Remodeling and Vascular Spaces in Bone

Authors


Abstract

In recent years, we have come to appreciate that the close association between bone and vasculature plays a pivotal role in the regulation of bone remodeling and fracture repair. In 2001, Hauge et al. characterized a specialized vascular structure, the bone remodeling compartment (BRC), and showed that the outer lining of this compartment was made up of flattened cells, displaying all the characteristics of lining cells in bone. A decrease in bone turnover leads to a decrease in surfaces covered with remodeling compartments, whereas increased turnover causes an increase. Immunoreactivity for all major osteotropic growth factors and cytokines including osteoprotegerin (OPG) and RANKL has been shown in the cells lining the BRC, which makes the BRC the structure of choice for coupling between resorption and formation. The secretion of these factors inside a confined space separated from the bone marrow would facilitate local regulation of the remodeling process without interference from growth factors secreted by blood cells in the marrow space. The BRC creates an environment where cells inside the structure are exposed to denuded bone, which may enable direct cellular interactions with integrins and other matrix factors known to regulate osteoclast/osteoblast activity. However, the denuded bone surface inside the BRC also constitutes an ideal environment for the seeding of bone metastases, known to have high affinity for bone matrix. Reduction in BRC space brought about by antiresorptive therapies such as bisphosphonates reduce the number of skeletal events in advanced cancer, whereas an increase in BRC space induced by remodeling activators like PTH may increase the bone metastatic burden. The BRC has only been characterized in detail in trabecular bone; there is, however, evidence that a similar structure may exist in cortical bone, but further characterization is needed.

INTRODUCTION

Bone remodeling is based on the concerted action of resorptive and formative cell populations to replace old bone with new bone and thus secure the integrity of the skeleton. This sequence has to be tightly regulated by both local and systemic factors, because significant deviations from a neutral balance between resorption and formation would mean severe accelerated bone loss or bone gains with possible disastrous consequences in terms of increased fracture risk or compression syndromes.

Angiogenesis is closely associated with bone resorption and bone formation.(1) Angiogenic factors such as vascular endothelial growth factor (VEGF) and endothelin are regulators of osteoclast and osteoblast activity,(1) but the formation of blood vessels also serve as a way of transporting circulating osteoblast(2) and osteoclast precursors(3) to sites undergoing active remodeling. Angiogenesis also constitutes a necessary precondition together with mechanical stimulation for bone induction during distraction osteogenesis.(4,5)

Five years ago, Hauge et al.(6) reported that bone remodeling takes place in specialized vascular structures, bone remodeling compartments (BRCs), which were lined by cells exhibiting osteoblastic staining characteristics on the outside (Fig. 1). The inner wall of the compartment was made up of the bone surface with either resorptive or formative cells, depending on the state of the remodeling cycle. Based on results obtained using a combination of immunohistochemical and enzyme staining, the authors concluded that the cells constituting the outer wall of the BRC were lining cells showing positive immunoreactivity for a whole array of osteoblastic markers (alkaline phosphatase, osteocalcin, osteonectin, type I collagen). Later analyses have shown positive staining for a variety of osteotropic growth factors (IGFs, TGFβ, BMPs), as well as the primary regulators of osteoclast differentiation (RANKL and osteoprotegerin [OPG]; Table 1).(7–10) The staining intensity of some of the osteoblast markers was actually more pronounced in cells covering the BRC than lining cells on quiescent surfaces. The cells were negative for the endothelial markers VEGF, Von Willebrand factor, and CD34, whereas marrow sinusoids showed positive staining for these markers.

Table Table 1. Osteoblastic and Endothelial Markers Detected on Cells Lining the BRC vs. Vascular Endothelial Cells as Assessed by Immuno- and Enzyme Histochemical Staining
original image
Figure Figure 1.

Different representations of BRC structures in cortical (top) and trabecular bone (bottom). In cortical bone, the BRC (outer demarcation by the broken line) is filled with erythrocyte ghosts (EG) and is located at the closing cone of the Haversian system situated over osteoblasts (OBs). A few osteoclasts (OCs) are also seen. CV denotes the central vessel of the Haversian system. In trabecular bone (bottom), the outer lining of the BRC is clearly discernible, demarcating a vascular structure on top of osteoblasts (OBs). Picture in top panel courtesy of Pierre Delmas, Lyon, France.

The study by Hauge et al.(6) also revealed that the bone surface covered with BRCs varied with bone turnover. A decrease in bone turnover caused a decrease in surfaces covered with remodeling compartments, whereas increased turnover led to an increase.

As discussed by Parfitt in an editorial,(11) it was still debatable whether all cells involved in remodeling arrived through the circulation. Whereas circulating osteoclast precursors were shown more than a decade ago, there is now increasing evidence that osteoblast lineage cells are also present in the circulation, although the precise concentration of these cells remains a subject of debate and further study.(2,12,13) Thus, the basis for the involvement of circulating precursor cells has been further strengthened.

Whereas the systemic hormonal regulation of the remodeling process has to occur by factors arriving at individual remodeling sites through the bloodstream, the way by which local regulatory factors exert their action on individual cell populations involved is still obscure. Over the last decades, however, we have increased our knowledge about the different growth factors and cytokines involved in local regulation of bone remodeling tremendously (Fig. 2). Apart from growth factors and cytokines, simple molecules like NO, as well as hypoxia and acidosis, have been shown to exert pronounced effects on bone remodeling balance and activity. NO exerts biphasic effects on osteoclast activity with low concentrations potentiating and high concentrations inhibiting bone resorption.(14) Similarly, osteoblastic growth and differentiation are inhibited by high concentrations of NO, whereas lower concentrations may play a role in regulating normal osteoblast growth and in mediating the effects of estrogens on bone formation, mechanotransduction, and bone anabolic responses.(14) Acidosis and hypoxia generally increase bone resorption(15–18) and inhibit bone formation.(19) Because hypoxia may cause acidosis through increased anaerobic metabolism, the two factors may act synergistically at the tissue level.(18) Hypoxia and acidosis also affect secretion of pro-angiogenic factors such as VEGF as outlined below.

Figure Figure 2.

Depiction of some of the main local regulatory factors operating at remodeling sites With osteoclasts (OCs) and osteoblasts (OBs). ILs, TNFs, TGFs, CSFs, IGFs, FGFs, PDGFs, BMPs are formed by both monocytic cells in the marrow space and bone cells in the BMU. NF-κB or RANKL and OPG are formed specifically by osteoblasts. Factors from the marrow space and factors liberated by endothelial cells (VEGF, endothelin, NO) are freely diffusing to receptors on osteoclasts or osteoblasts. The cellular responses in the BMU are further modulated by systemic hormones like estrogen (E2), PTH, active vitamin D,(1,25D) and thyroid hormone (T3).

The literature on local regulation of bone remodeling generally assumes that the local growth factors, cytokines, and even NO come either from cells in the marrow space or vascular cells having free access to the remodeling site without barriers or are produced by osteoclasts and osteoblasts at the remodeling site. The BRC concept implies that the all factors liberated from the cells or vessels in the marrow space exert their regulatory role either through diffusion through the outer layer of the BRC, transport through the bloodstream to the interior of the BRC, or indirectly through modulation of cell activity in the outer wall of the BRC.

POSSIBLE FUNCTIONS OF THE BONE REMODELING COMPARTMENT

The demonstration of a dome of cells displaying an osteoblastic phenotype covering remodeling sites demands reassessment of some of the basic ideas about the structure of remodeling sites and the interaction of cells at these sites with the marrow microenvironment and the vasculature.

The possible functions of this compartment could be manifold.

It could constitute a dome, under which local growth factors liberated from bone and implicated in the coupling process, such as TGF-β and IGF-II, could exert regulatory action at the level of individual bone multicellular units (BMUs). If the access to the marrow space was open, the very high levels of growth factors in the marrow microenvironment might offset eventual localized regulatory effects by these growth factors, crucial to osteoclast and osteoblast differentiation and the remodeling process. The same holds for acidosis and hypoxia. The confined space of the BRC would constitute an ideal environment for localized acidosis to exerts its actions without rapid buffering by proteins and other factors in the marrow space.

It could be the structure where coupling between osteoclasts and osteoblasts take place. The importance of the RANKL/OPG system for the regulation of osteoclast differentiation and action has been well established. This pathway involves presentation of osteoblastic, membrane-bound RANKL to the RANK receptor on osteoclast precursors through cell to cell contact. Because of the timing and sequence of bone resorption and bone formation, the two processes rarely occur within the same area, which makes the needed cell to cell contact between osteoclast precursors and active osteoblasts highly unlikely on a broader basis. Moreover, resorption precedes formation, which makes it even more unlikely that bone-forming osteoblasts may regulate bone osteoclast formation through the RANKL/OPG pathway. A more likely candidate would be the lining cell. As shown in animals by Silvestrini et al.(20) and Eriksen et al,(10) lining cells exhibit positive immunoreactivity for OPG and RANKL and might therefore be responsible for the cell to cell contact with osteoclast precursors. The most likely route by which these osteoclast precursors reach the BRC are through the circulation, where there seems to be an abundance of cells with the potential to differentiate into osteoclasts. Another possibility is, of course, diapedesis through the outer lining of the BRC, which would still secure local regulation of osteoclast differentiation within the BRC.

As previously discussed by Parfitt,(11) the existence of such a structure would also obviate the need for a “postal code” system ensuring that resorptive and formative cells adhere to areas on the bone surface where they are needed. Bone surfaces are generally covered by lining cells, which would prevent direct contact between bone cells and integrins or other adhesion molecules known to modulate cell activity. The BRC would be the only place where these cells (circulating osteoclasts and circulating osteoblast precursors) would be exposed to these matrix constituents, because the formation of the BRC involves detachment of lining cells from the bone surface. This mechanism does not rule out a role for homing factors like stromal cell derived factor 1 (SDF-1), which is abundantly expressed on endothelial cells in the interaction between BRC and circulating cells. Apart from effects as a “homing factor,” SDF-1 may also play a role in diapedesis of marrow cells from the marrow space into the BRC.(21)

The BRC may also be the structure through which mechanosensory signals from the osteocyte network are translated into changes in osteoclast and osteoblast activity on trabecular surfaces. Lining cells are connected to the osteocyte network through gap junctions between lining cells on quiescent surfaces and osteocyte canaliculi(22) (Fig. 3). These signals could be transmitted to the outer lining cell layer of the BRC, and the BRC would be a site where hormonal modulation of the mechanosensory input could take place. Signaling through the osteocyte network to surface lining cells with the subsequent transformation of the lining cells into cells lining a BRC would be a very effective way by which changes in mechanical stress to bone could trigger these remodeling events on the bone surface.(23,24)

Figure Figure 3.

Connections between the osteocyte network, lining cells, and the BRC. All cells in this network are connected with gap junctions, which may provide a pathway (block arrows), by which signals generated deep within bone may reach the surface and elicit remodeling events by osteoclasts (OCs) and osteoblasts (OBs) in response to mechanical stimuli. The response may be modulated by factors liberated from the vascular endothelium or marrow capillaries/sinusoids and paracrine factors (broken arrow) liberated from lining cells may also play a role.

Exposure to short-duration PTH peaks seem to elicit dedifferentiation of lining cells into osteoblasts. This dedifferentiation is not accompanied by increased proliferative activity in the osteoblast population or lining cell layer,(25) but is nevertheless thought to be crucial for the initial burst of bone-forming activity seen 3–6 months after PTH administration.(26,27) Whether the initial activation of lining cells only involves cells covering quiescent surfaces or also cells lining BRCs is still unknown.

The BRC provides ready access for regulatory factors produced outside bone to the BMU. Recent studies have characterized several circulating inhibitors of mineralization: matrix GLA protein (MGP), secreted phosphoprotein 24 (SPP 24), and fetuin. These factors are thought to inhibit further growth of calcium phosphate crystals in situations where hypersaturation may occur and secure removal of complexes from the circulation.(28) Whereas MGP is formed in bone, SPP 24 and fetuin are formed in the liver. These proteins would therefore only be able to regulate mineralization at individual BMUs if they could reach bony surfaces through the circulation. As highlighted by the group pioneering this work, the BRC is the only way by which these agents could reach denuded bone surfaces.(28)

The BRC may also play a crucial role in the spread of bone metastases. It is well established that apart from entering bone through frank in-growth, tumor cells reach bony surfaces through the circulation. The growth of metastatic cells in bone is enhanced by the so called “vicious cycle,” where PTH-related peptide (PTHrP) produced by tumor cells (e.g., breast cancer cells) induces increased local bone resorption and subsequent liberation of TGFβ from the bone matrix.(29) The local effects of TGFβ in the bone microenvironment are 2-fold: it enhances the growth of bone metastases and increases PTHrP formation from tumor cells further,(29) thus maintaining the vicious cycle. As shown above, one of the key components of the vicious cycle, TGFβ, is produced by the cells lining the BRC. Other key promoters of bone metastases such as interleukin (IL)-1 and IL-6 are also produced by the lining cell layer covering BRCs. It is therefore probable that the microenvironment in the BRC is highly conducive to metastatic seeding and the formation of the vicious cycle, further enabling growth of the bone metastasis. Moreover, the existence of a closed compartment would make vicious cycle formation easier because of the absence of interference with cytokine and growth factors from the marrow space. The BRC would also constitute an area where tumor cells would be exposed to high levels of integrins, which could further enhance their adherence.

It is now well established that bisphosphonates (BPs) reduce the number of skeletal events in breast cancer, prostate cancer, and myelomatosis, and intravenous bisphosphonates are used routinely in advanced cancer. There is still debate as to how much of the beneficial effects of bisphosphonates in advanced cancer are caused by inhibition of angiogenesis or to other, direct antitumor effects. In animal experiments, however, treatment with zoledronic acid only affects bone metastatic burden and not the growth of nonbone tumor cells.(30) As shown in the paper by Hauge et al.,(6) the number of BRCs correlates with the turnover state of the patients. One alternative pathway, by which BPs could exert their inhibitory effects on bone metastases, could be by simply reducing the surface of bone where possible seeding would be possible. This would be accomplished by reducing the number of BRCs operating (i.e., by inducing low turnover, which is the main effect of bisphosphonates on bone).

UNRESOLVED ISSUES

The demonstration of the BRC forming a barrier between the marrow space and individual remodeling sites also opens a number of new questions:

  • 1How do cells enter the BRC? Given the unequivocal presence of osteoclast precursor cells in bone marrow and peripheral blood,(31,32) it is likely that these cells arrive in the BRCs through the circulation. The situation is less clear for osteoblastic cells. Certainly, adherent bone marrow stromal cells could enter the BRC simply by diapedesis, although this still leaves unresolved the origin of osteoblastic cells in BMUs in cortical bone. A second possibility is that bone marrow nonadherent osteoblastic cells, as identified over a decade ago by Long et al.,(33) enter the marrow sinusoids and thus the BRCs through the marrow microcirculation, eventually egressing into the peripheral circulation.(2) Finally, the intimate association of the BRCs with the vasculature raises the possibility of resident cells in the vessels, such as pericytes,(34) entering the BRCs and differentiating into osteoblasts. Indeed, there is increasing evidence pointing to a common origin of osteoblasts and endothelial cells (meso-angioblasts).(35) The latter two scenarios would also provide an explanation for how osteoblasts access not only the BRCs in trabecular bone, but also the BMUs in cortical bone. Certainly, these possibilities are not mutually exclusive, and there may potentially be multiple pathways to the active osteoblast on the bone surface.
  • 2What is the initial event starting the remodeling sequence? Close contacts between vascular endothelium and BRCs were reported in the paper by Hauge et al.(6) Is the very first step, adhesion of a blood vessel to bone-lining cells at a site where targeted repair is needed. Conceivably, osteocyte apoptosis and possible release of osteotropic growth factors and cytokines could be attractants for blood vessels, which would subsequently initiate the formation of a resorptive BRC. However, as outlined above, the framework for signaling within the osteocyte–lining cell–BRC network could also be a way by which remodeling events on bony surfaces are triggered from damage accumulation or changes in mechanical strain within bone. If this is the sequence of events preceding initiation of bone remodeling, what are the vascular events preceding modeling activity (e.g., during the early phases of PTH therapy)?

As noted earlier, there is increasing evidence for a common lineage and close interactions between vascular endothelial cells and bone cells. Endothelial cells posses the capacity to drive differentiation of marrow stromal cell toward the osteoblastic phenotype,(36) and angiogenesis plays a pivotal role in fracture healing and intramembranous bone formation during distraction osteogenesis.(4,5) Endothelin and VEGF are involved in the signaling between vasculature and bone,(37) and VEGF and other angiogenetic factors reveal increased expression during de novo intramembranous osteogenesis.(5) Osteoblastic cells, as well as osteoclasts, possess receptors for VEGF and also produce VEGF,(38) expression of VEGF is closely associated with the early phases of bone modeling and remodeling events,(39) and it induces osteoblast chemotaxis(40) and differentiation.(41) Osteoblastic synthesis of VEGF is increased by hypoxia(42) and BMP-2, which has been invoked as a mechanism by which fracture repair and bone resorption is initiated.(43) Conversely, VEGF increases production of BMP-2 and BMP-4 in endothelial cells, further corroborating the tight interactions between the two cell types in bone formation.(1) Hypoxia also increases levels of other factors influencing endothelial proliferation, mainly members of the TGFβ, IGF, and fibroblast growth factor (FGF) growth factor families.(1) Acidic pH and high lactate levels decreases osteoblastic production of VEGF, which has been quoted as a possible mechanism by which angiogenesis is inhibited during ongoing bone resorption.(44)

Apart from its pivotal role in the regulation of osteoclast differentiation, the RANK/RANKL system also plays an important role in angiogenesis. VEGF upregulates the RANKL receptor (RANK) on endothelial cells, thus increasing the sensitivity of these cells to RANKL(45) and promoting angiogenesis. Conceivably, this upregulation of RANK would also enhance adhesion of mononuclear osteoclast precursors in marrow sinusoids and the BRC.

CONCLUSION

The demonstration of specialized vascular spaces in bone adds a new dimension to our understanding of bone biology in general and bone remodeling in particular. It is now well established that both osteoclastic and osteoblastic cells can be isolated from the peripheral circulation. However, it is still unknown whether these circulating precursors, vascular cells with osteogenic potential or cells entering the BRC from the adjacent marrow space through diapedesis, constitute the main pool from which more differentiated osteoclasts and osteoblasts are recruited. The size of the BRC compartment varies with the remodeling rate and is therefore also regulated by osteoactive drugs through their effects on bone turnover. How these variations of a vascular space in close contact to denuded bone surfaces affect the effects of these drugs on bone mass and the risk of bone metastases remains to be established. The BRC has only been characterized in detail in trabecular bone. As shown in Fig. 1, there is evidence that a similar structure may exist in cortical bone, but further characterization is needed. The interaction between the BRC and adjacent sinusoidal structures with endothelial cells is complex. Whether the osteoblast like cells of the outer wall of the BRC are real lining cells or transformed endothelial cells or pericytes also remains to be established.

Ancillary