The Bone Remodeling Compartment: A Circulatory Function for Bone Lining Cells

ILIAC BONE histomorphometry has from its beginning been plagued by technical problems. To preserve the discrimination between mineralized bone and osteoid and the fixation of tetracycline, sections must be cut without decalcification, which needs plastic embedding and special microtomes. The boundary between hard brittle bone and soft friable marrow has been especially contentious—general acceptance that most of this boundary is occupied by bone lining cells took many years, and there is still disagreement about whether an additional layer of cells invests the marrow.1 One of the pioneers of bone histomorphometry was Philippe Bordier, who died prematurely in 1977. His collaboration with Howard Rasmussen led to an unusual book2; its central theme was borrowed from Harold Frost,3 but it presented many novel observations and penetrating insights. One such observation was that bone lining cells (which they referred to as “mesenchymal cells”) persist as a canopy over sites of remodeling, separating both osteoclasts and osteoblasts from the marrow. This notion has always appealed to me, although it was generally ignored and only rarely was I convinced that I could see the canopy in sections from my own laboratory. Rasmussen and Bordier believed that the location of osteoclasts and osteoblasts beneath the canopy was a consequence of their origin from lining cells, but this is definitely wrong for osteoclasts,4 although it could still sometimes be true for osteoblasts.5 If, as is commonly assumed, osteoclasts arise from precursors in the adjacent hematopoietic marrow that wander directly to the bone surface, they would somehow have to burrow underneath the lining cells. Hauge et al.6 have now convincingly rediscovered the canopy and presented evidence that it provides a conduit for osteoclast precursors. Thus, these cells reach the bone surface via the circulation, not only in cortical bone,7 but also in cancellous bone. Cells made in the marrow for export to the circulation, which include preosteoclasts,7, 8 must traverse the thin wall of the branching sinusoids that empty into the central longitudinal vein.9 Structures resembling marrow sinusoids have been observed adjacent to trabeculae,10, 11 but they were not clearly demonstrated to be involved in bone remodeling. Hauge et al.6 have now identified a structure, which they refer to as the bone remodeling compartment (BRC), that resembles both the sinusoidal structures described by Burkhardt et al.10 and the canopy described by Bordier and Rasmussen.2 This structure is found almost exclusively at sites of bone remodeling. In the present paper this association was demonstrated in patients with primary hyperparathyroidism, but in earlier studies from this group the association was found also in renal osteodystrophy12 and osteoporosis.13 The cells forming the outer wall of the sinusoidal BRC, furthest from the bone and next to the marrow, were originally thought to be of endothelial origin,12 but these cells are negative for expression of CD34,6, 13, 14 von Willebrand factor,13, 14 and VEGF14 and positive for expression of alkaline phosphatase, osteocalcin, osteonectin,6 IGF-1 and −2,15, 16 TGF-β1, -β2, and -β3, and basic FGF.16 The cells are evidently of the osteoblast phenotype and are clearly the same cells as in the canopy of former bone lining cells described by Bordier and Rasmussen.2 To explain the full significance of these observations, it is necessary to distinguish between the vascular system, defined according to embryologic origin and current repertoire of gene expression, and the circulatory system, defined according to function. Ordinarily these are the same, but in some circumstances nonvascular cells may perform a circulatory function, referred to as vasculogenic mimicry.17 Like neoangiogenesis, this process is usually thought of in connection with the survival of neoplastic cells, but neoangiogenesis is an essential component of normal cortical bone remodeling,18 and it now appears that vasculogenic mimicry is an essential component of normal cancellous bone remodeling. Although final proof that the BRC performs a circulatory function is lacking, the circumstantial evidence for this conclusion is compelling. When originally seen12 they were believed to be sinusoids because their cells looked like endothelial cells, they contained numerous red cells (Fig. 1), and they resembled previously described paratrabecular sinusoids.10, 11 They are closely associated with marrow vessels14 with which they appear to communicate directly,19 and scanning electron microscopy (SEM) of vascular corrosion casts in lumbar vertebrae showed a close relationship between resorption cavities and vessels of diameter 10 to 15 μm.20 But the strongest reason for believing that the BRC has a circulatory function is that no other explanation is plausible for the presence of osteoclasts on the wrong side of the canopy (Fig. 1)! Bone remodeling compartment lying between bone marrow and bone, which are separated by a linear structure consisting of flat cells. The compartment contains many red cells, one osteoclast, several adjacent mononuclear cells that could be osteoclast precursors, several septa, and a fat cell, the origin of which is obscure. Figure provided by Drs. F. Melsen and E. Hauge and reproduced with their permission. Bone remodeling is carried out by temporary structures known as basic multicellular units (BMU), which maintain the same spatial and temporal relationships between their constituent cells for several months.21 The BRC is evidently an integral component of the BMU, and it must be constructed during the process of BMU origination. Hauge et al.6 suggest that after site selection and digestion of the endosteal membrane, the lining cells separate from the underlying osteocytes, presumably by disruption of the gap junctions between their processes.22 Incomplete retraction of these processes could give rise to the septa sometimes evident within the BRC (Fig. 1). Such separation would account for persistence of the canopy over the BRC but would not explain how the connection is made with the circulation. Perhaps, an existing para-trabecular sinusoid grows toward the site of origination (another form of neoangiogenesis) and fuses with the canopy as a prelude to establishing communication. Whether there is blood flow through the BRC, and if so, in what direction is unclear. Endothelial cells or associated pericytes may under some circumstances have osteogenic potential18 and could contribute to the canopy after changing their pattern of gene expression toward the osteoblast phenotype. As with any new concept there are many more questions than there are answers. The three dimensional organization of the BMU is much more clear in cortical bone than in cancellous bone because only in cortical bone can an almost complete BMU be readily captured in a single section,23 analogous to a whole nephron dissected out of a kidney. The individuality of cancellous BMU rests on strong but less direct evidence,24 and most people in the bone and mineral field have not yet incorporated the concept of hemiosteonal remodeling into their thinking.25 An important consequence of the search for the BRC is the accumulation of much more direct evidence that osteoclasts and osteoblasts are present simultaneously in different regions of the same BMU,6 and that the simple down and up model of cancellous bone remodeling, although useful for considering cellular mechanisms of bone loss,26 is no longer tenable. Inherent in the concept of hemiosteonal remodeling is that the BMU moves across the surface of cancellous bone, so that the sites of arrival of new preosteoclasts and osteoblasts are continually changing. The BRC must move at the same rate, as must its communication with the circulation. If the blood within the BRC is in direct contact with resorbing and forming surfaces, there seems to be no need for an area code mechanism to direct preosteoclasts to specific sites of diapedesis as is needed in cortical bone,7 but they would still need some form of chemoattractant mechanism to ensure their arrival at the apex of the cutting hemicone to join the existing osteoclast team. A major question raised by the BRC concept is: how do osteoblasts get to the bone surface? Osteoblasts arise from local connective tissue precursors cells that belong to the marrow stromal system.27 If the BRC is a closed system except for its connection with the marrow sinusoids, there seems to be no pathway for marrow derived preosteoblasts to reach sites of bone formation, and any such pathway would breach the barrier between the inside and the outside of the marrow vessels. There is no evidence that circulating precursor cells contribute significantly to bone formation as they do to bone resorption.18 The only way I can see out of this dilemma is that the canopy cells themselves are the origin of preosteoblasts, as Rasmussen and Bordier proposed2 and as Hauge et al. discuss at length.6 The BRC canopy cells have much stronger expression of the osteoblast phenotype than do the lining cells from which they began.6 Because the area covered by a lining cell is much larger (by a factor of at least 5) than the area covered by an osteoblast,1 and because neither BRC canopy cells nor lining cells have been observed to divide, there must be a flow of cells into the BRC canopy from another source. One such source has already been mentioned—the sinusoid endothelial cells involved in the genesis of the BRC—but it seems likely that marrow stromal cells, already known to be the major source of preosteoblasts,27 make the largest contribution. Whatever the explanation, it seems that coupling of formation to resorption in cancellous bone depends on local signals within the BRC,6 and not, as in cortical bone, on sequential changes in gene expression in the BMU capillary.18