Anda di halaman 1dari 46

Biomechanical and Molecular Regulation of Bone Remodeling

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Alexander G. Robling,1 Alesha B. Castillo,1,2 and Charles H. Turner2,3


1 3

Departments of Anatomy and Cell Biology, 2 Biomedical Engineering, and Orthopaedic Surgery, Indiana University Purdue University, Indianapolis, Indiana 46202; email: turnerch@iupui.edu

Annu. Rev. Biomed. Eng. 2006. 8:45598 First published online as a Review in Advance on April 3, 2006 The Annual Review of Biomedical Engineering is online at bioeng.annualreviews.org doi: 10.1146/ annurev.bioeng.8.061505.095721 Copyright c 2006 by Annual Reviews. All rights reserved 1523-9829/06/08150455$20.00

Key Words
mechanotransduction, osteoblast, osteoclast, osteocyte, bone density

Abstract
Bone is a dynamic tissue that is constantly renewed. The cell populations that participate in this processthe osteoblasts and osteoclastsare derived from different progenitor pools that are under distinct molecular control mechanisms. Together, these cells form temporary anatomical structures, called basic multicellular units, that execute bone remodeling. A number of stimuli affect bone turnover, including hormones, cytokines, and mechanical stimuli. All of these factors affect the amount and quality of the tissue produced. Mechanical loading is a particularly potent stimulus for bone cells, which improves bone strength and inhibits bone loss with age. Like other materials, bone accumulates damage from loading, but, unlike engineering materials, bone is capable of self-repair. The molecular mechanisms by which bone adapts to loading and repairs damage are starting to become clear. Many of these processes have implications for bone health, disease, and the feasibility of living in weightless environments (e.g., spaceight).

455

BASIC BONE BIOLOGY


Osteoclast: bone cell of hematopoetic origin that resorbs bone by secretion of acid and proteases Osteoblast: bone cell of mesenchymal origin that secretes unmineralized bone matrix (osteoid), which eventually mineralizes to yield mature bone Osteocyte: terminally differentiated osteoblast that has become entombed in the bone matrix it once elaborated

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Bone is a metabolically active tissue capable of adapting its structure to mechanical stimuli and repairing structural damage through the process of remodeling. The bones of most mammals have four surfacesor bone envelopesupon which the addition or removal of bone can occur: the periosteal, endocortical, trabecular, and Haversian (or intracortical) envelopes. Skeletal envelopes differ in their surface areato-volume ratios and in their response to certain stimuli. At each skeletal envelope, bone resorption and formation are executed and regulated by the bone cellsthe osteoclast and osteoblast, respectively (Figure 1). Tissue yielded by osteoblastic formation is maintained by the osteocytes and bone-lining cells, which are terminally differentiated relics of once-prolic osteoblasts. The cells that populate bone tissue are derived from different origins, and they proliferate and differentiate in response to different cues. Osteoblast-lineage cells, in addition to regulating bone formation, also regulate bone resorption via an elegant signaling axis that controls osteoclast generation and activity. Further, bone cells, most probably the osteocytes, are strain-sensitive cells and can transduce mechanical signals derived from mechanical loading into cues that ultimately result in reduced bone loss and enhanced bone gain. Below, we review the intricacies of these processes, from the level of molecular control of cells to organ-level properties.

Frostian Bone Modeling and Remodeling


Nearly 40 years ago, Frost began describing two distinct mechanisms by which different types of bone cells team up or work individually to achieve skeletal formation and/or renewal (1). These processesbone modeling and remodelingwork together in the growing skeleton to dene the appropriate skeletal shape, maintain proper serum levels of ions, and repair structurally compromised regions of bone. Bone modeling is a process that works in concert with bone growth and functions to alter the spatial distribution of accumulating tissue presented by growth (2, 3). For example, a growing childs muscle mass increases at a rate that outpaces accumulation of bone mass (4). Therefore, the tissue being deposited on a bone experiencing an increased or altered loading environment from (a) the growing and increasingly powerful muscles, (b) increasing body mass, and (c) a lengthening diaphysis must be positioned to optimally meet these rapidly evolving mechanical demands (2, 5). This is accomplished by modeling drifts, through which bone is selectively added or removed from existing surfaces with the goal of optimizing the geometry of the bone. Thus modeling can alter the size, shape, and position in tissue space of a typical long bone cross section by selectively inhibiting or promoting cellular activity at the resorptive and appositional surfaces accordingly (Figure 2a). Bone modeling at any surface involves osteoclast activation and subsequent resorption of bone (A R), or it involves osteoblast activation and subsequent formation of bone (A F), but not both at the same location. Once skeletal maturity is reached, modeling reduces to a trivial level compared with that which occurs during development (69). However, renewed modeling in the adult skeleton can occur in some

456

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 1 (a) Frontal thin section (4 m) through a mouse proximal tibia, just beneath the growth plate (primary spongiosa), illustrating the presence of active osteoclasts (stained pink) resorbing mineralized cartilage and bone (stained black). The osteoclasts are labeled using a reaction that involves tartrate-resistant acid phosphatase, an enzyme marker expressed in immature and mature osteoclasts. (b) Frontal thin section (4 m) through a rat proximal tibia, in the secondary spongiosa, illustrating the process of bone formation by a team of osteoblasts (white arrowheads). Bone is stained black, soft tissue and cells are stained blue. Intervening between the mineralized bone and the row of osteoblasts is a pale blue strip of tissueosteoidwhich represents freshly deposited matrix that has not yet incorporated mineral. As the osteoid seam advances behind the osteoblasts, some of the osteoblasts get trapped in their own osteoid matrix ( green arrow), which subsequently accumulates mineral (red arrow), eventually leading to a former osteoblast completely surrounded by mineralized bone ( yellow arrow), which is now considered an osteocyte.

disease states and in cases where the mechanical loading environment has been altered signicantly. Unlike modeling, which involves either resorption or formation (but not both) at a locus, bone remodeling always follows an activation resorption formation sequence (A R F) (10). Remodeling removes and replaces discrete, measurable packets of bone. On the intracortical envelope, these replacement packets of bone, or bone structural units (BSUs), comprise secondary osteons (Figure 2b). Bone is

www.annualreviews.org Bone Remodeling

457

Basic multicellular unit (BMU): temporary anatomic structure comprising osteoclasts and osteoblasts that replace older packets of bone with new bone tissue

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

remodeled by teams of cells derived from different sources, which are collectively called the basic multicellular unit (BMU). The BMU is a mediator mechanism bridging individual cellular activity to whole bone morphology (2). Intracortical BMUs maintain a distinctive three-dimensional structure as they move through long bone diaphyses in a nearly longitudinal orientation (11, 12). The leading region of the BMU is lined with osteoclastsspecialized cells capable of bone resorption. The diameter of the tunnel excavated by osteoclasts, which typically reaches roughly 250300 m, denes the cross-sectional size of the secondary osteon that will form in its wake (Figure 2b). Following closely behind the osteoclasts is a group of mononuclear cells that line the resorptive bay during the reversal phase (the period between resorption and formation). Although the exact function of these cells remains unclear, it is likely that they smooth off the scalloped periphery of the resorptive bay in preparation for the deposition of a reversal linea thin, mineral-decient, sulfur-rich layer of matrix

458

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

that separates an osteon from surrounding interstitial lamellae (13, 14). Behind the mononuclear cells, rows of osteoblasts (bone-forming cells) adhere to the reversal zone and deposit layers of osteoid (unmineralized bone matrix) centripetally. The size of the remodeling space constricts as more concentric osteonal lamellae are deposited and mineralized. At a specied point, deposition ceases leaving a Haversian canal in the center of the newly formed osteon. Remodeling on the trabecular and endocortical surfaces follows the same sequence of cellular events as described for the Haversian envelope, except that the cells do not dig and rell tunnels. Rather, they remove and replace pancake-like packets of bone scalloped from these surfaces (12). Because of the morphology of the remodeling BMU, where the osteoblast teams trail behind osteoclast teams and the entire structure moves as a unit, the resorption and formation processes are said to be coupled to one another. Coupling is a strictly controlled process in remodeling, ensuring that where bone is removed, new bone will be restored (15). The net amount of old bone removed and new bone restored in the remodeling cycle is a quantity called the bone balance. While coupling rarely is affected, bone balance can vary quite widely in many disease states; for example, in osteoporotic patients, resorption and formation are coupled but there is a negative bone balance, i.e., more bone is resorbed than is replaced by the typical BMU (16). BMUs are constantly remodeling bone tissue in the growing, adult, and senescent skeleton. Most metabolic bone diseases are manifest in one form or another
Figure 2 (a) Osteoclasts and osteoblasts arrange themselves into temporary anatomical structures called BMUs (basic multicellular units) that execute bone remodeling. BMUs are characterized by teams of osteoclasts that resorb bone in the characteristic cutting cone leading edge of the structure, followed by osteoblasts, which line the closing cone and centripetally deposit layers of new bone to rell the tunnel excavated by the osteoclasts. The new bone begins as unmineralized matrix (osteoid), which stains blue in the preparation shown. Eventually, the osteoid incorporates mineral, which stains black in the preparation shown. The upper half of the panel is a longitudinal section through an active BMU, and the lower half completes the structure in the form of a cartoon, illustrating the position of the osteoclasts (blue) and osteoblasts (green). The red structure represents an extending capillary that keeps pace with the cutting cone and feeds new osteoclast precursors to that region. (b) Photomicrograph of a cross section through the midshaft of a long bone, illustrating the structural consequences of tunneling BMUs. Marrow and periosteum have been removed for clarity. The modeling process (formation without prior resorption) produced the sheets of primary bone, indicated by the green star. The cortex resulting from modeling becomes remodeled soon after it is created, leaving behind osteons, the doughnut-shaped structures within the cortex ( yellow arrows). The fact that the osteons (products of remodeling) were created after the periosteal primary bone was formed (modeling) can be appreciated in panel c (a close-up view of the bracketed area in panel b) by the observation that the osteons cut through the layers of bone near the periosteum, rather than the periosteal layers wrapping around the osteons. Remodeling on the endocortical surface occurs by the same mechanism as osteon creation. Completed remodeling structures (BSUs) are visible on the endocortical surface, the borders of which are indicated by red arrows. It is clear that the endocortical bone packets shown were remodeled and not modeled, as indicated by the osteon fragments created by endocortical resorption prior to relling near the endocortical surface (c). Photomicrograph in panel a kindly provided by Dr. Matthew R. Allen, Indiana University School of Medicine.

www.annualreviews.org Bone Remodeling

459

Table 1

Histomorphometric outcomes for selected metabolic bone diseases Ac.f + + + + Sigma + + N/C Bone balance N/C2 +

Condition Osteoporosis Hyperparathyroidism Osteomalacia1 Corticosteroid-induced osteopenia


1 2

Though total bone volume is increased, the porportion of mineralized bone is reduced. N/C = no change.

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

from perturbations in remodeling BMU dynamics. The number of BMUs working in a specied volume of tissue at any one moment depends on several factors (17). The rst factor is the birth rate of new BMUs, which is also called the activation frequency (Ac.f ). A high activation frequency will result in a large number of active BMUs, and ultimately, a large number of secondary osteons. A second factor, which summarizes the longevity of individual BMUs, is related to the speed with which the BMU travels through tissue space. This factor is called the sigma period, so named for the Greek symbol originally used by Frost (1) to denote this quantity. The sigma period (RC ) quanties the number of days it takes for a BMU to completely remodel a xed two-dimensional slice through a region of bone. One can think of this concept as a spectator watching a parade pass by from a seat on the grandstand. In human cortical bone, it would take approximately 120 days for the entire BMU to pass through a plane, leaving a new osteon behind. Roughly 20 days would be spent initiating and increasing the diameter of the resorption cavity by the osteoclasts, followed by 10 days of reversal (relative quiescence), and nally, 90 days of centripetal deposition of bone matrix by the osteoblast teams. Sigma periods are frequently subdivided into resorption [RC(r ) ] and formation [RC( f ) ] periods as just described. Many bone diseases can be classied as to their effects on activation frequency and sigma periods. Table 1 provides several common diseases of bone and their associated effects on Ac.f and . For example, primary hyperparathyroidism is associated with an increase in Ac.f and a decrease in , resulting in rapid bone turnover (18).

Molecular Control of Bone Cell Differentiation and Fate


The previously described BMU comprises a collection of different cell types with different origins. The osteoclast teams that line the cutting cone are derived from hematopoetic stem cells residing mainly in the marrow and spleen (Figure 3a). Osteoclastogenesis begins when a hematopoetic stem cell is stimulated to generate mononuclear cells, which then become committed preosteoclasts and are introduced into the blood stream. This step requires expression of the Ets family transcription factor PU.1 and macrophage colony stimulating factor (M-CSF) (19, 20). The circulating precursors exit the peripheral circulation at or near the site to be resorbed, and fuse with one another to form a multinucleated immature osteoclast. Fusion of the mononuclear cells into a polykaryon (immature osteoclast) requires the presence of M-CSF and

460

Robling

Castillo

Turner

the receptor activator of nuclear factor B (RANK-L), a tumor-necrosis factor family member (21). Successful creation of the immature osteoclast is associated with the initiation of tartrate-resistant acid phosphatase (Trap) expression, an enzyme used later in more mature cells to assist in bone resorption (Figure 1a). Further differentiation of the immature osteoclast occurs only under the continued presence of RANK-L and also requires the expression of several genes, including the AP-1 member c-fos (22, 23), micropthalmia-associated transcription factor (MITF) (24, 25), and nuclear factor of activated T cells, calcineurin dependent 1 (NFAT-c1) (22, 26). Once the transition to mature osteoclast is reached, the bone-resorbing activity and survival of the mature osteoclast are regulated by RANK-L. The mature osteoclast engages in bone resorption via peripheral attachment to the matrix, employing the 3 integrin (27), which creates a microcompartment between the rufed basal border of the cell and the bone surface. H+ ions are pumped into the compartment by the osteoclast to solubilize the mineral component, followed by protease degradation of the organic matrix (28). This process leaves the characteristic Howships lacunae, seen in tunneling BMUs. Osteoblast development follows a different course, beginning with the local proliferation of mesenchymal stem cells residing in the marrow (and periosteum; Figure 3b). Expression of the transcription factors runt-related transcription factor-2 (Runx2), distal-less homeobox-5 (Dlx5), and msh homeobox homologue-2 (Msx2) is required to push the precursor cells toward the osteoblast lineage and away from the adipocyte, myocyte, and chondrocyte lineages also yielded by the mesenchymal stem cell (2933). The committed preosteoblast expresses type I collagen and bone sialoprotein (Bsp). Further differentiation of the preosteoblast into a mature, boneforming osteoblast phenotype requires the expression of Runx2, osterix (Osx), and several components of the Wnt signaling pathway (30, 3436). The mature osteoblast expresses the matrix proteins type I collagen (ColI) and osteocalcin (OC) and a key enzyme in the mineralization process, alkaline phosphatase (Alk Phos). As a row of active osteoblasts secretes unmineralized matrix (osteoid) and advances away from the bone surface, a small number of cells fall behind and become incorporated into the matrix (Figure 1b). These osteoblasts begin to generate long cytoplasmic processes to remain in communication with surrounding cells and upregulate expression of E11, an early osteocyte marker (37). At this point the cells are considered immature osteocytes (38). As the matrix matures and mineralizes, and the osteoid seam moves further away, the osteocyte becomes entombed in a bony matrix and begins to mature and express a new set of genes, including dentin matrix protein-1 (DMP-1), matrix extracellular phosphoglycoprotein (MEPE), and SOST (3943). Osteocytes make up the majority of bone cells.

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

The OPG/RANK-L Signaling Axis: A Molecular Rheostat for Bone Resorption


The effect of RANK-L on osteoclastogenesis was discussed briey in the previous section, but the signaling axis to which this molecule belongs plays such a major role in osteoclast biology that a separate discussion of its action is warranted. For many

www.annualreviews.org Bone Remodeling

461

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 3 Lineage of osteoclasts and osteoblasts. (a) Osteoclasts are derived from a hematopoeitic precursor in the bone marrow, spleen, or liver. Proliferation of mononuclear cells from the precursor population requires M-CSF. The blood-borne preosteoclasts, which express the surface antigen F4/80 (142, 143), enter the circulation and arrive at the site to be resorbed. They will fuse together into a polykaryon (immature osteoclast) only in the presence of M-CSF and RANK-L. The immature osteoclast begins to express TRAP, calcitonin receptor, and the beta-3 integrin. RANK-L and a host of transcription factors are required to push the cell into a mature osteoclast phenotype, which maintains expression of many of the same immature osteoclast markers. (b) Osteoblasts are derived from a mesenchymal stem cell, which can also give rise to adipocytes, myoblasts, and chondrocytes. Proliferating precursors are pushed toward the preosteoblast phenotype by the expression of Runx2, Dlx5, and Msx2. The preosteoblast expresses collagen I and bone sialoprotein. Further, Runx2 expression, but also osterix, and members of the Wnt signaling cascade (b-catenin, TCF/LEF1) are required to achieve a mature, matrix-producing osteoblast phenotype (Col I, osteocalcin, and alkaline phosphatase expression). Osteoblasts that become trapped in the matrix express E11, an early osteocyte marker, and eventually express DMP-1, Mepe, and Sost as the mature osteocyte phenotype is reached. The red bars on the bottom of the gure indicate the expression sequence of several osteoblast/osteocyte gene promoter fragments that are used to drive expression of proteins at different stages of differentiation in the osteoblast lineage.

462

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

years, it had been known that osteoclast differentiation in vitro from hematopoetic stem cells required coculture with cells of mesenchymal origin. Some factor was presented by the osteoblast/stromal cells that permitted osteoclast development (44). Furthermore, physical contact with stromal cells, and not secretion of some soluble compound into the media by stromal cells, was required for osteoclastogenesis (45). This conclusion was demonstrated most notably by Takahashi et al. (46), who showed that a mixed stromal/hematopoetic coculture produced osteoclasts, but the same two cell populations cultured separately (separated by a 0.45 m membrane) in the same dish failed to yield osteoclasts. The unknown factor turned out to be RANK-L, and the receptor on osteoclasts and their precursors was subsequently identied as RANK (47). At the same time, a soluble factor was found that inhibits the activity of RANK-L, which was identied as osteoprotegerin (OPG) (48). OPG is a soluble decoy receptor for RANK-L, and it functions to reduce osteoclastogenesis by competitively occupying the stromal RANK-L binding sites for RANK receptors on precursor and later-stage osteoclasts (Figure 4a) (49, 50). Consequently, stromal cells can control osteoclastogenesis in a positive direction by increasing the expression of RANK-L and decreasing the expression of OPG, or conversely, the proportions can be reversed to decrease resorptive activity. It had been known for many years that osteoclast development and activity were under the control of the osteoblast/stromal cell (44). The RANK-L/OPG signaling axis provides a mechanism for how such control is transduced by stromal cells. Factors that exhibit a strong effect on resorption (parathyroid hormone, prostaglandins, interleukins, vitamin D3, corticosteroids) all signal to the osteoblast/stromal cell, which then appears to translate each barrage of signals into an appropriate RANK-L and OPG output to control osteoclast development and resorption (51). Thus, the stromal cell can be thought of as a transducer of cytokine/hormonal signals for the osteoclast, where the input is the cytokine/hormonal/mechanical signal and the output is a RANK-L and OPG ratio (Figure 4b). With the discovery of RANK-L, RANK, and OPG, a revolutionary understanding of osteoclastogenesis was born.

Disuse: absence of the normal peak strains typical for a region of bone tissue; can be induced by immobilization, bed rest, or spaceight Microdamage: irreversible fatigue damage created in bone as a result of loading-induced stress risers, manifest as microscopic cracks, microfractures, and diffuse damage

BIOMECHANICAL CONTROL OF BONE REMODELING


Mechanical loading has profound inuences on bone remodeling. Disuse or lack of loading causes an acceleration of bone turnover, with bone resorption dominating bone formation and thus a rapid loss of bone mass. This type of bone loss is observed in astronauts who spend extended periods of time in the weightless environment of a space station or shuttle. Overuse of bone can cause damage to the tissue, which in turn stimulates bone remodeling. One of the important roles of bone turnover is to continuously replace and repair damaged bone tissue. Osteoclasts target regions of microdamage preferentially and remove compromised bone tissue. The damaged tissue is then replaced by new bone tissue. If damage accumulates faster than the tissue can be repaired, larger microcracks may develop and propagate to form a stress fracture. The increased number of bone remodeling sites with disuse or overuse is preceded by programmed cell death in osteocytes (52). The factors that initiate osteocyte death

www.annualreviews.org Bone Remodeling

463

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 4 (a) RANK-L signaling from stromal-derived cells induces osteoclastogenesis and activity via binding to RANK on the osteoclast (and preosteoclast). Downregulation of RANK-L, or inhibition of RANK-L availability via OPG (a soluble dummy receptor that binds RANK-L release), prevents new osteoclast generation and induces apoptosis in existing osteoclasts. RANK-L exists in both membrane-bound and soluble forms, but there is some evidence to suggest that the membrane-bound form is more potent (144). (b) Stromal cells control osteoclastic bone resorption in response to a variety of stimuli by adjusting the level of RANK-L and OPG expression. High RANK-L expression promotes osteoclastogenesis and survival, whereas high OPG expression results in low osteoclastogenesis and osteoclast apoptosis. The stromal cells can therefore be thought of as a rheostat in the resorptive process.

(apoptosis) are not well understood but may include direct damage to osteocytes via microcracks in the bone matrix or lack of convective uid ow during disuse. The effects of loading on bone remodeling follow a U-shaped curveremodeling is increased in disuse (insufcient loading) or overuse (overloading causing damage) (Figure 5). There is a range of loading within which bone remodeling is minimized. This is often called the physiological range. Periosteal bone formation, on the other hand, steadily increases with increased loading, as described in the following section.

464

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 5 Bone turnover (remodeling) increases in states of disuse (red) and overuse (orange). A physiological window falls between the extreme loading conditions, in which bone turnover is low. Bone formation at the periosteal bone surfaces is also affected by mechanical loading, but the relationship is different. Periosteal bone formation is very low in states of disuse and increases with increasing mechanical stimulus.

Mechanical Loading and Bone Formation


The mechanical properties of bone tissue are similar among mammals. For instance, the Youngs modulus of rat femoral bone is similar to that of human femoral bone, so the major difference between rats and humans is size. Long bones are for the most part thick walled tubes. This geometry allows bones to carry loads effectively but remain relatively light. The most effective way to strengthen a long bone is to add new bone tissue to the periosteal (outside) surface. Periosteal bone formation most effectively increases the second moment of area and the resistance to bending or torsional loading. Moreover, new bone formation is most effective if it occurs where the stress in the bone tissue is greatest, thus reducing stress hot spots and greatly diminishing the risk of structural failure. This is the goal of optimum structural design. Structural engineers have adapted several strategies to achieve optimum structure. Mattheck (53) demonstrated one approach to structural optimization with the analysis of a shaft with a rectangular aperture. This mechanical part developed a stress concentration in the llet when loaded in bending. An algorithm was employed that allowed the llet to grow in areas of high stress, thus changing the shape in the llet slightly. The stress concentration was eliminated in the optimized part and mechanical tests showed that the improved design sustained over 40-fold more loading cycles before failing. The structural optimization algorithm favored by Mattheck dictates that material should be added only where stresses are highest. In long bones, stresses are highest in certain regions along the periosteal surfaces. Recent studies from our laboratory show how long bones apply this algorithm in response to loading. Cyclic mechanical loads were applied axially along the ulna of adult rats three times per week for 16 weeks. The rat ulna has a natural curvature in the medial-lateral direction, so axial loads induce bending of the bone (Figure 6). Under load, the medial surface of the bone is

www.annualreviews.org Bone Remodeling

465

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 6 The effect of exercise on the ulna is simulated by applying axial loading on the forelimb of a rat (left panel). Axial loading causes the ulna to bow laterally so the medial surface is under compression and the lateral surface is under tension. New bone formation is clearly evident on the medial and lateral surfaces (top right). The red label indicates the beginning of the experiment. Loading caused new bone to be formed beyond the red label. Axial loading applied for 3 min/day, 3 days/week, for 16 weeks improved bone strength by 64% and the energy required to fracture the bone was 94% higher, yet the increase in bone mineral content was only 7% (bottom right). Reprinted from Reference 54 with permission from publisher.

subject to compressive stresses and the lateral surface is in tension. The ratio of compressive to tensile stress magnitude in the loaded adult rat ulna is about 1.5, indicating that the highest stress in the ulna occurs at the medial surface in compression. The pattern of bone formation induced by loading resembles the stress distribution, with more bone formation where the stresses are highest. The improvement in bone structure is evidenced by a 69% increase in second moment of area. The ulnar bone strength of loaded limbs was 64% greater than controls and energy absorbed before fracture increased by 94%, yet the improvement in bone mineral content (BMC) was only a modest 7% (54). Therefore, in this animal model, loading induced dramatic improvements in bone biomechanical properties, even with small changes in BMC. The structural efciency of the ulna was improved by bone formation, specically in highly stressed areas where it was most needed.

Wolff s Law and Optimum Structural Design of Bone


The rat ulna is not the only bone that resembles an optimum structure. Trabecular strut alignment has long been studied as an example of material optimization. The

466

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

engineering interest in trabecular structures stems from Wolff s law, based on the observation that trabecular trajectories within the proximal femur roughly align with the directions of maximum stresses (55) (Figure 7). It remains a matter of some debate whether the architecture of the proximal femur is determined solely by biomechanics or if it results from growth patterns during development. Nevertheless it is well established that trabeculae tend to align with maximum stresses in many bones, e.g., vertebral bodies, proximal tibia, and calcaneus. Trabecular alignment results in a material property called anisotropy, which means that the strength and stiffness of the bone varies with direction. Trabecular bone is stiffer and stronger in the direction of trabecular alignment. By becoming anisotropic, trabecular bone greatly increases its load-carrying capacity without increasing mass, thus improving structural efciency. Mechanical stress also improves bone strength by inuencing collagen alignment as new bone is being formed. Cortical bone tissue located in regions subject to predominantly tensile stresses has a higher percentage of collagen bers aligned along the bone long axis. In regions of predominant compressive stresses, collagen bers are more likely to be aligned transverse to the long axis (56). This observation has functional signicance because bone with more longitudinally oriented collagen bers has superior tensile properties, whereas transversely oriented collagen make better mechanical properties in compression (57, 58). There is some question as to whether the orientation of collagen bers in bone occurs through functional adaptation as the bone is being remodeled or is under genetic inuence during development. In an experiment where stresses were altered in the forelimbs of adult dogs for a period of a year, collagen orientation within the bone was changed. At the medial cortex, tensile stresses were increased substantially, which was associated with an increase in collagen alignment (measured as an anisotropy ratio in demineralized bone) (59). These results suggest that local mechanical stresses affect the collagen construction during bone remodeling.

Three Rules That Govern Bone Adaptation


A great deal of experimental evidence has been gathered in the past 30 years, and common threads have emerged that allow us to describe the concept of bone adaptation in mathematical terms. Of greatest importance are the following three rules. Rule 1: The dynamic stimulus. In 1971, Hert and his coworkers showed that dynamic, but not static strains, increased bone formation in rabbits (60). Dynamic strains thus appeared to be the primary stimulus of bone adaptation. These results were later conrmed by Lanyon & Rubin (61). More recently, it has been shown that both loading frequency and strain rate are important determinants of bone adaptation (62, 63). Loading will have no effect on bone formation unless it is applied at a frequency of 0.5 Hz or greater (62). Mechanical loading increases bone formation better if applied at higher frequency. For instance, loading of rat forelimbs produced over tenfold more bone formation when applied at 10 cycles/s (Hz), compared with application at 1 Hz (64) (Figure 8). Thus increasing loading frequency is one step toward more effective application of mechanical forces to promote osteogenesis. In another study, small loads were applied to the sheep hindlimbs at 30 Hz, 20 min per

www.annualreviews.org Bone Remodeling

467

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

468

Robling

Castillo

Turner

day, for a year (65). In this experiment, sheep stood on a vibrating plate with their hind legs. The results were encouraging but not spectacular. Trabecular bone volume was increased by over 30% in a small region of the femur (one of the bones subjected to loading) but overall bone density was not increased in the loaded bones, and there was no increase in bone mass at any cortical bone site. The limited anabolic response to vibration in cortical bone could be due to the fact that peak strains in the bone were very small (about 5 microstrain). The experimental results demonstrate that the stimulus for bone formation is increased if either the magnitude or frequency of loading is increased, so E = k1 f,
Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

(1)

where E is the strain stimulus, k1 is a proportionality constant, is peak-to-peak strain magnitude, and f is loading frequency in cycles per second. Equation 1 gives the strain stimulus for a sinusoidal loading waveform. However, the result can be generalized using the Fourier method that allows any periodic loading waveform to be expanded into a series of n sine waves at different amplitudes and frequencies. So in the general case, the strain stimulus is dened as
n

E = k1
i=1

i fi .

(2)

Equation 2 predicts experimental results fairly well, but tends to be less accurate at higher loading frequencies. Recent experiments have shown that the sensitivity of bone to loading changes little when loading frequency surpasses 10 Hz (66). However, loading at frequencies up to 90 Hz can have effects on bone formation (67). Rule 2: A case of diminishing returns. Extending the duration of skeletal loading does not yield proportional increases in bone mass. As loading duration is increased, the bone formation response tends to fade as if the cells become desensitized (Figure 9). The data from Figure 9 t a logarithmic relationship, such that
Figure 7 Differing trabecular architectures in the proximal femur near the hip joint demonstrated with plane lm X-ray images. The angle between the femoral shaft and neck usually falls within 125 to 130 (b). However, in some people this angle can exceed 140 , which is called coxa valga (a). Conversely the neck angle can sometimes be less than 110 (c). The customary forces on the femoral head from walking (arrows) cause stresses within the trabecular bone of the femoral neck (narrow section). In the normal hip, these stresses are mostly compressive in the inferior (bottom) region of the femoral neck and tensile in the superior (top) region. Trabecular struts are aligned with the principal stress directions and most densely spaced where the stresses are highest. The blue lines show primary compressive stress directions and the red bands depict tensile stresses. In the case of coxa valga (a) the stresses are primarily compressive and one can see the trabecular struts aligning with the major compressive band. When the femoral neck angle approaches 90 (c) the femoral neck is under large bending loads that result in tensile stresses superiorly and compressive stresses inferiorly. Dense trabecular bands arc along the primary tensile (red) and compressive (blue) stress trajectories. The trabecular density in the superior neck is greater when the neck is largely under bending loading. Adapted from Reference 145.

www.annualreviews.org Bone Remodeling

469

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 8 Periosteal bone formation rate (rBFR/BS) measured in rat ulnae subjected to loading at different peak magnitudes and frequencies. The magnitude of load was converted to peak compressive strain at the medial surface of the ulna. The threshold strain necessary to initiate new bone formation was reduced from 1820 microstrain at 1 Hz loading frequency to 650 microstrain at 10 Hz. Reprinted from Reference 64 with permission from publisher.

Bone formation = k2 log (N + 1),

(3)

where k2 is a proportionality constant and N is the number of loads on the bone during a training session. Equation 3 demonstrates that bone tissue sensitivity to mechanical loading is proportional to 1/(N+1). Thus, bone loses more than 95% of its mechanosensitivity after only 20 loading cycles. Presumably, bone cells will resensitize to loading if they are given a period of rest between loading bouts. Resensitization can occur in seconds or hours depending on the nature of the loading stimulus. We have observed that rats receiving 8, 4, 2, 1, 0.5, or 0 h of rest between loading bouts responded with 100% more bone formation when allowed at least 4 h of rest (Figure 10). The data from Figure 10 t the following relationship: Recovery (%) = 100 (1 et/), (4)

where t is the time between bouts and is a time constant approximately equal to 6 h. With a rest period of 4 h between loading bouts, loading-induced bone formation was almost doubled. After 24 h of rest, 98% of bone mechanosensitivity returns. In addition, we have shown that there are several different timescales, ranging from seconds to hours, required for resensitization to occur. For example, rats given 14 s between each load cycle formed 67% more bone than rats administered back-to-back cycles (68). Similar experiments from other groups have conrmed the osteogenic benet of short-term (on the order of seconds) recovery periods in restoring sensitivity to loaded bone (69).

470

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 9 Bone mass in the tibia of rats (closed circles) (146) or ulna of turkeys (open triangles) (147) increases after applied mechanical loading. However, the anabolic effect of loading saturates as the number of loading cycles increases. There is limited benet of additional loading cycles above approximately 40 cycles per day. Reprinted from Reference 148 with permission from publisher.

The interaction between the strain stimulus (f ) and duration of loading (number of loading cycles per day, N) can be described mathematically using either a logarithmic relationship or the exponential relationship derived by Carter et al. (70). We can combine Equations 2 and 3 into a new formula for the daily loading stimulus (S):
k

S
j =1

log(1 + Nj )E j ,

(5)

where k represents the number of daily loading conditions. After a loading session, the bone cells will be desensitized, and one must consider Equation 4 to estimate the responsiveness of the bone to a second loading session. Rule 3: Bone cells accommodate to routine loading. For bone to adapt to a new mechanical loading state, bone cells must have some memory of their previous mechanical environment to determine that the new environment is different and requires a response. Bone cells must process loading information locally because bone tissue is poorly innervated and, unlike many mechanoreceptor cells, cannot rely on the central nervous system to integrate and distribute information about mechanical signals. Most long bones are loaded in bending (71), creating a strain gradient across the bone section that has a value of zero along the neutral axis. To

www.annualreviews.org Bone Remodeling

471

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 10 Bone formation (rBFR/BS) on the endocortical surface of the rat tibia after applying loading in 4 bouts of 90-load cycles every second day. Either 0, 0.5, 2, 4, or 8 h of rest was inserted between loading bouts. As the rest period between the bouts was extended, the bone formation induced by mechanical loading increased. A limit is reached at approximately 4 h at which further lengthening of the rest period has little additional benet. Reprinted from Reference 68 with permission from publisher.

avoid excessive bone resorption at the neutral axis, the threshold for bone response must vary considerably across the bone section. Therefore, each mechanosensitive cell must have some stimulus threshold above which a mechanical signal causes a cellular response. This threshold is a product of local loading history. It is well-known that many cell types, including osteoblasts, reorganize their cytoskeletons in response to a mechanical stimulus (72). Cytoskeletal reorganization in turn changes a cells mechanosensitivity, allowing the cell to accommodate to its strain environment. Also, cellular mechanosensitivity may be altered through reworking the local environment around the cell (73). Osteocytes are considered to be part of the mechanosensory apparatus in bone, and any change in their extracellular microenvironment could change their mechanosensitivity. The extracellular environment of osteocytes does not remain constant, as these cells actively form and remove layers of matrix on the surface of their lacunae (74). An algorithm using the principle of cellular accommodation is derived from the general model for bone adaptation: = B( S0 ), t (5)

where is density in g/cc, t is time in weeks, B is a rate constant (g/cc-week), is the strain stimulus function, and S0 is the equilibrium strain stimulus (75, 76). Equation 5 can be expanded to correct for the effect of cellular accommodation:

472

Robling

Castillo

Turner

and

= B()( F (, t)), t F (, t) = S0 + ( S0 )(1 e t/ ), ( F ) dF = , dt

(6)

where F takes the form of a relaxation function derived from the differential equation, (7)

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

where is the time constant describing the rate at which accommodation takes place. Cellular accommodation results from the propensity of F to approach with time, thus nulling the driving force for density change. Equation 6 can be used to predict bone density in trabecular bone (76) or bone size and shape in long bone sections (75). Cellular accommodation has been demonstrated in animals using the ulna loading model described above (Figure 6). An experiment was performed in which the ulnae of rats were loaded with progressively decreasing loads or progressively increasing loads. The results showed the greatest changes in bone structural rigidity in groups with progressively decreasing loads (Figure 11), as is predicted by the cellular accommodation algorithm (77). This experiment demonstrates that the greatest inuence on bone formation is the initial loading stimulus, not the long-term accumulation of loading. In another experiment, a 15-week experimental period was divided into three time blocks: weeks 15, weeks 610, and weeks 1115. The rst group of rats was loaded during the rst and third period (with a rest period during the middle ve weeks), while a second group was loaded during all three periods. Despite receiving less mechanical stimulation (10 weeks of loading versus 15 weeks), the rst group of rats exhibited substantially greater biomechanical properties of the ulna than the group loaded for all three periods (78). Therefore, stopping loading for a period of time reverses the effects of cellular accommodation and restores mechanosensitivity.

Disuse and Bone Loss


A child growing in an environment devoid of skeletal loading will develop pencil-like long bones with diminished outer circumference and low bone mineral density (79). In addition, the lack of mechanical usage during growth also impairs the development of a long bones characteristic cross-sectional shape. For instance, an immobilized tibia will develop a fairly round cross section, rather than the typical triangular shape (80). Disuse of bone (low stress) causes reduced bone formation on periosteal surfaces and increased bone resorption and turnover on endocortical and trabecular surfaces. This combination of effects leads to rapid bone loss. Experiments by Uhthoff and Jaworski provided information about effects of disuse on different bone surfaces, how growing and mature skeletons respond differently to disuse, and regional differences in the amount of bone lost with disuse (8183). These experiments involved dogs with one forelimb placed in a spica cast and their bones were examined at various intervals up to 40 weeks after the cast was applied. The casted forelimb underwent profound bone loss resulting from disuse. In growing dogs, the effect of disuse was most apparent at the periosteal (outside) surfaces of long bones, where normal appositional bone

www.annualreviews.org Bone Remodeling

473

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 11 The progressively decreasing load group ( ) underwent ulna loading with a peak force of 13.5 N for the rst ve weeks, 11.3 N for the next ve weeks, and 9.0 N for the nal ve weeks. The progressively increasing load group ( ) received the loads in the opposite order compared with the decreasing load group. The total amount of load applied over the duration of the study was the same for both load groups. Percent change in the minimum second moment of area (Imin ) in each of the groups was recorded. The Imin plane was most affected by mechanical loading of the ulna. Imin was signicantly increased with both loading regimens, but the progressively decreasing load group increased Imin by over fourfold more than the progressively increasing load group. This experiment demonstrates that the greatest inuence on bone formation is the initial loading stimulus. Adapted from Reference 77.

formation is suppressed. This resulted in bones of smaller cross section and a greatly reduced second moment of area (Figure 12). In mature dogs, disuse causes frank bone loss involving accelerated bone resorption and turnover, with loss occurring mainly at the endosteal (inside) surface of long bones. Consequently, deprivation of loading to a bone from an older animal leads to rapid expansion of the marrow cavity (81). There is also a substantial increase in cortical porosity. In both young and mature dogs, bone loss was greatest in the most distal bones of the forelimb, suggesting that skeletal sites closest to the ground lose the most

474

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 12 Cross sections of canine metacarpal bones after 40 weeks of disuse. In bones from young dogs (still undergoing appositional growth; top), periosteal bone formation is suppressed by disuse. In bones from mature adult dogs, disuse caused accelerated bone resorption and remodeling at endocortical (inside) and trabecular (not pictured) surfaces. In addition, cortical porosity is increased. Reprinted from References 83 and 81 with permission from publisher.

bone with disuse (81). Conversely, the osteogenic effects of mechanical loading are greatest in bony regions closest to the ground (i.e., the distal-most aspects of weightbearing limbs) (84). One explanation for the site-specicity of the bone response is that the distal-most aspects of the limb are in closer contact with the ground and thus are loaded more directly. A second compelling explanation is that pressurization of the medullary cavity owing to the effects of gravity enhances appositional bone growth. The distal-most regions of weight-bearing limbs that have the largest anabolic response to exercise and the most severe catabolic response to disuse are also the furthest locations below the heart and are therefore subject to higher interstitial uid pressures. Therefore interstitial uid pressure may be important for maintaining bone mass. This concept is supported by the observation in human volunteers that cranial bones gain bone mass during long-term bed rest (85). Due to the relative change of the gravitational vector when a person lies down, body uids shift

www.annualreviews.org Bone Remodeling

475

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

and regions typically under high uid pressures, like the feet, are now subjected to lower pressure, and sites typically subjected to lower uid pressures, like the head, must adapt to a pressure increase, which in turn may increase bone formation (or decrease bone resorption) in the cranial bones. Bone loss is not equal throughout the skeleton with bed rest (Figure 13). Bone loss is greatest in lower extremities, particularly the calcaneus, where uid pressures decrease substantially with bed rest. It is not surprising that this trend is reversed with exercise. Bone density measurements in active soccer players, compared with nonathletes, show increased bone density at the hip but decreased bone density in the upper skull (86). Therefore, bone mass in the entire skeleton tends to be somewhat preserved because changes in bone mass at weight-bearing sites caused by disuse or exercise are partially offset by counteracting changes in bone mass at skeletal sites that do not bear load.

Overuse, Microdamage, and Bone Remodeling


During normal everyday activities such as walking and running, the skeleton is exposed to cyclic loading at forces well below what is required to break bone. As is true for any structural material exposed to repetitive loading, bone is susceptible to irreversible fatigue damage, resulting in the creation of microscopic cracks. The presence of bone microdamage in vivo was rst described by Frost (87) and has since been reported by several others (8891). Basal levels of microdamage are higher in cancellous than in cortical bone (90, 92), and microdamage increases exponentially with age, but does so more quickly in women than in men (87, 89, 90, 9294). Four types of damage commonly observed in bone are linear microcracks, small localized cracks in a cross-hatched pattern, diffuse damage, and fully fractured trabecular struts (microfractures) (92, 95) (Figure 14). The type of damage created in bone is often dependent on the mode of loading. Studies show that in compression, microcracks are linear and originate at stress risers or discontinuities. They are found in regions of high shear stress and generally run along slip planes that are oblique to the direction of loading, resulting in a crosshatched pattern. In tension, both diffuse damage and longer microcracks running along lamellar interfaces or cement lines are commonly observed (96, 97). In cortical bone, 80%90% of microcracks are contained within interstitial bone or traverse osteonal cement lines (89, 90), whereas the remaining cracks are contained entirely within secondary osteons (Figure 15). This observation is consistent with in vitro studies showing that microcracks are more easily initiated in interstitial bone and grow transverse to the direction of loading until they encounter a natural boundary such as an interlamellar interface or osteon cement line (98). In this regard, fatigue behavior of Haversian bone is similar to the behavior of a brous lamellar composite material in which embedded bers (osteons) impede crack growth. Microdamage is created at physiologic strain levels and strain rates and increases with continued loading (99, 100). This increase in damage is associated with a progressive loss of stiffness and strength as microcracks coalesce, forming larger macrocracks until eventual failure (101, 102). Consistent with these observations, overuse or exercise-induced bone injuries are common among athletes that undergo repetitive

476

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 13 Bone mineral change after 17 weeks of bed rest. Volunteers subjected to bed rest for 17 weeks lost a signicant amount of bone mass in their lower extremities. Bone loss was greatest in the calcaneus, but there was also signicant bone loss in the proximal femur and lumbar spine. Bone loss was not signicant in the arms and ribs. Interestingly, there was signicant gain in bone mass within the head. The pattern of bone mass change reects the coincident changes in interstitial uid pressure brought about by the change from an upright to supine position. Adapted from Reference 85.

www.annualreviews.org Bone Remodeling

477

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 14 Basic fuchsin-stained microdamage in a trabecular bone strut taken from bone core samples of the intertrochanteric region of the human femur. Arrows depict ( from left to right) (a) a linear microcrack, (b) cross-hatching, and (c) diffuse staining. Reprinted from Reference 92 with permission from publisher.

loading of their skeleton, such as military recruits (103), runners (104), and ballet dancers (105). Overuse injuries are often classied as stress or fatigue fractures and occur as a result of microdamage accumulation and loss of bone stiffness and strength. Bone is remarkable in that it has the innate ability to repair fatigue damage through the coordinated process of BMU-based remodeling, whereby osteoclasts remove packets of damaged bone and osteoblasts lay down new bone in its place. In addition, there is strong evidence that microdamage can initiate new remodeling BMUs that then spatially target regions of damage, indicating a repair mechanism that is directed rather than random in nature (Figure 16). Burr and colleagues demonstrated in dogs that new remodeling BMUs are 4 to 6 times more likely to be associated with fatigueinduced microcracks than by chance alone (106, 107). Furthermore, when cortical bone remodeling in dogs was suppressed 50%60% by treatment with bisphosphonates, the amount of microdamage increased ve to sixfold (108). Additional evidence for targeted remodeling was provided by Schafer and colleagues who showed that remodeling can be activated by fatigue-induced microdamage in rat cortical bone, an animal that does not normally undergo large amounts of cortical bone remodeling (109, 110). In these experiments, microdamage was spatially associated with apoptotic osteocytes and resorption cavities, suggesting that damage-induced osteocyte

478

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 15 Basic fuchsin-stained microcracks (indicated by white arrows) in human femoral cortical bone. (a) Microcracks contained within interstitial bone are typical in regions of tensile loading. In tension, continued propagation at the crack tip is impeded by an osteon cement line (left black arrow). (b) Microcracks traversing osteon cement lines (right black arrow) are typical in regions of compressive loading. In compression, propagation across interfaces is easier due to a tensile Poisson stress that tends to open the crack tip. Field width, 500 m. Unpublished images reproduced with permission from R.B. Martin, copyright 2000.

apoptosis may be involved in the recruitment of osteoclasts and initiation of new remodeling BMUs (Figure 17) (111). Human cortical bone can endure 410 million loading cycles at a physiologic peak strain level of 2000 before failure (112), suggesting that damage alone cannot account for the relatively short time (months) in which stress fractures develop in some athletes (104). Consequently, the remodeling process itself is believed to play a signicant role in the etiology of fatigue fractures. That is, while remodeling serves to remove fatigue damage, the associated transient increase in porosity may reduce bone strength and stiffness, thereby creating even more damage and remodeling with continued loading (Figure 18). This behavior can be characterized as a positive feedback control system and has been explored by Martin (113) using a mathematical model. The model examines the relationship between damage, activation frequency, porosity, and strain, and predicts the behavior of these variables under different loading conditions. The model assumes a cylindrical cortex of bone undergoing cyclic compressive loading and examines a representative single cross section of bone over

www.annualreviews.org Bone Remodeling

479

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 16 An interstitial microcrack stained with basic fuchsin (indicated by two arrows) lies in the path of resorbing osteoclasts at the front of a BMU-based cutting cone (indicated by one arrow). The cutting cone appears to be moving in the direction of the microcrack, suggesting targeted removal of microdamage. Unpublished images reproduced with permission from R.B. Martin, copyright 2000.

time. The stress-strain relationship is assumed linear. Damage is dened as total crack length per square millimeter of bone (mm/mm2 or 1/mm). The damage formation rate is proportional to the strain range raised to a power (q ) and to the loading rate (RL , cycles per day), both of which vary depending on activity. Thus, the damage formation rate is DF = kD RL q , (8)

480

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 17 Adult rat cortical bone showing a spatial association between (a) microdamage (Mdx) and TUNEL-stained osteocytes (indicated by arrows), which is a marker for apoptosis, and (b) a resorption cavity (Rs) and TUNEL-stained osteocytes, suggesting a link between damage-induced osteocyte apoptosis and the initiation of new remodeling BMUs. Reprinted from Reference 110 with permission from publisher.

where k D is the damage rate coefcient and depends on bone structure and loading conditions. The damage removal rate is proportional to the fraction of bone area removed by a remodeling BMU: D R = Dfa AO FS , (9)

where D is the amount of damage present, fa is the activation frequency (the number of newly initiated remodeling BMUs per area of bone per day), AO is the area of bone removed by a remodeling BMU, and FS is the damage repair specicity factor, which takes into consideration targeted versus random removal of damage by remodeling. The daily change in damage is DTODAY = DYESTERDAY + D F D R t, (10)

where t = 1 day. The relationship between damage and activation frequency is described by a sigmoidal function, where fa min is the observed minimum activation frequency when the system is in equilibrium, fa max is the maximum allowable activation frequency based on values observed in nature, k R is a coefcient that controls the slope of the curve, and D0 is the amount of damage present when the system is

www.annualreviews.org Bone Remodeling

481

Load

Cortical area

Stress
Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Porosity

Stiffness

Strain

a
Osteocyte-bone living cell network

Remodeling

Modeling

Damage removal

Damage

Damage formation

Figure 18 Physiologic model of fatigue fracture. When a long bone is loaded in axial compression, the load is transmitted through the cortical area to produce stress (stress = force/area). The stress produces strain, whose magnitude is inversely proportional to the elastic modulus or stiffness of the bone. Under normal loading conditions, the creation of damage and the removal of that damage by remodeling are in equilibrium. However, if the damage formation rate suddenly increases, such as during a period of rapid increase in loading frequency and/or magnitude, damage accumulates, which interrupts the normal function of the osteocyte network ( path a), resulting in osteocyte apoptosis. Osteocyte apoptosis then initiates additional remodeling. The associated transient increase in porosity owing to remodeling, and the existing damage itself, leads to a reduction in stiffness and increased strain, creating even more damage ( path b). With continued loading, there is potential for a rapid, uncontrollable and nonlinear increase in damage and remodeling, each feeding off the other until eventual failure.

482

Robling

Castillo

Turner

in equilibrium or when fa = fa min : fa = fa min fa max . fa min + ( fa max fa min ) exp[k R fa max (D D0 )/D0 ] (11)

The daily change in bone porosity is a function of the amount of bone gained and lost over time and is calculated using the number of resorbing (NR ) and relling (NF ) BMUs, which depends on fa , and the rate at which bone is lost (Q R ) or added (Q F ) at the level of the BMU: PTODAY = PYESTERDAY + (NR Q R NF Q F ) t,
Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

(12)

where t = 1 day. The elastic modulus or stiffness of bone varies as a function of porosity: E = E0 (1 P)3 , (13) where E0 is the initial modulus of bone. Resulting strain range () is a function of the applied load (L), the cross sectional area of bone (A) and stiffness (E): = L . AE (14)

Damage is also assumed to initiate periosteal bone apposition, which changes the cross-sectional area of bone. The model can be run with or without this response. All variables can be calculated on a daily basis and graphed as a function of time. When the model is run without a periosteal response at two different loading levels (RL = 109 and 185 cycles per day), all variables increase slowly over time and then rapidly increase, indicating failure. When the model is run with periosteal modeling at three higher loading levels (RL = 185, 225, and 321 cycles per day), damage, activation frequency, porosity, and strain range increase only transiently and rapidly return to their initial levels (Figure 19). Thus, structural changes owing to periosteal modeling can compensate for the decline in material integrity. What is striking about the response of this model is that only one additional loading cycle per day (RL = 322) is required before all variables rapidly increase without limit, indicating the intrinsic instability of the system. Hence there may be a loading threshold above which stress fractures develop rapidly.

Mechanotransduction Pathways
Osteocytes within the bone and lining cells on the bone surface have long been thought to be terminally differentiated osteoblasts that are metabolically inactive and have limited roles in bone biology. However, their abundance and connectivity make them a virtual antenna for detecting mechanical strains. Osteocytes are the most abundant cells in bone, and lining cells are the second most abundant. Together, they outnumber osteoblasts and osteoclasts by more than 20 to 1. Osteocytes are distributed throughout the bone tissue volume and have the ability to communicate with other bone cells through an extensive network of cellular processes connected at gap junctions. Lining cells cover more than 90% of the surfaces of adult bone (17) and they contain gap junction connections to osteocytes and osteoblasts. There is

www.annualreviews.org Bone Remodeling

483

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 19 Generalized nonlinear response of a system modeling the development of stress fractures. Damage (shown here), activation frequency, porosity, and strain all follow similar proles at different loading rates (RL = 185, 225, 321, and 322 cycles per day). With one additional cycle per day the model becomes unstable. Adapted from Reference 113.

experimental evidence in vivo showing that mechanical loads can stimulate responses from osteocytes and bone lining cells (Figure 20). Osteocytes probably do not respond directly to mechanical strain (deformation) of bone tissue, but respond indirectly to extracellular uid ow caused by loading. When cultured osteocytes are subjected to uid shear stress, they release several messengers, including prostaglandins and nitric oxide (111). The network formed by osteocytes is thought to sense local mechanical perturbations, yet cultured osteoblasts respond to mechanical stimuli even in the absence of osteocytes. Osteoblasts also secrete messengers (e.g., prostaglandins and nitric oxide), and express several growth factors. Mechanical stimuli also affect osteoclasts, but this effect appears to be indirect. Loading increases hydrostatic pressure within the bone marrow, which may activate marrow stromal cells to decrease osteoclast differentiation. When exposed to strain, preosteoblastic marrow stromal cells reduce expression of RANK-L, which in turn decreases osteoclast number (114). Consequently, cells of osteoblastic lineage appear to be mediators of the suppressive effects of mechanical stimuli on bone resorption. The mechanical loads applied to a bone generate mechanical signals at the cellular level. The exact nature of those signals is still unknown. The average strain of the bone matrix cannot exceed approximately 0.7% (7000 ) or the bone will be permanently damaged. Normal locomotor strains range from about 5002500 (115, 116). However, these strains represent tissue deformations averaged over an area of

484

Robling

Castillo

Turner

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 20 In situ hybridization for c-fos mRNA in rat vertebral bone. (a) Vertebral cortex hybridized for c-fos mRNA 1 h after applied mechanical loading. (b) Similar region of the vertebral cortex from a nonloaded control vertebra hybridized for c-fos mRNA. Reprinted from Reference 149 with permission from publisher.

several millimeters. Strains at the cellular level might be considerably higher. Strain will be concentrated immediately adjacent to an osteocyte lacunae and, in theory, cells can experience deformations three- to sevenfold greater than the average strain in the tissue. In as much as only dynamic strains generate bone formation in vivo, it seems that the important stimulus likely is tied to rate of applied strain. Rapidly applied strains in bone tissue promote extracellular uid ow and the interaction between this moving uid and bone cells is key to mechanotransduction (117). Because of the strain concentrations at osteocyte lacunae, the lacunae might act as pumps that push uid along the canaliculae when the bone tissue is loaded. Fluid ow along cell bodies or processes produces drag force, uid shear stress, and an electric potential called a streaming potential (or stress-generated potential). Each of these signals might activate bone cells, although cell culture experiments suggest that cells are more sensitive to uid forces than they are to electric potential (118, 119). Most bone surfaces are covered with bone lining cells that dedifferentiate into synthetic osteoblasts within 48 h after a mechanical stimulus (120, 121). About 96 h

Mechanotransduction: the conversion of a physical force into a biochemical/cellular response

www.annualreviews.org Bone Remodeling

485

after a mechanical stimulus, mineralization of new bone matrix is observed (122). Loading stimulates proliferation of osteoblast precursors, which then differentiate into osteoblasts. Mechanical loading also reduces rates of programmed cell death (apoptosis) in osteocytes (123) and probably in active osteoblasts (124). Therefore, loading may extend the period of matrix synthesis for each osteoblast.

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 21 Model for mechanotransduction in bone. Fluid shear on osteocytes (OCY) induces an inux of extracellular Ca2+ via voltage-sensitive (V) and perhaps mechanosensitive (M) channels. Shear stress also enhances ATP release, which binds to the purinergic receptors P2X (ionotropic) and P2Y (metabotropic). Signaling through P2Y is required for Ca2+ release from intracellular stores via a Gq PLC PIP2 IP3 pathway. ATP release causes PGE2 release through signaling downstream of the P2X7 receptor. PGE2 binds and signals through one of the EP receptors, probably EP4 and/or EP2, and ultimately results in enhanced bone formation. PTH signaling also appears to be required for mechanotransduction to occur, but the intracellular pathways involved are not well understood (question mark). Wnt signaling through the Lrp5 receptor, which acts through beta catenin (-cat) translocation to the nucleus, also appears to be important in mechanically induced bone formation. Pressure in the marrow cavity and/or uid shear forces on marrow stromal cells (MSC) may stimulate nitric oxide synthase (NOS) activity and nitric oxide (NO) release. NO is a strong inhibitor of bone resorption and probably acts by inhibiting RANK-L expression, while increasing osteoprotegerin (OPG) production (RANK-L enhances osteoclast differentiation, whereas OPG suppresses this process). OCY = osteocyte; OB = osteoblast; MSC = marrow stromal cell.

486

Robling

Castillo

Turner

Cellular mechanisms. Skeletal mechanotransduction involves ion channels (mechanosensitive and L-type) (125, 126), focal adhesions (72), and a putative G proteincoupled mechanotransducer (127). The ion channels have been identied mostly with fairly general pharmacological blockers, although it appears that the alpha 1C isoform of the L-type calcium channel is most likely one of the channels involved in mechanotransduction (128). Likewise, the key focal adhesion proteins involved in mechanotransduction have not yet been identied. It is possible that extracellular matrix connections to the cytoskeleton act mainly as ampliers of mechanical signals, rather than actual mechanotransducers (129). The ATP receptor

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Figure 22 Mechanical loading requires signaling through the Lrp5 receptor to initiate bone formation. This signal is initiated by Wnt, which binds to the receptor complex made up of Lrp5 and Frizzled. Wnt signaling can be blocked by Dickkopf related protein (Dkk) when it binds Lrp5 and Kremen. Sclerostin (Sclr) inhibits Wnt signaling by binding Lrp5 and secreted Frizzled-related protein (sFrp) binds to Wnt and prevents signaling. Each of these inhibitors blocks mechanotransduction and reduces bone formation.

www.annualreviews.org Bone Remodeling

487

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

P2Y2 is a candidate for the G proteincoupled mechanotransducer. This receptor couples ATP release from bone cells, which occurs within a minute after mechanical stimulation (130), with intracellular calcium signaling (131). The P2X7 receptor is an ATP-gated ion channel that may be a second major signaling pathway for mechanotransduction. P2X7 signaling is directly linked to prostaglandin release in bone cells (132). Prostaglandins and nitric oxide are the important biochemical mediators of mechanical loading in bone. Certain prostaglandins, particularly PGE2 , are anabolic with demonstrated capacity to stimulate osteoblast activity and new bone formation (133). The anabolic effects of PGE2 are mediated through the EP4 prostaglandin receptor (134), suggesting that signaling downstream from EP4 is important in bone matrix synthesis. Nitric oxide is a strong inhibitor of bone resorption, which works in part by suppressing the expression of RANK-L and increases expression of OPG that in turn leads to decreased recruitment of osteoclasts (135) (Figure 21). There are several examples of hormones that might amplify or transduce the effects of mechanical loading; these include parathyroid hormone (PTH) or the 134 PTH fragment (136), estrogen (137), and insulin-like growth factors (138). One key event linking mechanical loading to bone formation is Wnt signaling through the LRP5 receptor (Figure 22). Mice with nonfunctional Lrp5 receptors respond poorly to mechanical loading of the skeleton, with 88%99% reduced bone formation compared with controls (139). These mice are osteopenic and have reduced diameters of long bones. People with an activating mutation of LRP5 have high bone mass and increased bone strength (140). The mutation in the receptor is thought to inhibit the binding of an antagonist called Dickkopf related protein (DKK). A key step in the signaling cascade downstream from the LRP5 receptor is glycogen synthase kinase 3 (GSK3), which must be inactivated before nuclear signaling and gene expression can take place. Drugs that suppress GSK3, and thus simulate the effect of signaling through the LRP5 receptor, increase bone formation (141). In summary, the biological processes involved in bone mechanotransduction remain poorly understood, but several pathways are emerging from current research, including membrane ion channels, ATP signaling, and second messengers such as prostaglandins and nitric oxide. Some key signaling pathways include the P2X7 purinergic receptor, the EP2 or EP4 prostanoid receptors, and the LRP5 Wnt receptor.

SUMMARY POINTS 1. Bone remodeling is continuously active in skeletal tissue through all stages of life, and is managed by teams of osteoclasts and osteoblasts organized into temporary anatomical structures called basic multicellular units (BMUs). 2. Osteoblast and osteoclast proliferation, differentiation, and function are under tight molecular control; osteoclastogenesis is controlled by RANKL/OPG expression in osteoblast-lineage cells. 3. Mechanical stimulation of bone tissue accelerates periosteal bone formation in regions of high stress and effectively strengthens bones.

488

Robling

Castillo

Turner

4. The biochemical response to mechanical stimulation (mechanotransduction) involves a cascade of events including ATP and Ca2+ signaling, prostaglandin and nitric oxide release, and Wnt/Lrp5 signaling, which ultimately leads to new bone formation. 5. Bone tissue response to loading is governed by a set of rules involving the rate of applied strain, cellular desensitization, and accommodation. 6. Disuse results in suppressed periosteal apposition in growing bone and enhanced endocortical resorption in mature bone.
Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

7. Spaceight/weightlessness causes bone loss that may originate from uid pressure changes induced by a general uid shift in the body. 8. Repetitive loading of bone causes fatigue damage, manifest as microscopic cracks within the tissue. Damage stimulates and attracts new remodeling BMUs to replace damaged bone with new, tougher bone.

LITERATURE CITED
1. Frost HM. 1963. Bone Remodelling Dynamics. Springeld, IL: Charles C. Thomas 2. Frost HM. 1986. Intermediary Organization of the Skeleton. Boca Raton, FL: CRC Press 3. Jee WSS, Frost HM. 1992. Skeletal adpatations during growth. Triangle: Sandoz J. Med. Sci. 31:7788 4. Frost HM. 1997. Obesity, and bone strength and mass: a tutorial based on insights from a new paradigm. Bone 21:21114 5. Hillam RA, Skerry TM. 1995. Inhibition of bone resorption and stimulation of formation by mechanical loading of the modeling rat ulna in vivo. J. Bone Miner. Res. 10:68389 6. Frost HM. 1973. Bone Modeling and Skeletal Modeling Errors. Springeld, IL: Charles C. Thomas 7. Garn SM. 1970. The Earlier Gain and the Later Loss of Cortical Bone. Springeld, IL: Charles C. Thomas 8. Lazenby RA. 1990. Continuing periosteal apposition. II. The signicance of peak bone mass, strain equilibrium, and age-related activity differentials for mechanical compensation in human tubular bones. Am. J. Phys. Anthropol. 82:473 84 9. Lazenby RA. 1990. Continuing periosteal apposition. I. Documentation, hypotheses, and interpretation. Am. J. Phys. Anthropol. 82:45172 10. Partt AM. 1979. Quantum concept of bone remodeling and turnover: implications for the pathogenesis of osteoporosis. Calcif. Tissue Int. 28:15 11. Hert J, Fiala P, Petrtyl M. 1994. Osteon orientation of the diaphysis of the long bones in man. Bone 15:26977

www.annualreviews.org Bone Remodeling

489

12. Partt AM. 1994. Osteonal and hemi-osteonal remodeling: the spatial and temporal framework for signal trafc in adult human bone. J. Cell. Biochem. 55:273 86 13. Eriksen EF, Langdahl B. 1995. Bone remodeling and its consequences for bone structure. In Bone Structure and Remodeling, ed. A Odgaard, H Weinans, pp. 2536. London: World Sci. 14. Schafer MB, Burr DB, Frederickson RG. 1987. Morphology of the osteonal cement line in human bone. Anat. Rec. 217:22328 15. Partt AM. 2000. The mechanism of coupling: a role for the vasculature. Bone 26:31923 16. Eriksen EF, Mosekilde L, Melsen F. 1985. Effect of sodium uoride, calcium, phosphate, and vitamin D2 on trabecular bone balance and remodeling in osteoporotics. Bone 6:38189 17. Partt AM. 1983. The physiologic and clinical signicance of bone histomorphomstric data. In Bone Histomorphometry: Techniques and Interpretations, ed. RR Recker, pp. 14323. Boca Raton, FL: CRC Press 18. Eriksen EF, Mosekilde L, Melsen F. 1986. Trabecular bone remodeling and balance in primary hyperparathyroidism. Bone 7:21321 19. Tondravi MM, McKercher SR, Anderson K, Erdmann JM, Quiroz M, et al. 1997. Osteopetrosis in mice lacking haematopoietic transcription factor PU.1. Nature 386:8184 20. Yoshida H, Hayashi S, Kunisada T, Ogawa M, Nishikawa S, et al. 1990. The murine mutation osteopetrosis is in the coding region of the macrophage colony stimulating factor gene. Nature 345:44244 21. Franzoso G, Carlson L, Xing L, Poljak L, Shores EW, et al. 1997. Requirement for NF-kappaB in osteoclast and B-cell development. Genes Dev. 11:348296 22. Matsuo K, Galson DL, Zhao C, Peng L, Laplace C, et al. 2004. Nuclear factor of activated T-cells (NFAT) rescues osteoclastogenesis in precursors lacking c-Fos. J. Biol. Chem. 279:2647580 23. Wang ZQ, Ovitt C, Grigoriadis AE, Mohle-Steinlein U, Ruther U, Wagner EF. 1992. Bone and haematopoietic defects in mice lacking c-fos. Nature 360:74145 24. Partington GA, Fuller K, Chambers TJ, Pondel M. 2004. Mitf-PU.1 interactions with the tartrate-resistant acid phosphatase gene promoter during osteoclast differentiation. Bone 34:23745 25. So H, Rho J, Jeong D, Park R, Fisher DE, et al. 2003. Microphthalmia transcription factor and PU.1 synergistically induce the leukocyte receptor osteoclastassociated receptor gene expression. J. Biol. Chem. 278:2420916 26. Takayanagi H, Kim S, Koga T, Nishina H, Isshiki M, et al. 2002. Induction and activation of the transcription factor NFATc1 (NFAT2) integrate RANKL signaling in terminal differentiation of osteoclasts. Dev. Cell 3:889901 27. McHugh KP, Hodivala-Dilke K, Zheng MH, Namba N, Lam J, et al. 2000. Mice lacking beta3 integrins are osteosclerotic because of dysfunctional osteoclasts. J. Clin. Invest. 105:43340 28. Vaananen K. 1996. Osteoclast function: biology and mechanisms. In Principles of Bone Biology, ed. JP Bilezikian, LG Raisz, GA Rodan, pp. 10313. New York: Academic

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

490

Robling

Castillo

Turner

29. Bendall AJ, Abate-Shen C. 2000. Roles for Msx and Dlx homeoproteins in vertebrate development. Gene 247:1731 30. Ducy P, Zhang R, Geoffroy V, Ridall AL, Karsenty G. 1997. Osf2/Cbfa1: a transcriptional activator of osteoblast differentiation. Cell 89:74754 31. Komori T, Yagi H, Nomura S, Yamaguchi A, Sasaki K, et al. 1997. Targeted disruption of Cbfa1 results in a complete lack of bone formation owing to maturational arrest of osteoblasts. Cell 89:75564 32. Otto F, Thornell AP, Crompton T, Denzel A, Gilmour KC, et al. 1997. Cbfa1, a candidate gene for cleidocranial dysplasia syndrome, is essential for osteoblast differentiation and bone development. Cell 89:76571 33. Robledo RF, Rajan L, Li X, Lufkin T. 2002. The Dlx5 and Dlx6 homeobox genes are essential for craniofacial, axial, and appendicular skeletal development. Genes Dev. 16:1089101 34. Glass DA 2nd, Bialek P, Ahn JD, Starbuck M, Patel MS, et al. 2005. Canonical Wnt signaling in differentiated osteoblasts controls osteoclast differentiation. Dev. Cell 8:75164 35. Hu H, Hilton MJ, Tu X, Yu K, Ornitz DM, Long F. 2005. Sequential roles of Hedgehog and Wnt signaling in osteoblast development. Development 132:49 60 36. Nakashima K, Zhou X, Kunkel G, Zhang Z, Deng JM, et al. 2002. The novel zinc nger-containing transcription factor osterix is required for osteoblast differentiation and bone formation. Cell 108:1729 37. Wetterwald A, Hoffstetter W, Cecchini MG, Lanske B, Wagner C, et al. 1996. Characterization and cloning of the E11 antigen, a marker expressed by rat osteoblasts and osteocytes. Bone 18:12532 38. Nijweide PJ, Burger EH, Klein Nulend J, Van der Plas A. 1996. The osteocyte. In Principles of Bone Biology, ed. JP Bilezikian, LG Raisz, GA Rodan, pp. 11526. New York: Academic 39. Feng JQ, Huang H, Lu Y, Ye L, Xie Y, et al. 2003. The Dentin matrix protein 1 (Dmp1) is specically expressed in mineralized, but not soft, tissues during development. J. Dent. Res. 82:77680 40. Nampei A, Hashimoto J, Hayashida K, Tsuboi H, Shi K, et al. 2004. Matrix extracellular phosphoglycoprotein (MEPE) is highly expressed in osteocytes in human bone. J. Bone Miner. Metab. 22:17684 41. Toyosawa S, Shintani S, Fujiwara T, Ooshima T, Sato A, et al. 2001. Dentin matrix protein 1 is predominantly expressed in chicken and rat osteocytes but not in osteoblasts. J. Bone Miner. Res. 16:201726 42. van Bezooijen RL, Roelen BA, Visser A, van der Wee-Pals L, de Wilt E, et al. 2004. Sclerostin is an osteocyte-expressed negative regulator of bone formation, but not a classical BMP antagonist. J. Exp. Med. 199:80514 43. van Bezooijen RL, ten Dijke P, Papapoulos SE, Lowik CW. 2005. SOST/sclerostin, an osteocyte-derived negative regulator of bone formation. Cytokine Growth Factor Rev. 16:31927 44. Rodan GA, Martin TJ. 1981. Role of osteoblasts in hormonal control of bone resorptiona hypothesis. Calcif. Tissue Int. 33:34951

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

www.annualreviews.org Bone Remodeling

491

45. Jimi E, Nakamura I, Amano H, Taguchi Y, Tsurukai T, et al. 1996. Osteoclast function is activated by osteoblastic cells through a mechanism involving cellto-cell contact. Endocrinology 137:218790 46. Takahashi N, Akatsu T, Udagawa N, Sasaki T, Yamaguchi A, et al. 1988. Osteoblastic cells are involved in osteoclast formation. Endocrinology 123:26002 47. Anderson DM, Maraskovsky E, Billingsley WL, Dougall WC, Tometsko ME, et al. 1997. A homologue of the TNF receptor and its ligand enhance T-cell growth and dendritic-cell function. Nature 390:17579 48. Yasuda H, Shima N, Nakagawa N, Mochizuki SI, Yano K, et al. 1998. Identity of osteoclastogenesis inhibitory factor (OCIF) and osteoprotegerin (OPG): a mechanism by which OPG/OCIF inhibits osteoclastogenesis in vitro. Endocrinology 139:132937 49. Simonet WS, Lacey DL, Dunstan CR, Kelley M, Chang MS, et al. 1997. Osteoprotegerin: a novel secreted protein involved in the regulation of bone density. Cell 89:30919 50. Yasuda H, Shima N, Nakagawa N, Yamaguchi K, Kinosaki M, et al. 1998. Osteoclast differentiation factor is a ligand for osteoprotegerin/osteoclastogenesisinhibitory factor and is identical to TRANCE/RANKL. Proc. Natl. Acad. Sci. USA 95:3597602 51. Martin TJ. 2004. Paracrine regulation of osteoclast formation and activity: milestones in discovery. J. Musculoskelet. Neuronal Interact. 4:24353 52. Noble BS, Peet N, Stevens HY, Brabbs A, Mosley JR, et al. 2003. Mechanical loading: biphasic osteocyte survival and targeting of osteoclasts for bone destruction in rat cortical bone. Am. J. Physiol. Cell Physiol. 284:C93443 53. Mattheck C. 1997. Design in Nature: Learning from Trees. Berlin: Springer 54. Robling AG, Hinant FM, Burr DB, Turner CH. 2002. Improved bone structure and strength after long-term mechanical loading is greatest if loading is separated into short bouts. J. Bone Miner. Res. 17:154554 55. Wolff J. 1892. Das Gesetz der Transformation der Knochen. Berlin: Hirschwald 56. Riggs CM, Lanyon LE, Boyde A. 1993. Functional associations between collagen bre orientation and locomotor strain direction in cortical bone of the equine radius. Anat. Embryol. 187:23138 57. Ascenzi A, Bonucci E. 1967. The tensile properties of single osteons. Anat. Rec. 158:37586 58. Ascenzi A, Bonucci E. 1968. The compressive properties of single osteons. Anat. Rec. 161:37791 59. Takano Y, Turner CH, Owan I, Martin RB, Lau ST, et al. 1999. Elastic anisotropy and collagen orientation of osteonal bone are dependent on the mechanical strain distribution. J. Orthop. Res. 17:5966 60. Liskova M, Hert J. 1971. Reaction of bone to mechanical stimuli. 2. Periosteal and endosteal reaction of tibial diaphysis in rabbit to intermittent loading. Folia Morphol. (Praha) 19:30117 61. Lanyon LE, Rubin CT. 1984. Static vs dynamic loads as an inuence on bone remodelling. J. Biomech. 17:897905 62. Turner CH, Forwood MR, Otter MW. 1994. Mechanotransduction in bone: do bone cells act as sensors of uid ow? FASEB J. 8:87578

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

492

Robling

Castillo

Turner

63. Turner CH, Owan I, Takano Y. 1995. Mechanotransduction in bone: role of strain rate. Am. J. Physiol. Endocrinol. Metab. 269:E43842 64. Hsieh YF, Turner CH. 2001. Effects of loading frequency on mechanically induced bone formation. J. Bone Miner. Res. 16:91824 65. Rubin C, Turner AS, Bain S, Mallinckrodt C, McLeod K. 2001. Anabolism. Low mechanical signals strengthen long bones. Nature 412:6034 66. Warden SJ, Turner CH. 2004. Mechanotransduction in the cortical bone is most efcient at loading frequencies of 510 Hz. Bone 34:26170 67. Rubin C, Xu G, Judex S. 2001. The anabolic activity of bone tissue, suppressed by disuse, is normalized by brief exposure to extremely low-magnitude mechanical stimuli. FASEB J. 15:222529 68. Robling AG, Burr DB, Turner CH. 2001. Recovery periods restore mechanosensitivity to dynamically loaded bone. J. Exp. Biol. 204:338999 69. Srinivasan S, Weimer DA, Agans SC, Bain SD, Gross TS. 2002. Low-magnitude mechanical loading becomes osteogenic when rest is inserted between each load cycle. J. Bone Miner. Res. 17:161320 70. Carter DR, Fyhrie DP, Whalen RT. 1987. Trabecular bone density and loading history: regulation of connective tissue biology by mechanical energy. J. Biomech. 20:78594 71. Bertram JE, Biewener AA. 1988. Bone curvature: sacricing strength for load predictability? J. Theor. Biol. 131:7592 72. Pavalko FM, Chen NX, Turner CH, Burr DB, Atkinson S, et al. 1998. Fluid shear-induced mechanical signaling in MC3T3-E1 osteoblasts requires cytoskeleton-integrin interactions. Am. J. Physiol. Cell Physiol. 275:C1591601 73. Rubin C, Judex S, Hadjiargyrou M. 2002. Skeletal adaptation to mechanical stimuli in the absence of formation or resorption of bone. J. Musculoskelet. Neuronal Interact. 2:26467 74. McKee MD, Nanci A. 1996. Osteopontin at mineralized tissue interfaces in bone, teeth, and osseointegrated implants: ultrastructural distribution and implications for mineralized tissue formation, turnover, and repair. Microsc. Res. Tech. 33:14164 75. Cowin SC, Hegedus DH. 1976. Bone remodeling I: theory of adaptive elasticity. J. Elasticity 6:31326 76. Fyhrie DP, Schafer MB. 1995. The adaptation of bone apparent density to applied load. J. Biomech. 28:13546 77. Schriefer JL, Warden SJ, Saxon LK, Robling AG, Turner CH. 2005. Cellular accommodation and the response of bone to mechanical loading. J. Biomech. 38:183845 78. Saxon LK, Robling AG, Alam I, Turner CH. 2005. Mechanosensitivity of the rat skeleton decreases after a long period of loading, but is improved with time off. Bone 36:45464 79. Rodriguez JI, Palacios J, Garcia-Alix A, Pastor I, Paniagua R. 1988. Effects of immobilization on fetal bone development. A morphometric study in newborns with congenital neuromuscular diseases with intrauterine onset. Calcif. Tissue Int. 43:33539

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

www.annualreviews.org Bone Remodeling

493

80. Lanyon LE. 1980. The inuence of function on the development of bone curvature: an experimental study on the rat tibia. J. Zool. 192:45766 81. Jaworski ZF, Liskova-Kiar M, Uhthoff, HK. 1980. Effect of long-term immobilisation on the pattern of bone loss in older dogs. J. Bone Joint Surg. Br. 62B:10410 82. Jaworski ZF, Uhthoff HK. 1986. Reversibility of nontraumatic disuse osteoporosis during its active phase. Bone 7:43139 83. Uhthoff HK, Jaworski ZF. 1978. Bone loss in response to long-term immobilisation. J. Bone Joint Surg. Br. 60B:42029 84. Turner CH. 1999. Site-specic skeletal effects of exercise: importance of interstitial uid pressure. Bone 24:16162 85. Leblanc AD, Schneider VS, Evans HJ, Engelbretson DA, Krebs JM. 1990. Bone mineral loss and recovery after 17 weeks of bed rest. J. Bone Miner. Res. 5:843 50 86. Magnusson H, Linden C, Karlsson C, Obrant KJ, Karlsson MK. 2001. Exercise may induce reversible low bone mass in unloaded and high bone mass in weightloaded skeletal regions. Osteoporos. Int. 12:95055 87. Frost HM. 1960. Presence of microscpoic cracks in vivo in bone. Henry Ford Hosp. Med. Bull. 8:2535 88. Devas M. 1975. Stress Fractures. New York: Churchill Livingston. 89. Norman TL, Wang Z. 1997. Microdamage of human cortical bone: incidence and morphology in long bones. Bone 20:37579 90. Schafer MB, Choi K, Milgrom C. 1995. Aging and matrix microdamage accumulation in human compact bone. Bone 17:52125 91. Wenzel TE, Schafer MB, Fyhrie DP. 1996. In vivo trabecular microcracks in human vertebral bone. Bone 19:8995 92. Fazzalari NL, Forwood MR, Smith K, Manthey BA, Herreen P. 1998. Assessment of cancellous bone quality in severe osteoarthrosis: bone mineral density, mechanics, and microdamage. Bone 22:38188 93. Evans FG. 1976. Mechanical properties and histology of cortical bone from younger and older men. Anat. Rec. 185:111 94. Smith AB, Smith DA. 1976. Relations between age, mineral density, and mechanical properties of human femoral compacta. Acta. Orthop. Scand. 47:496 502 95. Koszyca B, Fazzalari NL, Vernon-Roberts B. 1989. Trabecular microfractures. Nature and distribution in the proximal femur. Clin. Orthop. Relat. Res. 244:208 16 96. Boyce TM, Fyhrie DP, Glotkowski MC, Radin EL, Schafer MB. 1998. Damage type and strain mode associations in human compact bone bending fatigue. J. Orthop. Res. 16:32229 97. Carter DR, Hayes WC. 1977. Compact bone fatigue damageI. Residual strength and stiffness. J. Biomech. 10:32537 98. Grifn LV, Gibeling JC, Martin RB, Gibson VA, Stover SM. 1997. Model of exural fatigue damage accumulation for cortical bone. J. Orthop. Res. 15:607 14

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

494

Robling

Castillo

Turner

99. Carter DR, Caler WE, Spengler DM, Frankel VH. 1981. Fatigue behavior of adult cortical bone: the inuence of mean strain and strain range. Acta Orthop. Scand. 52:48190 100. Schafer MB, Radin EL, Burr DB. 1989. Mechanical and morphological effects of strain rate on fatigue of compact bone. Bone 10:20714 101. Burr DB, Turner CH, Naick P, Forwood MR, Ambrosius W, et al. 1998. Does microdamage accumulation affect the mechanical properties of bone? J. Biomech. 31:33745 102. Carter DR, Hayes WC. 1977. Compact bone fatigue damage: a microscopic examination. Clin. Orthop. Relat. Res. 127:26574 103. Milgrom C, Giladi M, Stein M, Kashtan H, Margulies JY, et al. 1985. Stress fractures in military recruits. A prospective study showing an unusually high incidence. J. Bone Joint Surg. Br. 67:73235 104. Matheson GO, Clement DB, McKenzie DC, Taunton JE, Lloyd-Smith DR, MacIntyre JG. 1987. Stress fractures in athletes. A study of 320 cases. Am. J. Sports Med. 15:4658 105. Khan K, Brown J, Way S, Vass N, Crichton K, et al. 1995. Overuse injuries in classical ballet. Sports Med. 19:34157 106. Burr DB, Martin RB. 1993. Calculating the probability that microcracks initiate resorption spaces. J. Biomech. 26:61316 107. Burr DB, Martin RB, Schafer MB, Radin EL. 1985. Bone remodeling in response to in vivo fatigue microdamage. J. Biomech. 18:189200 108. Mashiba T, Hirano T, Turner CH, Forwood MR, Johnston CC, Burr DB. 2000. Suppressed bone turnover by bisphosphonates increases microdamage accumulation and reduces some biomechanical properties in dog rib. J. Bone Miner. Res. 15:61320 109. Bentolila V, Boyce TM, Fyhrie DP, Drumb R, Skerry TM, Schafer MB. 1998. Intracortical remodeling in adult rat long bones after fatigue loading. Bone 23:27581 110. Verborgt O, Gibson GJ, Schafer MB. 2000. Loss of osteocyte integrity in association with microdamage and bone remodeling after fatigue in vivo. J. Bone Miner. Res. 15:6067 111. Burger EH, Klein-Nulend J. 1999. Mechanotransduction in bonerole of the lacuno-canalicular network. FASEB J. 13(Suppl.):S10112 112. Pattin CA, Caler WE, Carter DR. 1996. Cyclic mechanical property degradation during fatigue loading of cortical bone. J. Biomech. 29:6979 113. Martin RB. 2001. The role of bone remodeling in preventing or promoting stress fracture. In Musculoskeletal Fatigue and Stress Fractures, ed. D Burr, C Milgrom, pp. 183201. Boca Raton, FL: CRC Press 114. Rubin J, Murphy T, Nanes MS, Fan X. 2000. Mechanical strain inhibits expression of osteoclast differentiation factor by murine stromal cells. Am. J. Physiol. Cell Physiol. 278:C112632 115. Burr DB, Milgrom C, Fyhrie D, Forwood M, Nyska M, et al. 1996. In vivo measurement of human tibial strains during vigorous activity. Bone 18:40510

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

www.annualreviews.org Bone Remodeling

495

116. Rubin CT, Lanyon LE. 1982. Limb mechanics as a function of speed and gait: a study of functional strains in the radius and tibia of horse and dog. J. Exp. Biol. 101:187211 117. Weinbaum S, Cowin SC, Zeng Y. 1994. A model for the excitation of osteocytes by mechanical loading-induced bone uid shear stresses. J. Biomech. 27:33960 118. Hung CT, Allen FD, Pollack SR, Brighton CT. 1996. What is the role of the convective current density in the real-time calcium response of cultured bone cells to uid ow? J. Biomech. 29:14039 119. Reich KM, Gay CV, Frangos JA. 1990. Fluid shear stress as a mediator of osteoblast cyclic adenosine monophosphate production. J. Cell Physiol. 143:100 4 120. Chow JW, Wilson AJ, Chambers TJ, Fox SW. 1998. Mechanical loading stimulates bone formation by reactivation of bone lining cells in 13-week-old rats. J. Bone Miner. Res. 13:176067 121. Turner CH, Owan I, Alvey T, Hulman J, Hock JM. 1998. Recruitment and proliferative responses of osteoblasts after mechanical loading in vivo determined using sustained-release bromodeoxyuridine. Bone 22:46369 122. Forwood MR, Owan I, Takano Y, Turner CH. 1996. Increased bone formation in rat tibiae after a single short period of dynamic loading in vivo. Am. J. Physiol. Endocrinol. Metab. 270:E41923 123. Noble BS, Reeve J. 2000. Osteocyte function, osteocyte death and bone fracture resistance. Mol. Cell Endocrinol. 159:713 124. Pavalko FM, Gerard RL, Ponik SM, Gallagher PJ, Jin Y, Norvell SM. 2003. Fluid shear stress inhibits TNF-alpha-induced apoptosis in osteoblasts: a role for uid shear stress-induced activation of PI3-kinase and inhibition of caspase-3. J. Cell Physiol. 194:194205 125. Li J, Duncan RL, Burr DB, Gattone VH, Turner CH. 2003. Parathyroid hormone enhances mechanically induced bone formation, possibly involving Ltype voltage-sensitive calcium channels. Endocrinology 144:122633 126. Rawlinson SC, Pitsillides AA, Lanyon LE. 1996. Involvement of different ion channels in osteoblasts and osteocytes early responses to mechanical strain. Bone 19:60914 127. Reich KM, McAllister TN, Gudi S, Frangos JA. 1997. Activation of G proteins mediates ow-induced prostaglandin E2 production in osteoblasts. Endocrinology 138:101418 128. Miyauchi A, Notoya K, Mikuni-Takagaki Y, Takagi Y, Goto M, et al. 2000. Parathyroid hormone-activated volume-sensitive calcium inux pathways in mechanically loaded osteocytes. J. Biol. Chem. 275:333542 129. Weinbaum S, Guo P, You L. 2001. A new view of mechanotransduction and strain amplication in cells with microvilli and cell processes. Biorheology 38:119 42 130. Genetos DC, Geist DJ, Liu D, Donahue HJ, Duncan RL. 2005. Fluid shearinduced ATP secretion mediates prostaglandin release in MC3T3-E1 osteoblasts. J. Bone Miner. Res. 20:4149 131. Jorgensen NR, Geist ST, Civitelli R, Steinberg TH. 1997. ATP- and gap

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

496

Robling

Castillo

Turner

132.

133.

134.

135.

136.

137. 138.

139.

140.

141.

142.

143.

144.

145. 146.

junction-dependent intercellular calcium signaling in osteoblastic cells. J. Cell Biol. 139:497506 Li J, Liu D, Ke HZ, Duncan RL, Turner CH. 2004. Osteogenesis after mechanical loading requires the P2X7 nucleotide receptor. J. Bone Miner. Res. 19(Suppl. 1):S9 Jee WS, Mori S, Li XJ, Chan S. 1990. Prostaglandin E2 enhances cortical bone mass and activates intracortical bone remodeling in intact and ovariectomized female rats. Bone 11:25366 Machwate M, Harada S, Leu CT, Seedor G, Labelle M, et al. 2001. Prostaglandin receptor EP(4) mediates the bone anabolic effects of PGE(2). Mol. Pharmacol. 60:3641 Fan X, Roy E, Zhu L, Murphy TC, Ackert-Bicknell C, et al. 2004. Nitric oxide regulates receptor activator of nuclear factor-kappaB ligand and osteoprotegerin expression in bone marrow stromal cells. Endocrinology 145:75159 Ma Y, Jee WS, Yuan Z, Wei W, Chen H, et al. 1999. Parathyroid hormone and mechanical usage have a synergistic effect in rat tibial diaphyseal cortical bone. J. Bone Miner. Res. 14:43948 Lee K, Jessop H, Suswillo R, Zaman G, Lanyon L. 2003. Endocrinology: bone adaptation requires oestrogen receptor-alpha. Nature 424:389 Gross TS, Srinivasan S, Liu CC, Clemens TL, Bain SD. 2002. Noninvasive loading of the murine tibia: an in vivo model for the study of mechanotransduction. J. Bone Miner. Res. 17:493501 Sawakami K, Robling AG, Pitner ND, Warden SJ, Li J, et al. 2004. Sitespecic osteopenia and decreased mechanoreactivity in Lrp5-mutant mice. J. Bone Miner. Res. 19(Suppl. 1):S38 Little RD, Carulli JP, Del Mastro RG, Dupuis J, Osborne M, et al. 2002. A mutation in the LDL receptor-related protein 5 gene results in the autosomal dominant high-bone-mass trait. Am. J. Hum. Genet. 70:1119 Kulkarni NH, Liu M, Halladay DL, Frolik CA, Engler TA, et al. 2004. An orally bioavailable Gsk3 a/b dual inhibitor increases markers of cellular differentiation in vitro and bone mass in vivo. J. Bone Miner. Res. 19(Suppl. 1):S57 Lean JM, Matsuo K, Fox SW, Fuller K, Gibson FM, et al. 2000. Osteoclast lineage commitment of bone marrow precursors through expression of membranebound TRANCE. Bone 27:2940 Takahashi N, Udagawa N, Tanaka S, Murakami H, Owan I, et al. 1994. Postmitotic osteoclast precursors are mononuclear cells which express macrophageassociated phenotypes. Dev. Biol. 163:21221 Wani MR, Fuller K, Kim NS, Choi Y, Chambers T. 1999. Prostaglandin E2 cooperates with TRANCE in osteoclast induction from hemopoietic precursors: synergistic activation of differentiation, cell spreading, and fusion. Endocrinology 140:192735 Pauwels F. 1980. Biomechanics of the Locomotor Apparatus. Berlin: Springer-Verlag Umemura Y, Ishiko T, Yamauchi T, Kurono M, Mashiko S. 1997. Five jumps per day increase bone mass and breaking force in rats. J. Bone Miner. Res. 12:148085

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

www.annualreviews.org Bone Remodeling

497

147. Rubin CT, Lanyon LE. 1984. Regulation of bone formation by applied dynamic loads. J. Bone Joint Surg. Am. 66:397402 148. Burr DB, Robling AG, Turner CH. 2002. Effects of biomechanical stress on bones in animals. Bone 30:78186 149. Lean JM, Mackay AG, Chow JW, Chambers TJ. 1996. Osteocytic expression of mRNA for c-fos and IGF-I: an immediate early gene response to an osteogenic stimulus. Am. J. Physiol. Endocrinol. Metab. 270:E93745

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

498

Robling

Castillo

Turner

Contents
Fluorescence Molecular Imaging Vasilis Ntziachristos p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 1 Multimodality In Vivo Imaging Systems: Twice the Power or Double the Trouble? Simon R. Cherry p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p35 Bioimpedance Tomography (Electrical Impedance Tomography) R.H. Bayford p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p63 Analysis of Inammation Geert W. Schmid-Schnbein p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p93 Drug-Eluting Bioresorbable Stents for Various Applications Meital Zilberman and Robert C. Eberhart p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 153 Glycomics Approach to Structure-Function Relationships of Glycosaminoglycans Ram Sasisekharan, Rahul Raman, and Vikas Prabhakar p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 181 Mathematical Modeling of Tumor-Induced Angiogenesis M.A.J. Chaplain, S.R. McDougall, and A.R.A. Anderson p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 233 Mechanism and Dynamics of Cadherin Adhesion Deborah Leckband and Anil Prakasam p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 259 Microvascular Perspective of Oxygen-Carrying and -Noncarrying Blood Substitutes Marcos Intaglietta, Pedro Cabrales, and Amy G. Tsai p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 289 Polymersomes Dennis E. Discher and Fariyal Ahmed p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 323 Recent Approaches to Intracellular Delivery of Drugs and DNA and Organelle Targeting Vladimir P. Torchilin p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 343 Running Interference: Prospects and Obstacles to Using Small Interfering RNAs as Small Molecule Drugs Derek M. Dykxhoorn and Judy Lieberman p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 377

Annual Review of Biomedical Engineering Volume 8, 2006

Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Stress Protein Expression Kinetics Kenneth R. Diller p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 403 Electrical Forces for Microscale Cell Manipulation Joel Voldman p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 425 Biomechanical and Molecular Regulation of Bone Remodeling Alexander G. Robling, Alesha B. Castillo, and Charles H. Turner p p p p p p p p p p p p p p p p p p p p p p p 455 Biomechanical Considerations in the Design of Graft: The Homeostasis Hypothesis Ghassan S. Kassab and Jos A. Navia p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 499
Annu. Rev. Biomed. Eng. 2006.8:455-498. Downloaded from arjournals.annualreviews.org by Indiana University - Indianapolis on 10/08/07. For personal use only.

Machine Learning for Detection and Diagnosis of Disease Paul Sajda p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 537 Prognosis in Critical Care Lucila Ohno-Machado, Frederic S. Resnic, and Michael E. Matheny p p p p p p p p p p p p p p p p p p p p 567 Lab on a CD Marc Madou, Jim Zoval, Guangyao Jia, Horacio Kido, Jitae Kim, and Nahui Kim p p p 601 INDEXES Subject Index p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 629 Cumulative Index of Contributing Authors, Volumes 18 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 643 Cumulative Index of Chapter Titles, Volumes 18 p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p p 646 ERRATA An online log of corrections to Annual Review of Biomedical Engineering chapters (if any, 1977 to the present) may be found at http://bioeng.annualreviews.org/

vi

Contents

Anda mungkin juga menyukai