Colony-stimulating factor-1 (CSF-1), also called macrophage colony-stimulating factor, is required for growth, differentiation, activation, and survival of cells of the mononuclear phagocytic system. This cytokine has been shown to be essential for osteoclast development as well as for inducing both proliferation and differentiation of osteoclast progenitors. It also sustains survival of mature osteoclasts and stimulates spreading and migration of these cells. In the present in vitro study, the formation of large tartrate-resistant acid phosphatase (TRAP)-positive cells with a high number of nuclei was observed when osteoclasts isolated from rat long bones were incubated with CSF-1. These large cells, cultured on plastic, bind calcitonin and form F-actin along the edges of the cells. Fusion to such large TRAP-positive multinucleated cells in the presence of CSF-1 and the formation of pits were also observed on dentine slices. Quantitative data obtained from cultures on plastic demonstrated that the number of osteoclasts slightly increased in the course of 72 h in the presence of 250 pM CSF-1, whereas it decreased rapidly after 24 h in the absence of CSF-1, which confirms that this cytokine is required for the survival of osteoclasts. The number of nuclei per osteoclast was maximal after 16 h of incubation with CSF-1, namely twice the value found in the absence of CSF-1. The maximal effect of the cytokine on the fusion process was observed at a concentration of 250 pM. A calculation of the medians of the average frequency of nuclei distribution per osteoclast resulted in four nuclei per osteoclast in the absence and six in the presence of CSF-1. Genistein and herbimycin A, inhibitors of tyrosine kinases, inhibited the fusion induced by CSF-1. The data suggest that CSF-1 induces osteoclast fusion and that tyrosine kinase(s) are involved in this process. The fusion process may continue throughout the entire life of an osteoclast.
COLONY-STIMULATING FACTOR-1 (CSF-1), also known as macrophage CSF, is one of the hemopoietic growth factors required for proliferation, differentiation, activation, and survival of the cells of the mononuclear phagocyte system.1 Its cellular effects are mediated via a high-affinity cell surface receptor that belongs to the tyrosine kinase receptor family and is encoded by the proto-oncogene c-fms.2
CSF-1 is essential in driving the development of osteoclasts and, consequently, in bone resorption (for review see 3). The osteopetrotic mouse of the type op proved to be a useful tool to study the function of CSF-1 in this process. In homozygous op/op mice, the synthesis of the cytokine is impaired4,5 due to a point mutation in the coding region of the growth factor gene.6 The op phenotype is characterized by a low number of macrophages and osteoclasts. This lack of osteoclasts causes the impaired bone resorption which leads to osteopetrosis.7,8 Daily injection of CSF-1 into op/op mice reversed the osteopetrotic phenotype. This provided conclusive evidence that the development of osteoclasts depends on this cytokine.9,10
Osteoclast precursor cells and mature osteoclasts contain transcripts encoding c-fms and express binding sites for CSF-1,11–13 suggesting that the cytokine not only supports osteoclastogenesis by a direct action on cells of the osteoclast lineage, but that it also acts on mature osteoclasts. Indeed, CSF-1 was shown to be necessary for the survival of osteoclasts in vitro. Furthermore, it increases spreading and migration of osteoclasts and thus reduces the number of bone-resorbing osteoclasts.14,15 Therefore, CSF-1 might be one of the factors by which the osteoblasts mediate the regulation of bone resorption through a direct action on osteoclasts.
In the present study, we incubated osteoclasts disaggregated from rat bone with CSF-1 and observed the formation of large osteoclasts with a high number of nuclei. The results suggest that CSF-1 induced a fusion between multinucleated osteoclasts and mononuclear cells and also between multinucleated osteoclasts.
MATERIALS AND METHODS
Animals and reagents
Wistar rats from our own breeding colony were used. Medium 199 and HEPES buffer were obtained from Seromed (Berlin, Germany), Iscove's modified Dulbecco's medium and herbimycin A from Gibco BRL, Life Technology Inc. (Gaithersburg, MD, U.S.A.), and hydroxyurea and genistein from Sigma (St. Louis, MO, U.S.A.). Salmon calcitonin was generously donated by Sandoz AG (Basel, Switzerland) and recombinant human CSF-1 by Cetus Oncology, Chiron Corporation (Emeryville, CA, U.S.A.). All other reagents were of analytical grade and obtained from Merck (Darmstadt, Germany).
Preparation of osteoclasts
Osteoclasts were isolated from tibiae and femurs of 1-day-old rats according to the method described by Chambers and Magnus.16 The tibiae and femurs of six rats were dissected and cleaned of adherent soft tissue and epiphyseal cartilage. The bones were curetted with a scalpel in 1 ml of Medium 199 containing 20 mM HEPES buffer, pH 7.35, but no sodium bicarbonate. The cell suspension was collected. The bone fragments were washed twice with 1 ml of Medium 199 to remove the remaining cells. After allowing the bone fragments to settle for 30 s, the supernatant was centrifuged at 250g for 5 minutes, and the pellet was resuspended in 0.6 ml of Medium 199 containing 10% heat inactivated fetal bovine serum (FBS). Aliquots of 50 and 10 μl were pipetted onto plastic coverslips 13 mm in diameter (Thermanox, Nunc Inc., Naperville, IL, U.S.A.) and on dentine slices 4 mm in diameter, respectively. The cells were allowed to adhere for 1 h at 37°C, then nonadhering cells were removed by washing. For further incubation, the coverslips and the dentine slices were transferred to 24-well Costar tissue culture plates (Costar Corp., Cambridge, MA, U.S.A.).
Incubation of osteoclasts with CSF-1
The osteoclasts on coverslips and on dentine slices were incubated in 24-well tissue culture plates with 0.5 ml of Iscove's modified Dulbecco's medium, pH 7.35, containing no bicarbonate, but additional 36 mM NaCl, 15% FBS, and 1 mM hydroxyurea at 37°C in normal atmosphere. Incubation time and concentration of CSF-1 varied from 4 to 72 h and from 0 to 750 pM, respectively, except in experiments with dentine slices, which were incubated at 0 and 250 pM CSF-1 for 20 h. At the end of the incubations, the coverslips and dentine slices were washed with phosphate-buffered saline (PBS) and fixed in 4% paraformaldehyde in PBS for 10 minutes. Tartrate-resistant acid phosphatase (TRAP) was stained using a kit from Sigma (kit No.386-A, Leukocyte acid phosphatase). The final concentration of tartrate was 100 mM. Osteoclasts and nuclei per coverslip were counted with a light microscope. The number of nuclei per osteoclast was calculated by dividing the number of nuclei per coverslip by the number of osteoclasts per coverslip. In Fig. 6, the TRAP-positive multinucleated cells with the same number of osteoclasts were counted. For F-actin staining, the fixed cells were permeated by 0.1% Triton X-100 in PBS for 5 minutes. F-actin was stained with 0.3 mM rhodamine-conjugated phalloidin. The distribution of F-actin was detected under a fluorescent microscope.
Backscattered electron microscopy
After culture, the dentine slices were treated with 0.05 M ammonium hydroxide for 30 minutes, cleaned by ultrasonication, and fixed in 4% glutaraldehyde for 2 h. The slices were dehydrated, critical-point dried, and coated with carbon. Backscattered electron images of the dentine slices were examined with a Hitachi S-2500CX scanning electron microscope.
Binding of [125I]calcitonin
Osteoclast preparations were incubated for 1 h with 0.2 nM [125I]calcitonin.17 For the determination of the nonspecific binding, an excess of 300 nM unlabeled calcitonin was added. After extensive washing, the cells were fixed in 4% paraformaldehyde/PBS, dipped in NTB-2 photoemulsion (Eastman Kodak Corp., Rochester, NY, U.S.A.), and exposed for 2 weeks at 4°C. Salmon calcitonin was labeled with [125I]NaI, using chloramin T as the oxidizing reagent.17
Significance between groups was analyzed using analysis of variance (ANOVA) (Students-Newman-Keuls multiple comparisons test). In Fig. 6, the values are not normally distributed. Therefore, the medians are given and significance between the two medians was analyzed using the Mann–Whitney test.
Formation of large osteoclasts in the presence of CSF-1
When osteoclasts isolated from rat femurs were incubated overnight in the presence of CSF-1, large osteoclasts with a high number of nuclei (Figs. 1C and 1D) were observed alongside with osteoclasts of normal size (the latter are not shown). The large osteoclasts stained for TRAP (Fig. 1C) and formed F-actin along the edges of the cells (Fig. 1D). No or only little F-actin was observed inside the cells. In the absence of CSF-1, the morphology of the osteoclasts was similar to that at the beginning of the culture. These cells stained strongly for TRAP (Fig. 1A). F-actin appeared inside the cells around the nuclei and in the podosomes at the edges of the cells (Fig. 1B). To demonstrate the osteclastic properties of the large osteoclasts, the binding of calcitonin was investigated. As seen in Fig. 2, these cells bind calcitonin. Formation of large TRAP-positive cells with a high number of nuclei could also be observed when isolated osteoclasts were incubated on dentine in the presence of CSF-1 (Fig. 3A). These cell preparations formed pits, some of which were as large as 60 μm in diameter (Fig. 3B). In the absence of CSF-1, the number of TRAP-positive multinucleated cells and of pits was lower (data not shown).
Time course of the fusion induced by CSF-1
To obtain quantitative data on the fusion process occurring in the presence of CSF-1, the number of osteoclasts and osteoclast nuclei per coverslip were counted in the presence and absence of CSF-1 during an incubation time of 72 h. As seen in Fig. 4, the number of osteoclasts slightly increased in the presence of 250 pM CSF-1, whereas in the absence of the cytokine, the number of osteoclasts remained approximately constant for 24 h but decreased rapidly thereafter. At 16 h, the time point chosen for the dose response (see below), the number of osteoclasts per coverslip in the presence of CSF-1 was slightly, but not significantly, higher than in its absence. The number of nuclei per osteoclast was calculated as a ratio (number of total osteoclast nuclei over number of osteoclasts) and is shown in Fig. 5. In the presence of CSF-1, the value reached the maximum after 16 h and decreased thereafter. In the absence of the cytokine, the nuclei per osteoclast decreased after a small increase within the first 4 h.
Dose response for the fusion induced by CSF-1
Since the highest number of nuclei per osteoclast was observed after 16 h of incubation with CSF-1, this time point was chosen for the dose response, as seen in Table 1. While the number of osteoclasts per coverslip was not significantly changed by CSF-1, the number of nuclei in these cells was significantly increased, resulting in a significant increase in the number of nuclei per osteoclast. The maximal effect was observed at a CSF-1 concentration of 250–750 pM.
Table TABLE 1. Effect of CSF-1 on the Number of Osteoclasts
Frequency of distribution of nuclei in the osteoclasts
The distribution of the nuclei in the osteoclasts in the presence and absence of CSF-1 after 16 h of culture is shown in Fig. 6. The variation in the number of nuclei per osteoclast is large. Osteoclasts with two to eight nuclei appeared in similiar numbers independent of the presence of CSF-1. The number of larger osteoclasts with more than 10 nuclei was increased by CSF-1. In the presence of CSF-1, osteoclasts with more than 50 nuclei were formed. The medians in the presence and absence of CSF-1 are six and four, respectively, and differ significantly. As seen in Fig. 5, the total number of osteoclasts per coverslip was 228 ± 13 in the presence, and 189 ± 17 in the absence of CSF-1 (mean ± SEM), thus not significantly different. In the presence of CSF-1, the number of osteoclasts with 2–8 nuclei was decreased by 15, and the number of osteoclasts with more than 8 nuclei was increased by 55. If the data are expressed as total number of nuclei, there were a total of 2328 nuclei in the presence of CSF-1 and 1188 in the absence of CSF-1. This higher number of nuclei was mainly present in osteoclasts with more than 8 nuclei, whereas the number of nuclei present in osteoclasts with 2–8 nuclei was decreased by 65 in the presence of CSF-1.
Effect of tyrosine kinase inhibitors on the CSF-1–induced fusion
To investigate the signal transduction pathway utilized by CSF-1, the effect of genistein and herbimycin A was tested. Genistein is an isoflavone which inhibits nonreceptor and receptor tyrosine kinases,18 whereas herbimycin A inhibits mainly nonreceptor tyrosine kinases, e.g., the members of the src-kinase family.19 Genistein decreased the fusion of osteoclasts induced by CSF-1 by about 60% at 100 μM and abolished it completely at 200 μM (Table 2). The number of osteoclasts was either not affected or increased by the drug. Table 3 demonstrates the effect of herbimycin A at a concentration of 875 nM. While in all the other experiments, CSF-1 did not significantly increase the number of osteoclasts during the 16 h of incubation, such an effect was observed in the experiments described in Table 3. Herbimycin A decreased both the number of osteoclasts and nuclei, resulting in a number of nuclei per osteoclast equal to that in the absence of CSF-1 and herbimycin A.
Table Table 2. Inhibition of CSF-1–Induced Fusion by Genistein
Table Table 3. Inhibition of CSF-1–Induced Fusion by Herbimycin A
In this study, isolated rat osteoclasts were incubated with CSF-1. The cytokine induced some of the osteoclasts to fuse with precursor cells and form large cells with a high number of nuclei. The large cells showed osteoclastic properties such as expression of TRAP, binding of calcitonin, and formation of F-actin along the edge of the cells, suggesting that these cells are osteoclast-like. It has been reported that when osteoclast-like cells cocultured with osteoblasts on collagen are allowed to adhere to plastic, glass, or dentine, they form F-actin rings, indicating polarization of the cell.20 The F-actin structures of the large cells appear to be similar to these rings. It is not clear whether these large cells resorb bone. When incubated on dentine slices under the same conditions as on cover slips, the formation of large TRAP-positive multinucleated cells and of pits could be observed in the presence of CSF-1. However, since these large cells are mixed with osteoclasts of normal size, it is not possible to demonstrate from which cells these pits originate. The large diameter of some of the pits may suggest that they are formed by large cells. The lower number of osteoclasts and pits found in the absence of CSF-1 is probably due to increased cell death in the absence of the cytokine. This observation, however, contradicts earlier reports,14 which describe how CSF-1 decreases bone resorption of isolated osteoclasts. Indeed, if osteoclasts were incubated for 6 h at slight acid conditions in modified essential medium containing FBS that was not heat inactivated for stimulation of bone resorption, CSF-1 inhibited pit formation with a similiar dose response as reported by Fuller et al.14 An inhibition of pit formation by CSF-1 was also observed when the osteoclasts were stimulated with parathyroid hormone during an incubation time of either 6 or 24 h (data not shown).
The results of Fig. 4 confirm the previously described positive effects of CSF-1 on the survival of mature osteoclasts. In the absence of the cytokine, however, the cells survived up to 24 h. This is longer than has been reported previously.14 In the present study, FBS was added to the incubation medium, whereas in a previous study only albumin was present, suggesting that serum supports the survival of mature osteoclasts to some degree.
It is known that CSF-1 exerts its effects on osteoclast precursor cells and on mature osteoclasts, supporting proliferation and differentiation of precursors21,22 and migration and survival of mature osteoclasts.14,23 In the present study, an effect of the growth factor on the proliferation of its target cells was prevented by cultivating the isolated cells in the presence of hydroxyurea, an inhibitor of DNA synthesis.22 At 16 h, the time point when most of the results were obtained, the number of osteoclasts in the presence of CSF-1 was slightly higher than in its absence, but mostly not significantly different (see Fig. 4 and Tables 1, 2, and 3). The number of nuclei in osteoclasts was more than double in the presence of CSF-1. This increase is not due to karyokinesis, because DNA synthesis is inhibited by hydroxyurea. It might be expected that the number of osteoclasts would decrease in the presence of CSF-1 when fusion occurs. The slight increase in the number of osteoclasts and that of the nuclei during the first 16 h of CSF-1 incubation is probably due to stimulation of the differentiation of mononuclear precursor cells and/or inhibition of programmed cell death by the cytokine.
The fusion process consists of two steps. First, two cells meet to get into cell–cell contact. Then, during the actual fusion process, the membranes join and rearrange, with two cells uniting into one. The known effects of CSF-1 on differentiation, survival, and migration of osteoclast precursors and mature osteoclasts increase the probability of cell–cell contact, thereby supporting the first step. It is not known whether contact of an osteoclast with another osteoclast or with a precursor cell is sufficient to induce the actual fusion process, or whether other biochemical processes, e.g., alteration of the membrane structure, are required. Probably cell–cell contact alone is not sufficient, and specific biochemical reactions do prepare the cell for fusion, as has been observed in macrophages where several proteins were demonstrated to be involved in the fusion process.24–27 During osteoclast differentiation in vitro, the involvement of protein kinase C, the requirement of proteins such as LFA-1, ICAM-1, and the expression of mannose residues on the membrane have been demonstrated during the fusion process.28–30 Recently, it was observed that the expression of osteopontin increases in stromal cells and in the TRAP-positive cells during osteoclast formation in vitro.31 The actual function of osteopontin in osteoclastogenesis, however, could, not be identified, but an effect of this protein on fusion is not excluded. CSF-1 increases the expression of osteopontin mRNA32 and this may be one of the mechanisms by which CSF-1 facilitates fusion.
Very recently, the function of CSF-1 in the development of monocytes and in bone resorption has been investigated in transgenic op/op mice expressing Bcl-2 in monocytes, a gene blocking programmed cell death.33 In these mice, which did not produce CSF-1, monocytes and macrophages developed normally. These results suggest that the main function of CSF-1 is to support survival of these cells, permitting them to respond to internal and external stimuli for differentiation independent of CSF-1. However, bone resorption was not completely restored. This result is in agreement with the possibility that osteoclastogenesis is regulated differently from macrophage ontogeny. Controversial data about the role of CSF-1 in the fusion process of osteoclast precursors have been reported before.22,34 Inhibition of CSF-1 action by antisera against CSF-1 and its receptor inhibited osteoclast formation, suggesting a function of this cytokine in this fusion process.22 However, the other observations33,34 may suggest that the main function of CSF-1 is to block programmed cell death and that the cytokine is not needed in the fusion process. Our data do not distinguish whether CSF-1 is required only for survival or also for the actual fusion process. They do suggest, however, that the fusion process occurs also in mature osteoclasts and continues throughout their entire life. By this process, osteoclasts may replace nuclei lost through apoptosis.
Two inhibitors of tyrosine kinases, genistein and herbimycin A,35 inhibit CSF-1–induced fusion. Genistein acts both on the receptor and nonreceptor tyrosine kinases.18 From studies with macrophages, it is known that binding of CSF-1 to its receptor leads to autophosphorylation of the tyrosine residues in the cytoplasmic domain of the receptor36 and subsequently downstream phosphorylation of various intracellular proteins.1 Thus, it is not surprising that genistein inhibits the CSF-1–induced fusion. Herbimycin A is a more specific inhibitor acting mainly on nonreceptor tyrosine kinases, e.g., c-Src.19 It inhibits bone resorption in vitro and in vivo, probably by inhibiting c-Src in osteoclasts.37 Thus, the effect of herbimycin A on the CSF-1–induced fusion suggests the involvement of tyrosine kinases downstream of the receptor.
In conclusion, in culture of isolated osteoclasts, CSF-1 induced the formation of large osteoclasts with a high number of nuclei. This process was dependent on tyrosine kinases. The effect of CSF-1 may be due to its action on cell survival and/or osteoclast differentiation of precursor cells. Whether CSF-1 has a direct effect on the actual fusion process remains unknown.
We thank Dr. W. Hofstetter and Dr. M.G. Cecchini for the scientific advise and Ms. I. Ryf for correcting the English. We are grateful to Chiron Corporation, Emeryville, CA, U.S.A. for the recombinant human CSF-1 and to Sandoz AG, Basel, Switzerland for salmon calcitonin. This work was supported by the Swiss National Science Foundation (Grant 31–39185.93) and by the ministry of Education, Science and Culture of Japan (Grant-in-Aid no. 07672028).