsoft Excel). All p values were compared with an α value of 0.05 for statistical significance.
All 28 mice tolerated the experimental procedures well. There were no problems with wound healing. The animals were killed in a CO2 chamber 14 days after surgery.
Qualitative and quantitative analyses of the mouse skulls were performed. In 2D and 3D μCT reconstruction, polyethylene particle–induced osteolysis was found to appear mainly in the range of the midline suture of the mouse skulls in comparison with the sham procedure as indicated in Fig. 1. Groups 1 and 3 seemed to show no differences in BV/TV values, when comparing the skulls of the wildtype mice (0.127 ± 0.015) with those of the α-CGRP–deficient animals (0.122 ± 0.021). Differences in BMD between groups 3 (0.128 ± 0.015) and 4 (0.122 ± 0.021) also seemed to be marginal. However, a distinct difference between groups 1 and 2 (0.105 ± 0.015) indicated a decrease in the osseous portion within the midline suture of the polyethylene particle group (Fig. 2).
Despite the absence of major alterations in BMD between wildtype and α-CGRP–null animals, we focused our attention on potential changes in bone surface (BS/BV) that are increased in specimens with an eroded surface. Measurements of trabecular constitution were largely neglected because of the specific nature of cranial bone, which lacks a trabecular meshwork. Whereas a statistically significant difference in BV/TV was observed between groups 1 and 2 (p = 0.035), there seemed to be no significant difference in BV/TV between groups 3 and 4 (p = 0.528). UHMWPE particles caused less reduction in bone volume (BV/TV) in group 4 (−4.68% less than group 3) compared with group 2 (−17.32% less than Group 1).
The eroded surface in the group 1 animals (0.07 ± 0.02 mm2) was much smaller than in the group 2 mice (0.13 ± 0.05 mm2), whereas the eroded surface was 0.06 ± 0.01 and 0.07 ± 0.02 mm2 for the group 3 and group 4 animals, respectively (p = 0.043). Statistical analysis showed significantly less osteolysis in α-CGRP–deficient mice (p < 0.001) compared with wildtype mice (Figs. 3 and 4).
The osteoclast number in group 1 (8.05 ± 7.50) was significantly lower than in group 2 (12.8 ± 9.8; p < 0.05), whereas no significant difference was observed for the osteoclast number in groups 3 (3.04 ± 3.86) and 4 (3.34 ± 2.99). Furthermore, significantly fewer TRACP+ cells were observed in α-CGRP–deficient mice (p < 0.001) compared with wildtype animals (Fig. 5).
In the calvarium of wildtype mice, the steady-state level of basal RANKL mRNA expression was 3.2-fold higher than that observed for α-CGRP–deficient animals. Similar observations were seen in the femur, where a 5.8-fold increase in RANKL mRNA expression was seen compared with α-CGRP–deficient mice. Additionally, a 2.6-fold higher basal RANKL mRNA level was also seen in the vertebral body of wildtype mice compared with α-CGRP–deficient animals (Fig. 6).
An increased inflammatory process was observed in the group 2 animals, indicating an osteolytic process after UHMWPE particle implantation. We identified RANKL protein on cells in areas of osteoclast-mediated resorption. RANKL expression was observed in group 1 adjacent to the bone surface. In contrast to group 2, we graded these sections as moderate. According to the grading scale, RANKL expression in group 2 was from moderate to high; this pattern was observed in all sections in this group. A consistent increase in RANKL expression was observed after treatment with polyethylene particles in both groups. In contrast, low staining was noted in groups 3 and 4. The low level of staining was concentrated equally in groups 1 and 4 on cells in areas of osteoclast-mediated bone resorption, suggesting that the reduced polyethylene (PE)-induced osteolysis in the absence of α-CGRP can be explained by decreased RANKL production (Fig. 7).
The role played by UHMWPE wear debris in periprosthetic osteolysis and loosening of total joint replacements, which ultimately results in their long-term failure, has been well established.(31) The process of prosthetic loosening is thought to be mediated by tissue resident macrophages.(31,32) This idea is supported by numerous histological retrieval studies that identified granulomatous tissue rich in macrophages and giant cells corresponding to the amount of intra- and extracellular wear debris. Taken together, the phagocytosis of particles leads to an inflammatory response. To elucidate the mechanisms by which prosthetic loosening occurs, interest has been focused in this study on the nervous system and bone lysis. Recently, Ahmed et al.(3) and Saxler et al.(14) published independent studies concerning sensory nerve fibers within the surrounding granulomatous tissue of loosened prostheses, suggesting that the nervous system may influence bone cells and thus implant loosening.
Imai and Matsusue(33) reported an interaction of CGRP-containing nerve fibers with osteoclasts and osteoblasts. These nerve fibers rapidly perished because of inflammatory cellular infiltration.(34) The CGRP receptor is expressed on the surface of human osteoblasts,(35,36) as well as on primary osteoclasts.(23) Villa et al.(37) reported an excitatory effect of CGRP on the proliferation of the pre-osteoblastic population. Valentijn et al.(38) prevented bone loss after gonadectomy associated with an injection of CGRP in rats. In fracture healing, CGRP and the peripheral innervation are important for the prevention of nonunion fractures.(39–41)
In our study, histological examination and μCT of mouse skulls determined that there is less osteolysis in α-CGRP–deficient mice compared with wildtype mice after implantation of polyethylene particles. The values for bone volume and bone surface in μCT substantiate the results observed in bone histomorphometry. There seems to be no significant difference comparing the skulls of the wildtype mice (group 1) with those lacking α-CGRP expression (group 3). With regard to groups 3 and 4, the mice showed only a slight difference, as indicated by a decreased bone volume. The difference between groups 1 and 2 was more pronounced, showing a clear decrease in bone volume.
The number of osteoclasts is decreased in α-CGRP–null animals with polyethylene particle implantation compared with wildtype littermates with polyethylene. This finding further supports the results obtained using both bone histomorphometry and μCT.
It was an unexpected finding that α-CGRP knockout mice did not show the expected extended osteolysis compared with their wildtype counterparts. Recently, Lerner(42) indicated that the effects of CGRP on the skeleton can be attributed to its effect on osteoblasts. Regarding the normal state of bone quality, α-CGRP–deficient mice generally suffer from osteopenia, as confirmed by histological examination of various bones in an experimental set up recently published by Huebner et al.(43) They examined lumbar vertebral bodies and tibias of three different mouse strains using μCT and histomorphometric analysis. First, they analyzed wildtype mice, and second, mice deficient in Calca−/−, which leads to a lack of α-CGRP and calcitonin (CT), and α-CGRP–deficient mice. They discovered decreased trabecular bone volume, leading to age-related progressive osteopenia. However, in the mice lacking both α-CGRP and CT, as is the state in the Calca −/−-deficient mouse strain, they detected increased trabecular bone volume and reduced osteopenia. The authors concluded from their observations that the two polypeptides derived from the Calca gene play a specific, but distinct, role in bone remodeling.
Similar results have been reported by Schinke et al.(22) using μCT. They detected a low bone mass phenotype in mice lacking α-CGRP and therefore concluded that α-CGRP is a physiological activator of bone formation.
What could be the reason for the observed decrease in osteolysis within the α-CGRP–deficient mice after UHMWPE implantation in our study?
In our experiments, we used the α-CGRP–null mouse generated by Lu et al.(25) These mice showed no obvious phenotypic difference from their wildtype littermates. Neuromuscular junction morphoplogy and receptors, sprouting in response to denervation, revealed no differences in α-CGRP–deficient animals. There is no significant difference in calcitonin levels compared with the wildtype mice. The same is true for PTH, total calcium, and inorganic phosphate.(22) Therefore, one explanation for decreased osteolysis could be linked to the osteoprotegerin (OPG)/RANK/RANKL system in α-CGRP–deficient mice. We examined α-CGRP–deficient and wildtype mice and observed a 3.2-fold increased basal RANKL mRNA level within the wildtype mice calvarium compared with the CGRP-deficient animals using RT-PCR.
The expression of RANKL protein at the sites of osteolysis has further been shown by immunochemistry. We showed RANKL expression especially close to TRACP+ multinucleated osteoclast-like cells. The immunochemistry used for these studies is unlikely to allow detection of soluble RANKL protein. RANKL bound by RANK and OPG also cannot be detected. Therefore, the RANKL detected in this study reflects only the unbound and presumably active membrane-associated form of the protein.
Osteoclasts have been reported to be responsible for osteolysis induced by macrophages. The activity of osteoclasts is dependent on RANKL.(44,45) This peptide stimulates precursor cells to differentiate into mature osteoclasts.(45,46) It stimulates osteoclast migration, fusion, and activation. The expression of RANKL may be the essential link to the polyethylene particle–induced osteolysis mediated by activated osteoclasts. It may cooperate with other pro-inflammatory cytokines and drive the differentiation and function of osteoclasts.
Perhaps RANKL stimulates the osteoclasts in wildtype mice and thereby leads to extended bone lysis compared with α-CGRP–null animals.
Another hypothesis is that the α-CGRP–deficient mice may overexpress osteoprotegerin. This cytokine is the antagonist of RANKL and effectively blocks the pre-osteoclastic action at all stages of osteoclast generation and function.(47–49) Further studies are needed to fully understand the RANK/RANKL/OPG system in α-CGRP knockout mice under the condition of polyethylene particle–induced osteolysis. Another possibility is that β-CGRP, another peptide from the calcitonin family, stimulates osteoblasts or inhibits osteoclasts. This peptide is still expressed by α-CGRP knockout mice. The influence of β-CGRP in polyethylene particle–induced osteolysis will have to be studied in upcoming experiments.
It should be taken into account that our study had certain limitations. This model uses a flat bone that is formed by membranous rather than endochondral ossification. Therefore, our results should be interpreted with caution.
However, we assume the local neurogenic microenvironment to be a significant factor in bone remodeling of loose implants. Although the extensive pathophysiology of aseptic loosening is not solely caused by neuropeptides, they may play an important role in the progress of implant loosening. In the case of α-CGRP, this should be confirmed in a further similar experimental set-up and an additional in situ application of α-CGRP. Under these conditions, the impact of α-CGRP in particle-induced osteolysis could be examined in detail, because we expect that there could be coherence between various neurotransmitters and particle-induced osteolysis.
This study was supported by AG Biomaterialien NRW/Ministerium für Forschung und Wissenschaft and IFORES/University of Duisburg-Essen. The authors thank Dr Ronald Emeson (Vanderbilt University) and Prof Michael Amling (University Medical Center Hamburg-Eppendorf) for providing the α-CGRP–deficient mice used for these studies, Kaye Schreyer for editorial assistance with the manuscript, and Priv Doz Dr Frank Henschke and Sylvia Marks for technical assistance during histological and immunhistochemical processing of the specimens.