Poole: When you have premature expression of the hypertrophic phenotype, do you see any evidence of angiogenesis?
Kronenberg: In high level chimeras, we can see vascular invasion. When the chimerism is low level, there is still hypertrophy, but there is no mineralization or vascular invasion. It seems that a critical mass of chondrocytes is needed to get mineralization, and this is the only setting in which we see the vascular invasion. There must be some signal build-up or removal of inhibitors, or something like that. The differentiation state is cell autonomous, but secondary phenomena require groups of cells.
Poole: That's interesting, because in development one tends to see hypertrophy in the forming epiphyses without mineralization. With hypertrophy, vasculariza-tion is then seen.
Kronenberg: Along those lines, what this would mean is that one needs enough cells together to create an environment or a strong enough signal or whatever.
Kingsley: One of the things that surprised me was that the chimeras that were made with just the receptor mutant cells show this ectopic bone collar, but they didn't show the internal mineralization of the hypertrophic cartilage. In contrast, when you made the chimeras with the double receptor Ihh knockout mutant cells, you didn't get the bony collar, but you seemed to get a much more prominent internal mineralization.
Kronenberg: The mineralization of 'ectopic' hypertrophic chondrocytes is variable. It seems to correlate most with the extent of chimerism. It is difficult to get high-level chimerism in the growth plate with PTH/PTHrP receptor knockout cells, because the knockout of the PTH/PTHrP receptor limits the proliferation of those cells. The normal cells are not limited at all. In fact, they extend, so as you go down the growth plate what you see, even in high-level chimeras, is that as soon as you get past the knockout hypertrophic cells almost all the cells are normal chondrocytes. You are thus selecting against the mutant cells. But if you make chimeras that have high enough chimerism, then you do see normal mineralization, but not with isolated hypertrophic cells even though they make type X collagen perfectly normally in a way that doesn't require any critical mass of cells. But they won't mineralize the matrix unless there is a critical mass.
Blair: Does that suggest a relationship to the vitamin D response?
Kronenberg: The whole vitamin D story is very mysterious. Two different groups have knocked out the vitamin D receptor. Marie Demay in the Massachusetts General Hospital Endocrine Unit has shown that these mice develop rickets and mineralization abnormalities, but the growth plate is 100% completely normal if you give the mice enough calcium and phosphate, even though they don't have any possible receptor-mediated vitamin D action. Thus, while there may well still be some abnormalities left to be seen in the growth plate of the vitamin D receptor knockout mice, in terms of in situ gene expression and histology, even without vitamin D receptor the growth plate functions completely normally if there is enough calcium and phosphate in the diet.
Beresford: Are you talking about the classical vitamin D receptor or the recently described membrane-bound form (Nemere et al 1998)?
Kronenberg: The classical one. Obviously, this is a wonderful model for studying the non-genomic actions of vitamin D.
Russell: Many years ago, we looked at bisphosphonates, such as etidronate, which inhibits mineralization, and found remarkable changes in the growth plate (Schenk et al 1973). Of course, by blocking mineralization you get an accumulation of the hypertrophic cells so that the hypertrophic zone becomes extraordinarily enlarged, and mineralization seems to be the trigger to vascular invasion.
Burger: I would like to raise this concept of cross-talk between these different molecules. This implies that there is transportation of the molecules preferentially through the cartilage matrix, otherwise you would get diffusion of these molecules, that are formed for instance in the interzone area, elsewhere. Do you have any idea how this works?
Kronenberg: I'm not alone in not understanding this! The way that the hedgehog family moves from cell to cell is really quite confusing and is not known. There is a fruit fly mutant toutvelu which has its mammalian equivalent in the exostosis genes that are involved in heparan sulfate synthesis. These genes are essential for hedgehog to be able to diffuse at all. Exactly how hedgehog gets through matrix instead of getting just handed from one cell to another, nobody knows. Hedgehog diffusion is a tightly regulated process. Cholesterol certainly has a lot to do with it. Chuang & McMahon (1999) have shown that there is a membrane protein (hedgehog interacting protein) that binds hedgehog and limits its diffusion. The receptor for hedgehog itself is present in sufficient amounts, both in fruitfly and in vertebrates, to limit how far hedgehog can travel. The diffusion ofligands has to be thought of as a biologically regulated event. It is not just a physicochemical phenomenon, certainly not with hedgehog, nor with the majority of paracrine factors. I have come into this area from the perspective of an endocrinologist, where hormones can diffuse all over the place, and then they diffuse away — their main regulatory control after their secretion is their rapid diffusion, metabolism and destruction. In paracrine biology, things are quite different: moving three cells instead of five cells changes everything, and there are 10 other levels of regulation of simple diffusion.
Burger: The matrix also changes in the area of chondrocyte hypertrophy.
Kronenberg: So what this is showing is that the hedgehog is moving sideways — patched expression shows that — and is acting on these perichondrial cells, as well as in some presumed cascade eventually leading to increased PTHrP production. The problem is not that different from the mysteries of how Sonic hedgehog gets to the fibroblast growth factors (FGFs) in the limb bud, and we know that some of that is through the inhibition of bone morphogenetic protein (BMP) inhibitors locally (Zufiiga et al 1999). In analogous fashion sonic hedgehog in chicken affects BMP action the lateral plate mesoderm, hundreds of cell layers away, by stimulating the production of the BMP antagonist, caronte, which then diffuses (Yokouchi et al 1999). It seems likely that IHH isn't moving very far here either, and yet it must be at the start of some cascade, perhaps acting through BMP or transforming growth factor (TGF)^.
Tickle: We've been working on an interesting chicken mutant called Talpid3, which has a defect in responding to hedgehog signalling, in that the high level Patched expression is not induced in response to hedgehog signals in the limb bud. This has led us to various hypotheses. The mutant has limb polydactyly, but in the context of your work it's also interesting that it lacks bone. We are currently looking at the idea that this failure of high-level Patched induction is downstream of all hedgehog signalling, and so this could fit quite nicely with the idea that IHH signals bone formation.
Kronenberg: Is it missing all bone?
Hall: I thought that the membrane bones were present, but not the endochondral bone.
Kronenberg: The Ihh knockout has vertebrae, skull and scapula, but it doesn't have limb bones.
Karsenty: Vertebrae are endonchondral.
Hall: There's also the interesting mouse mutant brachypod, where the fibula delays hypertrophy for several weeks: it doesn't hypertrophy until well after birth, and then osteogenesis begins when hypertrophy is switched on. This would again be an interesting model in which to look at Ihh and PTHrP expression patterns. I think brachypod is BMP based.
Kingsley: Yes, it is a mutation in a member of the BMP family called Gdf5.
I have a question about the expression of PTHrP normally, which is periarticular. Of course, articular cartilage never hypertrophies. I was struck by the chimeras that you showed, that it looked like mutant cells that don't have the PTH receptor are excluded from the blue surface of the bone. Again, you can only show some of the pictures, but in the ones that you did show the chimeras that were made with wild-type had a lot of blue contribution to the very surface of the skeletal element, and in the ones that you showed right next to it that were the PTHrP mutant cells, all the blue that was along the surface in the wild-type chimeras was missing. Does this suggest that PTHrP and this receptor signalling pathway plays an important role in maintaining the undifferentiated state of the articular surface on the top of the bone?
Kronenberg: Either that is not so, or we may have missed something very interesting, because I haven't noticed this. But I think what you have just suggested is generally not the case. That is, what seems quite striking to us is that in the first few cell layers closest to the articular surface, the receptor knockout cells and normal cells behave pretty much identically. We like that idea, because this is a region that doesn't have PTH/PTHrP receptors. We, therefore, wouldn't have expected normal and mutant cells to behave differently there. This is a layer that seems to be expressing FGFR1 and as soon as you get past that, you start making PTH/PTHrP receptors, and that's where the columns stop in the mutants.
Kingsley: What is your current thinking on why direct effects of IHH on chondrocytes were missed earlier? In the first papers, the Patched receptor was not detected in chondrocytes and all the effects were postulated to go through the perichondrium and then to PTHrP. There are now expression reasons to think that the receptors are there in chondrocytes, and there's also phenotypic evidence for direct effects, and yet when the first experiments were done adding IHH to the bone cultures of the PTHrP in knockout mice, it appeared that eliminating PTHrP eliminated all responsiveness to IHH?
Kronenberg: We didn't look at BrdU incorporation in the bone explants. What we know for sure is that the growth plates didn't grow: there were not more chondrocytes. I think the difference between the in vivo data and the bone explant data is that in one experiment you're going from normal levels in IHH to none (Ihh knockout vs. wild-type mice), and in the other you're going from normal levels of IHH to even higher levels (bone explants). I've been assuming that going from normal IHH to even more doesn't do much to proliferation. There is a dose— response curve, and the normal level of IHH must be near the top of the proliferative response. Therefore the phenotype is dominated by the effect of the increase in PTHrP production. The two over-expression paradigms taught the same lesson: the effect in the chicken wing and of adding hedgehog to bone explants both led to the dramatic suppression of hypertrophy, without a dramatic proliferative effect.
Ornit^: One other issue we have to keep in mind, is that either IHH or PTHrP could effect the expression of an unknown FGF ligand, which must be present in the growth plate.
Kronenberg: That is certainly possible. We're studying these two signalling systems — PTHrP and IHH — not because we think they are the only two important ligands controlling how long the growth plate is but because our own interests led us in this direction. Other signalling systems are likely to be very important as well.
Chen: The other reason could be that IHH is necessary but not sufficient to stimulate chondrocyte proliferation. Therefore, if you are just adding IHH it doesn't stimulate proliferation. But if you knock it out proliferation is affected.
Morriss-Kay: I'd like to ask David Ornitz to expand on what he just said, about the possible relationship between FGFR3 function in chondrocyte development and the PTHrP|IHH system that Henry Kronenberg described. These two signalling systems appear to be studied rather separately, as far as I can see from the literature, yet clearly they're both acting on the same developmental process.
OrnitIn the data that I showed yesterday, activating FGFR3 mutations suppresses IHH signalling. That's one way you can tie into this system.
Morriss-Kay: Can you describe what you think is happening in normal development in terms of the relationship between these two systems?
OrnitIn normal development, FGFR3 is probably upstream of this feedback loop. I think this feedback loop is likely to be a key regulator of the length of the growth plates. There may be some direct effects of FGFR3 on proliferating chondrocytes, but just as easily it could be indirect effects by modulating the activity of the hedgehog—PTHrP pathway. There is some in vitro evidence that FGFR3 can have a direct effect, because when the activated receptor is expressed in chondrocyte cell lines in culture it can slow down the growth of those cells.
Mundlos: We have talked about proliferation, but isn't apoptosis another factor that regulates the length of the growth plate?
Kronenberg: Yes, it certainly does. We know that the hypertrophic cells are vividly TUNEL positive and have turned off Bcl2 expression. If you express more PTHrP and delay that part of the differentiation programme, you're also delaying apoptosis. In experiments in which Schipani and Juppner expressed a constitutively active PTH|PTHrP receptor in a transgenic mouse driven by the collagen II promoter, it is possible to have PTHrP-like action in proliferative chondrocytes and in hypertrophic chondrocytes. In that setting, the hypertrophic chondrocytes don't die and they fill up the bone marrow space. Apoptosis is part of the chondrocyte differentiation programme, and if you slow down the programme, you stop apoptosis.
Burger: Do you get interference with the excavation of the primitive marrow cavity, and influx of osteoclasts in that situation?
Kronenberg: We wanted that to be the case in the PTHrP and PTH/PTHrP receptor knockout mice because PTHrP is an important stimulator of osteoclastogenesis. When we've looked in our mutants, however, we haven't been able to see any evidence of decreased osteoclast development or action. It seems that at least during this modelling stage the osteoclasts needed for bone development are not importantly regulated by PTHrP.
M. Cohn: Goff and Tabin showed a few years ago that retroviral overexpression of Hoxdl 1 and Hoxd13 in hindlimb buds had an effect on leg bone growth. They found that bone length was altered because of an effect on cells in the growth plate rather than in the early limb bud. How do Hox genes interface with the IHH|PTHrP loop?
Kronenberg: I don't know this paper, but it sounds like something that we could study pretty easily.
Was this article helpful?