Regulating the Regulators of Chondrocyte Hypertrophy

The development, growth, and maintenance of a functional skeleton relies on the existence of several different types of chondrocytes with markedly different fates. The articular chondrocytes on the surface of joints are expected to persist throughout life, producing extracellular matrix components which provide a cushion to distribute load on bone at joints. In contrast, it is the changing dynamics of the underlying cartilage at the epiphyseal growth plate that permits a very different function: growth and development. Chondrocytes of this region undergo a series of changes including hypertrophy, release of extracellular matrix vesicles, and alterations in production of extracellular matrix proteins, prior to mineralization and replacement by bone. As chondrocytes progress toward terminal differentiation, morphological, and biosynthetic changes occur. Proliferating chondrocytes exit the cell cycle and become flattened before fully maturing into round, hypertrophic chondrocytes which secrete and organize a different extracellular matrix, characterized by high levels of alkaline phosphatase (ALP), diminished levels of collagens type II and IX, and production of a new hypertrophic chondrocyte-specific product, type X collagen.1-3 In addition to normal bone formation during development of the skeleton, this process of endochondral ossification has also been implicated in fracture repair and such pathologic processes as osteoarthritis and ectopic bone formation. The fundamental importance of chondrocyte maturation in the growth, development, and repair of the skeleton has led to intense investigation of the growth factors that regulate terminal differentiation of chondrocytes. Although several growth factors have been indentified which regulate chondrocyte maturation, it remains unclear how these factors interact to coordinate the process leading to bone elongation and endochondral ossification. In this issue, Grismud and colleagues propose a role for bone morphogenetic protein-6 (BMP-6) as a positive regulator of chondrocyte maturation. Their report also provides insight into how parathyroid hormone-related protein (PTHrP) and indian hedgehog, other growth factors known to modulate chondrocyte hypertrophy, may interact with BMP-6 in the growth plate. BMPs belong to the transforming growth factor-β (TGF-β) superfamily of secreted growth and differentiation polypeptides, which also includes activins, growth and differentiation factors, and Mullerian inhibiting substance.4, 5 For over 30 years, it has been known that BMPs are capable of inducing the formation of new cartilage and bone when implanted extraskeletally.6 However, it was not until 1988 that three of the proteins involved (BMP2–4) were cloned.7 Although their name stresses their critical impact on the skeletal system, more than 20 members of the BMP family have been identified and determined to have a broader biological impact than just skeletogenesis. This impact includes establishment of basic embryonic patterning and growth and differentiation of nearly all body systems including the cardiovascular, gastrointestinal, respiratory, nervous, and urogenital systems and the integument (see review by Hogan).7 As their name implies, most BMPs can initiate new cartilage and bone formation in a cascade of events reminiscent of endochondral bone formation. BMP-2, BMP-4, BMP-5, and BMP-7 have all been shown to possess the ability to promote bone formation in vivo.4, 8-10 Additional evidence with cultured cells also point to considerable overlap in BMP function; recombinant BMP-2, BMP-4, BMP-6, and BMP-7 have all been shown to induce both hypertrophy in cultured chondrocytes and osteogenesis from mesenchymal stem cells.5, 11 However, information derived from BMP localization studies and null mutations in mice suggests more specific roles for individual BMPs in normal development. The temporal and spatial patterns of expression of BMPs in embryonic mouse forelimbs support roles for BMP-2, BMP-4, BMP-6, and BMP-7 in skeletal formation, and multiple BMPs, including BMP-2, BMP-4, BMP-6, and BMP-7, are produced by chondrocytes undergoing endochondral ossification.12-16 Recent detailed analysis of expression patterns for BMPs in the embryonic mouse humerus by Solloway et al.16 show that BMP-7 is expressed in both proliferating chondrocytes and the perichondrum, BMP-4 is found in chondrocytes of the transition zone immediately adjacent to the mature hypertrophic chondrocytes as well as in the perichondrium, while BMP-2 and BMP-6 are coexpressed by hypertrophic chondrocytes. The coexpression of BMP-2 and BMP-6 is consistent with the hypothesis that the physiologically most active forms may be heterodimers of BMP-2/6 and BMP-2/7.17 Mice bred with homozygous null mutations in BMP-2 and BMP-4 die early in embryogenesis, before the process of skeletogenesis occurs.18, 19 While emphasizing the importance of some BMPs in early embryonic patterning and morphogenesis, the lethality means that these mutants cannot contribute information about the role of BMP-2 and BMP- 4 in bone and cartilage formation. Other BMP knockouts have been more enlightening. Loss of function mutation of BMP-5 results in short ears, defects in the sternum and ribs, with relatively minor alterations in the appendicular skeleton,20, 21 whereas mutation of the BMP family member GDF-5 results in shortened appendicular skeleton and relatively unaffected axial skeleton.22 BMP-7 null mice have severe kidney and eye defects but relatively minor skeletal pattern abnormalities of the rib cage, skull, and hindlimbs.23, 24 Mice carrying a null mutation for BMP-6 are viable and fertile with no overt defects in tissues known to express BMP-6 mRNA.16 However, careful examination of skeletogenesis in late gestation embryos showed a delay in ossification of the sternum. The defect in this specific skeletal element was attributed to early expression of only BMP-6 within the sternum; BMP-2 was found in overlapping expression patterns with BMP-6 in other skeletal elements and might functionally compensate for the loss of BMP-6 in the rest of the skeleton. In contrast to the positive role BMPs have in promoting chondrocyte hypertrophy, PTHrP acts to slow the rate of chondrocyte maturation and maintain chondrocytes in a proliferative state. A role for PTHrP in modulating proliferation of growth plate chondrocytes was suggested by in vitro studies in 1993.25 Subsequently, analysis of PTHrP homozygous knockout mice showed that absence of PTHrP during embryonic development causes chondrodysplasias, with the mice dying shortly after birth with abnormalities in endochondral bone formation.26, 27 The growth plates of these mice were characterized by a disorganized columnar architecture and a shortened proliferative/resting zone, indicating that chondrocytes prematurely exit the proliferative zone in order to hypertrophy and subsequently become calcified. An opposite effect is seen with PTHrP overexpression. Transgenic mice in which PTHrP overexpression is targeted to chondrocytes show delayed bone development with an accumulation of prehypertrophic chondrocytes,28 and PTHrP treatment of murine limb explants blocks terminal differentiation of growth plate chondrocytes.29 While both PTHrP and fibroblast growth factor-2 (FGF-2) are capable of reducing type X collagen mRNA levels and ALP activity in cultured avian cephalic sternal chondrocytes, only PTHrP could decrease these parameters of maturation in cells which had undergone further maturation.30 PTHrP localization studies also point to a role in maintaining a cartilage phenotype during development. PTHrP is present in high concentrations in fetal and postnatal epiphyseal cartilage, and protein expression in the growth plate is concentrated within a thin layer of chondrocytes immediately adjacent to the hypertrophic zone.31 In both the epiphyseal and growth plate regions, PTHrP expression is diminished in the mature animal.32 All of these data point to PTHrP as a major factor which diminishes the rate of chondrocyte hypertrophy and subsequent endochondral bone formation. Mice that lack the PTH/PTHrP receptor show a phenotype similar to that seen with PTHrP knockouts, but the effects are more severe and lead to early embryonic death. The opposite effect on bone formation is seen in patients with Jansen's metaphyseal chondroplasia, a naturally occurring, activating mutant of the PTHrP receptor.33 These individuals show a phenotype consistent with delayed endochondral ossification. In 1996, Vortkamp et al.29 proposed that PTHrP was a downstream mediator of another negative regulator of chondrocyte hypertrophy, Indian hedgehog (Ihh). The hedgehog proteins are a relatively small group of proteins identified as signaling molecules involved in early embryonic patterning, with sonic hedgehog (Shh) serving as a key regulator of anterior–posterior patterning during limb development. The related Ihh is expressed in cartilage, and in the growth plate it colocalizes with BMP-6 at the region of postproliferative chondrocytes adjacent to hypertrophic cells.29, 34 As seen with overexpression of PTHrP, overexpression of Ihh or Shh delays chondrocyte hypertrophy, and misexpression of Ihh within the chick wingbud has been shown to disrupt endochondral ossification.29 Vortkamp et al. therefore proposed a model in which PTHrP reciprocally inhibits Ihh expression, thus forming a negative feedback loop to moderate the rate of chondrocyte hypertrophy within the growth plate. They also demonstrated that overexpression of Ihh in developing chick wings blocked BMP-6 expression, which argues that Ihh acts upstream of both PTHrP and BMP. The ability of Ihh to induce both PTHrP and BMPs would allow for a tightly controlled progression of maturation within the growth plate. In this issue of JBMR, Grismud and colleagues explore the relationship of these three growth factors during maturation of chicken sternal chondrocytes in culture. They confirm that PTHrP is able to inhibit parameters of maturation, including type X collagen mRNA induction, ALP activity, increased chondrocyte size, and decreased proliferation. In addition, they demonstrate that PTHrP also abolishes expression of BMP-6 mRNA. The inhibitory effects on chondrocyte maturation could be rapidly reversed by the removal of PTHrP from the cultures and furthermore, exogenous BMP-6 could circumvent the inhibitory effects of PTHrP on maturation. Although many BMPs, including BMP-6, are able to autoregulate their mRNA expression, BMP-6 was unable to induce its own expression in the presence of PTHrP despite its ability to induce other parameters of maturation. These observations form the basis for their conclusion that PTHrP suppression of chondrocyte hypertrophy occurs via inhibition of BMP-6 levels, and that hypertrophy can proceed despite PTHrP-induced suppression of endogenous BMP-6 if the cells are provided with exogenous BMP. In addition, the authors provide evidence that ascorbate, which has long been known to accelerate maturation of chondrocytes,35-37 increases BMP-6 expression. Where does Ihh fit into this picture? The authors provide evidence that addition of BMP induces high-level Ihh expression, even in the presence of PTHrP. They therefore conclude that PTHrP inhibition of Ihh in the feedback loop of these two inhibitors of maturation is indirect and that BMP-6 is a positive mediator of Ihh expression in hypertrophic chondrocytes. The ability of BMP-6 to induce Ihh may be direct or indirect by increasing the maturation state of the cell and in turn increasing Ihh. Kronenberg et al.38 have proposed a local negative feedback loop in which postproliferative chondrocytes secrete Ihh, which increases secretion of PTHrP to slow hypertrophy and thereby slow production of cells which secrete Ihh. Under this scenario as modified by Grimsrud et al.,39 BMP-6 would be added to this feedback system: PTHrP suppresses BMP-6 and hypertrophy, low BMP suppresses Ihh expression, and low Ihh leads to decreased PTHrP, which permits BMP-induced hypertrophy until Ihh levels rise. The report by Grimsrud et al.39 does not address the question of where the PTHrP is produced. In situ hybridization studies by Lee et al.31 suggested that it was synthesized by perichondrial cells close to the articular surfaces, which would argue that Ihh from growth plate cells must travel a relatively long distance to activate PTHrP expression. Chung et al.40 have recently presented in situ hybridization data from mouse chimeric growth plates in which the proliferative zone of growth plates contained both normal proliferating chondrocytes and ectopic hypertrophic chondrocytes lacking PTH/PTHrP receptor. In nonchimeric growth plates, the Patched receptor for Ihh was present in proliferating chondrocytes, and this receptor was up-regulated in the proliferating chondrocytes of the chimeras. This would imply that Ihh can signal to nearby proliferative cells and might not need to make its way to the articular perichondrium. In support of this hypothesis, the chimeras showed increased PTHrP mRNA in both the perichondrium and proliferating chondrocytes. In addition, unpublished data from our laboratory indicate that proliferating chick sternal chondrocytes in culture produce PTHrP prior to hypertrophy. This new information suggests that the Ihh–PTHrP feedback loop to delay hypertrophy involves two adjacent populations of cells, the Ihh-producing prehypertrophic chondrocytes and the PTHrP-producing proliferative cells. It also highlights one of the complications arising from in situ hybridization studies: the site of maximal expression may not necessarily be the most physiologically relevant region. While the paper by Grimsrud et al.39 implicates BMP-6 both as an autocrine inducer of hypertrophy and a modulator of the rate of hypertrophy, there are many unanswered questions concerning the mechanism of its action. BMPs, like other members of the TGF-β superfamily, have been shown to transduce signals through transmembrane heteromeric receptor complexes consisting of type I and type II serine/threonine kinase receptors.41 In vertebrates, two type I BMP receptors (BMPR-IA and BMPR-IB) interact with one type II receptor (BMPR-II) to mediate signaling for BMP-2, BMP-4, and BMP-7. These receptors have differing affinities for various members of the BMP family, which may further amplify the diversity of biological effects possible with ligand and receptor combinations. BMP-7 binds to BMPR-IB with higher affinity than BMPR-IA and can also utilize the type I receptor ALK2, which was previously described as an activin receptor.42, 43 As Grimsrud et al.39 point out, the receptor activated by BMP-6 is unknown. Recent efforts to identify specific roles for different BMP receptors have relied heavily on overexpression of dominant negative and constitutively active mutant receptors. Maturation of chick sternal chondrocytes is prevented by expression of dominant negative BMP type II receptor and inhibited to a lesser extent by dominant negative BMPR-IB,44 which argues that BMP signaling is required for hypertrophy. However, since overexpression of any type I receptor might lead to absence of available type II receptor, this does not necessarily implicate the IB receptor in BMP-6 action. Overexpression of constitutively active BMP receptors in developing chick embryos indicates that the activated IA receptor both downregulates Ihh and induces expression of PTHrP. This does not fit the pattern seen by Grimsrud et al.39 in BMP-6–treated cultured chondrocytes, which would argue against BMP-6 working via BMPR-IA. Our recent data using constitutively active BMPR-IB45 suggest that it is the IB receptor which is critical for inducing chondrocyte hypertrophy. It is also possible that there are other BMP receptors yet to be discovered. It is also important to note that, at present, we do not know the implications of signaling through type IA versus type IB receptors. There are three Smad proteins, Smad 1, Smad 5, and Smad 8, which have been identified as intracellular signaling molecules activated by BMPs.46 It is possible that different BMP type I receptors have a preference for different Smad proteins, leading to transcriptional activation of different sets of genes. What is clear is that, while we have discovered much about the secreted signaling molecules which activate different types of chondrocytes, we still have much to learn about the mechanisms by which these effects are translated into the chondrocyte changes needed for endochondral bone formation.