Smad6/Smurf1 overexpression in cartilage delays chondrocyte hypertrophy and causes dwarfism with osteopenia

Bone morphogenetic proteins (BMPs) were originally identified as secreted signaling molecules that could induce ectopic endochondral bone formation when implanted subcutaneously. Subsequent molecular cloning experiments have revealed that the BMP family consists of various molecules including the growth and differentiation factor subfamily, and belongs to the TGF-β superfamily. BMP family members have diverse biological activities during the development of various organs and tissues, as well as during embryonic axis determination (Hogan, 1996).

Members of the TGF-β superfamily transduce their signals through two types of serine/threonine kinase receptors, types I and II (Heldin et al., 1997; Shi and Massague, 2003). Upon ligand binding, type I and type II receptors form a tetramer consisting of two pairs of type I and type II receptors. Type II receptors phosphorylate type I receptors. Then, type I receptors phosphorylate downstream targets such as Smads. In vertebrates, seven type I receptors and five type II receptors have been found so far. Among them, three type I receptors, BMP type IA receptor (BMPR-IA, or activin receptor-like kinase [ALK-3]), BMPR-IB (ALK-6), and ALK-2, mediate BMP signaling. Smads are the major downstream targets of type I receptors for TGF-β/BMP superfamily proteins (Heldin et al., 1997). Eight Smads have been identified in mammals and are classified into three subgroups. Receptor-regulated Smads (R-Smads) are phosphorylated at SSXS motifs at their COOH terminus by type I receptors. Smad1, Smad5, and Smad8 are R-Smads that transduce BMP signals, and Smad2 and Smad3 are responsible for TGF-β and activin signaling. Phosphorylated R-Smads form heteromers with the common-partner Smad (Smad4), and translocate into the nucleus where they interact with transcriptional factors to bind directly or indirectly to specific DNA sequences for the activation of gene transcription.

BMP signaling is subject to delicate regulation at multiple levels: extracellularly, at the membrane site, and intracellularly (Balemans and Van Hul, 2002). In the extracellular space, several molecules antagonize BMPs. Among these antagonists, noggin is expressed in cartilage and binds to BMPs and prevents them from interacting with their receptors. At the intracellular level, inhibitory Smads, Smad6 and Smad7, inhibit phosphorylation of R-Smads by competing with R-Smads for binding to phosphorylated type I receptors. In particular, Smad6 appears to inhibit BMP signaling, whereas Smad7 associates stably with TGF-β receptor and BMP receptor complexes and inhibits the TGF-β– or BMP-mediated phosphorylation of R-Smads (Hanyu et al., 2001). In addition, ubiquitin-dependent protein degradation plays key roles in Smad signaling. Smad ubiquitin regulatory factor 1 (Smurf1) and Smurf2 induce the ubiquitination and degradation of Smad1 and Smad5 (Zhu et al., 1999; Zhang et al., 2001). Furthermore, Smurf1 and Smurf2 interact with nuclear Smad7 and induce the nuclear export of Smad7. Smurf–Smad7 complexes then associate with type I receptor for TGF-β and enhance its turnover (Kavsak et al., 2000; Ebisawa et al., 2001). Recent biochemical analyses have shown that Smurf1 binds to BMP type I receptors via Smad6 and Smad7, and that it induces the ubiquitination and degradation of these receptors (Murakami et al., 2003). Thus, Smad6 and Smurf1 cooperatively down-regulate BMP signals by degradation of R-Smads as well as BMP receptors. However, the physiological function of Smurfs is unknown.

During development, the limb skeleton is formed through endochondral bone formation (Erlebacher et al., 1995). Mesenchymal cells initially undergo condensation followed by differentiation of cells within these condensations into chondrocytes. Chondrocytes then proliferate and produce ECM to form primordial cartilage. Shortly after the primordial cartilage formation, proliferating chondrocytes in the central region of the cartilage undergo terminal differentiation to hypertrophic chondrocytes. Hypertrophic chondrocytes exit the cell cycle and synthesize an ECM that is different in composition from that of proliferating cartilage. The hypertrophic cartilage is invaded by blood vessels along with osteoblasts, osteoclasts, and hematopoietic cells to form primary ossification centers. Within these centers, the hypertrophic cartilage matrix is degraded, hypertrophic chondrocytes die, and osteoblasts replace the disappearing cartilage with trabecular bone (Olsen et al., 2000). Then, bone formation and maintenance are performed by a balance between the new apposition of bony matrix by osteoblasts and resorption by osteoclasts.

Smad proteins have been identified in growth plate cartilage (Flanders et al., 2001). In vitro analyses have shown that Smad6 regulates the chondrocytic phenotype (Valcourt et al., 2002; Nishihara et al., 2003). However, the physiological roles of Smad6 and Smad signaling in normal endochondral bone formation have not been determined. Here, we generated transgenic mice overexpressing Smad6 or Smurf1 in chondrocytes under the control of the α2(XI) collagen chain gene (Col11a2) promoter/enhancer sequences. We found that overexpression of Smad6 does not significantly affect chondrocyte proliferation, but significantly delays chondrocyte hypertrophy, which may lead to postnatal dwarfism with osteopenia. By using double-transgenic mice, we also found that Smurf1 supports Smad6 function in vivo.

We ligated the Smad6 cDNA with a 5′ FLAG epitope to Col11a2 promoter/enhancer sequences to construct the transgene Col11a2-Smad6 (Fig. 1 A). Generation zero (G₀) embryos of Smad6 transgenic mice were initially killed, in case Smad6 transgenic mice were lethal. 17 of 129 embryos were genetically positive for the transgene. Smad6 transgenic and normal embryos appeared similar. The Smad6 transgenic mice survived after birth, so several independent transgenic mouse lines could be established. The phenotypes of all transgenic mouse lines were similar, with differences in the degree of abnormalities between lines.

1 wk after birth, the Smad6 transgenic mice started to develop dwarfism (Fig. 1, B–H). Our data were collected from three transgenic founders derived from microinjections, transgenic pups with severe phenotypes generated from two mosaic founders (lines 138 and 139), and transgenic offspring with mild phenotypes generated from two transgenic founders (lines 165 and 199). Dwarfism was more severe in the transgenic mice of line 139 than of line 138 (Fig. 1 B). Phenotype severity closely correlated with transgene expression levels as shown by RT-PCR (Fig. 1 C). Quantitative real-time RT-PCR using total RNAs extracted from limb buds showed that the amount of Smad6 mRNAs of transgenic line 139 was threefold that of normal mice, and double that of transgenic line 138. At 3 wk after birth, the average crown–rump length of transgenic mice (line 199) was ∼20% shorter than that of normal mice (Fig. 1 D). Transgenic mice (line 199) weighed an average of 30–40% less (Fig. 1 E), and the average length of the skeletal components was 20–30% shorter compared with normal littermates (Fig. 1, F–H).

Northern blotting demonstrated Smad6 expression in the limb buds of transgenic mice. Transgene Smad6 mRNA was ∼2 kb in length, which was smaller than endogenous Smad6 mRNA due to shorter 5′ and 3′ untranslated regions (Fig. 2 A). Immunoblotting demonstrated the expression of a 70-kD FLAG-tagged Smad6 protein in limb buds of Smad6 transgenic mice (Fig. 2 B, open arrow). Immunohistochemistry using anti-Smad6 antibody showed more intense signals for Smad6 proteins in forelimb chondrocytes of Smad6 transgenic mice (Fig. 2, D, F, H, and J) than from wild-type mice (Fig. 2, C, E, G, and I) from 13.0 through 18.5 days post coitus (d.p.c.).

We examined Smad signaling in chondrocytes by immunohistochemistry using an antibody that recognizes only phosphorylated forms of Smad1, Smad5, and Smad8. Phospho-Smad1/5/8 immunoreactivity was reduced in transgenic cartilage sections from 16.5 d.p.c. limbs (Fig. 3, B and D) compared with the wild type (Fig. 3, A and C). We cultured explants of metatarsal primordial cartilage from 15.0 d.p.c. embryos. Recombinant human BMP2 (rhBMP2) proteins were added to the culture media and explants were histologically analyzed 2 h later. rhBMP2 dramatically increased phospho-Smad1/5/8 immunoreactivity in explants from wild-type mice (Fig. 3, E and G), but not from Smad6 transgenic mice (Fig. 3, F and H). These results suggested that Smad signaling was inhibited in the transgenic cartilage by an excess of Smad6.

We analyzed the bone structure of transgenic mice. Micro-CT analysis of the humerus revealed that the primary spongiosa was significantly more hypoplastic and disorganized in Smad6 transgenic mice (Fig. 4, B and D) than in normal littermates (Fig. 4, A and C). The average bone volume per total tissue volume of primary spongiosa was 30–40% smaller than that of normal littermates (Fig. 4 E).

Differences in bone formation were further examined by bone histomorphometric analysis. Trabecular bone volume was significantly decreased in Smad6 transgenic mice (Fig. 4 F). The osteoblast surface per bone surface in transgenic mice was significantly decreased (Fig. 4 G). We analyzed dynamic changes in bone formation and mineralization by injecting tetracycline and calcein at 2-d intervals. The distance between the two consecutive labels was significantly decreased in transgenic mice (Fig. 4, H–J). The mineralization surface was decreased, although not significantly (Fig. 4 K). These data showed a significantly decreased bone formation rate (Fig. 4 L). As for osteoclastic bone resorption, the osteoclast number per bone surface, osteoclast surface per bone surface, and erosive surface per bone surface were significantly increased (Fig. 4, M–O).

To elucidate the mechanism by which osteoclastic bone resorption was activated in Smad6 transgenic mice, we performed bone marrow cell culture and analyzed dexamethasone/parathyroid hormone-induced osteoclastogenesis. The number of tartrate-resistant acid phosphatase (TRAP)–positive multinucleated cells in the culture prepared from transgenic mice was equivalent to that of normal mice (Fig. 4, P–R). The resorption of hydroxyapatite by cultured osteoclast prepared from transgenic mice was essentially normal (Fig. 4 S). Next, we examined osteoclast formation activities in spleen cell culture in the presence of RANK ligand. RANK ligand–induced osteoclastogenesis of spleen cells from Smad6 transgenic mice was normal, as indicated by the number of TRAP-positive multinucleated cells and the resorption of hydroxyapatite (unpublished data). These results suggested that both osteoclast precursors and osteoclast-supporting activities of osteoblast/stromal cells were normal in the bone marrow of Smad6 transgenic mice.

Because the transgene was expressed specifically in cartilage, we analyzed the skeleton from earlier stages of development. Whole-mount in situ hybridization using a type II collagen gene (Col2a1) antisense cRNA probe showed that the pattern and intensity of signals did not obviously differ between Smad6 transgenic and normal mice at either 12.5 (Fig. 5, A and B) or 13.0 (Fig. 5, C and D) d.p.c. This suggested that mesenchymal condensation was essentially normal in the transgenic mice. At 13.5 d.p.c., the size and shape of each cartilaginous skeletal component stained with Alcian blue of transgenic mice were essentially identical to those of normal littermates (Fig. 5, E and F). The central regions of metatarsals were mineralized in normal mice as they were stained with Alizarin red S, but not in transgenic mice at 16.5 d.p.c. (Fig. 5, G and H, arrowheads) and at 18.5 d.p.c. (Fig. 5, I and J, arrowheads). The size of mineralized tissue in the humerus and femur was considerably smaller in transgenic mice than in normal mice (Fig. 5, G and H, arrows). At 3 wk of age, skeleton of transgenic mice was much smaller than that of normal mice (Fig. 5, K and L).

We further examined endochondral bone formation by histological means. At 13.5 d.p.c. when condensed mesenchymal cells differentiate into chondrocytes, chondrocytes did not obviously differ histologically between normal and transgenic mice (Fig. 6, A and B). At 14.5 d.p.c., proliferative chondrocytes exit the cell cycle and start terminal differentiation into hypertrophic chondrocytes at the center of each skeletal component (Fig. 6 C). On the other hand, transgenic cartilage from the humerus contained proliferative but not hypertrophic chondrocytes (Fig. 6 D). Hybridization in situ showed that proliferative chondrocytes in Smad6 transgenic mice expressed Col2a1 mRNA (Fig. 6 F), like those of normal mice. Northern blotting of limb bud extracts showed that expression levels of the Sox9 gene, Col2a1, and type IX collagen α1 chain mRNAs were similar between transgenic and normal mice at 13.5–19.5 d.p.c. (unpublished data). However, the transgenic cartilage lacked type X collagen–positive (Fig. 6 H) and osteopontin-positive (Fig. 6 J) cells, whereas cartilage from normal mice expressed these genes (Fig. 6, G and I). In addition, von Kossa staining revealed mineralization in the normal (Fig. 6 K) but not in the transgenic (Fig. 6 L) mouse humerus. At 16.5 d.p.c., normal mice formed ossification centers in the humerus (Fig. 6 M). However, although a zone of hypertrophic chondrocytes was present in the transgenic humerus, an ossification center had not formed (Fig. 6 N). At 18.5 d.p.c., the transgenic skeletal element was composed of cartilage at both ends and bone at the center, like the wild type (PFig. 6, O and P). These results suggest that Smad6 overexpression delayed chondrocyte hypertrophy and ossification by ∼3 d during development of the mouse humerus.

For further analysis of chondrocyte proliferation in Smad6 transgenic mice, we performed BrdU labeling at embryonic (Fig. 7, A and B) and postnatal (Fig. 7, C–F) stages. There were not significant differences in BrdU labeling indexes between chondrocytes of normal and Smad6 transgenic mice at 16.5 d.p.c. (Fig. 7 G) and at 3 wk of age (Fig. 7 H).

Once hypertrophic chondrocytes formed at later stages of development, populations of hypertrophic chondrocytes of transgenic mice did not significantly differ from those of normal littermates (Fig. 7, I, J, and M). However, populations of hypertrophic chondrocytes in Smad6 transgenic mice decreased after birth (Fig. 7, K and L). The mean height of zones of hypertrophic chondrocytes was significantly less than that of normal littermates at 3 wk after birth (Fig. 7 N).

To investigate BMP signaling in chondrocytes, we organ cultured primordial metatarsal cartilage at 15.0 d.p.c. Phase-contrast microscopy could distinguish the zones of proliferative and mineralized hypertrophic cartilage in the cultures (Fig. 8 A), which was confirmed by histology (Fig. 8, B and C). At the start of culture, metatarsal rudiments from Smad6 transgenic mice were indistinguishable from those of normal mice (Fig. 8, D and E). After 4 d of culture in control medium, the length of hypertrophic cartilage in normal rudiments was increased (Fig. 8 F), whereas transgenic rudiments lacked hypertrophic cartilage (Fig. 8 G). Culture in the presence of rhBMP2 resulted in excessive outgrowth of the proliferative cartilage at both ends of the rudiments and enhanced formation of a hypertrophic center in the normal rudiments (Fig. 8 H). In transgenic rudiments incubated with rhBMP2, proliferative cartilage expanded (Fig. 8 I) like that of normal mice (Fig. 8 H), but formation of the hypertrophic center was limited (Fig. 8 I) compared with that of normal mice (Fig. 8 H). These findings were confirmed by morphometric analysis of the rudiments (Fig. 8, J and K), indicating that Smad6 overexpression inhibited chondrocyte hypertrophy induced by rhBMP2.

To investigate in vivo function of Smurf1, we generated transgenic mice overexpressing Smurf1 in chondrocytes. We prepared the transgene construct by ligating Smurf1 cDNA with a 5′ FLAG epitope to Col11a2 promoter/enhancer sequences (Col11a2-Smurf1; Fig. 9 A). Smurf1 transgenic mice appeared normal in all respects, as they grew normally after birth and were fertile. Northern blotting demonstrated Smurf1 expression in transgenic limb buds (Fig. 9 B). Transgenic 3-kb Smurf1 mRNA was shorter than endogenous Smurf1 mRNA due to differences at both the 5′ and 3′ untranslated regions. Exogenous rhBMP2 added to Smurf1 transgenic cartilage in organ cultures of metatarsal rudiments at 15.0 d.p.c. caused proliferative cartilage outgrowth and hypertrophic center formation as in normal mice. The mean area of proliferative cartilage in Smurf1 transgenic rudiments (0.44 ± 0.03 mm²) did not significantly differ from that of normal littermates (0.47 ± 0.07 mm²; n = 6, P = 0.08). The mean length of hypertrophic cartilage in Smurf1 transgenic rudiments (0.26 ± 0.02 mm) also did not significantly differ from that of normal littermates (0.26 ± 0.01 mm; n = 6, P = 0.75). These results suggest that Smurf1 overexpression does not significantly affect either chondrocyte proliferation or hypertrophy induced by rhBMP2.

We then produced transgenic offspring overexpressing both Smad6 and Smurf1 in chondrocytes (Smad6/Smurf1 double-transgenic mice) by mating Smurf1 transgenic mice with Smad6 transgenic mice of line 199 to test cooperative function of Smad6 and Smurf1. The phenotypes of the double-transgenic mice were similar to but more severe than those of Smad6 transgenic mice. At 16.5 d.p.c., mineralization was evident in the metatarsals of normal and Smurf1 transgenic mice (Fig. 9, C and D, arrowheads), but not in Smad6 transgenic and Smad6/Smurf1 double-transgenic mice (Fig. 9, E and F, arrowheads). The mineralized area in the femur was significantly shorter in Smad6 transgenic mice than in normal mice. Furthermore, the mineralized area in the femur was significantly shorter in Smad6/Smurf1 double-transgenic mice than in Smad6 transgenic mice. (Fig. 9, C–G). Histological analysis revealed that ossification was further delayed in the double-transgenic mice compared with the Smad6 transgenic mice (Fig. 9, H and I). Labeling cartilage with BrdU revealed little differences in chondrocyte proliferation among normal littermates and the single- or double-transgenic mice. The labeling indexes at 16.5 d.p.c. were 7.66 ± 1.51, 7.86 ± 1.05, 7.65 ± 1.12, and 7.31 ± 3.74 cells/0.01 mm² cartilage in normal littermates, Smurf1 transgenic, Smad6 transgenic, and Smad6/Smurf1 double-transgenic mice, respectively (no significant difference; n = 5, P > 0.05).

Mice lacking Smad6 develop cardiovascular abnormalities (Galvin et al., 2000). Ectopic endochondral bone forms in the heart of mutant mice, suggesting that Smad6 plays certain roles in endochondral bone formation in vivo. However, the physiological roles of Smad6 and Smad signaling in normal endochondral bone formation are poorly understood. In this work, we generated transgenic mice overexpressing Smad6 in chondrocytes. Smad6 overexpression resulted in a delayed chondrocyte hypertrophy and mineralization in endochondral ossification. Smad1/5/8 phosphorylation was inhibited in Smad6 transgenic cartilage, suggesting that Smad signaling was impaired in these mice.

Smad6 appears to block BMP signaling, whereas Smad7 blocks that of both TGF-β and BMP (Hanyu et al., 2001). Skeletal mineralization is delayed and skeletal patterning is defective in mice lacking growth and differentiation factor 5 (Storm et al., 1994), BMPR-IB (Yi et al., 2000), or BMP7 (Luo et al., 1995; Jena et al., 1997). The similarity of delayed mineralization between these mutant mice and Smad6 transgenic mice suggests that BMP signals are blocked by Smad6 during endochondral ossification. This notion was confirmed by our results that chondrocyte hypertrophy and mineralization induced by rhBMP2 was inhibited in cartilage explants from Smad6 transgenic mice. Because the phosphorylation of Smad1/5/8 induced by rhBMP2 was inhibited in explants from Smad6 transgenic mice, we concluded that Smad6 regulates chondrocyte hypertrophy through the inhibition of Smad1/5/8 phosphorylation, thus down-regulating BMP signaling in endochondral bone formation.

The most fundamental abnormality during chondrocyte differentiation in Smad6 transgenic mice was a delay in chondrocyte hypertrophy in humeri at 14.5 d.p.c. This delay was accompanied by the persistent expression of the Col2a1 gene and the retarded expression of the type X collagen gene (Col10a1). Extensive analyses in vitro have shown that BMP signals promote chondrocyte hypertrophy, and BMP-responsive cis-acting elements have been identified in the promoter sequence of the Col10a1 gene (Volk et al., 1998; Drissi et al., 2003). These in vitro analyses and our in vivo results collectively suggest that Smad6 overexpression blocks BMP signaling, thus preventing transcriptional activation of the Col10a1 gene.

Smurf1 binds Smads 1 and 5 and promotes their degradation (Zhu et al., 1999). Smurf1 and Smad6 form complexes and inhibit BMP signaling through the ubiquitin-dependent degradation of BMP receptors as well as of R-Smads (Murakami et al., 2003). Smurf2 may also exhibit functions similar to Smurf1. To examine the in vivo function of Smurf1, we generated transgenic mice expressing Smurf1 in chondrocytes and did not find obvious abnormalities. These results suggest that sufficient Smurf1 already exists in normal chondrocytes. When apparently normal Smurf1 transgenic mice were mated with Smad6 transgenic mice, the endochondral ossification of progenies overexpressing both Smad6 and Smurf1 was more delayed than in transgenic mice overexpressing only Smad6. It is likely that the Smad6 transgenic mice have far less Smurf1/2 than Smad6. In Smad6/Smurf1 double-transgenic mice, Smurf1 derived from the transgene might compensate for this shortage, thus supporting the activities of a large amount of Smad6. When Smurf1 transgenic mice were mated with those of the Smad6 transgenic line 165 in which the expression level was low, the phenotypic severity of the resultant double-transgenic progeny did not differ from those of Smad6 transgenic mice of line 165 (unpublished data). In Smad6 transgenic mice of this line, endogenous Smurf1/2 fully supported the activities of endogenous Smad6 and that derived from the transgene. From these lines of discussion, the expression level of Smad6 appears to be critical in the regulation of conversion from proliferative chondrocytes to hypertrophic chondrocytes. Actually, the expression level of Smad6 is decreased in the transitional zone between proliferative chondrocytes and hypertrophic chondrocytes (Flanders et al., 2001), suggesting that critical regulation of Smad6 expression is responsible for this conversion. The expression level of inhibitory Smads seems to be consistently and strictly regulated through autoregulatory negative feedback during signal transduction of the TGF-β/BMP superfamily because the inhibitory Smad mRNA is induced by TGF-β stimulation (Heldin et al., 1997).

The most apparent phenotype of Smad6 transgenic mice was postnatal dwarfism and osteopenia. Dynamic bone histomorphometric analysis revealed that osteoblastic bone formation decreased and that osteoclastic bone resorption increased in Smad6 transgenic mice. However, results from cultured bone marrow cells suggested normal osteoclast-supporting activities of osteoblasts/stromal cells prepared from Smad6 transgenic mice. These in vitro results suggest that the increased in vivo osteoclastic bone resorption in transgenic mice was not due to an autonomous abnormality within bone marrow cells. This notion was consistent with the observation that the transgene was specifically expressed in chondrocytes. We speculate that abnormal activities of osteoblasts and osteoclasts might be associated with the dysfunction in cartilage during endochondral bone formation. However, we could not rigorously exclude leaky transgene expression in cells in osteoblast lineage, and thus the possibility that osteoblasts have a primary malfunction.

BrdU labeling revealed that proliferation of chondrocytes in Smad6 transgenic mice was normal at the embryonic stage and at the postnatal stage. Therefore, postnatal dwarfism in Smad6 transgenic mice might develop through a different mechanism from that in transgenic mice overexpressing activated FGF receptor 3 in cartilage, in which chondrocyte proliferation is normal in embryos but postnatally decreased (Naski et al., 1998).

The onset of chondrocyte hypertrophy was delayed by ∼3 d during the development of Smad6 transgenic mice. Once the zone of hypertrophic chondrocytes formed, the height of the zone of hypertrophic chondrocytes was essentially normal, suggesting that the population of hypertrophic chondrocytes is strictly regulated at late embryonic stages. We did not determine which mechanism maintains the population of hypertrophic chondrocytes in Smad6 transgenic embryos. However, this population decreased in the transgenic mice after birth. It is likely that reduced hypertrophic chondrocyte population lead to a deficiency in signals required for coordination of growth and bone formation after birth. Hypertrophic chondrocytes are known to produce various factors, including angiogenic factors (Karsenty and Wagner, 2002). A reduction in production of such factors may well lead to dysfunction of osteoblast/osteoclast activities, thus resulting in impaired bone growth and osteopenia.

It is also possible that abnormal chondrocyte hypertrophy could result in the impaired structure of the ECM of hypertrophic cartilage. An impaired matrix might not provide a suitable scaffold for osteoblasts and osteoclasts to replace cartilage with bone. This speculation remains to be examined.

Previously, we generated transgenic mice expressing noggin in chondrocytes under the control of the identical Col11a2 promoter/enhancer sequences used in this paper (Tsumaki et al., 2002). We examined 62 G₀ founder mouse embryos for the noggin transgene, and 7 of them displayed a severe phenotype and almost completely lacked cartilage formation during development. On the other hand, 17 of 129 G₀ embryos were genetically positive for the Smad6 transgene, and skeletal abnormalities were either relatively minor or almost absent. Although we could not exclude the possibility that the phenotypic difference between noggin and Smad6 transgenic mice might be related to the different level of expression of the transgenes, these results suggest that noggin tends to affect cartilage more than Smad6 when expressed in chondrocytes in transgenic mice.

In addition, chondrocyte proliferation is inhibited in mice lacking BMPR-IB (Yi et al., 2000) and in cartilage explants cultured in the presence of noggin (Minina et al., 2001). Reports have consistently indicated that BMP signaling stimulates chondrocyte proliferation during endochondral bone formation at the embryonic stages of transgenic mice overexpressing BMPs in cartilage (Tsumaki et al., 1999, 2002) and in organ culture of cartilage rudiments in the presence of BMPs (De Luca et al., 2001; Minina et al., 2001).

On the other hand, chondrocyte proliferation appeared normal in Smad6 transgenic mice, as indicated by BrdU labeling. For explanation of the discrepancy between the findings of chondrocyte proliferation obtained from Smad6 transgenic mice and those of other papers, we considered four possibilities. First, we could not exclude the possibility that the expression level of Smad6 transgene was not sufficient to block BMP signaling completely, although immunohistochemical analysis showed strong expression of the transgene (Fig. 2, C–J) Second, Smad signaling might not be blocked by Smad6 alone. Certainly, phenotypes of Smad6/Smurf1 double-transgenic mice were more severe than that of Smad6 transgenic mice. However, it is still milder than that of noggin transgenic mice, and chondrocyte proliferation remained normal, as indicated by BrdU labeling. Third, BMP signals might be mediated by signaling pathways other than Smad proteins. In certain cell types, various MAPKs have been reported to mediate BMP pathways (Iwasaki et al., 1999). Existence of such pathways might account for the discrepancy between chondrocyte proliferation in BMPR-IB–deficient mice and Smad6 transgenic mice. There is likelihood of the fourth possibility as follows: BMPs and noggin are secreted and diffuse. In addition to direct binding to chondrocytes, these proteins might exert indirect effects on chondrocytes. For example, BMPs act on cells around cartilage, and in return these cells secrete factors, affecting chondrocyte proliferation. Thus, the addition or overexpression of BMPs/noggin modulate chondrocyte proliferation directly and indirectly. On the other hand, Smad6 overexpression in chondrocytes might block only the direct effect of BMPs on chondrocytes.

In conclusion, our data on inhibition of endochondral bone formation in Smad6 and Smurf1 transgenic mice suggest a role for Smad signaling in skeletogenesis and growth. By down-regulating Smad1/5/8 phosphorylation and BMP signals, Smad6 plays an important role in regulation of chondrocyte hypertrophy and synergistically cooperates with Smurf1 in vivo.

The α2(XI) collagen gene–based expression vector, 742lacZInt, contains the Col11a2 promoter (−742 to +380), an SV40 RNA splice site, the β-galactosidase reporter gene, the SV40 polyadenylation signal, and 2.3 kb of the first intron sequence of Col11a2 as an enhancer (Tsumaki et al., 1996). To create a Smad6 transgene, a 1.5-kb DNA fragment covering the entire coding region of mouse Smad6 cDNA tagged with a FLAG sequence at the NH₂ terminus was prepared. The FLAG-tagged Smad6 cDNA was cloned into the NotI sites of 742lacZInt expression vectors by replacing the β-galactosidase gene to create Col11a2-Smad6. For the Smurf1 transgene, a FLAG-tagged Smurf1 cDNA was cloned into NotI sites of the expression vectors 742lacZInt by replacing the β-galactosidase gene to create Col11a2-Smurf1.

The plasmids Col11a2-Smad6 and Col11a2-Smurf1 were digested with EcoRI and PstI to release the inserts. Transgenic mice were produced by microinjecting each of the inserts into the pronuclei of fertilized eggs from F1 hybrid mice (C57BL/6x DBA) as described previously (Tsumaki et al., 1996). Transgenic embryos were identified by PCR assays of genomic DNA extracted from the placenta or skin. Genomic DNA was amplified by transgene-specific PCR using primers derived from mouse Smad6 cDNA (5′-CAAGATCGGTTTTGGCATACTG-3′) and from the SV40 poly(A) signal region (5′-TCACTGCATTCTAGTTGTGGTTTGTCC-3′) to amplify a 411-bp product for Smad6 transgenic mice. To discriminate Smurf1 transgenic mice, genomic DNA was amplified by transgene-specific PCR using primers derived from mouse Smurf1 cDNA (5′-ATGGACTACAAGGACGATGATGACAAGG-3′ and 5′-AGGGGCTGGTTCCTCCATGAAGCAG-3′) to amplify a 570-bp product. Smad6/Smurf1 double-transgenic pups were generated by mating Smad6 and Smurf1 transgenic mice.

Mice were dissected and fixed in 100% ethanol overnight, and then stained with Alcian blue followed by Alizarin red S solution according to standard protocols (Peters, 1977).

The humeri from 5-wk-old transgenic mice and normal littermates were dissected and analyzed using a micro-focus X-ray CT system (SMX-100CT-SV; Shimadzu). The region extending from the proximal growth plate to the metaphyseal part of the humerus on 350 slices was scanned at a width of 6.75 μm per slice. The data were reconstructed to produce images of the humerus using 3-D visualization and measurement software (Vay Tek, Inc.). Bone mass was quantified by selecting 90 consecutive slices distal to the proximal growth plate (0.6 mm in length). Trabecular parameters in the metaphysis were determined using image analysis software (TRI/3D-BON; RATOC).

Embryos were dissected using a stereomicroscope (model SMZ645; Nikon), fixed in 4% PFA, processed, and embedded in paraffin. Serial sections were stained with hematoxylin and eosin, with safranin O/fast green/iron hematoxylin, or by the von Kossa reaction. Dynamic histomorphometric indices were determined by double-fluorescence labeling in vertebral bodies. 4-wk-old normal and transgenic mice were administered i.p. with tetracycline (20 mg/kg body weight; Sigma-Aldrich), followed by calcein label (10 mg/kg body weight; Wako Chemicals) 2 d later. After 24 h, the mice were killed. Bones were fixed with ethanol and embedded in methylmethacrylate. Sections were cut and viewed using a fluorescence microscope (Eclipse E1000; Nikon). The Niigata Bone Science Institute (Niigata, Japan) performed histomorphometric analyses. Immunohistochemistry proceeded using a rabbit pAb against Smad6 (1:200 dilution; Zymed Laboratories) and a rabbit pAb against phospho-Smad1/5/8 (1:200 dilution; Cell Signaling Technology). Immune complexes were detected using streptavidin-peroxidase staining and Histofine SAB-PO kits (Nichirei). Images were acquired using a microscope (Eclipse E1000; Nikon) with a digital camera system (DXM1200; Nikon).

Pregnant mice bearing 16.5 d.p.c. embryos and 3-wk-old mice were i.p. injected with BrdU labeling reagent (10 μl/g body weight; Zymed Laboratories). 2 h later, the mice were killed. Embryonic limb buds and tibia of the 3-wk-old mice were dissected and sectioned. Incorporated BrdU was detected using a BrdU staining kit (Zymed Laboratories) to distinguish actively proliferating cells. Tissue sections were measured using a micrometer, and the average number of BrdU-positive cells/mm² cartilage ± SD was calculated.

Total RNA extracted from the limb buds of 14.5–18.5 d.p.c. transgenic and normal embryos using RNeasy Mini Kits (QIAGEN) was fractionated by electrophoresis through formaldehyde agarose gels and transferred onto Nytran^® membranes (Schleicher & Schuell Bioscience). Complementary DNAs (cDNAs) were labeled with [³²P]dCTP using Prime-it^® II kits (Stratagene). The membranes were hybridized with ³²P-labeled Smad6 cDNA and rehybridized with ³²P-labeled probes for mouse α1(II) collagen, α1(IX) collagen, and Sox9.

Total RNAs were digested with DNase to eliminate any contaminating genomic DNA before real-time quantitative RT-PCR. 2 μg of total RNA was reverse transcribed into first-strand cDNA using OmniScript^® reverse transcriptase (QIAGEN) and an oligo(dT)12-18 primer. The PCR amplification proceeded in 20 μl containing 1 μl of cDNA, 2 μl of SYBER Green™ Master Mix (QIAGEN), and 10 pmol of primers specific for Smad6 (5′-GATCCCCAAGCCAGACAGT-3′ and 5′-AGCCTCTTGAGCAGCGCGAGTA-3′) to generate a 126-bp product (GenBank/EMBL/DDBJ accession no. NM008542). The cDNA was amplified by 35 cycles using a LightCycler^® quick system (Roche) according to the following protocol: 94°C for 15 s, 60°C for 20 s, and 72°C for 6 s, each with a temperature transition rate of 20°C, according to the manufacturer's instructions.

Digoxigenin-11 UTP-labeled single-strand RNA probes were prepared using a DIG RNA labeling kit (Boehringer) according to the manufacturer's instructions. We generated antisense and sense probes using α1(II) collagen, α1(X) collagen, and osteopontin cDNAs. Hybridization proceeded as described previously (Hirota et al., 1992; Conlon and Herrmann, 1993). A Genius detection system (Boehringer) detected signals according to the manufacturer's instructions.

Limb buds of 14.5 d.p.c transgenic and normal embryos were lysed with RIPA buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, and 10 mM DTT) supplemented with protease and phosphatase inhibitor cocktails (Sigma-Aldrich). The positive control consisted of COS7 cells transfected with the expression construct FLAG-Smad6 that was also lysed with RIPA buffer. The cell lysates were incubated with anti-Smad6 antibody (Zymed Laboratories) for 3 h at 4°C followed by an incubation with protein G–agarose beads (Roche) for 3 h at 4°C. After five washes with lysis buffer (20 mM Hepes, 150 mM NaCl, 10% glycerol, 1.5 mM MgCl₂, 1 mM EGTA, and 100 μM orthovanadate), 1× SDS sample buffer was added to the agarose beads. The samples were incubated for 5 min at 95°C, fractionated by 10% SDS-PAGE, transferred onto nitrocellulose membranes (Bio-Rad Laboratories), and Western blotted against anti-FLAG M2 mAb (Sigma-Aldrich). Immunocomplex bands were visualized using the ECL Western blotting detection system (Amersham Biosciences).

Osteoclast formation and bone resorbing activity were determined using the modified method described in Azuma et al. (2000). In brief, bone marrow cells prepared from the femurs and tibias of 10-wk-old transgenic mice or wild-type control mice were suspended in α-modified essential medium containing 10% FBS, and were cultured in 48-well plates (10⁶ cells/0.5 ml per well) for 7 d in the presence of 0.1 μM dexamethasone (Sigma-Aldrich) and 0.01 μM recombinant human parathyroid hormone (Peptide Institute, Inc.). Cells were then fixed and stained for TRAP using a TRAP staining kit (Hokudo) according to the manufacturer's recommendation. The number of multinucleated TRAP-positive cells with more than three nuclei was counted under a microscope (Eclipse TE300; Nikon). To examine calcified matrix resorption activity, 5 × 10⁵ cells were cultured in 16-well hydroxyapatite-coated slides (Osteologic; Becton Dickinson) for 14 d, and the resorption area was calculated by computer-assisted image analysis.

Metatarsal rudiments were cultured as described previously (Haaijman et al., 1997). Metatarsal rudiments were dissected from transgenic and normal embryos at 15.0 d.p.c. and cultured in α-modified essential medium without nucleosides (Invitrogen), supplemented with 0.05 mg/ml ascorbic acid (Sigma-Aldrich), 0.3 mg/ml l-glutamine (Merck), 0.05 mg/ml gentamycin (Invitrogen), 0.25 mg/ml Fungizone^® (Invitrogen), 1 mM β-glycerophosphate (Merck), and 0.2% FBS (GIBCO BRL) in a humidified atmosphere of 5% CO₂ in air at 37°C. 1 d after starting the cultures, the rudiments were incubated in 400 μl of the same medium containing 500 ng/ml of rhBMP2 (Yamanouchi Pharmaceutical Co., Ltd.) or without rhBMP2 for a further 3 d. For immunohistochemical analysis using anti-phosphorylated Smad1/5/8, rudiments were sectioned before and after a 2-h incubation with rhBMP2. Images of rudiments obtained under inverted phase-contrast microscopy were analyzed morphometrically (Eclipse TE300; Nikon). Areas of proliferative cartilage and length of hypertrophic cartilage were measured using NIH Image software (National Institutes of Health, Bethesda, MD).

We are grateful to Tomoatsu Kimura and Riko Nishimura for critical discussion. We are indebted to Yuko Shikauchi, Yuri Inada, and Kaori Sudo for their commitment to this investigation. We thank Akemi Ito for the histomorphometric analysis and Hideki Tsuboi for bone marrow cell culture.

This work was supported in part by Scientific Research grant 15390458 from the Ministry of Education, Science and Culture of Japan; Health and Labor Sciences Research Grants of Japan; and the Grant of Japan Orthopaedic and Traumatology Foundation, Inc. (0126).