Noggin, Cartilage Morphogenesis, and Joint Formation in the Mammalian Skeleton

See allHide authors and affiliations

Science  29 May 1998:
Vol. 280, Issue 5368, pp. 1455-1457
DOI: 10.1126/science.280.5368.1455


Noggin is a bone morphogenetic protein (BMP) antagonist expressed in Spemann's organizer. Murine Noggin is expressed in condensing cartilage and immature chondrocytes, as are many BMPs. In mice lacking Noggin, cartilage condensations initiated normally but developed hyperplasia, and initiation of joint development failed as measured by the expression of growth and differentiation factor–5. The maturation of cartilage and Hoxd expression were unaffected. Excess BMP activity in the absence of Noggin antagonism may enhance the recruitment of cells into cartilage, resulting in oversized growth plates; chondrocytes are also refractory to joint-inducing positional cues.

The bones of the developing limb bud are formed by condensations of chondrocytes followed by endochondral ossification. Postembryonic growth continues at the growth plates, at the ends of the bones (1). A series of inductive events determines the sizes and shapes of individual limb skeletal elements (2). Many growth factors of the BMP family have been implicated in limb growth and patterning (3). BMP2 and BMP4, for example, when overexpressed during chick limb development alter the size and shape of the long bones, probably by increasing recruitment of nondifferentiated limb mesenchyme into the chondrogenic pathway (4, 5). The joints are formed after the initial cartilage condensation and are first recognized histologically by an increase in cell density. Cell death and cavitation follows (1). The location of the joints defines the layout of the skeleton. Growth and differentiation factor–5 (GDF-5), a divergent member of the BMP family, is implicated in joint specification through its expression in prospective joints and its disruption in thebrachypodism mouse mutation (6).

BMP activities may be modulated not only through gene expression and protein processing, but also by interaction with antagonists such as Noggin and Chordin (7). Here we show that Noggin expression is essential for proper skeletal development; excess BMP activity in the Noggin null mutant resulted in excess cartilage and failure to initiate joint formation.

A null mutation was made in the mouse gene Noggin by replacing the coding sequence with the lacZ gene, deleting all but 10 NH2-terminal residues of Noggin. The effect of this mutation on the central nervous system and somite patterning is discussed elsewhere (8). This null mutation should cause overactivity of BMP proteins, because the BMP antagonist, Noggin, has been removed.

Although heterozygous embryos are normal, all skeletal structures are abnormal in the homozygous mutant, with striking defects in the vertebrae, ribs, and limbs (Fig. 1). The severity of the axial defects increases caudally. The skull and cervical vertebrae are nearly normal, but the thoracic vertebrae are fused, fail to close dorsally, and do not develop a neural arch, and the lumbar and tail vertebrae are missing. The ribs are abnormally few and branched, and sternal bands fail to fuse completely (9). Alizarin red staining of bone shows that the schedule of ossification for both axial and limb skeleton in the homozygous mutant appears normal. Although defects in the axial skeleton suggest that Noggin affects cartilage development, the early action of Noggin in midline patterning complicates interpretation of the phenotype (8). Consequently, we focused our analysis on the limb.

Figure 1

Skeletal abnormalities in Nogginhomozygous mutants. Skeletons, with forelimbs removed, from wild-type (A) and mutant (C) 18.5-dpc (days post coitum) embryos were stained with alcian blue for nonmineralized cartilage and alizarin red for mineralized cartilage and bone (21). The forelimbs are shown in (B) and (D), respectively. In (D), continuous ossification from the radius to humerus is indicated by the solid arrow and a cartilaginous spur by the open arrow.

The mutant limbs are shorter than those of the wild type and broadened along the anterior-posterior axis (Fig. 1D). Fusion of the joints is particularly evident at the elbow, where the radius and humerus are joined by a continuous ossification. The digits have secondary fusions and occasional cartilaginous spurs, and they lack joints.

We analyzed normal Noggin expression in heterozygous embryos using the lacZ transgene (Fig. 2A) (10). In situ hybridization (11) of a Nogginantisense probe to whole embryos or sectioned limbs confirmed that lacZ staining reflects authentic Noggin gene expression (9, 12) and correlates with the sites of abnormal development of the skeleton. Expression of Noggin in forelimb cartilage of the 16.5-dpc heterozygote is widespread but is extinguished in the developing joint cavity and in maturing hypertrophic cartilage (Fig.2B). Expression at the articular surface persists into adult stages (12).

Figure 2

Expression of Noggin and GDF-5 in the developing skeleton. Heterozygous embryos at 13.5 (A) or 16.5 (B) dpc were stained to detect lacZ activity (10). A section of a forelimb is shown in (B): j, joint; p, proliferating zone; h, hypertrophic cartilage. In panels (C) through (N) GDF-5 expression was detected by whole-mount in situ hybridization (blue) (11); Noggin/lacZ expression in heterozygotes (+/–) and homozygotes (–/–) was detected with salmon-gal staining (pink) (22). Bars are 1 mm. All limbs are right hindlimbs with anterior at top. (C through E) 11.5 dpc; (F through H) 12.5 dpc; (I throughK) 13.5 dpc; (L through N) 14.5 dpc. Wild type (+/+) (C, F, I, L); heterozygous (+/–) (D, G, J, M); homozygous (–/–) (E, H, K, N). GDF-5 transcripts are first detected in the developing phalanges at 12.5 dpc [(F) and (G)] but are not detected in the homozygous mutant limb (H). GDF-5 expression is peripheral to the sites of cartilage condensation that are revealed by Noggin expression, except where joints are to be formed (I, J, L, M). The overgrowth of cartilage in the mutant limb is seen by the extensive area of pink (lacZ) stain in (N), and is confirmed histologically (P). The arrow in (N) marks a cartilaginous spur. GDF-5 expression is shown in sectioned limbs from (O) wild-type and (P) homozygous mutant 16.5-dpc embryos (11).

To distinguish whether joints were absent from the maturing skeleton as a consequence of excess chondrocyte growth and secondary fusions, or a failure to divide the initial condensation, we examined the onset of gene expression in the joint. Noggin (pink) and GDF-5 (blue) expression were analyzed in 11.5- to 14.5-dpc embryos (Fig. 2, C to N). The initial chondrocyte condensations of heterozygous and homozygous mutant mice are indistinguishable. At first, GDF-5 is expressed around and at the borders of the condensations. As cartilage develops, GDF-5 expression is up-regulated in the prospective joints (Fig. 2). By 13.5 dpc (Fig. 2, I to K), the mass of cartilage expressing Noggin or seen histologically (19) is greater in null mutant embryos (Fig.2K) and seems to occur at the expense of the residual limb mesenchyme. At 13.5 dpc, GDF-5 expression is apparent in prospective joints of the phalanges in normal embryos (Fig. 2, I and J) but not the homozygotes, even though GDF-5 is expressed in the interdigital mesenchyme (Fig.2K), where it may also have a role in cell death (13). By 14.5 dpc, much of the growth of the limb in the long axis is confined to the growth plates, and the continued failure to specify joints, especially in the digits, probably leads to the slightly shorter length of mutant limbs at birth (Fig. 1, B and D). Examination of sectioned material confirmed that, whereas GDF-5 is normally expressed in all joints at 16.5 dpc (1), expression is absent from similar positions in the homozygous mutant (Fig. 2, O and P).

Elevated BMP activity in the Noggin null mutant may cause misregulation of BMP-dependent genes including the BMP genes themselves, which are up-regulated in the notochord of Noggin mutant embryos (8). BMP6 is expressed in the hypertrophic zone of cartilage and in the joints (14) (Fig. 3A). BMP6 expression is normal in the Noggin mutant (Fig. 3B), as is expression of BMP2, BMP4, and BMP5 (12). Therefore, defects in the Noggin null mutant are likely to result from the lack of Noggin protein and the consequent inability to locally antagonize BMP activity, rather than from changes in the expression patterns of members of the BMP family.

Figure 3

Normal maturation of cartilage in the phalanges. BMP6 expression (arrows) in sections of (A) wild-type and (B) null mutant limbs. Ihh expression in (C) wild-type and (D) mutant limbs. Ihh marks maturing cartilage in the zone between prospective joints (14). (E) Hoxd13 expression at 12.5 dpc in wild-type (+/+) and mutant (–/–) embryos.

Other markers of cartilage maturation, such as Indian hedgehog (Ihh; Fig. 3, C and D) (15) and parathyroid hormone-related protein (PTHRP) (12, 15), are also expressed in Noggin mutants, albeit in an expanded cartilage field unbroken by normal discontinuities at the joints. Thus, ossification proceeds on schedule (Fig. 1). A similar extended expression of Ihh occurs in chick embryos with elevated BMP concentrations (5). There, the failure of digital joint formation may also produce a single extended cartilage element, which matures synchronously.

Because failure of joint formation in the Noggin null mutant is correlated with a failure to up-regulate GDF-5 expression in the presumptive joint regions, the absence of joints in the mature limb cannot be a consequence of the secondary fusion of condensations, but must result from the failure to specify the joint. Only a subset of joints is affected in the brachypodism mouse, which has a mutation in the gene encoding GDF-5 (1, 6), suggesting that additional factors are involved in joint initiation.

A possible cause of the failure to specify joints in the mutant limb might be the loss of positional information that would normally specify where GDF-5 expression should initiate. To date there is no information on the upstream regulation of GDF-5, but changes in joints and skeletal elements also result from mutations in genes encoding Hox proteins (16). Elevated BMP activity in the Noggin null mutant could cause changes in Hox expression patterns (17, 18). However, at the time when joints in the phalanges are being specified, Hoxd13 and Hoxd11 expression is unaffected in the Noggin mutant, indicating that the positional information is unaltered (Fig. 3E) (9,12).

The limb phenotype of the Noggin null mutant is similar to the effects resulting from overexpression of BMPs in the developing chick limb (4, 5). BMP4 and BMP7 are expressed in the perichondrium (4, 5), which is a source of stem cells for appositional growth as well as a barrier to cartilage expansion (19). Application of excessive BMP2 or BMP7 to the developing chick limb bud before mesenchyme condensation induced apoptosis (5) and a subsequent loss of skeletal elements. Differentiating chondrogenic cells were unaffected. If the BMPs were applied later, the skeleton resembled the abnormal skeleton of the mouse Noggin null mutant. Thus, we conclude that Noggin acts to regulate BMP activity after chondrogenesis has initiated. The increased BMP activity increases the recruitment of cells into cartilage, expanding the cartilage at the expense of other tissues and causing larger growth plates (9).

Joint identity may require a particular balance of BMP activities to make chondrocytes competent for joint formation. Currently the best candidates for regulators of expression of GDF-5 and other BMPs are members of the HoxA and HoxD complexes (2, 20). Mutations in the genes encoding these proteins cause limb defects, and the phenotypes, which can include growth defects and an absence of particular elements, have been interpreted to result from changes in the rates of cell proliferation or condensation or both (17). In the absence of Noggin, and the consequent absence of regulated BMP activity, we propose that chondrocytes are unable to respond to instructions from the Hox complex, leading to a failure to turn on genes like the one encoding GDF-5. Thus, it is apparent that a balanced combination of positive and negative factors is required to define and shape each skeletal element.

  • * To whom correspondence should be addressed. E-mail: harland{at}


Stay Connected to Science

Navigate This Article