BoneKEy-Osteovision | Commentary

Your partner makes all the difference



DOI:10.1138/2001013

It's “all in the family” acquired some new relevance with the recent publication of two papers back-to-back in Nature Medicine. In these two complementary papers, osteosclerotic mice were created when either of two different members of the AP-1 (activator protein-1) sub-family of leucine zipper-containing transcription factors was overexpressed in transgenic mice: Fra-1 (), a Fos-related protein encoded by the c-Fos target gene Fosl1 (referred to as fra-1) (), or deltaFosB (ΔFosB), a naturally occurring splice variant of FosB (). In both cases, the transgenic mice appeared normal at birth, but with time increased bone formation was evident throughout the skeleton and the increase in bone mass was quite remarkable - much greater than what has been seen in other treatments or genetic manipulations that have been found to increase bone.

Were there hints that overexpression of at least certain of the AP-1 family members would have this striking capacity to increase bone formation? Yes and no! It has been well-known for some time that AP-1 family members play a role in regulation of proliferation and differentiation in bone cells, and the promoters of several genes involved in bone formation contain AP-1 consensus sequences, e.g., that for osteocalcin (). Several regulators of bone formation induce AP-1 expression in osteoblast precursors (e.g., ()). However, no bone abnormalities have been reported in fosB knockout mice (), or in mice overexpressing FosB, Fra-2, c-Jun or JunB (). On the other hand, c-fos knockout mice develop osteopetrosis due to a lack of osteoclasts (), a phenotype that is rescued by overexpression of Fra-1 (), but Fra-1-deficient mice have no detectable defects in either osteoclast or osteoblast differentiation in vivo ((); discussion in ()). Notably, however, mice overexpressing c-Fos develop osteosarcomas as well as chondrosarcomas ().

In both of these new transgenic mice, the osteosclerosis results from a marked increase in bone formation without a compensatory increase in bone resorption. Both papers provided strong evidence to support the cellular mechanism by which Fra-1 and ΔFosB increase bone formation. First, it should be said that the transgenic approaches used did not target overexpression solely to osteoblasts, i.e., Fra-1 was overexpressed in various mesenchymal tissues via the major histocompatability complex (MHC) class I antigen H2-kb promoter with 3′ long terminal repeat (LTR) sequences of the FBJ-murine sarcoma virus, while ΔFosB was expressed via a fragment of the neuron-specific enolase (NSE) promoter that also supported overexpression in osteoblasts, chondrocytes, adipocytes, skin and spleen, but not osteoclasts. Nevertheless, in both the Fra-1- and ΔFosB-overexpressing mice, the osteosclerotic phenotype derived from a cell autonomous modulation of osteoblast lineage cells that could be reproduced in other in vivo assays (see, e.g., bone transplant studies in ()) and was manifested in vitro by accelerated and more abundant mineralized bone nodule formation and accompanying increased osteoblast-associated gene expression (e.g., ALP, osteocalcin). In addition, no alteration in osteoblast proliferation was detected in cultured osteoblastic cells derived at least from the Fra-1 overexpressing mice.

Because adipogenesis was decreased concomitant with increased osteogenesis in the ΔFosB mice, and osteoblasts and adipocytes are thought to share a common progenitor and/or plasticity between phenotypes (), the quite compelling suggestion was made that ΔFosB overexpression promotes osteogenesis at the expense of adipogenesis () (Fig. 1). However, the adipocyte phenotype was not seen in the Fra-1-overexpressing mice, suggesting that the increased commitment or differentiation of mesenchymal precursors to osteoblasts is an event independent of the repression of differentiation of adipocytes; caution in interpreting apparently inverse relationships between these two lineages when both bi-/multipotential progenitors may reside in the same populations as restricted monopotential progenitors has been discussed elsewhere (). Nevertheless, it will be of great interest to dissect precisely when during the osteoblast differentiation sequence these AP-1 family members are working (see below). The data to date support the hypothesis that Fra-1 and ΔFosB share some common target genes in osteoblast lineage cells while having different targets and functions in some other lineages, e.g., the adipocyte lineage (Fig. 1). The lack of the adipocyte phenotype in the Fra-1-overexpressing mice, together with the fact that both the Fra-1 and the ΔFosB overexpression phenotypes are cell autonomous and intrinsic to the osteoblast phenotype, also appears to rule out a role for leptin, an adipocyte-derived protein that is a negative regulator of bone formation requiring the hypothalamus/central control (), in the observed osteosclerosis. Similarly, although increased bone formation is seen in mice lacking osteocalcin () and osteocalcin is regulated in part by AP-1(), osteocalcin expression is not decreased in either of these AP-1 transgenic models. The fact that much smaller increases in bone formation were seen in both the leptin signaling-deficient (ob/ob and db/db) and osteocalcin-deficient mice also suggests that changes in regulation of these genes do not account for the osteosclerosis. Although exhaustive analysis has not yet been done, there was also no evidence found for increased expression of growth factors with functional AP-1 binding sites, e.g., IGF-I or TGF-β, or BMP-2 ().

The molecular mechanism is now the main question to be addressed. The transcription factor AP-1 is a dimer composed of members of the Fos, Jun, and ATF family of proteins. It has been hypothesized that the ability of members of this family to regulate many and diverse biological processes is because they can form different functional dimers with distinct properties. Whereas Jun proteins can form homodimers and heterodimers with Fos and ATF proteins, Fos proteins (c-Fos, FosB, Fra-1, Fra-2) form stable dimers with Jun proteins (c-Jun, JunB, JunD) (). ΔFosB is a naturally occurring splice variant of FosB, but it is further truncated to the Δ2ΔFosB isoform in osteoblasts as in some other cell types. Interestingly, it is the Δ2ΔFosB isoform, rather than ΔFosB itself, which appears to be responsible for the increased bone formation seen in the ΔFosB transgenic mice. As raised above, elucidation of developmental time-specific aspects of the mechanism may shed light, since differentiation stage-specific alternative splicing of fosB mRNA and selective initiation site use of ΔFosB appears to be what regulates osteoblast development and the increased bone formation seen in the ΔFosB transgenic mice (). Notably, neither Fra-1 nor ΔFosB (or Δ2ΔFosB) possesses the typical C-terminal transactivation domain (), suggesting that the transcriptional potential of either resides in or depends on their heterodimerizing partners or coactivators. However, no significant differences were found in expression levels of other AP-1 family members in Fra-1-overexpressing osteoblasts and although JunD was speculated to be involved in the ΔFosB transgenic phenotype, the authors of both papers acknowledge that their data do not yet allow selection between one of two different models - a model of activation versus a model of repression.

In an activation model, and although both Fra-1 and ΔFosB/Δ2ΔFosB are expressed in normal developing osteoblasts, heterodimerization need not be with a normal or physiologically-relevant partner but may occur with another nuclear protein required for osteoblast development: the new heterodimer may have higher affinity for DNA and/or show higher transcriptional activity. Notably, the partner is unlikely to be Cbfa1, because the osteosclerotic phenotype was not rescued when the Fra-1 transgenics were prepared on a Cbfa1 -/- background ().

A repression model could involve heterodimerization of overexpressed Fra-1 or ΔFosB with, and titering out of, a transcriptional inhibitor of osteoblast differentiation (see, Fig. 1 in ()). Since neither Fra-1 () nor ΔFosB () are normally required for osteoblast development in vivo, the repression model requires that excess Fra-1 and ΔFosB heterodimerize with a protein with which they don't normally heterodimerize. That expression levels of at least some members of the AP-1 family have perhaps surprisingly exquisite ability to differentially regulate biological processes is seen in experiments done to assess whether Fra-1 can functionally substitute for c-Fos. When Fosl1 was knocked-in to mice with disrupted c-Fos, Fra-1 fully complemented the lack of c-Fos in bone development (i.e., the osteopetrosis was rescued), but the rescue was gene dosage dependent, suggesting that osteoclast differentiation is sensitive to threshold levels of Fra-1 (). In addition, ectopic expression of Fra-1 in Fosl1 knock-in mice did not recapitulate the osteosclerotic phenotype of Fra-1- and ΔFosB-overexpressing transgenic mice, perhaps because of the 5-10-fold higher Fra-1 expression in the MHC class I promoter driven Fra-1 transgenic mice compared to the knock-in mice. Thus, the dimerizing partners and confirmation of an activation versus repression model of the transcriptional regulation of osteoblast development in the osteosclerotic mice are of intense interest.

Transcription factors that regulate bone formation are only beginning to be understood and are growing in number (). The two recent papers have re-emphasized the value of assessing multiple different family members in multi-gene families in both under- and over-expression models. They provide additional examples of the utility of in vivo/mouse studies to augment what we learn from in vitro analyses. For example, in addition to effects on osteoblasts, Fra-1 potentiates osteoclast differentiation in vitro through impinging on the RANK/RANKL pathway (). However, in neither Fra-1- nor ΔFosB-overexpressing mice was an increase seen in osteoclast surfaces or number, or in urinary deoxypyridinoline crosslinks, nor were transgenic osteoblasts in vitro altered in their ability to support osteoclastogenesis or osteoclast activity. This suggests that the transgenic osteoblastic cells are somehow deficient in or show decreased ability to upregulate osteoclast activity in vivo although not in vitro, an observation that underscores yet again the regulatory complexity and fine balance between cells of the two main bone lineages and their activities. It is striking that while neither Fra-1 nor ΔFosB appears normally to control osteoblast development, altering expression levels of either of them, in this case increasing expression, does have profound consequences for the entire postnatal skeleton, i.e., bones formed by intramembranous (e.g., calvaria) or endochondral (e.g., tibia and femur) ossification. The studies further predict that elucidation of the molecular pathways and partners by which Fra-1 and ΔFosB are working in osteoblasts will provide new impetus to the goal of establishing novel anabolic targets for therapeutic approaches to diseases such as osteoporosis.


Creative Commons License This work is licensed under a Creative Commons Attribution-Noncommercial-No Derivative Works 3.0 United States License.