Bone is composed of compact bone and cancellous bone. In long bones, the shaft (cortical bone) consists of compact bone, and the inside of the shaft (trabecular bone), which is a three-dimensional lattice of branching bony spicules, consists of cancellous bone. Compact bone is mature bone, because it is composed of densely packed, highly organized collagen fibrils with high mineralization, and is relatively resistant to osteolysis. In contrast, cancellous bone is less mature, because it is composed of loosely organized collagen fibrils with low mineralization, and it is easily resorbed and plays an important role in calcium homeostasis (Marks and Odgren,2002).
Runt-related transcription factor 2 (Runx2) is a transcription factor that belongs to the Runx family and is involved in many aspects of skeletal development (Komori,2005). Upon forming a heterodimer with core binding factor β (Cbfβ), Runx2 acquires DNA-binding activity and regulates transcriptional activity (Kundu et al.,2002; Miller et al.,2002; Yoshida et al.,2002; Kanatani et al., 2006). There are two Runx2 isoforms, type I Runx2 and type II Runx2, which have different N-termini, and type I Runx2 is more dependent on Cbfb than type II Runx2 for their functional activities (Kanatani et al., 2006). Runx2-deficient mice lack osteoblasts and show a complete lack of bone formation, demonstrating that Runx2 is essential for osteoblast differentiation (Komori et al.,1997; Otto et al.,1997). Runx2 also plays important roles in chondrocyte maturation, maintenance of the chondrocyte phenotype, and vascular invasion into cartilage (Komori,2005; Zelzer et al.,2001). Furthermore, Runx2 regulates RANKL and OPG expression stimulating osteoclast differentiation (Enomoto et al.,2003). These findings indicate that Runx2 functions as a key molecule in skeletal development.
The DNA-binding sites of Runx2 in major bone matrix protein genes including the Col1a1, osteopontin, bone sialoprotein, and osteocalcin genes, have been identified, and Runx2 induced the expression of these genes or activated their promoters (Ducy et al.,1997,1999; Sato et al.,1998; Harada et al.,1999; Javed et al.,1999; Kern et al.,2001). Furthermore, the expression of dominant-negative (dn)-Runx2 under the control of osteocalcin promoter completely abrogated the expression of major bone matrix protein genes in postnatal bone development (Ducy et al.,1999). These findings indicate that Runx2 plays an important role in the expression of major bone matrix protein genes. In transgenic mice that express Runx2 under the control of the Col1a1 promoter, however, the cortical bone was composed of woven bone due to the inhibition of osteoblast differentiation at the late stage, which was shown by the reductions in the mRNA expression of alkaline phosphatase (ALP), Col1a1, osteocalcin, and matrix metalloproteinase 13 (MMP13), all of which are normally upregulated according to the degree of osteoblast differentiation (Liu et al.,2001; Geoffroy et al.,2002). Therefore, the function of Runx2 in postnatal bone development is controversial.
To clarify the function of Runx2 in osteoblasts, we inhibited the function of Runx2 using dn-Runx2 in in vitro and in vivo studies. We show here that the maturity and turnover rate of bone are determined by the level of functional Runx2, and that Runx2 is involved in bone loss in estrogen deficiency. However, Runx2 was not essential for maintenance of major bone matrix protein gene expression in postnatal bone development and maintenance.
Dn-Runx2 Inhibited ALP Activity in Osteoblast Precursors But Did Not Suppress Bone Matrix Protein Gene Expression in Mature Osteoblastic Cells in Vitro
Using two cDNAs, one containing the runt domain with the N-terminus of type I Runx2 (long form) and the other containing only the runt domain (short form) (Fig. 1A), we examined their dominant-negative (dn) activity against Runx2. Both forms had no capacity for transcriptional activation in luciferase assays using oligonucleotides containing 6 repeats of the consensus Runx2 binding sequence, but both inhibited Runx2-dependent transcription dose-dependently (Fig. 1B and C). First, we examined their dominant-negative effects on ALP activity, which is an early marker of osteoblast differentiation, using primary calvarial cells as osteoblastic precursors and an immature mesenchymal cell line, C2C12, which differentiates into osteoblasts in the presence of rhBMP-2 (Fig. 1D and E). Adenoviral introduction of Runx2 upregulated ALP expression in primary calvarial cells and in rhBMP-2-treated C2C12 cells, whereas adenoviral introduction of either form of dn-Runx2 downregulated ALP expression in comparison with the level of ALP expression in enhanced green fluorescence protein (EGFP)-expressing vector-infected cells. These findings indicate that Runx2 induces an early marker of osteoblast differentiation, while dn-Runx2 inhibits it, and that the two forms of dn-Runx2 have similar efficiencies in vitro. We previously observed that transgenic mice overexpressing either form of dn-Runx2 under the control of the Col2a1 promoter exhibited inhibition of chondrocyte maturation, although the strong dominant-negative effect was observed more frequently among transgenic mice expressing the long form of dn-Runx2 (Ueta et al.,2001, unpublished observation).
We next examined the effect of adenoviral introduction of dn-Runx2 on endogenous Col1a1 and osteocalcin expression in vitro. As mature osteoblasts express high levels of Col1a1 and osteocalcin in vivo, we analyzed their expression in mature osteoblastic cells. Primary calvarial cells and the immature osteoblastic cell line MC3T3-E1 were cultured for 10 days after confluence to generate mature osteoblastic cells. We also used the mature osteoblastic cell line MLO-A5. In primary calvarial cells, MC3T3-E1 cells, and MLO-A5 cells, adenoviral introduction of Runx2 upregulated osteocalcin expression but had no effect on Col1a1 expression in comparison with the respective level in EGFP-expressing cells, while adenoviral introduction of dn-Runx2 had no effect on Col1a1 nor osteocalcin expression (Fig. 1F–H). These findings indicate that exogenously introduced Runx2 upregulates osteocalcin expression in mature osteoblastic cells, but that Runx2 is not absolutely required for the steady-state expression of Col1a1 and osteocalcin in mature osteoblastic cells in vitro.
Generation of dn-Runx2 Transgenic Mice
We generated transgenic mice that express dn-Runx2 under the control of the 2.3-kb promoter of mouse Col1a1 (Fig. 2A). We previously reported that the promoter directs transgene expression to immature and mature osteoblasts (Liu et al.,2001). Two lines of dn-Runx2 transgenic mice were established from two F0 mice, and they highly expressed the long form of dn-Runx2. The two established lines of transgenic mice showed significant levels of transgene expression, as confirmed by Northern blot and Western blot analyses (Fig. 2B and C). The two lines of transgenic mice showed similar phenotypes, and detailed data on transgenic mouse line 1 are reported. In transgenic mouse line 1, the transgene expression was much stronger than endogenous Runx2 expression (Fig. 2D). To confirm that the protein encoded by the transgene had a dominant-negative effect on Runx2 function in vivo, dn-Runx2 transgenic mice were mated with transgenic mice that expressed Runx2 under the control of the same Col1a1 promoter, which were described previously (Liu et al.,2001). The Runx2 transgenic mice showed osteopenia with fractures, reduced osteocalcin expression, and a reduced number of osteocytes (Liu et al.,2001) (Fig. 2E–H). However, the double transgenic mice, which expressed both the Runx2 and dn-Runx2 transgenes, showed no fractures and had a normal level of osteocalcin expression and normal osteocyte number, although the mice were mildly osteopenic (Figs. 2E–H). The mild osteopenic phenotype of the double transgenic mice indicates that the expression level of dn-Runx2 was not high enough to inhibit exogenous Runx2 completely, because Runx2 and dn-Runx2 bound a Runx recognition sequence with similar efficiencies (unpublished data). These findings indicate that dn-Runx2 expression inhibited Runx2 function in vivo and restored the late stage of osteoblast differentiation and their transition to osteocytes.
Age-Dependent Increase in Bone Mass in dn-Runx2 Transgenic Mice
On X-ray and peripheral quantitative computed tomography (pQCT) analyses of long bones, the trabecular bone density of dn-Runx2 transgenic mice and wild-type mice at 4 weeks of age did not significantly differ (Fig. 3A and B). However, the trabecular bone density was significantly higher in the dn-Runx2 transgenic mice than in wild-type mice at 10 weeks and at 7 months of age. Histological analysis also showed an increased amount of trabecular bone in dn-Runx2 transgenic mice at 7 months of age (Fig. 3E). The dn-Runx2 transgenic mice showed transient reductions in the mineralization and volume of cortical bone, as shown by the reductions in mineral density and thickness compared with the respective values in wild-type mice, at 10 weeks of age (Fig. 3C and D).
In bone histomorphometric analysis, the trabecular bone volume (BV/TV) in dn-Runx2 transgenic mice was similar to that in wild-type mice up through the young adult age (10 weeks of age) but it increased thereafter, and the trabecular bone volume in dn-Runx2 transgenic mice was 50% higher than that in wild-type mice at 7 months of age (Fig. 4A). The thickness of newly deposited matrix (osteoid thickness; O.Th) was lower in dn-Runx2 transgenic mice, with a significant difference seen at 10 weeks of age, and the osteoblast number (N.Ob/B.Pm) was significantly reduced at 4 weeks of age compared with those in the wild-type mice. The reduction in osteoblast number seemed to be caused by decreased osteoblast proliferation, because osteoblast proliferation was increased in both type I and type II Runx2 transgenic mice (Kanatani et al., 2006). Osteoclast parameters, including the eroded surface (ES/BS) and osteoclast number (N.Oc/B.Pm), were similar and bone formation rate (BFR/BS) was not significantly different between the dn-Runx2 transgenic mice and wild-type mice. To determine the level of bone resorption, we examined the urinary deoxypyridinoline level, a marker of bone resorption, and found that it was significantly reduced in the dn-Runx2 transgenic mice (Fig. 4B). These findings indicate that in dn-Runx2 transgenic mice, the bone mass of the trabecular bone gradually increased with aging due to reduced bone resoption, even though the parameters of osteoclasts remained at normal levels.
Trabecular Bone Is Highly Mineralized and the Collagen Fibrils Are Densely Packed in dn-Runx2 Transgenic Mice
To investigate the reason why bone resorption was reduced in dn-Runx2 transgenic mice, we examined the characteristics of bone by transmission electron microscope (TEM) at 12 weeks of age (Fig. 5). Mineralization of matrix vesicles in the trabecular bone, which is typically observed in the mineralization of osteoid, was seen in wild-type mice (Fig. 5A). However, in the dn-Runx2 transgenic mice, the osteoid was thin, the mineralization surface was flat, and mineralization of matrix vesicles was rarely seen (Fig. 5B). The extent of mineralization in the trabecular bone was higher in dn-Runx2 transgenic mice than in wild-type mice (Fig. 5C and D). Furthermore, the collagen fibrils were loosely deposited in a random orientation in the trabecular bone of wild-type mice, while they were densely and regularly packed in the trabecular bone of dn-Runx2 transgenic mice (Fig. 5E and F). These characteristics of the trabecular bone of dn-Runx2 transgenic mice were similar to those seen in the cortical bone of both wild-type and dn-Runx2 transgenic mice (Fig. 5G and H, and data not shown). These findings indicate that the trabecular bone in dn-Runx2 transgenic mice had characteristics of compact bone, which explains why bone resorption was reduced in dn-Runx2 transgenic mice.
Expression of Endogenous Runx2 During Bone Development and Maintenance
As the expression of dn-Runx2 in osteoblasts resulted in the maturation of the trabecular bone, we examined the expression of endogenous Runx2 after birth in wild-type mice. We compared the Runx2 expression with Col1a1, osteopontin, and osteocalcin expression by in situ hybridization using wild-type long bones (Fig. 6). Most of the osteoblasts at newborn stage and 1 week of age were immature and expressed Col1a1, osteopontin, and Runx2, while osteocalcin-positive mature osteoblasts were restricted to cortical bone (Fig. 6A–H). At 4 weeks of age, the expression pattern of Runx2 overlapped with that of osteopontin or osteocalcin (Fig. 6I–L). At 8 weeks of age, osteocalcin-positive mature osteoblasts expressed Runx2 moderately and osteopontin weakly (Fig. 6M–P), but Runx2 and osteopontin expression decreased during aging and was faintly detected in many osteocalcin-positive mature osteoblasts at 12 weeks of age and at 10 months of age (Fig. 6Q–X).
We also examined the expression of Runx2, osteopontin, and osteocalcin proteins in wild-type mice by double labeling using anti-Runx2, anti-osteopontin, and anti-osteocalcin antibodies (Fig. 7). At 1 week of age, preosteoblasts in the periosteum of mandible expressed Runx2 but not osteopontin and osteocalcin (Fig. 7A–C, asterisks). Inside of mandible, however, both osteopontin-positive immature osteoblasts and osteocalcin-positive early mature osteoblasts expressed Runx2 (Fig. 7B, arrows; Fig. 7C, arrowheads). In the femur, preosteoblasts in the perichondrial region surrounding proliferating and prehypertrophic chondrocytes expressed Runx2 but not osteopontin and osteocalcin (Fig. 7D–F, asterisks). Immature osteoblasts surrounding the hypertrophic chondrocyte layer expressed Runx2 and osteopontin but not osteocalcin (Fig. 7E, arrows). Osteocalcin-positive early mature osteoblasts, which expressed Runx2, appeared in the metaphyseal cortical bone (Fig. 7F, arrowhead). In the diaphysis of the femur, both osteopontin-positive immature osteoblasts and osteocalcin-positive early mature osteoblasts expressed Runx2 (Fig. 7H, arrows; 7I, arrowheads). In the metaphysis of femur at 4 weeks of age, osteopontin- positive immature osteoblasts strongly expressed Runx2, while osteocalcin-positive mature osteoblasts weakly expressed Runx2 (Fig. 7K, arrows; 7L, arrowheads). In the diaphysis, we observed an apparent reduction in Runx2 and osteopontin protein expression compared with their mRNA expression (Fig. 6), and Runx2 and osteopontin proteins were barely detectable in most of the osteocalcin-positive late mature osteoblasts (Fig. 7M–O, arrows). These findings show that Runx2 is expressed in preosteoblasts, in which osteopontin and osteocalcin are not expressed, is strongly expressed in osteopontin-positive immature osteoblasts, and then is expressed in osteocalcin-positive early mature osteoblasts, but that Runx2 expression is reduced in osteocalcin-positive late mature osteoblasts.
Expression of Major Bone Matrix Protein Genes in the Osteoblasts of dn-Runx2 Transgenic Mice Was Not Impaired
We examined the expression of bone matrix protein genes including Col1a1, osteopontin, and osteocalcin by Northern blot analysis and in situ hybridizatoin. Col1a1 and osteocalcin expression were significantly reduced and osteopontin expression was marginally reduced in the dn-Runx2 transgenic mice at 4 weeks of age (Fig. 8A–C). However, it is considered that the reduced expression levels of these matrix protein genes were mainly due to the reduced number of osteoblasts in dn-Runx2 transgenic mice, because their expression decreased in parallel with the reduction in the number of osteoblasts at 4 weeks of age (Figs. 4A, 8A–C). In situ hybridization analysis confirmed that there were reduced numbers of Col1a1-positive, osteopontin-positive, and osteocalcin-positive cells at 4 weeks of age (Figs. 8D–I). At 10 weeks of age, the dn-Runx2 transgenic mice expressed similar levels of Col1a1, osteopontin, and osteocalcin compared with the respective levels in wild-type mice as demonstrated by Northern blot analysis (Fig. 8A–C) and in situ hybridization (Fig. 8J–O). These findings showed that individual osteoblasts in dn-Runx2 transgenic mice expressed these matrix protein genes at nearly normal levels, and indicate that Runx2 is not absolutely required for maintenance of the expression of the major bone matrix protein genes in postnatal bone development and maintenance.
The expression levels of RANKL and OPG, which stimulate and inhibit osteoclast differentiation, respectively, did not significantly differ between the dn-Runx2 transgenic mice and wild-type mice at 4 and 10 weeks of age by real-time reverse transcription (RT)-PCR (Fig. 8P, Q).
Bone Resorption Was Not Enhanced After Ovariectomy in dn-Runx2 Transgenic Mice
As bone resorption was reduced in dn-Runx2 transgenic mice, the response of the trabecular bone to estrogen depletion was investigated. We performed ovariectomy at 4 months of age, and examined the urinary deoxypyridinorine level one week later (Fig. 9A). Although there was no significant difference in the urinary deoxypyridinorine level between wild-type and dn-Runx2 transgenic mice that underwent a sham operation, probably due to an increase in bone mass in dn-Runx2 transgenic mice at this age, the urinary deoxypyridinoline level in ovariectomized wild-type mice was significantly higher than those in sham-operated wild-type mice and ovariectomized dn-Runx2 transgenic mice. In contrast, the urinary deoxypyridinoline levels were similar between sham-operated and ovariectomized dn-Runx2 transgenic mice. Five weeks after ovariectomy, the femurs were compared with those from sham-operated mice by X-ray, pQCT, and bone morphometric analyses. After ovariectomy, the trabecular mineral content of femurs apparently decreased in wild-type mice but not in dn-Runx2 transgenic mice (Fig. 9B–D). The eroded surface, osteoclast surface, and osteoclast number were significantly higher and the bone formation rate was marginally higher after ovariectomy in the wild-type mice, although these parameters did not significantly change after ovariectomy in the dn-Runx2 transgenic mice (Fig. 9E–H). These results show that ovariectomy did not enhance bone resorption nor bone formation in dn-Runx2 transgenic mice.
As bone resorption was not enhanced after ovariectomy in dn-Runx2 transgenic mice, we examined the function of Runx2 in osteoclastogenesis by coculture of calvarial cells and bone marrow cells. Adenoviral introduction of Runx2 to wild-type calvarial cells increased TRAP-positive cells and resorbed area, while adenoviral introduction of dn-Runx2 reduced them, indicating that Runx2 positively regulates osteoclast differentiation and bone resorption (Fig. 9I–K).
These findings suggest that Runx2 induces bone loss in the estrogen-deficient state not only by regulating bone maturity but also by mediating the enhancement of osteoclastogenesis after estrogen depletion.
Less Anabolic Effect of Parathyroid Hormone (PTH) in dn-Runx2 Transgenic Mice
To examine the effect of PTH in dn-Runx2 transgenic mice, we injected rat PTH(1-34) three times a day at a dose of 80 μg/kg for 18 days. A previous report showed that the same protocol using human PTH(1-34) decreased bone mineral density in rats (Frolik et al.,2003). After the treatment with PTH, however, we observed significant increases in trabecular bone volume, trabecular number, trabecular thickness, and cortical bone volume in wild-type mice (Fig. 10). In dn-Runx2 transgenic mice, trabecular bone volume and thickness increased after the PTH treatment, but the increase in the trabecular bone volume was less than that of wild-type mice and the trabecular number and cortical bone volume were not increased (Fig. 10B–E). These findings indicate that the osteoblasts in dn-Runx2 transgenic mice are less responsive to the anabolic effect of PTH, suggesting that Runx2 is involved in the anabolic effect of PTH. The involvement of Runx2 in the anabolic effect of PTH was also reported previously (Krishnan et al.,2003).
We investigated the function of Runx2 in osteoblasts using dn-Runx2 in in vitro and in vivo studies. Adenoviral introduction of dn-Runx2 suppressed ALP activity in immature mesenchymal cells, but it did not suppress osteocalcin and Col1a1 expression in mature osteoblastic cells in vitro. In Runx2 transgenic mice, dn-Runx2 restored osteoblast differentiation at the late stage and their transition to osteocytes, which were inhibited in Runx2 transgenic mice. Dn-Runx2 transgenic mice showed increased mineralization and increased volume of the trabecular bone, and the trabecular bone had the appearance of compact bone. These findings indicate that dn-Runx2 increased the volume of trabecular bone by promoting the formation of mature bone, which is relatively resistant to bone resorption. In accordance with these findings, endogenous Runx2 was highly expressed in immature osteoblasts but down-regulated during osteoblast maturation in wild-type mice. Furthermore, dn-Runx2 conserved the trabecular bone after estrogen depletion by promoting bone maturation as well as inhibiting osteoclastogenesis. Moreover, dn-Runx2 transgenic mice showed less response to the anabolic effect of PTH. Thus, the level of Runx2 determines bone maturity and the bone turnover rate, and is responsible for the bone loss observed in the estrogen-deficient state and at least partly for the anabolic effect of PTH. However, as the expression levels of major bone matrix protein genes in individual osteoblasts of the dn-Runx2 transgenic mice were not significantly reduced, our findings also indicate that Runx2 is not essential for maintenance of the expression of major bone matrix protein genes in postnatal bone development and maintenance.
In dn-Runx2 transgenic mice, the trabecular bone had characteristics of compact bone, in that the bone was highly mineralized and collagen fibrils were densely and regularly deposited, and the trabecular bone was refractory to bone resorption. In contrast, the cortical bone of Runx2 transgenic mice is mainly composed of less mineralized, immature bone, so-called woven bone, in which collagen fibers run in all directions, and is prone to bone resorption (Liu et al.,2001; Geoffroy et al.,2002). Furthermore, Runx2 was strongly expressed in immature osteoblasts but downregulated during osteoblast maturation in wild-type mice (Figs. 6, 7). In postnatal bone development, therefore, Runx2 plays an important role in maintaining a supply of immature osteoblasts, leading to the formation of immature bone that is easily resorbed, whereas Runx2 has to be suppressed for osteoblast differentiation at the late stage and compact bone formation. Thus, bone maturity and the bone turnover rate are determined by the level of functional Runx2. In contrast to our present finding that suppression of Runx2 in osteoblastic cells resulted in an increase in trabecular bone volume, selective deficiency of a Runx2 isoform (type II Runx2) results in severe osteopenia (Xiao et al., 2005), indicating that insufficiency of Runx2 at the beginning of osteoblast differentiation leads to osteopenia in which both bone formation and bone resorption are reduced with more severe impairment in bone formation.
Runx2 is considered to play important roles in the regulation of expression of major bone matrix protein genes including Col1a1, osteopontin, bone sialoprotein, and osteocalcin and in postnatal bone development (Ducy et al.,1997,1999; Sato et al.,1998; Harada et al.,1999; Javed et al.,1999; Kern et al.,2001). Adenoviral introduction of Runx2 in mature osteoblastic cells in vitro upregulated osteocalcin expression but not Col1a1 expression, while adenoviral introduction of dn-Runx2 did not suppress the expression of Col1a1 nor osteocalcin, which were highly expressed in the mature osteoblastic cells (Fig. 1F–H). These findings, combined with previous findings, indicate that Runx2 is able to up-regulate the expression of major bone matrix protein genes in osteoblastic cells and in immature mesenchymal cells, although Runx2 is not absolutely required for the steady-state expression of these genes in mature osteoblastic cells. Moreover, the expression levels of major bone matrix protein genes in individual osteoblasts of dn-Runx2 transgenic mice were not reduced (Figs. 4A and 8A–C). Therefore, our findings indicate that Runx2 is not essential for maintenance of the expression of major bone matrix protein genes in mature osteoblasts in postnatal bone development and maintenance. The down-regulation of endogenous Runx2 during osteoblast maturation also supports the conclusion (Figs. 6, 7). As Runx2 induces the expression of major bone matrix protein genes in immature mesenchymal cells in vitro (Ducy et al.,1997; Harada et al.,1999), however, it is likely that Runx2 triggers the expression of these genes in mesenchymal progenitors in vivo.
Interestingly, the trabecular bone in dn-Runx2 transgenic mice did not significantly decrease after ovariectomy (Fig. 9B–D). In the trabecular bone, neither the parameters of bone resorption nor bone formation were enhanced in dn-Runx2 transgenic mice after ovariectomy (Fig. 9A,E–H), indicating that Runx2 is highly involved in the high bone turnover in the estrogen-deficient state. It is considered that the trabecular bone was conserved after ovariectomy in dn-Runx2 transgenic mice due to its compact bone-like structure and the absence of enhancement of osteoclastogenesis. We previously reported that in vitro Runx2 induces RANKL expression and inhibits OPG expression in calvarial cells derived from Runx2 null (−/−) mice (Enomoto et al.,2003). However, we could not detect evidence that Runx2 regulated RANKL and OPG expression in either Runx2 or dn-Runx2 transgenic mice (Liu et al.,2001) (Fig. 8P,Q). These findings suggest that Runx2 regulates RANKL and OPG expression only during the differentiation stage of mesenchymal cells. Dn-Runx2 expression in osteoblastic cells suppressed osteoclastogenesis after ovariectomy (Fig. 9E–G). However, the molecular mechanisms through which Runx2 mediates the enhancement of osteoclastogenesis in the estrogen-depleted state need to be further investigated. Estrogen inhibits osteoclastogenesis, and the estrogen receptor α interacts with Runx2, enhancing transcriptional activation by Runx2 (Manolagas and Jilka,1995; Pacifici,1996; McCarthy et al.,2003). As Runx2 was involved in the enhancement of osteoclastogenesis and bone resorption after ovariectomy (Fig. 9), estrogen seems to inhibit osteoclastogenesis through a mechanism other than Runx2-dependent transcriptional activation. As our findings indicate that Runx2 is responsible for bone loss in the estrogen-deficient state, dn-Runx2 transgenic mice will be a useful model for the study of mechanisms involved in the development of menopausal osteoporosis.
Cell Culture and Adenoviral Transfer
Mouse immature mesenchymal cell lines C2C12 and C3H10T1/2 and the mouse pre-osteoblastic cell line MC3T3-E1 were purchased from RIKEN Cell Bank (Tsukuba, Japan). Mouse mature osteoblastic cell line MLO-A5 was provided by Dr. Lynda Bonewald (The University of Texas Health Science Center, San Antonio, TX). Primary calvarial cells derived from mouse embryonic day 18.5 embryos were prepared as described previously (Komori et al.,1997). To induce osteoblast differentiation of primary calvarial cells and MC3T3-E1 cells, the medium was supplemented with 10 mM β-glycerophosphate and 100 μg/ml ascorbic acid. To induce osteoblast differentiation of C2C12 cells, the cells were treated with recombinant human BMP-2 (rhBMP-2) (100 ng/ml), which was provided by Yamanouchi Pharmaceutical Co. Ltd (Tokyo, Japan). Adenovirus vectors expressing type II Runx2, dn-Runx2 (short form), dn-Runx2 (long form), or EGFP were generated as previously described (Enomoto et al.,2003). Cells plated in collagen-coated 24-well plates were incubated with adenovirus for 2 hr. The infected cells were cultured for 7 days and stained for ALP as previously described (Komori et al.,1997).
Type II Runx2, dn-Runx2 (short form) that contains the runt domain and the nuclear localization sequence, and dn-Runx2 (long form) that contains the runt domain with the N-terminal domain of type I Runx2 and the nuclear localization sequence, were subcloned into the pSG5 vector (Stratagene, La Jolla, CA). A pGL3 vector containing 6 repeats of the consensus Runx2 binding sequence (6 × OSE2) was used as the reporter vector (Harada et al.,1999). pRL-CMV was used as a control vector. Each expression vector (3 ng), the reporter vector (200 ng), and the control vector (2 ng) were cotransfected into C3H10T1/2 cells in 48-well multiplates using FuGENE 6 (Roche Diagnostics). An increasing amount of each dn-Runx2 expression vector, 200 ng of the reporter vector, and 2 ng of the control vector were cotransfected into stable C3H10T1/2 transfectants of type II Runx2. The total amount of DNA for transfection was adjusted to 250 ng using an empty pSG5 vector. After 48 hr, the luciferase activity of the cell lysate was assayed using the Luciferase Reporter Assay System (Promega, Madison, WI). The transfection efficiency was normalized by quantifying renilla luciferase activity.
Generation of dn-Runx2 Transgenic Mice and Runx2/dn-Runx2 Double Transgenic Mice
To generate transgenic mice with osteoblasts that express dn-Runx2, dn-Runx2 cDNA (long form) was inserted into the mammalian expression vector pNASSβ (CLONTECH) by replacing the β-galactosidase gene at the Not I sites, and the 2.3-kb osteoblast-specific promoter region of mouse Col1a1 (Rossert et al.,1995) was inserted into pNASSβ at the XhoI site. The DNA fragment containing the 2.3-kb Col1a1 promoter, an intron from SV40, dn-Runx2 cDNA, and polyadenylation signal from SV40, was injected into the pronuclei of fertilized eggs from C57BL/6 X C3H F1 (B6C3H F1) mice. Transgene integration and expression were identified by Southern blot and Northern blot analyses, respectively, using the cDNA coding the runt domain of Runx2 as a probe. Transgenic lines were maintained in a B6C3H F1 background. The dn-Runx2 transgenic mice were mated with transgenic mice that expressed Runx2 under the control of the Col1a1 promoter (Liu et al.,2001) to generate Runx2/dn-Runx2 double transgenic mice. To investigate the effect of estrogen deficiency on bone, dn-Runx2 transgenic mice and their wild-type littermates at 4 months of age were ovariectomized or sham-operated and sacrificed at 5 weeks after the surgery. The urinary excretion of deoxypyridinoline was measured using an ELISA kit (Metra Biosystems). Results were expressed as deoxypyridinoline/creatinine (nmol/mmolCr). Prior to the study, all experiments were reviewed and approved by the Animal Care and Use Committee of Nagasaki University Graduate School of Biomedical Sciences.
Real-Time PCR and Northern Blot Analyses
Real-time PCR was performed using cDNA (7.5 ng total RNA equivalent) and the following primers as previously described (Enomoto et al.,2004): RANKL, 5′-CAAGCTCCGAGCTGGTGAAG-3′ and 5′-CCTGAACTTTGAAAGCCCCA-3′; OPG, 5′-AAGAGCAAACCTTCCAGCTGC-3′ and 5′-CACGCTGCTTTCACAGAGGTC-3′. We normalized the obtained CT (cycle number at which amplification threshold of detection was reached) values to that of rodent Gapdh (Applied Biosystems) expression by the ΔΔCT method. The mean ΔΔCT was converted to relative expression value by the equation, 2−ΔΔCt, and the range was calculated by the equation, 2(−ΔΔCt+StdevΔΔCt). Northern blot analysis was performed using cDNA probes of runt, Col1a1, osteopontin, osteocalcin, and GAPDH as previously described (Inada et al.,1999). The relative intensities of the bands were measured using NIH Image software.
Western Blot Analysis
Nuclear extracts were prepared from the maxillae and mandibles of newborn mice using NE-PER reagent (Pierce). Proteins (10 μg/well) were resolved by SDS-12/25% gradient polyacrylamide gel electrophoresis. The blots were first incubated with a polyclonal rabbit antibody against the N-terminal domain of type I Runx2 (Ueta et al.,2001), and then with horseradish peroxidase-conjugated anti-rabbit IgG (New England Biolabs).
X-ray and pQCT Analyses
Long bones were dissected from sacrificed mice and subjected to X-ray exposure using a Micro-FX1000 (Fuji Film Inc., Tokyo, Japan). For pQCT analysis, femurs were fixed with 70% ethanol and analysis was performed using an XCT Research SA (Stratec Medizintechnick). Parameters of the trabecular bone and cortical bone were analyzed using the threshold value, 395 and 690 mg/cm3, respectively.
For histological analyses of the long bones, mice were sacrificed at newborn stage, 1 week, 4 weeks, 8 weeks, 10 weeks, 12 weeks, 7 months, and 10 months of age. To prepare the sections for hematoxylin and eosin (H-E) staining and in situ hybridization, mice were fixed in 4% paraformaldehyde/0.1M phosphate buffer, and the long bones were decalcified in 0.5M EDTA/10% glycerol buffer (pH 7.5) and embedded in paraffin. Sections (7 μm thick) were stained with H-E or subjected to in situ hybridization using probes for Col1a1, osteopontin, osteocalcin, and Runx2 as described previously (Inada et al.,1999). For assessment of dynamic histomorphometric indices, mice were injected twice with calcein at a dose of 0.16 mg/10 g body weight, and analysed at 4 weeks, 10 weeks, or 7 months of age, or at 5 weeks after ovariectomy. The mice received the two injections at 6 and 2 days before sacrifice. The long bones were fixed with 70% ethanol, and the undecalcified bones were embedded in glycolmethacrylate. Three-micrometer–thick longitudinal sections from the distal parts of femurs were stained with toluidine blue and analyzed using a semiautomated system (Bone Histomorphometric System, System Supply). For ultrastructural analysis, specimens were immersed in a mixture of 2% paraformaldehyde and 2.5% glutaraldehyde in 0.067M cacodylate buffer (pH 7.4), post-fixed in a mixture of 1% osmium tetraoxide and 1.5% potassium ferrocyanide, and then embedded in epoxy resin (Taab). The ultrathin sections were examined under a TEM (Hitachi H-7000, Hitachi).
Serial sections were incubated with anti-Runx2 monoclonal antiboby (MBL), processed with Histofine Simple Stain MAX-PO(M) (Nichirei), and treated with phosphate buffer containing 0.05% 3,3′-diaminobenzidine(DAB), 0.01% nickel chloride, and 0.01% H2O2. For double labeling, the sections were incubated with rabbit anti-mouse osteocalcin polyclonal antibody (Takara) or rabbit anti-mouse osteopontin polyclonal antibody (IBL), processed with Histofine Simple Stain MAX-PO(R) (Nichirei), and treated with phosphate buffer containing 0.05% DAB and 0.01% H2O2. The sections were counterstained with methyl green.
In Vitro Osteoclastogenesis
Bone marrow cells were obtained from the tibias and femurs of 4- to 6-week-old C57BL/6 male mice, and primary osteoblastic cells were prepared from calvariae of C57BL/6 newborn mice. The osteoblastic cells were infected with EGFP-, Runx2-and-EGFP-, or dn-Runx2-and-EGFP-containing adenovirus. Bone marrow cells (5 × 105/ml) and osteoblastic cells (5 × 104/ml) were co-cultured in 200 μl α-MEM containing 10% heat-inactivated FBS in the presence of 10−8 M 1α, 25-dihydroxyvitamin D3 [1α, 25(OH)2D3] (Wako) and prostaglandin E2 (10−6 M) (Wako) on 96-well plates for 6 days. Media were changed every other day. Osteoclast formation was evaluated by TRAP staining. Cultured cells were fixed with 4% paraformaldehyde for 30 min, washed in 0.2% Triton X-100 at room temperature for 5 min, and incubated in acetate buffer (pH 5.0) containing naphthol AS-MX phosphate (Sigma), fast red-violet LB salt (Sigma), and 50 mM sodium tartrate. For the evaluation of bone resorption, the bone marrow cells and osteoblastic cells were co-cultured on dentin slices (Wako) for 6 days, the slices were stained with Coomassie Brilliant Blue R, and the resorbed area was analyzed using NIH image software.
PTH Administration and Micro-CT Analysis
Rat PTH(1-34) (Sigma) was administered to experimental animals subcutaneously as three injections per day over 8 hr at a dose of 80 μg/kg (total 240 μg PTH/kg/day) for 18 days. Control animals were given vehicle alone. Animals were then sacrificed and bone samples were collected to study trabecular bone mass and micro-architecture. Quantitative micro-computed tomography (micro-CT) was performed with a micro-CT system (μCT-20; Scanco Medical). Data from scanned slices was used for three-dimensional analysis to calculate femoral morphometric parameters, including trabecular and cortical bone volume density (bone volume [BV]/tissue volume [TV]), trabecular thickness (Tb.Th = 2 × BV/bone surface [BS]) and trabecular number [Tb.N = (BV/TV)/Tb.Th]. Trabecular bone parameters were measured on distal femoral metaphysis. Approximately 4.8 mm (0.5 mm far from the growth plate) were cranio-caudally scanned and a total of 400 slices with 12-μm increments were taken. Cortical BV/TV was measured on femoral diaphysis and proximal metaphysis.
Statistical analyses were performed using a Student's t-test. P < 0.05 was considered to be significant.
We thank B. de Crombrugghe for the Col1a1 promoter, L. Bonewald for MLO-A5, Y. Ito for Runx2 antibody, Yamanouchi Pharmaceutical Co. Ltd for rhBMP-2, F. Ikeda and H. Murayama for technical advice, K. Nonaka (Elk Corporation) for performing pQCT analysis, N. Komatsu and K. Kawagoe for maintaining mouse colonies, and A. Kakiya for secretarial assistance.