Epidermal Growth Factor Receptor-deficient Mice Have Delayed Primary Endochondral Ossification Because of Defective Osteoclast Recruitment*

The epidermal growth factor receptor (EGFR) and its ligands function in diverse cellular functions including cell proliferation, differentiation, motility, and survival. EGFR signaling is important for the development of many tissues, including skin, lungs, intestines, and the craniofacial skeleton. We have now determined the role of EGFR signaling in endochondral ossification. We analyzed long bone development in EGFR-deficient mice. EGFR deficiency caused delayed primary ossification of the cartilage anlage and delayed osteoclast and osteoblast recruitment. Ossification of the growth plates was also abnormal resulting in an expanded area of growth plate hypertrophic cartilage and few bony trabeculae. The delayed osteoclast recruitment was not because of inadequate expression of matrix metalloproteinases, including matrix metalloproteinase-9, which have previously been shown to be important for osteoclast recruitment. EGFR was expressed by osteoclasts, suggesting that EGFR ligands may act directly to affect the formation and/or function of these cells. EGFR signaling regulated osteoclast formation. Inhibition of EGFR tyrosine kinase activity decreased the generation of osteoclasts from cultured bone marrow cells.

Skeletal elements develop by two distinct mechanisms: intramembranous and endochondral ossification (1). Endochondral ossification is a process by which a cartilaginous template is first formed and then replaced by bone. During embryogenesis, condensations of mesenchymal cells form, within which chondrocytes develop, proliferate, and differentiate to form a cartilage template that contains distinct zones of resting, proliferative, and hypertrophic chondrocytes. The proliferation and differentiation of chondrocytes within the cartilage template are spatially ordered, with proliferating cells at the two ends of the template and progressively more mature cells forming hypertrophic cartilage in the middle. Hypertrophic chondrocytes secrete a specialized extracellular matrix (ECM) 1 containing collagen X, which becomes calcified. Endochondral ossification begins with the invasion of the calcified hypertrophic cartilage by blood vessels, accompanied by osteoclasts and osteoblasts (primary ossification center). The function of osteoclasts is to remove the hypertrophic cartilage ECM and that of osteoblasts is to replace it with bone ECM. Longitudinal bone growth is accomplished by the continuing proliferation and maturation of chondrocytes at the ends of the cartilage template (the growth plates) to form more hypertrophic cartilage and its continual removal and replacement by bone (growth plate ossification or formation of primary spongiosa). Normal endochondral bone development requires the exquisite coordination of hypertrophic cartilage formation, vascular invasion, and the development and function of osteoclasts and osteoblasts (2,3).
EFGR expression has been detected in the axial and appendicular skeleton at the bone-cartilage junction (17), suggesting a role for this signaling pathway in endochrondral bone formation. Overexpression of EGF in transgenic mice using the ␤-actin promoter results in growth retardation and overproliferation of osteoblasts, consistent with a role for EGFR in osteoblastic cell growth (18). A recent study showed that EGFR-deficient mice have impaired endochondral ossification, probably secondary to a defect in hypertrophic chondrocyte maturation and osteoblastic cell proliferation (19). However, in cultured fetal rat long bones, EGF stimulates bone resorption, suggesting that EGFR signaling also plays a role in osteoclast function (20). In this study, we showed that impaired recruitment of osteoclasts contributed to the impaired endochondral bone formation in EGFR-deficient mice and that EGFR signaling is necessary for osteoclast formation from bone marrow progenitors.
Histological Analyses-The generation of the EGFR-null allele by homologous recombination in ES cells was previously described (11). EGFRϩ/ϩ and EGFRϩ/Ϫ mice were genotyped by PCR for the targeted allele. Mice heterozygous for the null allele (EGFRϩ/Ϫ) were mated and the day of the vaginal plug is designated as E0.5. Pregnant mice were sacrificed at E16.5 and E18.5, the embryos were removed, and the long bones dissected for analyses. In some experiments, long bones from newborn pups from heterozygous mating are collected. EGFRϪ/Ϫ embryos or newborn pups were recognized by their obvious phenotype of open-eyed (15). Bones were fixed in 4% paraformaldehyde in phosphatebuffered saline overnight and decalcified in 0.5 M EDTA (pH 7.4) for 1-3 days at 4°C prior to processing for paraffin embedding. E16.5 bones were not decalcified. For general morphology, sections were stained with Masson Trichrome stains using a kit from Sigma according to the instructions provided by the manufacturer.
Isolation of Bone Marrow Mononuclear Cells and Osteoclast Formation Assay-Bone marrow mononuclear cells were isolated from 6 -8week-old CD1 mice by a modified procedure from a previously described method (22). Briefly, mice were killed by cervical dislocation. Femora and tibiae were dissected free of soft tissue. The ends of the bone were cut, and bone marrow cells were flushed out into a cell culture dish by slowly injecting MEM at one end of the bone using a sterile 27-gauge needle. The cell suspension was filtered through a cell strainer (Falcon, 70 m nylon) and pelleted by centrifugation at 1000 rpm at 4°C. Cells were then resuspended in MEM containing 10% fetal calf serum, plated in a 100-mm cell culture dish at a density of 1 ϫ 10 7 cells/ml, and incubated at 37°C in 5% CO 2 overnight. The next morning the nonadherent cells were collected, centrifuged, and purified on a Ficoll-Hypaque gradient. The monocyte cell layer was aspirated carefully from the medium and washed with phosphate-buffered saline. The cells were counted, resuspended in MEM containing 2.5% fetal bovine serum, and placed in a 24-well plate at 4 ϫ 10 4 cells/ml. To each of these wells, growth factor (25 ng/ml M-CSF and 25 ng/ml RANKL) and either vehicle (Me 2 SO) or AG1478 at different concentrations (1.25, 2.5, and 5 M) were added. The culture medium was changed every 3 days. After 6 days of culturing, osteoclast formation was evaluated by quantification of tartrate-resistant acid phosphatase (TRAP)-positive multinucleated osteoclast cells as described below.
TRAP Staining-The osteoclast preparations were stained for TRAP activity using a leukocyte acid phosphatase kit from Sigma. Briefly, after culturing for 6 or 8 days, cells were rinsed with phosphatebuffered saline, fixed with 37% formaldehyde (formalin) in acetonecitrate buffer for 1 min, and stained according to the instructions provided by the manufacturer. All the osteoclasts in one well were counted under the microscope after counterstaining with hematoxylin as TRAPϩ cells containing at least three nuclei. The results were expressed as mean Ϯ S.D. of triplicate samples. TRAP staining was also performed on the paraffin sections according to the instructions provided by the manufacturer. The determination of the numbers and distribution of TRAPϩ cells in longitudinal sections of bones were performed as described previously (23).
Analysis of EGFR mRNA in Osteoclasts by RT-PCR-Total RNAs from cultures of osteoclast cells, and from EGFRϪ/Ϫ and EGFRϩ/ϩ embryonic heads were isolated using TRIzol Reagent (Invitrogen) according to the instructions provided by the manufacturer and dissolved in 50 l of nuclease-free water. Concentration of the RNA preparations was quantified by absorbance at 260 nm. EGFR mRNA expression was determined using one tube access reverse transcription (RT)-PCR system (Promega). Total RNA (1 g) was added to a RT mixture containing 1ϫ AMV/Tfl reaction buffer, 0.2 mmol/liter dNTP mixture, 1 mmol/liter MgSO 4 , 0.1 units/l AMV reverse transcriptase, 0.1 units/l Tfl DNA polymerase, and 1 mol/liter of each EGFR primer in a total volume of 50 l. The EGFR forward primer (5Ј-CTGCCAAGGCACAAGTAACA-3Ј) spans nucleotides 304 -323 of the mouse EGFR gene and the reverse primer (5Ј-ATTGGGACAGCTTGGATCAC-3Ј) spans nucleotides 783-802. The RT reaction was carried out at 48°C for 40 min. PCR was carried out for 40 cycles of 94°C for 30 s, 56°C for 30 s, 72°C for 30 s. PCR products were analyzed on a 1.5% agarose gel.
Northern Blotting-Total RNAs from cultured osteoclasts, EGFRϪ/Ϫ and EGFRϩ/ϩ embryonic heads were isolated as described above. Total RNA (20 g) was electrophoresed through a 1% denaturing formaldehyde-agarose gel, transferred to Hybond-Nϩ membrane (Amersham Biosciences), and then cross-linked by UV irradiation. Blots were prehybridized for 1 h at 68°C in QuikHyb TM reagent (Stratagene, La Jolla, CA) and then hybridized with a random primed 32 P-labeled EGFR probe overnight at 68°C. Blots were washed at a final stringency of 60°C in 0.2ϫ SCC ϩ 0.1% SDS and then exposed to Hyperfilm MP (Amersham Biosciences) at Ϫ70°C. The EGFR probe used was provided by Dr. Janice Liu at the University of Washington, Seattle, WA (24). The probe contains residues 969 -1242 of the rat EGFR cDNA.
Bone Marrow Cell Culture Proliferation Assay-Bone marrow mononuclear cells were isolated as described above and seeded in 96-well culture plates at a density of 20,000 cells/well with MEM supplemented with 2.5% fetal bovine serum, 25 ng/ml M-CSF, 25 ng/ml RANKL. Either vehicle (Me 2 SO) or different concentrations of AG1478 (1.25, 2.5, and 5 M) were added to the wells. Cells were then incubated at 37°C in 5% CO 2 . After 48 h, the number of viable cells was measured by the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide uptake method. Briefly, 10 l of a 5 mg/ml 3-(4,5-dimethylthiazol-2-yl)-2,5diphenyltetrazolium bromide solution was added to 100 l of culture medium in each well and the cells were incubated at 37°C in 5% CO 2 for 4 h. 100 l of 10% SDS in 0.01 N HCl was then added to each well overnight and the color reaction was determined by absorbance at 570 nm. Six wells were used for each treatment, and experiments were repeated 3 times.
Bone Resorption Assay-Osteoclast resorption was performed on calcium phosphate-coated discs (BD Biosciences Biocoat Osteologic discs). Bone marrow cells were isolated and cultured on osteologic discs in the presence of M-CSF and RANKL in MEM with 10% fetal calf serum. After osteoclasts form, the medium was changed to resorption medium (MEM adjusted to pH 7.0 with HCl with 10% fetal calf serum), and either Me 2 SO or AG1478 (5 M) was added. After 2 days, the discs were bleached to remove cells, washed in water, and air-dried. Resorption pits were visualized under light microscopy. Images were taken with a digital camera and analyzed using Adobe Photoshop. Images were visualized in gray scale and inverted. Areas of resorption pits were outlined, and under the histogram function, the percentage of pixels in the top 25% of the gray scale range contained in the outlined area was calculated as the percent resorption area. Statistical analysis was done using the Student's t test.

EGFRϪ/Ϫ Mice Show Delayed Primary Endochondral Ossification and Lengthened Growth Plate Hypertrophic Cartilage
Zones-EGFR-null mice generally die within the first postnatal day because of severe respiratory distress (14). Therefore, we studied skeletal development during embryonic development.
In the data that follows, we show the humerus; however, similar phenotypes were seen in the radius, ulna, and in the hind limbs. Primary ossification in the wild type and EGFRϩ/Ϫ bones occurred normally with humeri at E16.5 showing completed invasion of capillaries into the calcified hypertrophic cartilage (HC), with the resultant removal of the middle section of the HC and replacement of this area with vascularized tissues (Fig. 1A). In contrast, in the EGFRϪ/Ϫ humeri, the middle section of the EGFRϪ/Ϫ HC remained intact, indicating delayed primary ossification (Fig. 1B).
At E18.5, ossification in the wild type/heterozygous humeri continued in the longitudinal direction resulting in the formation of an area of trabecular bone (primary spongiosa) and a growth plate that contained an area of hypertrophic cartilage of relatively small size (Fig. 1C). However, at E18.5, the EGFRϪ/Ϫ humeri showed a lengthened HC zone at the growth plate and ossification that had not proceeded very far longitudinally (Fig. 1D). This delay in ossification continued until birth, with continuing accumulation of HC at the growth plates in newborn EGFRϪ/Ϫ mice compared with their wild type or heterozygous littermates (Fig. 1, E and F). Trabecular bone formation was also impaired in EGFRϪ/Ϫ mice. In E18.5 wild type or heterozygous humeri, the primary spongiosa area was large and contained many long trabeculae (Fig. 1G), but in the EGFRϪ/Ϫ humeri, there were only a few short trabeculae (Fig.  1H). This impairment in bone formation persisted until birth, and the differences in trabecular bone were also seen in newborn mice (Fig. 1, I and J). We concluded that primary ossification of the cartilage templates and the subsequent ossification of the growth plates of the long bones are impaired in the EGFRϪ/Ϫ mice. This occurs in the absence of overall growth retardation in utero. Because the EGFRϪ/Ϫ mice die soon after birth, and those that survive for a few days to weeks are severely growth retarded, we did not analyze postnatal development.
Because heterogeneous mice did not show a bone phenotype, the EGFR wild type and heterozygote embryos were used interchangeably in the results that follow. For simplicity they are referred to as wild type.
EGFRϪ/Ϫ Mice Have Delayed Osteoclast Recruitment into Hypertrophic Cartilage-We next determined the EGFR-dependent mechanisms in endochondral ossification. A key event in primary ossification of the hypertrophic cartilage anlage is the recruitment of osteoclasts. Mononuclear hematopoietic precursors are disseminated via the bloodstream and deposited in the mesenchyme surrounding the bone rudiments. There, they proliferate and differentiate into TRAPϩ cells that are the precursors of multinucleated osteoclasts (3). These (pre)osteoclasts invade the HC together with blood vessels and initiate the resorption of HC (25). During this migration into HC the mononucleated (pre)osteoclasts fuse together to form the multinucleated mature osteoclasts. We saw many TRAPϩ cells within the middle section of the HC in the E16.5 wild type humeri ( Fig. 2A). In contrast, TRAPϩ cells were found mainly at the periphery of the HC in the EGFRϪ/Ϫ humeri (Fig. 2B). Quantification of the number of these cells on serial sections showed a significant difference in the number of TRAPϩ cells inside versus outside the calcified HC between wild type and EGFRϪ/Ϫ bones (Fig. 2G). There were no apparent differences in the size of the TRAPϩ cells. There were also no differences in the number of nuclei per TRAPϩ cell between wild type and EGFRϪ/Ϫ mice (data not shown). The difference in the number of TRAPϩ cells inside the calcified HC between EGFRϪ/Ϫ and wild type bones diminished by E18.5 (Fig. 2, C-G). These results indicate that the delayed primary ossification of EGFRdeficient HC is coupled with a delay in osteoclast recruitment.
Impaired Bone Formation in EGFRϪ/Ϫ Mice Is Due in Part to Delayed Osteoblast Recruitment-During primary ossification, concurrent with vascular invasion and osteoclast recruitment, osteoblasts also migrate from the bony collar into the hypertrophic cartilage. Because bone formation was also impaired in the EGFRϪ/Ϫ humeri, we asked whether the recruitment of osteoblasts into the EGFRϪ/Ϫ HC was also delayed. We assayed for the expression of the transcription factor Cbfa-1, a marker of osteoblast differentiation (26 -28) and for the expression of osteocalcin, a marker of mature osteoblasts (29). In E16.5 wild type and heterozygous humeri, many Cbfa-1-positive cells were found in the middle of the HC zone, whereas in the EGFRϪ/Ϫ bones, these cells were found more at the periphery (Fig. 3, A and B). Similarly, osteocalcin-positive cells were found in the middle of the HC zone of E16.5 wild type humeri, but they were largely limited to the periphery in the EGFRϪ/Ϫ bones (Fig. 3, C and D). Thus the delay in primary ossification of EGFRϪ/Ϫ HC is also coupled to a delay in recruitment of osteoblasts. However, by E18.5, there were many osteocalcin-positive cells in EGFRϪ/Ϫ bones at the cartilage bone junction (Fig. 3, E and F). Yet, there were still few trabecular spicules in these bones, and few osteocalcin-positive cells on these spicules (Fig. 3, E and F). The deficiency in trabecular spicules persisted until birth (see Fig. 1, G-J). These data suggested that development of osteoblast in EGFR deficiency might be normal, but their subsequent proliferation, survival, and/or function might be impaired. The functional impairment is not because of a deficiency in osteoblast differentiation as measured by collagen I expression, however, because in situ hybridization showed abundant collagen type I mRNA expression at the EGFRϪ/Ϫ growth plates (Fig. 3, G  and H).
Osteoclasts Express EGFR-The delay in osteoclast recruitment into EGFRϪ/Ϫ HC may be because of either a direct effect of EGFR signaling in osteoclasts and/or osteoclast precursors or an indirect effect because of EGFR function on other cells that in turn regulate osteoclasts. To determine whether EGFR ligands can act directly on osteoclasts, we determined expression of EGFR by osteoclasts. We isolated wild type bone marrow cells from CD1 mice and cultured them in vitro with M-CSF and RANK ligand to induce the formation of osteoclasts. We then determined the expression of EGFR in cultured osteoclasts by RT-PCR and Northern blotting. Total RNAs were isolated from primary cultures of osteoclasts and from the heads of wild type and EGFRϪ/Ϫ E18.5 embryos for positive and negative controls, respectively. RT-PCR was performed using a forward primer in exon 1 and a reverse primer in exon 4 to distinguish wild type EGFR mRNA and any residual mRNA from the targeted allele, which has a disrupted exon 2. We observed the expression of EGFR mRNA in osteoclasts , and newborn (E and F) wild type or heterozygous (A, C, and E) and EGFRϪ/Ϫ (B, D, and F) mice stained for TRAP activity. In E16.5 wild type humerus, many TRAPϩ cells were detected in the vascularized hypertrophic cartilage (A, arrows). However, in E16.5 EGFRϪ/Ϫ humerus, most of the TRAPϩ cells were found at the periphery of the hypertrophic cartilage (B, arrows). In E18.5 (C and D) and newborn mice (E and F), there were just as many TRAPϩ cells inside the bone rudiment in EGFRϪ/Ϫ humerus (D and F, arrows) as in the wild-type humerus (C and E, arrows). Bar, 100 m. G, quantification of the number of TRAPϩ cells in wild type and EGFRϪ/Ϫ humeri at different stages. Horizontal bars show mean counts of TRAPϩ cells found either outside the calcified hypertrophic cartilage at the perichondrium/periosteum or inside the calcified hypertrophic cartilage. At E16.5, there is a significant difference in the total number of TRAPϩ cells found outside versus inside the calcified hypertrophic cartilage between wild type and EGFRϪ/Ϫ mice (p Ͻ 0.05). A and B), osteocalcin (C-F), and collagen type I (G and H) antisense probes. In E16.5 wild type humerus, many Cbfa-1-positive cells were found in the middle section of the hypertrophic cartilage (A, arrows), whereas these cells are found largely at the periphery of the hypertrophic cartilage in EGFRϪ/Ϫ humerus (B, arrows). Similarly, osteocalcin (oc) expressing cells are found at the periphery of hypertrophic cartilage in E16.5 EGFRϪ/Ϫ (D, arrows) and in the middle of wild type hypertrophic cartilage (C, arrows). By E18.5, there were abundant osteocalcin-positive cells in the metaphysis of EGFRϪ/Ϫ humerus. However, these cells are found mainly at the cartilage-bone junction (F, arrows), and very few are found in the primary spongiosa, which also contain very few trabecular spicules (F, arrowhead). In contrast, in the E18.5 wild type humerus there are abundant osteocalcin-positive cells both at the cartilage-bone junction (E, arrows) and in the primary spongiosa on trabecular spicules (E, arrowheads). Collagen type I expression was found in the metaphysis in both wild type and EGFRϪ/Ϫ E18.5 humerus (G and H, arrows). Bar, 200 m. and wild type embryonic heads (Fig. 4A). As expected, no PCR product corresponding to EGFR was seen in the EGFRϪ/Ϫ embryonic heads. On Northern blot, two transcripts of ϳ9.6 and 5 kb were found in osteoclast cultures and wild type embryonic heads (Fig. 4B). No transcripts were found in EGFRϪ/Ϫ embryonic heads. These data suggest that EGFR ligands may act directly on osteoclasts to affect their function.

Delayed Primary Osteoclast Recruitment in EGFR-null Mice Is Not Because of Deficiency in MMPs-Previous studies
showed that MMPs were necessary for the migration of (pre)osteoclasts into calcified hypertrophic cartilage during primary endochondral ossification (23). In particular, MMP-9 (gelatinase B) is required for the timely recruitment of these cells. MMP-9-deficient mice show a delay in osteoclast recruitment into the calcified hypertrophic cartilage during primary endochondral ossification of the metatarsals (25). Because EGFR signaling can modulate the cellular expression of MMPs, the delay in osteoclast recruitment into the EGFRϪ/Ϫ HC may be because of deficiency in the expression of MMPs. To test this hypothesis, we analyzed the expression of several MMPs by in situ hybridization. The expression of MMP-9 and MMP-14 (MT1-MMP), which are highly expressed in osteoclasts (30 -32), was consistent with the TRAP staining results. In E16.5 wild type humeri, MMP-9 expression was found in cells corresponding to osteoclasts in the vascularized bone marrow cavity and at the cartilage-bone junction (Fig. 5, A and C). In the E16.5 EGFRϪ/Ϫ humeri, MMP-9 expression was found in cells at the periphery of the HC, consistent with the location of osteoclasts in these skeletal elements at this time (Fig. 5, B and  D). There were no significant differences in the amount of MMP-9 mRNA per cell, judged by the intensity of the signal. By E18.5, there were just as many MMP-9 expressing cells at the cartilage-bone junction of the growth plate and in the bone marrow cavity in the EGFRϪ/Ϫ bones compared with wild type bones. This is consistent with the observation that the differences in the number of osteoclasts between wild type and EGFRϪ/Ϫ bones have diminished by this time. Substrate gel FIG. 4. RT-PCR and Northern blot analysis of EGFR mRNA in cultured osteoclasts. A, RT-PCR with EGFR-specific primers of total RNA isolated from cultured osteoclasts, wild type embryonic heads, and EGFRϪ/Ϫ heads. The expected 499-bp PCR product was seen in RNA from osteoclast and wild type embryonic head, but not in EGFRϪ/Ϫ head. B, Northern blot of total RNA isolated from cultured osteoclasts, wild type embryonic heads, and EGFRϪ/Ϫ heads hybridized with a 32 P-labeled EGFR probe. Two transcripts, 9.6 and 5.0 kb in size, were detected in RNA from osteoclasts and wild type embryonic head, but not from EGFRϪ/Ϫ head. analyses showed that there was slightly less MMP-9 protein in extracts of EGFR-null bones compared with wild type (Fig. 5I). There was no apparent change in the ratio of active versus latent forms of MMP-9 proteins. These results indicated that MMP-9 expression and activity in osteoclasts were not significantly altered by EGFR deficiency, even though there might be an overall decrease in the level of MMP-9 because of decreased osteoclast number. Thus deficiency in MMP-9 activity was unlikely to be the primary cause for the delayed recruitment of osteoclasts into EGFR-null HC. Similar results were observed for the expression of MMP-14 (MT1-MMP). In E16.5 wild type humeri, MMP-14 was expressed by cells with similar distribution to osteoclasts, at the cartilage-bone junction and in the bone marrow cavity (Fig. 6, A and C). In contrast, cells expressing MMP-14 were located at the periphery of the unvascularized HC in E16.5 EGFRϪ/Ϫ bones (Fig. 6, B and D). At E18.5 the number and location of MMP-14 expressing cells were comparable in wild type, heterozygous, and EGFRϪ/Ϫ bones (data not shown).
MMP-13 (collagenase-3) may also be important for the vascularization of hypertrophic cartilage during primary endochondral ossification. It can degrade native collagen, which is a major component of the hypertrophic cartilage ECM (33,34). MMP-13 is highly expressed in hypertrophic chondrocytes and osteoblasts (35) and therefore may act to degrade collagens in the hypertrophic cartilage ECM during endochondral ossification. In E16.5 wild type humeri, MMP-13 was expressed by late hypertrophic chondrocytes and by cells at the vascularization front, some of which may be osteoblasts (Fig. 6, E and G). In the E16.5 EGFRϪ/Ϫ bones, there was high expression of MMP-13 in hypertrophic chondrocytes in the calcified HC (Fig. 6, F and  H). These results indicated that deficiency in MMP-13 expression was not the cause for the delayed primary ossification and osteoclast recruitment into EGFRϪ/Ϫ hypertrophic cartilage.
Osteoclast Formation from Bone Marrow Cells Is Attenuated by EGFR Inhibitor-The delay in osteoclast recruitment into EGFRϪ/Ϫ hypertrophic cartilage may be because of deficiency in either their formation or their migration. To address the role of EGFR signaling in osteoclast formation, we inhibited EGFR signaling during the formation of osteoclasts from bone marrow cells in vitro using the EGFR tyrosine kinase inhibitor AG1478 (36). Wild type bone marrow cells from CD1 mice were isolated and cultured in the presence of M-CSF and RANK ligand to induce the formation of osteoclasts. The addition of AG1478 in these cultures attenuated osteoclast formation (Fig. 7). Cultures treated with vehicle showed the formation of multinucleated TRAPϩ cells characteristic of osteoclasts (Fig. 7A). Addition of increasing concentrations of AG1478 decreased the number of multinucleated TRAPϩ cells that were formed in a dose-dependent manner (Fig. 7, B-D), reaching over 90% inhibition at 5 M AG1478 (Fig. 7E). Addition of exogenous EGF or transforming growth factor-␣ had no additional effects on the formation of osteoclasts in these cultures (data not shown).
Several genes have been shown to be important for osteoclast formation from hematopoietic stem cells (37). We asked whether any of these genetic pathways was downstream of EGFR signaling. Quantitative real time RT-PCR of RNA isolated from bone marrow cells cultured in the presence of M-CSF and RANK ligand with or without AG1478 showed no change in the expression of PU.1, c-fos, TRAF-6, MITF, and c-src with AG1478 treatment (data not shown). Thus the regulation of osteoclast formation by EGFR signaling does not appear to be mediated through these pathways. To determine whether the inhibition of osteoclast formation by AG1478 was because of an effect on cell proliferation, we assayed for cell proliferation in the AG1478-treated bone marrow cell cultures. Wild type bone marrow cells were isolated and cultured for 2 days in the presence of M-CSF and RANK ligand with or without AG1748. Treatment with AG1478 resulted in a dosedependent decrease in 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide uptake, indicative of decreased cell number (Fig. 7F). This suggested that inhibition of EGFR signaling by AG1478 led to decreased cell proliferation. This may account in part for the decrease in osteoclast formation caused by AG1478 treatment.
Osteoclast Resorptive Activity Is Not Modulated by EGFR Inhibitor-To determine whether EGFR signaling modulates osteoclast function, we tested whether osteoclast resorptive activity is inhibited in the presence of the EGFR tyrosine kinase inhibitor AG1478. Osteoclasts derived from bone marrow progenitor cells were allowed to form on calcium phosphate-coated discs and subsequently switched to resorptive media in the presence of AG1478 or vehicle control for 2 days. Quantitative analyses of the areas of resorption showed that there were no significant differences in the percent resorption area between control and treated osteoclasts (Fig. 8). Therefore EGFR signaling does not appear to modulate osteoclastic bone resorption.

DISCUSSION
In this study we have investigated the function of EGFR signaling in endochondral bone formation by analyzing long bone development in EGFR-null mice. Our data showed that in EGFR-null mice primary ossification of the calcified hypertrophic cartilage anlage was delayed and was associated with a delay in the recruitment of osteoclasts and osteoblasts. Forma- tion of the primary spongiosa was also abnormal leading to accumulation of growth plate HC and decreased trabecular bone mass. The delay in osteoclast recruitment might be be-cause of inadequate formation of osteoclasts resulting from EGFR deficiency.
EGFR Signaling Is Necessary for Normal Osteoclast Recruitment and Function during Endochondral Ossification-A critical step in endochondral ossification is the invasion of capillaries into the calcified hypertrophic cartilage zone in the diaphysis of the cartilage anlage. Vascularization of calcified HC is accompanied by the recruitment of osteoclasts and osteoblasts. We found that EGFR deficiency resulted in delayed recruitment of osteoclasts into calcified HC during primary ossification. However, this delay was only temporary, consistent with the model that the lack of EGFR signaling causes defective formation of osteoclasts from precursor cells surrounding the bone rudiments, thus requiring a longer time for sufficient osteoclasts to accumulate. Alternatively, the migration of osteoclasts into calcified HC may also be defective.
Even though osteoclast recruitment into EGFRϪ/Ϫ HC reached the same level as wild type mice at E18.5, ossification of growth plate HC was still not complete, as evidenced by the lengthened HC zones of the EGFRϪ/Ϫ growth plates up until birth. This may be because of the accumulation of HC caused by the delay in primary ossification, and the inability of the subsequently normal number of osteoclasts to overcome the initial difference and therefore ossification in the longitudinal direction would always be behind in the EGFR-null bones. Alternatively, growth plate ossification may also be abnormal because of either abnormal formation of growth plate HC or abnormality in its removal. A recent study also reported increased growth plate HC in the EGFRϪ/Ϫ mice (19). The authors found expression of EGFR in chondroblasts but no differences in growth plate chondrocyte proliferation, and suggested that EGFR negatively regulated hypertrophic chondrocyte maturation. On the other hand, the removal and ossification of EGFR- null growth plate HC may also be abnormal, proceeding at a slower rate resulting in accumulation of growth plate HC. We found expression of EGFR on mature osteoclasts, suggesting that these cells can respond directly to EGFR ligands. The function of EGFR in osteoclasts is not known, but it may be to stimulate their proteolytic activity. EGFR ligands have been found to stimulate osteoclastic bone resorption in vivo and in vitro in bone organ cultures (20,38,39).
MMPs are necessary for the recruitment of osteoclasts into calcified hypertrophic cartilage during primary endochondral ossification (25). The craniofacial and lung developmental defects of EGFR-null mice may be because of alteration in MMP activity (15,16). We found that down-regulation MMP-9, MMP-13, and MMP-14 was unlikely to be the primary mechanism for the delayed recruitment of osteoclasts into EGFRϪ/Ϫ HC. However, this does not exclude EGFR regulation of other ECM degrading proteinases that are important in osteoclast function. In addition, because MMP-13 acts synergistically with MMP-9 (25), the initial delayed recruitment of osteoclasts expressing MMP-9 could functionally blunt HC ECM degradation and exacerbating the delay in HC vascularization.
Vascular endothelial growth factor (VEGF) is important for vascularization of and osteoclast recruitment into calcified hypertrophic cartilage of the developing bones (25,40). We found no significant difference in VEGF expression in EGFR-null bones by in situ hybridization. 2 This suggests that the delayed osteoclast recruitment into EGFR-null HC is not because of deficiency in VEGF expression. However, because MMP-9 may regulate the bioavailability of VEGF, the initial delay in osteoclast recruitment may cause an initial decrease in MMP-9 activity leading to decreased bioavailable VEGF, which leads to further delay in vascularization and recruitment of osteoclasts.
EGFR Deficiency Leads to Impaired Trabecular Bone Formation-Trabecular bone mass is decreased in the EGFRϪ/Ϫ bones. This may partly be because of a delay in the initial recruitment of osteoblasts during primary ossification. However, trabecular spicules continued to be deficient until birth. Fewer osteoblasts were found in the EGFRϪ/Ϫ primary spongiosa even though they are abundant at the cartilage bone junction. This suggests that formation of osteoblasts in the EGFRϪ/Ϫ bones may be normal, but that their subsequent proliferation/survival and/or function may be impaired. Studies of the LRP5 (low density lipoprotein receptor-related protein 5)-deficient mice showed that osteoblast differentiation and proliferation were differentially regulated (42). Sibilia et al. (19) reported that primary EGFRϪ/Ϫ calvarial osteoblast cultures showed decreased proliferation potential and increased differentiation as measured by their ability to form bone nodules in vitro (19). Thus the decreased trabecular bone mass in the EGFRϪ/Ϫ mice was likely because of decreased osteoblast proliferation. A direct effect of EGFR signaling on osteoblasts is supported by previous studies showing that EGFR was expressed in osteoblasts in vivo (17,43), and that EGF stimulated osteoblast proliferation in vitro (44,45).
EGFR Signaling Is Necessary for the Formation of Osteoclasts-Previous studies have identified two essential factors for osteoclastogenesis: M-CSF and RANKL (37). M-CSF is necessary for the generation of the monocyte/macrophage cell lineage and RANKL for their differentiation into osteoclasts. We found that induction of osteoclast formation from bone marrow cells in the presence of RANKL and M-CSF was significantly inhibited in the presence of the EGFR tyrosine kinase inhibitor AG1478. This is consistent with previous studies suggesting a role for EGFR ligands in osteoclast formation. Addition of ei-ther transforming growth factor-␣ or EGF increased the formation of multinucleated osteoclasts from cultured human bone marrow cells (46). In our bone marrow cell cultures addition of exogenous EGF or transforming growth factor-␣ had no additional effect on osteoclast formation. This may be because there were already saturating amounts of endogenous EGFR ligands in these cultures, or that the endogenous EGFR was transactivated by other ligand-receptor signaling pathways (47). The role of EGFR signaling in osteoclast formation may be a direct effect on osteoclast precursors to stimulate either their proliferation or differentiation. Alternatively, EGFR ligands may act on other bone marrow cells to stimulate production of either secreted or cell-surface factors that in turn act in a paracrine fashion on osteoclast precursors to effect their growth and differentiation. Further studies are needed to distinguish between these two possibilities.