EXTra hit for mouse osteochondroma

Osteochondromas are the most common benign bone tumors that can occur either singularly or in hereditary multiple osteochondromas syndrome (MO; previously known as hereditary multiple exostoses). Osteochondromas histologically resemble the growth plate and are located preferentially at the long bones. MO is an autosomal dominant disorder caused by mutations in either EXT1 at 8q24 (65%) or EXT2 at 11p11-13 (35%). After the cloning of these genes in 1995 and 1996, respectively, it has taken some time for their function to be known. The breakthrough came unexpectedly in 1998 from a totally different field. By searching for genes involved in the biosynthetic pathway of glycosaminoglycans, it was discovered that EXT1 was a glycosyltransferase involved in heparan sulfate (HS) biosynthesis (1). The challenge since that time was to resolve the exact mechanism by which dysfunctional EXT and defective HS result in osteochondroma formation. The conditional mouse model for Ext1 presented by Jones et al. (2) in PNAS puts some of the controversies to rest. The authors provide compelling evidence that loss of heterozygosity (LOH) is required for mouse osteochondroma formation and suggest that a proliferating chondrocyte is its cell of origin. In addition, the authors show a heterogeneous, mosaic composition of the cartilaginous cap.

Haploinsufficiency Versus LOH

A longstanding controversy in the literature is whether osteochondromas in patients with MO develop through haploinsufficiency or whether they need two hits. Most germ-line mutations found in EXT are nonsense, frame-shift or splice-site mutations leading to a premature stop codon inactivating the gene. A classic tumor-suppressor gene function was, therefore, assumed. According to Knudson’s two-hit model (3) and analogous to other tumor predisposing conditions, both alleles of EXT would need to be inactivated for osteochondroma formation. LOH of EXT1 and/or EXT2 was shown in solitary and hereditary osteochondromas supporting the two-hit model (4, 5). The absence of LOH in a proportion of osteochondromas, however, resulted in a longstanding discussion in the literature as to whether osteochondromas could arise through haploinsufficiency (i.e., if loss of one copy resulting in 50% activity is sufficient to cause osteochondromas) (6).

Previous traditional knockout mouse models for Ext1 and Ext2 were unable to resolve the controversy. Homozygous Ext1 and Ext2 knockout mice were not viable. Mice carrying a heterozygous (Ext1^+/−) or a hypomorphic (Ext1^gt/gt; activity reduced to ∼20%) Ext1 mutation did not show osteochondromas, although bones were shortened and growth plates were aberrant (7). Heterozygous inactivation of Ext2 showed ectopic bone growths at the ribs of the mice but not at the long bones (8). Jones et al. (2) used a doxycyclin-inducible cre-Col II regulated knockdown of Ext1. This way, the Ext1 gene can be completely inactivated, specifically in cartilaginous, collagen II–expressing cells on administration of doxycyclin. This is an elegant way of studying somatic LOH. Doing so, Jones et al. (2) showed that clonal inactivation of Ext1 indeed generates osteochondromas of the long bones and that their mouse phenotype strongly resembles human MO.

The homozygous loss of Ext required for osteochondromagenesis suggests that EXT acts according to the classical two-hit model for tumor-suppressor genes (Fig. 1). Inactivation of a tumor-suppressor gene is generally believed to give cells a growth advantage, contributing to tumor formation. However, in our own experience as well as in those of others, EXT^−/− cells are hard to grow in culture, and knockdown of EXT1 in multiple myeloma leads to reduced growth and increased apoptosis (9), which conflicts with a tumor-suppressor function for EXT, which is puzzling. One could speculate that EXT mutant cells require HS from neighboring chondrocytes to be capable of tumor formation.

Fig. 1.

Model for osteochondroma formation. A second hit in EXT in a proliferating cell of the growth plate near the bony collar results in osteochondroma formation in MO patients. The cells probably lose their polarity and grow out of the bone through a defective bony collar as a result of disturbed hedgehog signaling. In the cartilaginous cap, EXT^−/− cells are intermingled with EXT^+/− cells (mosaicism). In solitary osteochondromas, two somatic hits are required in a single chondrocyte.

On the Cell of Origin

Candidates for the cell of origin of osteochondroma include growth-plate chondrocytes, perichondrial cells, and cells of the Groove of Ranvier. The latter is a fibrochondrosseous structure encircling the growth plate containing chondro- and osteoprogenitor cells. Using the osterix instead of the collagen II promoter, Jones et al. (2) show absence of an MO phenotype when the Ext1 clonal inactivation is induced in osteoblasts. Their results suggest a growth-plate chondrocyte to be the cell of origin; this would seem plausible, because there is a high turnover in the growth plate, rendering the cells prone to acquire a secondary inactivating mutation. However, their results do not completely rule out the chondroprogenitor cells in the Groove of Ranvier because they can also express collagen II.

Mosaicism in the Cartilaginous Cap

So why is the second mutational event in EXT not always detected in human osteochondromas? Jones et al. (2) provide a plausible explanation for this. Using HS immunohistochemistry combined with laser-capture microdissection to isolate DNA from single clones of cells, they elegantly show that, in a proportion of osteochondromas, wild-type cells with functional Ext1 are being integrated in the mutated cartilaginous tumor tissue. If these results can be verified in human osteochondromas, this would indicate that the cartilaginous cap of osteochondroma is mosaic. The detection of a second mutational event would then strongly depend on the balance between EXT mutated and wild-type cells. A mosaic composition of the cartilaginous cap would also raise a novel conceptual angle toward osteochondroma. Osteochondromas were initially considered a disturbance in the orientation of normal bone growth. Later on, they were regarded as a true neoplasm based on genetic aberrations that could be detected in the cartilaginous cap. Monoclonality is generally considered a feature of neoplasia, whereas reactive and hyperplastic lesions are mostly polyclonal. Mosaicism in osteochondroma would question whether osteochondroma should be considered a true neoplasm.

If EXT^−/− cells need the surrounding normal cells to survive, this would be similar to fibrous dysplasia of bone, a benign bone lesion caused by somatic mutations in the GNAS gene. Fibrous dysplasia was shown to be mosaic; mutant clones can only survive and form fibrous dysplasia-like lesions in the mouse when combined with normal cells, otherwise the mutant clones are lost (10).

Future Perspectives

Future studies are required to determine how the complete inactivation of EXT and the resulting absence of HS in a

Loss of heterozygosity (LOH) is required for mouse osteochondroma formation.

chondrocyte can cause osteochondroma formation. Jones et al. (2) propose a model in which the loss of HS gives the cell a proliferative advantage and the adjacent perichondrium fails to form a bony collar. In chondrocytes, defective HS, as a result of EXT inactivation, mainly affects the range of hedgehog signaling (7). Because hedgehog is important for the formation of the bony collar, disturbed signaling may cause a defect in the bony collar. In addition, based on zebrafish models, it was suggested that EXT^−/− cells lose their ability to respond to polarity signals (11). This may cause EXT^−/− cells to grow out of the bone and recruit normal cells to form an osteochondroma (Fig. 1).

The main complication of osteochondroma is malignant transformation to secondary peripheral chondrosarcoma, which is estimated to occur in up to 5% of osteochondromas. It is generally assumed that additional somatic genetic events are required for malignant transformation. The mouse model developed by Jones et al. (2) offers exciting possibilities to investigate additional genetic events generating secondary chondrosarcoma. In addition, it offers opportunities to develop therapeutic as well as preventive pharmacological strategies to improve outcome for patients with MO.