Burger: You have demonstrated nicely how the osteoblast-mediated osteoclast generation is part of the haematopoietic system — it uses the same mechanisms as developing leukocytes. You saw homologous molecules to ODF in other systems, including in dendritic cells. Is it also functioning as a decoy molecule there?
Suda: At first we thought that ODF was present only in bone, but this was not the case, because ODF was found in dendritic cells and T cells as well. ODF therefore may be important not only for osteoclast differentiation, but also for immune responses.
Beresford: In the Amgen experiment that you described, are you talking about a complete rescue of the osteopetrotic phenotype or just the appearance of a few osteoclasts?
Suda: When TNFa is administered into RANK knockout mice, they start to generate osteoclasts. These osteoclasts, however, cannot resorb bone. IL-l is necessary for TNFa-induced osteoclasts to resorb bone.
Beresford: Following on from that, I was interested in your suggestion that there is a TNF/IL-l dependent pathway which becomes very important in osteoclastogenesis in rheumatoid arthritis and related conditions. Have you tested that formally in an animal model? If you take an animal model of chronic rheumatoid arthritis with excessive joint destruction and you infuse OPG into those animals, you do not see any improvement. If I have understood correctly, this argues against ODF having an important role in the maintenance of chronic joint destruction; it's actually a TNF/IL-l-mediated pathway.
Suda: We postulate that the ODF-dependent pathway is important preferentially in physiologic conditions, whereas the ODF-independent pathway is important in pathological conditions. However, I cannot exclude the possibility that the ODF-dependent pathway is important in pathological conditions as well.
Russell: OPG is being developed for clinical use. It seems to work in just about everything that has been tried, including arthritis models and tumour-driven bone destruction.
Beresford: So this TNF/IL-1 'auxiliary pathway' may not be that critical in the real world setting.
Suda: We need further in vivo experiments to prove our hypothesis.
Russell: One of the interesting things about osteoclasts is the duration of their action and their lifespan. It seems increasingly that they have a finite lifespan which can be altered in disease states — they undergo apoptosis. What is known in relation to the RANK system about whether those factors influence lifespan? M-CSF is meant to.
Suda: The lifespan of a mature osteoclast is less than 24 h in the absence of ODF and M-CSF. When ODF or M-CSF is present, osteoclasts can survive for several days. In in vivo conditions many osteoblastic stromal cells are present around osteoclasts. Osteoblastic stromal cells are capable of producing ODF and M-CSF.
Burger: If this is true, how do you explain the tunnelling that osteoclasts can make in secondary osteomes. They can dig a tunnel in compact bone, which is not lined with osteoblasts or lining cells. Where does the ODF come from there?
Suda: In our hands, ODF acts as a membrane-bound factor; we do not have any evidence that ODF acts as a soluble factor. When osteoblastic cells and spleen cells are co-cultured, but separated by a membrane filter, no osteoclasts are formed.
Burger: But how do you explain this in vivo observation that osteoclasts are able to tunnel deep into the compact bone, and once inside they cannot encounter osteoblasts, stromal cells or lining cells?
Blair: There have been studies looking at that. The active site is full of proteinases and an acid pH. When they encounter osteoblasts, the cells are immediately apoptotic (Elmardi et al 1990, Taniwaki & Katchburian 1998). So they are unlikely to have this sort of cell surface interaction. On the other hand, at their basolateral surfaces, the osteoclasts do maintain interaction with stromal cells. The question would be whether there would be enough time in this tunnelling activity for the stromal cells to follow on. Osteoblasts are quite actively proliferative and expand to cover the available surface. I would suspect that osteoblast—osteoclast interactions at their lateral surfaces would be the major mechanism.
Suda: Osteocytes may play a similar role in stimulating osteoclast function.
Beresford: To my knowledge mice don't exhibit cortical tunnelling. Anyway in those species in which the process does occur, there is no evidence of any physical separation between the cutting cone and the cells that follow on in its wake. The idea that you have osteoblasts tunnelling through bone in isolation is actually not true: they are always in close proximity to other cells and/or their processes. I think the potential is there.
Blair: The tunnelling is mainly an artefact of hyperparathyroidism in humans with renal failure.
Burger: It is also seen in animals. The mouse is just too small and too short-lived for remodelling to take place, but it can occur if it is needed.
Blair: In humans or animals that have mature skeleton, there is a complex microarchitecture, particularly at the ends of the long bones, which is very important. If you look at where the bone is resorbed by osteoclasts, and bone is laid down by osteoblasts, this architecture is maintained. It is difficult to see how this kind of general expression of ODF/RANK ligand is managing this higher architectural behaviour.
Burger: The higher-order control can only be explained if mechanical stimuli are considered.
Suda: I would like to propose a new concept. Bone loss by osteoblast-induced osteoclasts can be followed by osteoblastic bone formation. However, TNF-a/IL-l-induced bone loss cannot be followed by osteoblastic bone formation, since ODF is induced on the plasma membrane of osteoblastic stromal cells in response to several bone resorbing factors. TNFa acts directly on osteoclast progenitors. I think that some interaction between Cbfal and RANK ligand/ODF may be involved in this connection.
Kronenberg: That is an interesting model. What it potentially leaves out is the other model whereby osteoclastic release of growth factors in bone matrix by itself might be able to stimulate bone formation. If that is an important stimulus in bone formation, then you would predict that TNF would release those factors during osteoclastic resorption. So if you don't see formation in inflammatory disease, either that release of growth factors is not important after all, or else inflammatory mediators can block the action. So, the relationship between bone formation and resorption may be very complicated.
Karsenty: If this mediator released by the osteoclasts that we are talking about in the in vitro assay is important in controlling bone formation, wouldn't you expect that in animal models where there are no osteoclasts that there will be no bone formation? We should not lose sight that in the absence of osteoclasts bone formation is not affected.
Beresford: The work that you did with the inducible osteoblast ablation model is very nice (Corral et al l998), but it does not preclude the possibility that cells earlier in the lineage are influencing osteoclast differentiation.
Karsenty: I was thinking about the osteoclasts in terms of function, not differentiation.
Beresford: I think you have to distinguish between a skeleton that is actually growing and modelling, and one that's reached a steady-state. The interactions between the two cell populations may be fundamentally different in these two contexts.
Karsenty: Our experiment was not designed to study differentiation, but function. In the absence of bone formation, the function of the osteoclasts was not affected. This is different from differentiation. This work led us to imagine a novel regulation of bone remodelling, which in vivo seems to be more powerful than the alternative mechanism.
"Hall: You mentioned mononuclear osteoclasts just once in your talk. Are they functional? And is this a way of cutting short the time dependence that you need for an interaction of osteoclasts with osteoblasts?
Suda: Tim Chambers has shown that mononuclear osteoclasts can also resorb bone.
Hall: Certainly, if one goes away from mammals to organisms such as fish, mononuclear resorption is the dominant situation. I had the impression that it's very rare in mammals.
Poole: It is very much a feature of rheumatoid arthritis, particularly with respect to regions at the cartilage—bone interface.
Blair: There is a question that has been puzzling me. In the normal lymph node there are T cells that bear lots of RANK ligand/ODF, there is plenty of CSF-1 and there are macrophage precursors. Why is it that there are no osteoclasts?
Suda: That's an important question. At present, we do not have any good answers to explain the exclusive localization of osteoclasts in bone. Russell: Presumably the stem cells aren't there.
Blair: But you can get dandy osteoclasts from blood or spleen macrophages. Burger: My question at the beginning of the discussion was related to that: is the decoy system also working to prevent this going on? Otherwise, you would have many osteoclasts.
Suda: Yes, many drug companies are interested in developing OPG/OCIF as a new drug. They also would like to get a small molecule that has the same inhibitory activity. I understand that Roussel-Uclafis now trying to develop such a molecule.
Was this article helpful?