From Estrogen-Centric to Aging and Oxidative Stress: A Revised Perspective of the Pathogenesis of Osteoporosis

Abstract

Estrogen deficiency has been considered the seminal mechanism of osteoporosis in both women and men, but epidemiological evidence in humans and recent mechanistic studies in rodents indicate that aging and the associated increase in reactive oxygen species (ROS) are the proximal culprits. ROS greatly influence the generation and survival of osteoclasts, osteoblasts, and osteocytes. Moreover, oxidative defense by the FoxO transcription factors is indispensable for skeletal homeostasis at any age. Loss of estrogens or androgens decreases defense against oxidative stress in bone, and this accounts for the increased bone resorption associated with the acute loss of these hormones. ROS-activated FoxOs in early mesenchymal progenitors also divert ß-catenin away from Wnt signaling, leading to decreased osteoblastogenesis. This latter mechanism may be implicated in the pathogenesis of type 1 and 2 diabetes and ROS-mediated adverse effects of diabetes on bone formation. Attenuation of Wnt signaling by the activation of peroxisome proliferator-activated receptor γ by ligands generated from lipid oxidation also contributes to the age-dependent decrease in bone formation, suggesting a mechanistic explanation for the link between atherosclerosis and osteoporosis. Additionally, increased glucocorticoid production and sensitivity with advancing age decrease skeletal hydration and thereby increase skeletal fragility by attenuating the volume of the bone vasculature and interstitial fluid. This emerging evidence provides a paradigm shift from the “estrogen-centric” account of the pathogenesis of involutional osteoporosis to one in which age-related mechanisms intrinsic to bone and oxidative stress are protagonists and age-related changes in other organs and tissues, such as ovaries, accentuate them.

Increased oxidative stress is strongly implicated in the biology of aging and the pathogenesis of age-related diseases. Recent evidence indicates that oxidative stress is also a fundamental mechanism of the age-dependant decline of bone mass and strength and that loss of estrogens exaggerates the effects of aging on bone by decreasing defense against oxidative stress. Moreover, the balance between the generation of reactive oxygen species versus defense against them by FoxO-activated transcription programs is critical for bone homeostasis throughout life. Attenuation of Wnt signaling by PPARγ activation by oxidized lipids and an increase in endogenous glucocorticoids with age are two additional mechanisms contributing to skeletal involution. This new knowledge provides a paradigm shift from the traditional “estrogen-centric” account of the pathogenesis of osteoporosis to one in which aging per se is inexorably the protagonist.

• I. Introduction

• II. The Traditional Estrogen-Centric Perspective of the Pathogenesis of Osteoporosis

• III. Aging as a Pivotal Determinant of Loss of Bone Mass and Strength

• IV. Aging and Oxidative Stress

• V. Defense Mechanisms against Oxidative Stress

  • A. Enzymatic
  • B. The FoxO transcription factors
  • C. β-Catenin as a pivot in the regulation of oxidative stress-induced transcription programs

• VI. Organismal Aging, Oxidative Stress, and Skeletal Homeostasis

• VII. The Antiosteoporotic Effects of Estrogens and their Antioxidant Properties

  • A. ROS, estrogens, and osteoblasts
  • B. ROS, estrogens, and the generation and apoptosis of osteoclasts

• VIII. Diabetes, Oxidative Stress, and Osteoporosis

• IX. Lipid Oxidation, Oxidative Stress, PPARγ, and the Link between Osteoporosis and Atherosclerosis

• X. Aging, Endogenous Hyperglucocorticoidism, and Bone Strength

  • A. Cell autonomous effects of glucocorticoids on bone
  • B. Glucocorticoids and bone strength
  • C. Angiogenesis and bone
  • D. Bone hydration

• XI. Oxidative Stress and Nutrient-Dependent Deacetylases (Sirtuins) as Therapeutic Targets

• XII. Summary and Conclusions

I. Introduction

I am well aware that scarcely a single point is discussed in this volume on which facts cannot be adduced, often apparently leading to conclusions directly opposite to those at which I have arrived. A fair result can be obtained only by fully stating and balancing the facts and arguments on both sides of each question.”

Charles Darwin, “The Origin of Species by Means of Natural Selection.”

The decline of ovarian function at menopause leads to loss of bone mass, and estrogen replacement prevents such loss (1,2,3). During the last 60 yr, these two seminal clinical observations have shaped the thinking of basic and clinical investigators on the pathogenesis and thereby the treatment of osteoporosis, overwhelmingly more than, and despite, any other clinical observation or experimental discovery. As a result of this estrogen-centric paradigm, the “postmenopausal” or “type I” form of the disease and its treatment with estrogens or estrogen-related compounds preoccupied the field for most of this period. However, advancements in our understanding of the basic biology of the aging process, the effects of aging and age-related oxidative stress (OS) on bone, genetic discoveries in animal models and humans, the demonstration of a similar function of the same genes in different organs, and a better grasp of integrative physiology and pathophysiology along with the serious shortcomings of estrogen-based therapies, strongly dictate the need for a reappraisal of our traditional ideas about the pathogenesis of osteoporosis. The goal of this review is to highlight these new developments and provide a fresh look at the pathogenesis of osteoporosis, not in the traditional context of a single disease entity, but rather as comorbidity with other accompaniments of old age—which often are present in the same patient. In addition, this review aims to provide an understanding of osteoporosis that is reconciled with what is known about the biology of aging and the mechanisms of other degenerative disorders—which inexorably share molecular pathogenetic mechanisms related to aging per se—such as insulin resistance and atherosclerosis.

II. The Traditional Estrogen-Centric Perspective of the Pathogenesis of Osteoporosis

According to the estrogen-centric paradigm of the pathogenesis of osteoporosis, bone mass remains unchanged between the attainment of its peak value in the third decade of life and menopause in women, or the age of 55 in men. Therefore, loss of bone mass in adulthood must be triggered by the loss of sex steroids (3,4). The postmenopausal bone loss is followed within 4–8 yr by a less sharp rate of continuous decline. This latter phase of loss has been traditionally ascribed to “old age,” and it is thought to occur primarily in cortical sites, as opposed to the faster and self-limiting postmenopausal loss that occurs primarily in cancellous sites (3). Men lose cortical and cancellous bone at a rate that is practically identical to that of the “old age”-dependent loss in women, although they do not experience an abrupt decline of sex steroids and the corresponding rapid phase of bone loss (4). Despite these obvious clues for a role of “old age” in osteoporosis, the molecular and cellular mechanisms of the adverse effects of aging on bone and the extent to which sex steroid deficiency contributes to age-related bone loss and to the less sharp rate of skeletal decline in the late postmenopausal years have remained completely unknown—until recently. On the other hand, several lines of evidence have lent credence to the notion that estrogen deficiency is a seminal mechanism of the age-related bone loss not only in females, but also in males (4,5,6). Support for this view has been provided from extrapolation of three types of evidence: 1) genetic evidence from men with estrogen receptor (ER) α or aromatase mutations (7); 2) results of short-term clinical experimentation with administration of aromatase inhibitors (8); and 3) cross-sectional correlations between free serum E₂ levels and bone mineral density (BMD) or bone remodeling markers. However, more recent longitudinal studies of the propositus of genetic ERα mutations indicate that these individuals experience a failure of normal bone development and growth, which is of course a different situation than that of the vast majority of patients with osteoporosis who have normal bone development and growth, but who start losing bone after the attainment of peak bone mass (9). Furthermore, it remains unknown whether, and to what extent, small differences of estrogen or androgen levels in the circulation (10) contribute to changes of bone markers or to BMD determinations in elderly males (11), especially because similar small differences do not seem to cause BMD differences in pre- or perimenopausal women (12).

Additionally, it has been suggested that effects of estrogens on peripheral calcium metabolism, specifically the kidney (13) and intestine (14,15), play a role in preventing secondary hyperparathyroidism (and the associated increase in bone resorption) that sometimes accompanies involutional osteoporosis. However, hyperparathyroidism is not a feature of estrogen deficiency in experimental animal models or in humans with primary hypogonadism or estrogen deficiency caused by anorexia nervosa (16,17,18).

III. Aging as a Pivotal Determinant of Loss of Bone Mass and Strength

In contrast to the earlier ideas that bone mass remains unchanged between the attainment of its peak and menopause in women or the age of 55 in men (19,20), large epidemiological studies have now clearly established that in both women and men bone loss begins as early as the early part of the third decade—immediately after peak bone mass and long before any change in sex steroid production (21) (Fig. 1). In agreement with the epidemiological data, volumetric BMD analysis at the tibia and spine, on a large age- and sex-stratified population sample using quantitative computed tomography, demonstrates that there is substantial trabecular bone loss in both sexes during sex steroid sufficiency (Table 1) (22). Nonetheless, in women the loss of trabecular bone in the spine accelerates substantially after the menopause, as does the rate of fractures at the wrist, spine, and hip (23), attesting to the adverse role of estrogen deficiency on skeletal homeostasis and its contribution to the acceleration of the age-associated bone loss. Cortical bone loss begins to decline after the age of 50 in both sexes, albeit at a faster rate in women than men.

Figure 1.

A, Bone loss begins in the third decade of life in both sexes. The data are from the Epidemiological Follow-up Study cohort of the First National Health and Nutrition Examination Survey (NHANES), a nationally representative sample of noninstitutionalized civilians who were followed for a maximum of 22 yr. A cohort of 2879 Caucasian men (1437 in the bone density subsample) aged 45–74 yr at baseline (1971–1975) were observed through 1992. [From A. C. Looker et al.: Osteoporosis Int 8:468–489, 1998 (21). Reproduced with permission of the International Bone & Mineral Society and the IBMS BoneKEy where this graphic depiction of the data is provided online.] B, Age is a more critical determinant of fracture risk than bone mass in humans. Data are from a follow-up of 521 Caucasian women over an average of 6.5 yr with repeated bone mass measurements at the radius. A total of 138 nonspinal fractures in 3388 person-years were detected, and the incident fractures were cross-classified by age and bone mass. The incidence of fracture was then fitted to a log-linear model in age and bone mass. [From S. L. Hui et al.: J Clin Invest 81:1804–1809, 1988 (24). Reproduced with permission from The American Society for Clinical Investigation.]

Table 1.

Substantial trabecular bone loss occurs in young adult women and men

Age groups	Median changes in volumetric BMD (%/yr)
	DR-Trab	DT-Trab	LS-Trab	DR-Cort	DT-Cort
Females
21–49 yr	−0.40	−0.24	−1.61	−0.04	0.00
50+ yr	−0.55	−0.24	−2.60	−0.48,	−0.36,
Males
21–49 yr	−0.38	−0.40	−0.84	−0.07	−0.08
50+ yr	−0.38	−0.17	−1.85	−0.32	−0.15

Data are from a study of 375 females and 325 males between the ages of 21 and 97 yr, performed by peripheral/central quantitative computed tomography. [From B. L. Riggs et al.: J Bone Miner Res 23:205–214, 2008 (22). Kindly contributed by Dr. Sundeep Khosla and reproduced with the permission of the Journal.] DR, Distal radius; DT, distal tibia; Trab, trabecular; Cort, cortical; LS, lumbar spine.

Difference from zero,

^aP < 0.05;

^bP < 0.001.

^cSex difference < 0.05.

Notwithstanding the evidence for an age-related decline of bone mass independent of sex steroid status, it is now widely appreciated that osteoporosis (the term used to define decreased bone mass per unit volume of anatomical bone) is not the disease, but rather is one of many factors responsible for the compromised bone strength that predisposes to an increased risk of fracture in the fragility syndrome that has inexactly become synonymous with only one of its features. Indeed, strong evidence for a pivotal contribution of aging to fractures, independent of bone mass, was highlighted almost 20 yr ago by the work of Hui et al. (24), showing that at the same BMD, a 20-yr increase in age is accompanied by a 4-fold increase in fracture risk (Fig. 1). Increased propensity to falls due to age-related decline in neuromuscular function is without doubt a factor for the age-related increase in fracture risk. However, there are also age-related changes in the bone itself, which contribute to the increase in fracture risk for the same BMD with an increase in age (24). For example, type I collagen is structurally complex and can deteriorate as it ages, with changes such as loss of cross-linking between the component chains (25). Collagen can also be damaged by accumulation of advanced glycation end-products (26), another general feature of the aging process. Such changes could account for the age-related decline in cortical bone tensile strength (27). Defective collagen cannot be repaired, so the bone containing it must be replaced by remodeling.

The importance of non-mass factors is demonstrated by several lines of evidence. First, a fracture at any site increases the risk of a subsequent fracture at any other site (28). Second, only a small part of the reduction in fracture incidence in response to anticatabolic therapy can be accounted for by the increase in bone mass (29). Third, many of the genetic effects on bone strength are mediated by factors other than bone mass (30,31). Non-mass factors include disrupted architecture (32), changes in bone mineral and matrix (25), delayed repair of fatigue microdamage (33), excessive turnover (34), and inadequate bone size (35). The most recently appreciated qualitative factor is loss of osteocytes, which makes an independent contribution to vertebral bone strength, both in patients with vertebral fractures (36) and in mice (37). Osteocyte death may disrupt the signals necessary for microdamage repair (38) and also lead to long-term changes in bone hydration, as will be discussed in Section X.D.

Collectively, the above lines of evidence indicate that the estrogen-deficiency paradigm of the pathogenesis of osteoporosis is incomplete. That being said, there is little doubt that loss of bone is due in part to age-related changes in other organs and tissues, such as the ovary (estrogen deficiency), the adrenal gland (glucocorticoid excess or hyperresponsiveness), the kidney [loss of nephrons, reduced synthesis of calcitriol, calcium malabsorption, and secondary hyperparathyroidism (39), which also has other causes], and muscle (sarcopenia, inactivity, reduced mechanical loading). However, the evidence I discussed in this section along with the universality of age-related bone loss indicates that there must be additional age-related mechanisms intrinsic to bone. Excessive accumulation of reactive oxygen species (ROS)—the radical forms of oxygen—contributes to age-related changes in many tissues, and bone is unlikely to be an exception (40,41). In the remainder of this review, I will address some of these mechanisms and highlight evidence for the contribution of OS to the development of osteoporosis.

IV. Aging and Oxidative Stress

To date, more than 50 yr since it was proposed by Harman (42), the most enduring theory of aging stipulates that OS resulting from an increase in intracellular ROS is the major determinant of aging and lifespan (43,44,45), as well as the culprit in several degenerative disorders associated with aging (41). Reactive nitrogen species are also pathogenetic culprits in some of these conditions, but for the sake of space and because very little is known about reactive nitrogen species and osteoporosis, this review will focus on ROS.

Formation of ROS is an inescapable consequence of life in an oxygen-rich environment and occurs primarily in the mitochondria from the escape of electrons passing through the electron transport chain during aerobic metabolism—the process that is fueled by nutrients such as glucose and is responsible for the formation of ATP (43,46). ROS are also generated during fatty acid oxidation or in response to external stimuli, such as inflammatory cytokines, growth factors, environmental toxins, chemotherapeutics, UV light, or ionizing radiation. Free electrons are added to molecular oxygen to generate superoxide (O₂^·−), hydrogen peroxide (H₂O₂), and the hydroxyl radical (OH^·−). Of these species, H₂O₂ has the highest oxidative activity, the highest stability, and the highest intracellular molar concentration.

As a consequence of its properties, H₂O₂ represents a critical signal for the replicative capacity of regenerative cells, apoptosis, and global changes in gene expression leading to aging and aging-related diseases (43). The production of H₂O₂ is amplified considerably by the adapter protein p66^shc, which is released from an inhibitor complex in the inner mitochondrial membrane in response to a variety of proapoptotic stimuli and acts as a redox enzyme catalyzing the reduction of O₂ to H₂O₂ through electron transfer from cytochrome c (47,48,49). H₂O₂ in turn causes opening of the permeability transition pore, swelling, and apoptosis. Deletion of the p66^shc gene enhances cellular resistance to apoptosis induced by H₂O₂ or UV light, and mice deficient in p66^shc not only exhibit increased resistance to OS but also have a 30% increase in lifespan (48). Importantly, p66^shc sensitizes cells to proapoptotic stimuli by activating thymoma viral protooncogene 1 (Akt), phosphorylating/inactivating FoxO transcription factors, and preventing the induction of antioxidant/free radical scavenging genes [such as manganese superoxide dismutase (MnSOD)] (50,51).

Until recently, it was assumed that ROS are exclusively harmful by-products of aerobic life that damage proteins, lipids, and DNA leading to cell death, and that it was important for cells to eliminate them. However, extensive new evidence indicates that at levels lower than those that cause OS and damage, ROS, like H₂O₂, are deliberately produced within cells (and removed) to trigger physiological signaling cascades by causing reversible oxidations of proteins at cysteine residues (52). Such reversible amino acid oxidations regulate ERKs, jun N-terminal kinases (JNKs), p38 MAPK, phosphatidyl inositol 3 kinase (PI3K)/Akt, activator protein-1, p53, tyrosine phosphatases, proteases, molecular adaptors and chaperones, the Wnt/β-catenin signaling pathway, as well as the activity of transcription factors including nuclear factor κB (NF-κB), the glucocorticoid and estrogen receptors, and hypoxia inducible factor-1α (HIF-1α) (52).

V. Defense Mechanisms against Oxidative Stress

A. Enzymatic

To protect against oxidative insults, organisms ranging from prokaryotes to mammals utilize a network of overlapping mechanisms that involve both enzymatic reactions and altered gene transcription. OS occurs when the rate of generation of ROS exceeds the capacity of the cell to detoxify them. OS increases with advancing age, in part because the ability of cells to scavenge ROS decreases progressively with lifespan (44,53). Of the most important antioxidant enzymes, various forms of SOD catalyze the conversion of O₂^·− to H₂O₂, and catalases convert H₂O₂ to water and oxygen. Alternative mechanisms of ROS detoxification involve reactions with thiol-containing oligopeptides with redox-active sulfhydryl moieties, the most abundant of which are glutathione (GSH) and thioredoxin (Trx). GSH peroxidase (Gpx) converts peroxides to harmless alcohols in a reaction in which it oxidizes GSH to the disulfide GSSH; and GSH reductase (GSR) converts GSSH back to GSH (54). The peroxiredoxin family of enzymes provides a similar defense mechanism using Trx as substrate.

B. The FoxO transcription factors

FoxOs, a subclass of a large family of forkhead proteins characterized by the presence of a winged-helix DNA binding domain called Forkhead box, represent another major cell defense mechanism against oxidative insults. In mammals, this subclass comprises four members: FoxO1 (or FKHR), FoxO3 (or FKHRL1), FoxO4 (also called AFX), and FoxO6 (55). FoxO1, -3, and -4 show broad, overlapping patterns of expression in developing and adult tissues, whereas FoxO6 is restricted to specific structures of the developing brain. FoxOs shuttle between the cytoplasm and the nucleus depending on the phosphorylation of specific sites by distinct sets of kinases (Fig. 2). In the setting of stimulation by growth factors, such as insulin and IGFs, FoxO1, -3, and -4 are subject to Akt-mediated phosphorylation, which results in their nuclear export and inhibition of FoxO-mediated transcription (56,57,58,59,60). As will be discussed in Section VIII, exclusion of FoxOs from the nucleus is essential for glucose homeostasis because it is critical for the ability of insulin and other growth factors to stimulate β-cell proliferation and survival and thereby the expansion of β-cell mass in the pancreas (61,62,63).

Figure 2.

Schematic illustration of the shuttling of FoxOs between the cytoplasm (in the setting of growth factor stimulation, e.g., insulin) and the nucleus in response to oxidative stress (exemplified by H₂O₂). Note the distinct site of phosphorylation and cytoplasmic retention of FoxOs by the insulin- initiated Ras/PI3K/Akt cascade, as opposed to the sites of phosphorylation and nuclear localization of FoxOs in response to the H₂O₂-induced JNK activation cascade. The structure in the bottom left depicts a mitochondrion and its electron transport chain along with the tricarboxylic acid cycle that is responsible for ATP synthase activation and the generation of ATP.

On the other hand, OS promotes the translocation of FoxOs into the nucleus by a mechanism that involves activation of the JNK or macrophage stimulating 1 kinases by ROS and the phosphorylation of FoxOs in different sites than those activated by Akt. OS also promotes other posttranslational modifications of FoxOs, including ubiquitylation and acetylation (64). These changes in turn lead to the activation of FoxOs and thereby to FoxO-mediated OS responses by regulating the transcription of antioxidant enzymes such as MnSOD and catalase as well as genes involved in cell cycle, DNA repair, and lifespan (64,65). Through these effects, FoxOs promote mammalian cell survival by inducing cell cycle arrest and quiescence in response to OS, and they also regulate longevity in model organisms (66,67,68). Importantly, stress stimuli override the sequestration of FoxO in the cytoplasm by growth factors, both in mammalian cells and in Drosophila (69,70). FoxOs can also induce apoptosis through FoxO-mediated regulation of several proapoptotic genes (56,71). The role of FoxOs in cell death seems counterintuitive considering their role in the protection against OS. However, elimination of damaged or abnormal cells by FoxO-induced apoptosis evidently plays an important role in the ability of these transcription factors to promote tumor suppression and longevity (72).

The importance of FoxOs in mammalian biology has been recently highlighted by the profound effects of the deletion of individual FoxOs or the combined deletion of FoxO1, -3, and -4 in mice. FoxO1 is indispensable during development because FoxO1-null mice die at embryonic d 10.5 due to defects in angiogenesis (73,74). FoxO3-null mice become infertile due to premature activation of the ovarian follicles and subsequent oocyte exhaustion (75). These mice also develop lymphoproliferative disorders, consistent with a role of FoxO3 in the promotion of cell cycle arrest (76). In addition, they exhibit defects in glucose uptake in line with a role for FoxO family members in glucose metabolism (77). Studies of FoxO-deficient mice have revealed that FoxOs are also indispensable for normal erythropoiesis and hematopoietic stem cell quiescence and survival, because of their ability to prevent excessive accumulation of ROS (78,79). Specifically, FoxO3 null mice die rapidly when exposed to OS because of failed erythropoiesis (79). Moreover, combined conditional somatic deletion of FoxO1, -3, and -4 using the Mx-Cre⁺ transgene results in increased ROS levels in hematopoietic stem cells (HSCs). Increased ROS enforces cell fate decisions, driving HSCs into the cell cycle and terminal differentiation at the expense of self-renewal (78). This phenotype was completely reversed by administration of the antioxidant N-acetyl cysteine (NAC), indicating that FoxOs maintain HSC quiescence and survival by handling free radicals constantly generated in HSCs.

Extensive evidence also indicates that FoxOs restrict angiogenesis. This fact has been convincingly demonstrated by the striking overproduction of new blood vessels in mice with conditional deletion of the three major FoxO genes (80). The suppressive effect of FoxOs on angiogenesis results from their ability to suppress endothelial cell proliferation and survival (80). One of the potential mechanisms of this effect is direct binding of FoxOs to HIF-1α on some target genes and prevention of HIF-1α interaction with p300, a required transcriptional coactivator (81). In addition, FoxOs stimulate the expression of CITED2, a factor that also inhibits HIF-1α and p300 interaction (82). As will be discussed in Section X.D, these molecular mechanisms may be directly relevant to the suppression of angiogenesis and bone formation by old age and glucocorticoid excess because both of these conditions increase OS and FoxO activity.

C. β-Catenin as a pivot in the regulation of oxidative stress-induced transcription programs

During the last few years, Wnts, a large family of secreted glycoproteins, and their signaling pathway have received major attention by bone biologists because of genetic evidence that loss or gain of function mutations of the low-density lipoprotein receptor-related protein 5 or 6 (LRP5 or LRP6), the coreceptors for Wnts, are responsible for the dramatic decrease or increase of bone mass associated with the osteoporosis pseudoglioma syndrome and the hereditary high-bone-mass trait in humans, respectively (83,84,85,86). In line with evidence for a major role of Wnts in the regulation of bone mass, increased Wnt/β-catenin signaling is evidently a normal physiological response of bone to changes in mechanical load (87). Furthermore, osteocytes—former osteoblasts that are entombed in the mineralized matrix and are responsible for sensing and adapting bone to changes in mechanical needs—exert their potent influence on bone mass by controlling Wnt/β-catenin signaling thanks to their ability to produce and secrete sclerostin, a potent Wnt/β-catenin inhibitor (88,89,90). Loss of function mutations of SOST—the gene encoding sclerostin—cause the high bone mass disorders Van Buchem’s disease and sclerosteosis (91,92), whereas mice overexpressing SOST exhibit low bone mass (88). Moreover, a neutralizing antisclerostin antibody dramatically increases bone formation in rodents and humans and is currently under development as an anabolic bone therapy for patients with osteoporosis (93). The ability of Wnts to regulate bone mass results from a critical role of β-catenin in the commitment of multipotential mesenchymal progenitors to the osteoblastic lineage and from the prevention of the apoptosis of both uncommitted osteoblast progenitors and differentiated osteoblasts by β-catenin-dependent and -independent signaling cascades (94,95). Unexpectedly, the Wnt coreceptor LRP6 is also a key element of the PTH signaling that regulates osteoblast activity (96). Thus, binding of PTH to its receptor PTH1R induces the association of LRP6 with PTH1R and stabilization of β-catenin. And, an increase in the amount of β-catenin in osteoblasts in response to PTH is evidently responsible for the ability of the hormone to increase bone formation in the rat model. This evidence notwithstanding, the effect of the activating LRP5 mutation on bone formation was recently shown to result not from direct effects on bone cells, but indirectly through the inhibition of serotonin synthesis in the duodenum (97).

Besides bone formation, the Wnt/β-catenin signaling pathway affects many other biological processes ranging from embryonic development, patterning, and postembryonic stem cell fate to insulin secretion in adulthood and cancer (98). β-Catenin mediates canonical Wnt signaling by binding to and activating members of the T cell factor (TCF)/lymphoid enhancer factor transcription factor family (99). β-Catenin also plays a very important role in the defense against OS, by virtue of the fact that in addition to its important role in mediating TCF/lymphoid enhancer factor transcription, β-catenin is an essential coactivator of FoxOs (100,101,102,103,104). OS, as opposed to insulin and growth factor signaling, promotes FoxO binding to β-catenin and activation of FoxO transcription. The interaction between β-catenin and FoxOs is evolutionarily conserved as evidenced by the fact that in Caenorhabditis elegans the β-catenin ortholog, BAR-1, is required for OS-induced expression of the FoxO ortholog DAF-16 target gene sod-3 and for resistance to oxidative damage (100).

VI. Organismal Aging, Oxidative Stress, and Skeletal Homeostasis

The histological hallmark of age-related bone loss in humans and animals is a decline in mean wall thickness—an index of the amount of bone made by each team of osteoblasts during bone remodeling (105,106). This is due mainly to a deficit in the number of osteoblasts rather than their biosynthetic capacity, and it is thought to result from decreased osteoblastogenesis or increased apoptosis (1,107,108). Despite this important histological insight, the molecular and cellular mechanisms of this adverse effect of aging on bone have until fairly recently remained elusive.

In agreement with the clinical and epidemiological evidence that aging per se is a pivotal mechanism of the decline of bone mass and strength, several lines of recent evidence, primarily from the mouse model, strongly suggest that the OS that underlies physiological organismal aging is an important pathogenetic mechanism of the age-related bone loss and strength. It is important to note here that in a difference from humans, mice do not undergo acute loss of estrogen in middle life, making them an invaluable experimental tool for dissecting the effects of old age from the effects of sex steroid deficiency.

Despite the fact that rodents do not experience acute estrogen loss, a decline of BMD starting as early as 2–3 months of age has been established in several strains of mice (109,110), as well as in rats (111). As shown in Fig. 3, loss of bone mass with age is present in all strains of mice examined and involves both the cancellous and cortical compartment, as determined by micro-computed tomography. However, cortical bone loss occurs later in life than trabecular bone loss. Importantly, C3H, one of the strains depicted in Fig. 3, is resistant to ovariectomy-induced loss of cortical bone (112), but not to the aging-associated loss, highlighting the overriding importance of the age-related mechanisms.

Figure 3.

Both trabecular and cortical bone mass decrease with age in mice. Micro-computed tomography analysis at the distal femoral metaphysis (left) and the femoral midshaft (right) of virgin female mice (n = 6 to 10 per group) was performed at the indicated time points. At the bottom of the right graph is a schematic representation of the changes of the femoral cortex with age, depicting the enlargement and thickening during the growth period (from 1 to 4 months) and the thinning after the attainment of peak bone mass due to endosteal resorption. BV, Bone volume; TV, total volume. [Unpublished data from M. Bouxsein and V. Glatt, Beth Israel Deaconess Medical Center, Harvard Medical School, generously provided for the purpose of this review article.]

Studies from our group have shown that progressive loss of bone strength and mass with age in sex steroid-sufficient female or male C57BL/6 mice is temporally associated with decreased osteoblast and osteoclast numbers and decreased bone formation rate as well as increased osteoblast and osteocyte apoptosis (113). Most importantly, the age-dependent decline of bone mass and strength is temporally linked with increased ROS levels and decreased GSR activity in the bone marrow, as well as a corresponding increase in the phosphorylation of p53 and p66^shc—the adapter protein that amplifies mitochondrial ROS generation and influences apoptosis and lifespan in mice (47). In agreement with our findings, studies by others have determined that both osteoblast numbers and bone formation are decreased in mice treated with an inhibitor of GSH (114). Moreover, murine models of premature aging and signs of oxidative damage exhibit osteoporotic features (115,116). Furthermore, as in mice, an association between OS and a decrease in BMD (117,118,119,120,121,122), as well as an effect of antioxidants on bone resorption (123,124), has been noted in several human clinical studies.

As discussed in Section V.B, OS activates FoxO transcription factors, which in turn combat OS by activating genes involved in free radical scavenging, DNA repair, and apoptosis; and β-catenin—a factor required for osteoblast differentiation (125)—is essential for ROS-induced FoxO activation (100). In C57BL/6 mice, the increase in OS and the decreased bone formation with increasing age are associated with increased FoxO-target gene expression and a decrease in β-catenin/TCF-target gene expression on bone. Moreover, in osteoblastic cell models, OS induces the association of FoxOs with β-catenin and promotes FoxO-mediated transcription at the expense of β-catenin/TCF-mediated transcription and osteoblast differentiation (101). The negative effect of ROS on TCF transcription is abrogated by raising the level of β-catenin, suggesting that a limited pool of active β-catenin is diverted from TCF to FoxO transcription under stress conditions (102,104). In line with these ideas, mice with targeted expression of Wnt10b in the bone marrow or mice carrying the LRP5 G171V activating mutation in osteoblasts have increased bone mass and no evidence of age-related loss of bone mass or strength, respectively (126,127). Hence, diversion of the limited pool of β-catenin from TCF- to FoxO-mediated transcription in osteoblastic cells represents a cell-autonomous mechanism of β-catenin/TCF antagonism. FoxOs are also known to suppress the expression and transcriptional activity of peroxisome proliferator-activated receptor (PPAR) γ, which is a potent repressor of osteoblastogenesis (128,129,130). Together, these observations suggest that FoxOs play an important role in bone biology by enabling the maintenance of a physiologically appropriate lifespan of mature osteoblasts through their oxidative defense activities. In addition, FoxOs may control the generation of new osteoblasts from their mesenchymal stem cell progenitors by modulating their proliferation and/or differentiation through their antioxidant properties or via modulating the activity of other transcription factors such as β-catenin or PPARγ.

Based on the evidence that FoxOs translate environmental stimuli, such as OS and hormonal changes, into dynamic gene expression programs involved in many physiological as well as pathological processes (64,65,131), and that decreased defense against OS is responsible, at least in part, for the adverse effects of aging on the murine skeleton, we have investigated in my laboratory the impact of the genetic manipulation of FoxOs on skeletal homeostasis in mice (132). Specifically, we hypothesized that if one were to remove an important defense mechanism against OS from bone cells, one may recapitulate at least some of, the adverse effects of aging on bone in young mice. We found that conditional deletion of the three broadly expressed genes, FoxO1, -3, and -4, in young adult mice or targeted overexpression of FoxO3 in osteoblasts leads to significant changes in bone mass. Specifically, loss of FoxO function in 3-month-old mice for a period of 5 wk results in increased OS in bone as well as increased osteoblast and osteocyte apoptosis and an osteoporotic phenotype characterized by decreased bone mass at both cancellous and cortical sites. The increased osteoblast apoptosis following the triple FoxO deletion is cell autonomous and is the result of increased OS because the rate of osteoblast apoptosis of the FoxO-deficient osteoblasts in ex vivo cultures can be reverted to normal by the addition of the antioxidant NAC. In sharp contrast to the adverse skeletal effects of the loss of FoxO function, overexpression of FoxO3 in mature osteoblasts leads to opposite effects: decreased OS and osteoblast apoptosis, and increased osteoblast number, bone formation rate, and vertebral bone mass. In agreement with findings from this model, the adverse effects of aging on cancellous and cortical bone can be reproduced in young mice by the administration of the prooxidant paraquat (M. Almeida, E. Ambrogini, S. C. Manolagas, unpublished data).

Osteoblast generation in ex vivo cultures of osteoblastic progenitors derived from the bone marrow of FoxO-deficient mice is attenuated compared with the control mice. This could result from either an increase in the apoptosis of these progenitors or attenuation of their differentiation. In support of the latter explanation, deletion of FoxOs causes an increase of PPARγ, the nuclear receptor that stimulates adipogenesis (133), whereas it represses osteoblastogenesis (130) and is tonically suppressed by FoxOs (128,129).

The evidence from the genetic analysis of the role of FoxOs in bone provides strong support for the view that ROS are seminal signals for the fate of osteoblasts and that inappropriate increase of ROS adversely affects bone formation. The FoxO family of transcription factors defends against such an increase by constantly up-regulating free radical scavenging and DNA-repair enzymes, thereby representing an indispensable homeostatic mechanism for skeletal health. In addition, FoxOs may control the generation of new osteoblasts from their mesenchymal stem cell progenitors by modulating their proliferation and/or differentiation through their antioxidant properties or via modulating the activity of other transcription factors such as β-catenin or PPARγ, but this topic is discussed further in Section I.X.

The antioxidant defense provided by FoxOs is ultimately overwhelmed by high level of OS or OS-activated pathways that interfere with the activity of FoxOs, or both. Thus, extensive evidence indicates that ROS-induced p66^shc activation leads to FoxO inactivation via Akt-mediated FoxO phosphorylation (50,51). Moreover, ROS-induced stimulation of NF-κB may lead to inhibitory κB kinase-mediated phosphorylation and inhibition of FoxO3 activity, at least in part by targeting it to ubiquitin-dependent degradation (134). Increased phosphorylation of p66^shc as well as increased activity of NF-κB are common features of old age and sex steroid deficiency (113,135,136,137). Furthermore, work from the author’s group which will be discussed in Sections IX and X has elucidated that age-associated increases in lipid oxidation and endogenous glucocorticoid production and sensitivity stimulate osteoblast apoptosis by both ROS-dependent and ROS-independent mechanisms. Hence, a plethora of counteracting mechanisms can account for overwhelming the ability of FoxOs to thwart osteoblast apoptosis in old age.

VII. The Antiosteoporotic Effects of Estrogens and their Antioxidant Properties

In addition to the evidence that the OS that underlies physiological organismal aging is a pivotal pathogenetic mechanism of the age-related bone loss and strength, evidence that will be summarized in this section suggests that the antiosteoporotic effect of estrogens results, at least in part, from their ability to protect against OS. In fact, loss of estrogens or androgens accelerates the effects of aging on bone by decreasing defense against OS.

The hallmark of the acute loss of sex steroids (as in menopause or after ovariectomy of humans and animals) is an increase in the rate of bone remodeling, resulting from an increase in both osteoclastogenesis and osteoblastogenesis and a corresponding increase in bone resorption and formation, albeit the former exceeds the latter, most likely because estrogen deficiency prolongs the lifespan of osteoclasts but decreases the lifespan of osteoblasts. Consistent with this evidence, estrogens protect the adult skeleton against bone loss by slowing the rate of bone remodeling (turnover) and by maintaining a focal balance between bone formation and resorption (1,138,139). Indeed, slowing of bone remodeling is evidently due to the attenuating effects of sex steroids on the birth rate of osteoclast and osteoblast progenitors (140,141). Maintenance of a focal balance between formation and resorption, on the other hand, is explained by the opposite effects of estrogens on the lifespan of osteoclasts and osteoblasts/osteocytes: a proapoptotic effect on osteoclasts and an antiapoptotic effect on osteoblasts and osteocytes (142,143,144). Despite these advances, it has remained unknown whether estrogen deficiency contributes to age-related bone loss—and if so, how.

Strikingly, the exact same increases in ROS levels and p53 and p66^shc phosphorylation and the decrease of GSR activity observed with advancing age in C57BL/6 mice are caused by the removal of the gonads in 5 month-old female or male mice (Fig. 4). Moreover, all these changes are reversed in the gonadectomized animals by replacement with estrogens or a nonaromatizable androgen or by the administration of the antioxidant NAC (113). More telling, NAC prevents the gonadectomy-induced loss of bone as well as the increase in osteoblast and osteocyte apoptosis as effectively as the replacement with estrogens or androgens. Similar to these findings, Lean et al. (145) have shown earlier that GSH and Trx, the major thiol antioxidants, and GSR and Trx reductase, the enzymes responsible for maintaining them in a reduced state, fall substantially in 2-month-old rat and mouse bone marrow after ovariectomy and are rapidly normalized by exogenous E₂. Moreover, administration of NAC, ascorbate, or catalase prevents ovariectomy-induced bone loss in these models, whereas l-buthionine-(S,R)-sulfoximine (BSO), a specific inhibitor of GSH synthesis, causes substantial bone loss. Likewise, Muthusami et al. (146) found that ovariectomy in rats increases H₂O₂ and lipid peroxidation in the femurs and decreased SOD, Gpx, and GSH S transferase; and H₂O₂ levels and SOD activity are inversely correlated.

Figure 4.

Advancing age (A) and loss of sex steroids (B) cause similar changes in oxidative stress. ROS and GSR activity were measured in the bone marrow aspirates, and the phosphorylation status of p53 and p66^shc was determined by Western blot analysis in vertebral lysates from female or male C57BL6 mice at the indicated ages. Ovariectomy (OVX) or orchidectomy (ORX) was performed at 5 months of age, and analysis was done 6 wk later. Bars indicate mean ± sd; n = 4 mice per group. AFU, Arbitrary fluorescence units; veh, vehicle. *, P < 0.05 compared to 4 months or OVX or ORX + vehicle. [Modified from M. Almeida et al.: J Biol Chem 282:27285–27297, 2007 (113).]

In line with the view that the beneficial effects of estrogens on bone result, at least in part, from their antioxidant properties, beneficial effects of estrogens in several other tissues, such as heart, arteries, central nervous system, lens epithelial cells, fat, liver, and oviducts, are also shown to result from improved defense against OS (147,148,149,150,151,152,153,154,155,156,157,158,159). In fact, practically identical to the evidence in bone, estrogens prevent cardiomyocyte apoptosis and the development of congestive heart failure in mice by increasing the expression of Trx, Trx reductases, and Trx reductase activity in the heart (160). Furthermore, long-term estrogen treatment prevents the activation of apoptosis signal-regulating kinase 1 and its downstream effectors, JNK and p38 MAPK. Estrogen-treated cardiomyocytes are much more resistant to angiotensin II-induced apoptosis; and the antiapoptotic and cardioprotective effects of E₂ are blocked by an ER antagonist (ICI 182,780) and by a Trx reductase inhibitor (azelaic acid), indicating that long-term estrogen treatment improves congestive heart failure by antioxidant mechanisms. On the other hand, adverse effects of estrogens on breast cancer, the uterus, and spermatogenesis may be due to increased ROS production or decreased antioxidant defense (161,162,163).

A. ROS, estrogens, and osteoblasts

Earlier work from the author’s group has indicated that estrogens and androgens control osteoblast and osteoclast apoptosis by a mechanism that is distinct from that requiring direct interaction of their receptors with DNA (hormone response element) or protein/protein interaction between the receptor and other transcription factors. Instead, the effect of estrogens on the apoptosis of either cell type is the result of an extranuclear action of the classical receptors that cause activation of cytoplasmic kinases, including ERKs, and kinase-dependent changes in the activity of transcription factors (142,144,164). Additionally, the number of osteoblast progenitors, as measured by colony forming units-osteoblast (CFU-OB), increase after loss of estrogens in mice (165), and this change is partially preserved in mice treated with antiresorptive drugs like bisphosphonates, indicating that bone resorption (and the release of growth factors from the bone matrix) is not required for the increase in osteoblast precursors. Therefore, estrogens must suppress osteoblastogenesis by direct actions on early osteoblast precursors. Furthermore, most murine CFU-OBs are early transit-amplifying progenitors (i.e., dividing cells with limited self-renewal capacity), and their replication is indeed attenuated by estrogens (141).

Prompted by the observations that estrogens diminish OS in bone and bone marrow and attenuate the prevalence of mature osteoblast apoptosis as well as osteoblastogenesis, Almeida et al. (166) have searched for the molecular mechanism of these effects using as tools a mouse model bearing an ERα knock-in mutation that prevents binding to DNA and several osteoblast progenitor cell models expressing the wild-type ERα. The ability of estrogens to diminish the generation of ROS, stimulate the activity of GSR, and to decrease the phosphorylation of p66^shc, osteoblastogenesis (in ex vivo bone marrow cultures), and osteoblast number and apoptosis is fully preserved in these mice, indicating that the DNA-binding function of the ERα is dispensable for all these effects. Consistent with the attenuation of osteoblastogenesis in this animal model, E₂ attenuates bone morphogenetic protein (BMP)-2- induced gene transcription and osteoblast commitment and differentiation in murine and human osteoblastic cell lines, as well as in primary cultures of calvaria or bone marrow-derived osteoblastic cells from C57BL/6 mice. The inhibitory effect of the hormone on BMP-2 signaling results from an ERα-mediated activation of ERKs and the phosphorylation of Smad1 at the linker region of the protein, which leads to Smad1 proteasomal degradation. In agreement with these findings, the number of CFU-OBs in the bone marrow is attenuated in C57BL/6 mice treated for 28 d with NAC (167). These results indicate that the effects of estrogens on OS and on the birth and death of osteoblasts do not require the binding of ERα to DNA response elements, but instead they result from the activation of cytoplasmic kinases.

In agreement with the in vivo evidence that estrogens or androgens prevent bone loss and the apoptosis of osteoblasts and osteocytes via antioxidant effects, in both primary bone marrow-derived cell cultures and osteoblastic cell lines, E₂ or dihydrotestosterone (DHT) attenuate osteoblasts apoptosis induced by etoposide, H₂O₂, or TNF and suppress p66^shc phosphorylation. These effects are mediated by a mechanism that involves direct antioxidant actions of either class of sex steroids and is dependent on Src and MAPK kinase kinases (113). Similarly, estrogens and selective ER modulators suppress the H₂O₂-induced apoptosis of the osteocyte-like cell line MLO-Y4 (168). Moreover, estrogens attenuate OS-induced IL-6 and TNFα (two cytokines implicated in estrogen deficiency- induced bone loss) production by stromal/osteoblastic cells by attenuating NF-κB activation (169). In line with this evidence, an increase in ROS, particularly H₂O₂, or depletion of GSH suppress osteoblastic differentiation of bone marrow cells, also in an ERK-dependent manner (170); and osteoblast apoptosis induced by OS in vitro is attenuated by glutaredoxin 5 (171). Moreover, the progressive increase in OS with age is associated with increased IL-1 and IL-6 mRNA expression in vertebral lysates, as well as an increase in the number of osteoclast progenitors. Importantly, the effects of aging are reproduced in young mice after administration of the GSH inhibitor BSO or the prooxidant agent paraquat (169). Furthermore, administration of the antioxidant NAC to orchidectomized mice decreases osteoclast progenitor numbers as effectively as the administration of DHT or E₂. In addition, E₂ or DHT potently suppress H₂O₂-induced p66^shc phosphorylation, as well as the phosphorylation of IκB and NF-κB transcriptional activation and also suppress the production of osteoclastogenic cytokines like TNF and IL-6 in two bone marrow-derived stromal cell models (OB6 and UAMS-32) that support osteoclastogenesis. Strikingly, silencing p66^shc by short hairpin RNA attenuates the ability of H₂O₂ to induce IκB phosphorylation and NF-κB transcriptional activity and dramatically reduces the levels of IL-6 mRNA as well as the activation of IL-6 by H₂O₂. These results demonstrate that p66^shc is an essential mediator of the stimulating effects of OS on NF-κB activation, cytokine production, and osteoclastogenesis. The ability of either estrogen or androgen to attenuate the effects of OS on NF-κB activation, cytokine production, and osteoclastogenesis results from their shared ability to suppress p66^shc phosphorylation (169) (Fig. 5).

Figure 5.

ROS-activated signals affecting the genesis and lifespan of osteoblasts and osteoclasts and the counter-regulatory actions of sex steroids. In osteoclast precursors, RANKL-induced activation of RANK stimulates ROS production, which is essential for osteoclastogenesis. In addition, mitochondria biogenesis coupled with activation of the transferrin receptor (TIR1) by the iron-transferrin (Fe-Tf) complex stimulates mitochondria respiration and ROS production, which are also essential for osteoclast activation. Estrogens and androgens, acting via their respective receptors (ERα and AR), attenuate both osteoclastogenesis and survival by stimulating GSH and Trx reductase in an ERK-dependant manner (113,145,172,176,177). In osteoblastic cells, p66^shc is an essential mediator of the effects of oxidative stress on apoptosis, NF-κB activation, and cytokine production. Estrogens and androgens attenuate these effects by suppressing p66^shc phosphorylation in an ERK-dependent manner (113,169).

B. ROS, estrogens, and the generation and apoptosis of osteoclasts

Osteoclasts are acid-secreting polykaryons that have high energy demands and contain abundant mitochondria. Mitochondria and ROS, particularly H₂O₂, play a crucial role in osteoclast differentiation and function. Indeed, ROS or depletion of GSH increases osteoclast number and resorption in vitro and in vivo by stimulating receptor activator of NF-κB ligand (RANKL) and TNFα expression through ERK and NF-κB activation (172,173,174,175). Moreover, osteoclast formation upon stimulation of bone marrow-derived monocytes/macrophages with the RANKL—a sine qua non osteoclastogenic signal—increases ROS through a cascade involving TNFα, TRAF6, Rac1, and nicotinamide adenine dinucleotide phosphate oxidase (NADPH). The antioxidant NAC and inhibition of NADPH (Nox1), the enzyme that generates ROS, prevent all these effects as well as RANKL-induced JNK, p38, and ERK activation and osteoclastogenesis (176). Similarly, PTH- and IL-1-stimulated bone resorption is inhibited by both natural and recombinant SOD (172). Gpx1, the enzyme primarily responsible for the intracellular degradation of H₂O₂, is highly expressed in osteoclasts, and its expression is increased in bone marrow macrophages by RANKL and in osteoclasts by E₂. Overexpression of Gpx in osteoclast progenitors abolishes osteoclast formation and suppresses RANKL-induced NF-κB activation and increased resistance to oxidation of dihydrodichlorofluorescein by exogenous H₂O₂ (175). Furthermore, mitochondrial biogenesis is integrated with osteoclast differentiation. Specifically, mitochondrial biogenesis coupled with iron uptake through TfR1 and iron supply to mitochondrial respiratory proteins represent a fundamental pathway linked to osteoclast activation and bone metabolism (177).

Consistent with the requirement of mitochondria-produced ROS in osteoclast generation and the in vivo evidence that estrogens or androgens prevent bone loss via antioxidant effects, E₂ or DHT attenuates osteoclastogenesis and stimulates osteoclast apoptosis by a mechanism that involves up-regulation of GSR (113). Similar to the case with their antiapoptotic effects on osteoblasts, the antiosteoclastogenic and proapoptotic effects of E₂ or DHT on osteoclasts in vitro are exerted in an Src and MAPK kinase-dependent mechanism. E₂ also increases GSH, GSR, and Trx reductase in osteoclast-like cells in vitro (114,145,175). Furthermore, in vitro NAC prevents osteoclast formation and NF-κB activation, whereas BSO and H₂O₂ have the opposite effect, consistent with the evidence that a certain level of ROS generation is essential for osteoclast generation.

Recently, Nakamura et al. (178) reported that female, but not male, mice in which the ERα has been specifically deleted in mature osteoclasts (ERα^ΔOc/ΔOc) exhibit decreased cancellous bone due to an increased number of osteoclasts resulting from the loss of a cell autonomous proapoptotic effect of estrogens on osteoclasts, which is mediated by an increase in Fas ligand (FasL) production by osteoclasts. These mice did not lose additional bone after ovariectomy. Based on these results, Nakamura et al. (178) have suggested that the proapoptotic effect of estrogens on osteoclasts is the sole molecular basis of the osteoprotective property of estrogens. In vitro studies by Krum et al. (179) have confirmed the role of FasL in the effect of estrogens on osteoclast apoptosis, but unlike Nakamura et al. (178), Krum et al. have concluded that estrogens increase the transcription of the FasL gene not in osteoclasts, but in osteoblasts, and that FasL must therefore act in a paracrine fashion to stimulate osteoclast apoptosis.

In the author’s laboratory, ERα was deleted in very early osteoclast progenitors including cells of the monocyte/macrophage lineage in mice, using a Lysozyme M promoter (180). These mice exhibited a 2-fold increase in osteoclast progenitors in the marrow and the number of osteoclasts in cancellous bone, along with a decrease in cancellous bone mass. After ovariectomy, these mice failed to exhibit the expected increase in osteoclast progenitors, the number of osteoclasts in bone, and further loss of cancellous bone. However, they lost cortical bone indistinguishably from their littermate controls. Mature osteoclasts lacking ERα were resistant to the proapoptotic effect of E₂. However, the effects of estrogens on osteoclasts were unhindered in mice bearing an ERα knock-in mutation that prevented binding to DNA. Moreover, a polymeric form of estrogen that is not capable of stimulating the nuclear-initiated actions of ERα was as effective as E₂ in inducing osteoclast apoptosis in cells with the wild-type ERα. These results contradict the conclusions of Nakamura et al. (178) and Krum et al. (179) because they demonstrate that estrogens attenuate osteoclast generation and lifespan via cell autonomous effects mediated by DNA-binding independent actions of ERα. Elimination of these effects is sufficient for loss of bone in the cancellous compartment in which complete perforation of trabeculae by osteoclastic resorption precludes subsequent refilling of the cavities by the bone-forming osteoblasts. However, additional effects of estrogens on osteoblasts, osteocytes, and perhaps other cell types are required for their protective effects on the cortical compartment, which comprises 80% of the skeleton.

Another difference from the results of Nakamura et al. (178) is that we have been unable to elicit a stimulatory effect of E₂ on FasL production in primary cultures of murine osteoclasts. Lack of an effect of estrogens on FasL production has also been reported by others in the osteoclast-like cell line RAW 264.7, although estrogens do enhance caspase-3 activity in Fas-induced apoptosis of mature osteoclasts (181). Furthermore, it has been shown that endogenous FasL does not have a role in the apoptosis of mature osteoclasts and has only a minimal effect on the apoptosis of osteoclast progenitors from C57BL/6 mice (182,183). In addition, bone marrow from mice without functional Fas or FasL have similar osteoclastogenic potential as bone marrow from wild-type mice. On the other hand, Wu et al. (184) have shown not only that Fas is expressed in osteoclasts but also that its expression increases during differentiation. Moreover, these workers found that mice lacking Fas have decreased BMD and increased osteoclast number, whereas mice deficient in FasL show no changes in BMD but show a significant increase in osteoclast number.

Be that as it may, our own work (180) has confirmed the observations of Nakamura et al. (178) that E₂ fails to induce the apoptosis of osteoclasts derived from FasL-deficient mice. Strikingly, in our hands these cells are also resistant to the proapoptotic effect of the nonaromatizable androgen DHT. Depending on the cell type, ROS may be required for FasL-induced apoptosis (186) and may play no role (187) or even act as an antiapoptotic signal in Fas-activated cells (188). Taking all the available evidence together, it is unlikely that the ability of estrogens, an estrogen polymer incapable of initiating nuclear actions, and DHT to promote osteoclast apoptosis in a FasL- dependent fashion results from an estrogen response element-dependent stimulation of FasL expression. Instead, it is more likely that FasL provides a tonic stimulatory signal for osteoclast apoptosis that is potentiated by estrogens or androgens via several nontranscriptional mechanisms: 1) attenuation of the osteoclastogenic and antiapoptotic effect of RANKL, secondary to decreasing ROS production by stimulating GSR activity; 2) attenuation of the antiapoptotic effect of NF-κB via protein-protein interaction of the ERα with NF-κB (138,189,190,191,192), which for example will attenuate the ability of NF-κB to inhibit TNF receptor type I-induced apoptosis via augmentation of the synthesis of cellular caspase-8 (FLICE) like inhibitory protein (193,194); and 3) suppression of the transcriptional activity of c-jun (195). Last but not least, the view that stimulation of FasL is unlikely to be the mechanism by which estrogens promote osteoclast apoptosis is supported by the evidence that estrogens have an antiapoptotic effect on osteoblasts (142,143,164), although murine and human osteoblasts do undergo Fas-mediated apoptosis in response to FasL (196,197,198,199,200). This conundrum is particularly incongruent with the contention of Krum et al. (179) that osteoblasts are the source of FasL for the apoptosis of osteoclasts.

If both aging and loss of sex steroids exert their adverse effects on bone by oxidative damage, how can sex steroid deficiency cause an increase in bone turnover associated with increased osteoclastogenesis and osteoblastogenesis, increased osteoclast and osteoblast numbers, and increased resorption and formation—albeit unbalanced—whereas aging mice and elderly individuals (without vitamin D deficiency and secondary hyperthyroidism) exhibit a low rate of bone remodeling associated with a decrease in osteoblast number and bone formation and no increase in osteoclast number? As will be discussed in Sections IX and X, two age-dependent pathogenetic mechanisms that increase OS, namely PPARγ activation by oxidized lipids and endogenous hyperglucocorticoidism, compounded by the loss of the antioxidant protection of estrogens, may over time thwart the effect of the acute loss of estrogens on remodeling and convert the high remodeling state of acute estrogen loss to the low remodeling of old age. In support of these ideas, hyperglucocorticoidism does indeed override the effects of gonadectomy in mice (201). Moreover, cortisol concentration and the rate of bone loss are inversely related in healthy elderly men and women even after adjustments for adiposity, smoking, alcohol consumption, dietary calcium, activity, as well as serum testosterone and E₂ levels (202,203). Additionally, polymorphisms of 11β-hydroxysteroid dehydrogenase (11β HSD) type 1 are strongly associated with osteoporosis, independently of sex steroids (204).

VIII. Diabetes, Oxidative Stress, and Osteoporosis

Similar to aging, both type 1 and type 2 diabetes mellitus adversely affect skeletal homeostasis and increase the risk of fractures, primarily by increasing osteoblast apoptosis and suppressing bone formation. Indeed, obese and non-obese diabetic rats and mice exhibit impaired bone formation and enhanced apoptosis of osteoblastic cells (205,206,207,208,209,210). In addition, both streptozotocin-induced diabetic mice, an animal model of type 1 diabetes, and spontaneously diabetic Torii rats, an animal model of type 2 diabetes, have low-turnover osteopenia associated with increased OS; and markers of OS are inversely associated with the histomorphometric parameters of bone formation. Overexpression of Trx-1 in transgenic mice attenuates streptozotocin-induced diabetic osteopenia (211). Moreover, impaired bone formation and decreased serum levels of osteoblast markers, including osteocalcin and bone-specific alkaline phosphatase, are typical features of the bone disease caused by type 1 and 2 diabetes in men and women of all ages (212,213,214,215).

It is now amply documented that OS plays a major role in the development of both type 1 and type 2 diabetes, in part by accelerating the death of pancreatic β-cells (216,217,218). In fact, the pancreatic islet is the tissue least endowed with intrinsic antioxidant defense mechanisms (49). Insulin, glucagon-like peptide-1, or IGF signaling induces PI3K/Akt activation and thereby the exclusion of FoxOs from the nucleus (56,57,58,59,60). Exclusion of FoxOs is critical for the ability of insulin and the other growth factors to stimulate β-cell proliferation and survival and thereby expansion of β-cell mass in the pancreas (Fig. 2) (61,62,63).

Hyperglycemia and insulin resistance, on the other hand, increase ROS production by accelerating mitochondria respiration and activating NADPH (193,216,217,218). Increased production of proinflammatory cytokines, such as TNF and interferon-γ, as well as of free fatty acids in type 2 diabetes also contributes to increased OS (219), as does the formation of increased advanced glycation end-products and glucose autoxidation. As depicted in Fig. 2, OS induces the phosphorylation of JNK, which overrides several effects of insulin including the Akt-induced FoxO phosphorylation, thereby providing a mechanism whereby OS decreases insulin sensitivity (70,220). In line with this, deletion of JNK through genetic knockout or through a JNK-inhibitory peptide improves insulin sensitivity in mice (221,222).

Consistent with the evidence reviewed above, haploinsufficiency of FoxO1 restores insulin sensitivity and rescues the diabetic phenotype in insulin-resistant mice by reducing hepatic expression of glucogenetic genes and increasing adipocyte expression of insulin-sensitizing genes (62). Conversely, a gain of function mutation of FoxO1 targeted to liver and pancreatic β-cells increases glucose production by the liver and impairs β-cell compensation resulting in diabetes. Similarly, enforced expression of FoxO1 in skeletal muscles impairs glycemic control after glucose or insulin administration (223). Furthermore, in line with the critical role of ROS in diabetes, the antioxidant NAC protects against diabetes in diabetic fatty rats and db/db mice and preserves insulin content and insulin gene expression (224,225). And, insulin resistance is prevented or retarded by overexpression of ROS scavenging enzymes such as SOD and catalase (219,226). Genetic deletion of p66^shc prevents hyperglycemia-induced endothelial dysfunction (49,227).

Increased production of ROS is the main cause of the development of insulin resistance by TNFα and glucocorticoids (219), two agents that induce rapid bone loss. Moreover, gene expression analysis in human osteoblasts exposed to glucocorticoids reveals that a significant number of transcripts related to OS are altered (228). The proapoptotic action of glucocorticoids and TNFα in cells of the osteoblast lineage is also mediated by ROS (229). Likewise, glucocorticoids and TNFα promote FoxO activity in a ROS-dependent manner and inhibit Wnt-induced osteoblastogenesis. Most importantly, administration of glucocorticoids to mice leads to an increase in ROS levels in the bone marrow, and this effect is prevented by NAC (229).

In addition to its role in bone homeostasis, Wnt/β-catenin signaling plays a critical role in glucose and lipid metabolism as well as atherosclerosis. Thus, a single missense mutation in LRP6, the coreceptor for the Wnt-signaling pathway, has been genetically linked with diabetes and osteoporosis, as well as early coronary artery disease, hyperlipidemia, and hypertension (230). Moreover, TCF4, the partner of β-catenin in the canonical Wnt-signaling pathway has recently emerged as the strongest type 2 diabetes susceptibility gene (231,232,233,234). In agreement with the genetic evidence in humans, β-catenin regulates glucagon-like peptide-1 secretion in mice (235). In addition, overexpression of Wnt10b in two murine models of obesity with marked hyperinsulinemia and insulin resistance (the ob/ob and the lethal yellow agouti) results in improved glucose homeostasis due to improved insulin sensitivity (236). Therefore, as we originally proposed elsewhere (102) and others have since concurred (104,237,238), antagonism of Wnt signaling by OS with increasing age, through the diversion of β-catenin from TCF- to FoxO-mediated transcription, may be a common mechanism that contributes to the development of not only involutional osteoporosis but also other diseases like insulin resistance, hyperlipidemia, coronary artery disease, and neurodegenerative disorders, all of which are more prevalent with advancing age.

Type 2 diabetics have low bone turnover and increased fracture risk, but they generally have normal or unchanged bone mass. This raises the possibility that mechanical strains on bone imposed by the increased bone mass index in such patients counteract the suppressive effects of insulin resistance on bone formation. Abnormal bone quality due to abnormal collagen or other components of the bone matrix, perhaps caused by advanced glycation end-product formation accelerated by hyperglycemia and OS, may contribute to the increased fracture risk in this condition (239).

IX. Lipid Oxidation, Oxidative Stress, PPARγ, and the Link between Osteoporosis and Atherosclerosis

A vast literature dating back to 1979 has established a critical role for lipid oxidation in the development of atherogenesis (240). In addition, epidemiological evidence shows that atherosclerosis and osteoporosis are linked (241,242,243). Thus, bone loss and vascular calcification progress in parallel with advancing age, and aortic calcification is inversely related to bone density and directly related to fractures in postmenopausal women (241). In fact, women in the highest quartile for increased atherosclerosis exhibit four times greater yearly bone loss than women in the lowest quartile.

Aging increases the differentiation of mesenchymal stem cells (MSCs) into adipocytes in many tissues, including the bone marrow (244). Increased marrow adiposity on the other hand, is associated with age-related bone loss (245,246,247). Moreover, extensive evidence accumulated during the last decade shows that one of the mechanisms responsible for reduced osteoblast production during aging is diversion of early MSCs into the adipocyte lineage at the expense of osteoblasts. Lineage commitment of MSCs to adipocytes is strongly dependent on the activation of the nuclear hormone receptor PPARγ by its ligands, which include oxidized lipids, prostaglandin J2, as well as thiazolidinediones (248,249,250,251,252,253,254,255).

The importance of PPARγ in skeletal homeostasis has been recently highlighted by evidence that mice lacking PPARγ in fat or bearing only one copy of the PPARγ gene exhibit increased bone mass associated with increased osteoblastogenesis and decreased adipogenesis (130,256). Therefore, endogenous PPARγ ligands must be acting as negative regulators of bone formation. In line with this view, activation of PPARγ by oxidized lipids or rosiglitazone promotes the development of adipocytes at the expense of osteoblasts in vitro (257). In addition, rosiglitazone induces bone loss in mice and humans, and this effect is associated with increased marrow adiposity, decreased osteoblast number and bone formation rate, and increased osteoblast and osteocyte apoptosis (258,259,260).

Commitment of multipotential progenitors to a specific lineage involves not only activation of a particular differentiation program but also suppression of programmed cell death. Cells actively suppress programmed cell death by continuous synthesis of specific genes, such as the α4 subunit of protein phosphatase 2A, indicating that apoptosis is the default cell fate (261). Consistent with this, antiapoptotic genes such as Mcl-1 and Bcl-2 are required for the survival of hematopoietic and melanocyte stem cells and mediate the effects of cytokines on commitment to a specific lineage (262,263,264). Hence, PPARγ-mediated commitment to the adipocyte lineage may also involve stimulation of apoptosis of the alternative osteoblast lineage. In line with this thinking, thiazolidinediones stimulate the apoptosis of osteoblasts in vitro (265) and in vivo (258).

Strong evidence in support of a link between the generation of oxidized lipids and osteoporosis has recently been provided by genetic and biochemical studies of the lipoxygenase Alox15 gene (255,266,267). As depicted in Fig. 6, Alox15 adds oxygen to polyunsaturated fatty acids (PUFAs) and, in the process, the catalytic iron is reduced to the ferrous form. The hydroperoxide product decomposes into a hydroxy derivative to generate a stable oxidation product, which can then bind to PPARγ to exert the skeletal effects. In the process of this decomposition, a hydroxy radical is released (268). Recent work by Almeida et al. (269) in our group indicates that during aging, Alox15 activity increases, as does lipid oxidation, and this probably contributes to the increased osteoblast apoptosis and other cellular changes leading to osteoporosis. Alox15 catalyzes the formation of several ligands of PPARγ, including 9-HODE, 13-HODE, 12-HETE, and 15-HETE (267,270) (Fig. 6). To make matters worse, ROS are generated by Alox15, and ROS in turn further oxidize PUFAs by nonenzymatic means (270). Hence, the chain reaction initiated by ROS has the potential to amplify the original oxidative signal by two to three orders of magnitude, resulting in the generation of large amounts of lipid peroxidation products. To add to this vicious cycle, hydroperoxy-PUFAs generated either enzymatically by lipoxygenases or by the actions of ROS, nonenzymatically decompose into α,β-unsaturated aldehydes, of which 4-hydroxynonenal (4-HNE) is a prototype (271). Such α,β-unsaturated aldehydes indirectly increase ROS by reacting with GSH, thereby depleting cells of this critical component of the coupled GSR/Gpx antioxidant system (272,273).

Figure 6.

Oxidized fatty acids, their degradation products, and bone formation. With advancing age, increased Alox15 expression and increased OS promote lipid peroxidation by adding oxygen to PUFAs, like linoleic acid. The hydroperoxide product of this reaction, 9-HPODE, decomposes into a hydroxy derivative 9-HODE, which binds to and activates PPARγ. Hydroxy radicals generated during this process add to the redox burden of the cell and can further stimulate PUFA peroxidation by nonenzymatic means. Importantly, 9-HPODE is also converted to 4-HNE, itself a potent prooxidant agent. Oxidized PUFAs activate PPARγ and promote its association with β-catenin, resulting in β-catenin degradation. ROS-activated FoxOs divert β-catenin from TCF- to FoxO-mediated transcription and thereby attenuate the restraining effect of β-catenin/TCF on the transcription of PPARγ. The combination of decreased Wnt signaling and increased PPARγ levels as well as PPARγ ligands leads to attenuation of osteoblastogenesis and increased osteoblast/osteocyte apoptosis along with increased adipogenesis (at some sites), thereby suppressing bone formation.

Alox15-null mice exhibit increased bone mass. Conversely, a mouse strain that bears a mutant allele of the Alox15 gene (DBA/2) characterized by 20-fold elevated levels of the enzyme exhibits decreased bone mass compared with C57BL/6 mice or DBA/2 congenic mice bearing a portion of chromosome 11 containing the wild-type allele of C57BL/6. Importantly, E₂ decreases oxidized low-density lipoproteins (274), and administration of an Alox15 inhibitor attenuates ovariectomy-induced bone loss and increases bone mass in mice with osteoporosis due to overexpression of IL-4, a stimulator of Alox15 expression (266). These observations strongly suggest that Alox15 generates oxidized lipids that bind to and activate the restraining effect of PPARγ on osteoblast production and bone formation.

Advancing age in C57BL/6 mice causes OS in bone and bone marrow, increases osteoblast apoptosis, and decreases bone formation (113). Advancing age also causes an increase in the expression of Alox15, PPARγ, and 4-HNE (269). Importantly, 4-HNE is a very potent inducer of p66^shc phosphorylation and osteoblastic cell apoptosis. Using cultured cell models, Almeida and co-workers (269) have also unraveled an oxidized PUFA-activated ROS/FoxO/PPARγ/β-catenin cascade, which explains how a rise in oxidized lipids caused increased OS, increased PPARγ expression, and reduced canonical Wnt signaling in osteoblasts and osteoblast progenitors (Fig. 6). As depicted in the model, lipid oxidation initiates this cascade by generating 4-HNE, which increases ROS and thereby activates FoxO. This results in diversion of β-catenin from proosteogenic TCF-mediated transcription to antioxidant FoxO-mediated transcription, as previously described for H₂O₂-induced OS (101). The decrease of β-catenin not only attenuates canonical Wnt signaling, but also unleashes the expression of PPARγ, which is normally suppressed by β-catenin/TCF transcription (275,276,277). The increase in PPARγ levels serves as an additional β-catenin sink by sequestering it and promoting its proteasomal degradation (278). Increased lipoxygenase and PPARγ expression is probably autoamplified, as evidenced by the increased expression of PPARγ, Alox12, and Alox15 in muscle cells stimulated with oxidized PUFAs (94). Considering the facts that canonical Wnt signaling increases (94,95,279,280) whereas Alox15 and PPARγ decrease osteoblast number (130,266), the findings of Almeida et al. (269) strongly suggest that the ROS/FoxO/PPARγ/β-catenin cascade accounts, at least in part, for the osteoblast deficit and the loss of bone with age. As discussed in Section V.C, FoxOs attenuate OS by stimulating the production of antioxidant enzymes, and β-catenin is a partner in this effect. Therefore, a decrease of the absolute amount of β-catenin, secondary to its degradation by PPARγ, should also compromise the ability of osteoblast progenitors to mount an antioxidant response, reducing further the generation of osteoblasts. Such a decrease in osteoblast number coupled with increased production of adipocytes is of course an invariable feature of old age in mice and humans.

In summary, the evidence reviewed in this section indicates that oxidized fatty acids and their degradation products in bone increase with advancing age secondary to elevated levels of Alox15 and ROS. This leads to activation of PPARγ, which increases adipogenesis at the expense of osteoblastogenesis. In addition, oxidized fatty acids and their degradation products stimulate apoptosis of osteoblasts and osteoblast progenitors via PPARγ-dependent and -independent mechanisms. Lipid radicals are detoxified by the Gpx\GSR and Trx peroxidase\reductase enzyme systems to form hydroxy fatty acid derivatives, and in the process of handling these radicals overall oxidative defense is lowered. As discussed in Sections VI and VII, both estrogen deficiency and aging decrease the activity of antioxidant enzymes, such as GSR and Trx reductase. Therefore, estrogen deficiency may potentiate the adverse effects of aging not only on bone but also on lipid metabolism as a result of decreased antioxidant defense and releasing the suppressive effect of estrogens on low-density lipoprotein production (281). In support of these ideas, increased generation of oxidized lipids by Alox15 has been linked with several age-related diseases including atherosclerosis and Alzheimer’s disease (282,283,284,285). In addition, lipid peroxidation in atherosclerotic plaques increases with age in humans in association with both nonenzymatic (ROS) and enzymatic (lipoxygenase) mechanisms (285).

X. Aging, Endogenous Hyperglucocorticoidism, and Bone Strength

A. Cell autonomous effects of glucocorticoids on bone

As opposed to earlier ideas that the adverse effects of glucocorticoids on bone are secondary to effects on calcium handling and the production of PTH or gonadal steroids, it is now quite clear that the deleterious effects of glucocorticoid excess on bone result from direct effects on osteoblasts (and their progenitors), osteocytes, and osteoclasts (286,287,288,289,290). Indeed, glucocorticoid excess suppresses osteoblastogenesis, strongly and rapidly stimulates osteoblast and osteocyte apoptosis, and prolongs the lifespan of osteoclasts. Evidently, the proapoptotic effect of glucocorticoids on osteocytes is initiated by detachment of their cellular processes from the lacunar-canalicular walls as a result of disengagement of integrins from proteins of the extracellular matrix (291). Glucocorticoids cause this detachment-induced apoptosis, also known as anoikis, via a glucocorticoid receptor-mediated rapid inside-out signaling leading to activation of the Pyk2 kinases that oppose the prosurvival signals of the association of integrins with focal adhesion kinases.

To address the contribution of the direct effects of glucocorticoids on osteoblasts and osteocytes to the adverse effects of these agents on bone mass and strength, O’Brien et al. (37) from our group have generated transgenic mice in which osteoblasts and osteocytes are protected from the direct effects of glucocorticoids, by virtue of the fact that they overexpress 11β-HSD type 2, an enzyme that inactivates glucocorticoids. Short-term administration of glucocorticoids to these mice induced equivalent bone loss to that observed in wild-type mice. However, the increase in osteoblast apoptosis that occurs in wild-type mice is prevented in the transgenic mice. Consistent with this, osteoblasts, osteoid area, and bone formation rate are significantly higher in glucocorticoid-treated transgenic mice compared with glucocorticoid-treated wild-type mice. Glucocorticoid-induced osteocyte apoptosis is also prevented in transgenic mice. Strikingly, the loss of vertebral compression strength observed in glucocorticoid-treated wild-type mice was prevented in the transgenic mice, despite equivalent bone loss. These results demonstrate that excess glucocorticoids directly affect bone-forming cells in vivo. Furthermore, these results have strongly suggested that glucocorticoid-induced loss of bone strength results in part from increased death of osteocytes, independently of bone loss.

B. Glucocorticoids and bone strength

Glucocorticoid excess, similar to aging, decreases bone strength disproportionately to its adverse effect on bone mass (292). In fact, the decline of bone strength with glucocorticoids precedes the decline in bone mass (292,293,294). Aging in humans blunts glucocorticoid feedback inhibition of ACTH (295), stimulates conversion of inactive to active glucocorticoids (296), and increases endogenous glucocorticoid production (202). Hyperglucocorticoidism in turn induces endothelial OS and vasoconstriction (297). Furthermore, both aging and glucocorticoid excess cause a reduction in blood flow and the volume of water present in the skeleton (298,299,300,301,302,303). Specifically, less water can be evaporated from older human bone (304), and the water fraction of human vertebrae decreases with age (305). Bone water volume decreases with glucocorticoid administration (302), and skeletal blood flow decreases with aging in rats (303). In addition, both aging and glucocorticoid excess decrease vascular endothelial growth factor (VEGF) in osteoblasts (306,307,308).

C. Angiogenesis and bone

Blood vessels are an essential component of the basic multicellular unit (BMU)—the anatomical structure responsible for bone remodeling—and are most likely the conduits by which both osteoclast and osteoblast precursors reach the site that is targeted for remodeling (309,310) (Fig. 7). Furthermore, bone formation almost always takes place in close proximity to capillaries (311). Thus, during development and fracture healing, the differentiation of osteoblasts requires blood vessel development. Recently, Sacchetti et al. (312) demonstrated that the pericytes (also known as mural cells) associated with the abluminal surface of vessels and sinusoids of the bone marrow are the MSC progenitors of osteoblasts. Remarkably, pericytes also participate in the development of blood vessels by producing an extracellular matrix to support endothelial cells and by providing growth factors like TGF-β. This dual functionality may contribute to the close linkage of vascularization and bone formation. Another link may be provided by growth factors produced by endothelial cells, including BMPs and endothelins (311).

Figure 7.

Photomicrograph of a cross-section of a murine, cancellous BMU comprising osteoclasts and osteoblasts along with two capillaries containing erythrocytes [courtesy of Robert S. Weinstein from the University of Arkansas for Medical Sciences]. Osteoclasts are identified by their discrete tartrate-resistant acid phosphatase-positive red granules, and the osteoblasts by their large nuclei with multiple nucleoli and underlying light blue osteoid. An abundance of osteocytes can be seen embedded individually (blue staining cells) within the mineralized bone (green) surrounding the BMU. Methyl green and tartrate-resistant acid phosphatase staining of undecalcified bone viewed with Nomarski differential interference contrast microscopy (X630).

Blood vessels develop during tissue growth, wound healing, and tissue remodeling to deliver oxygen and nutrients and remove toxic metabolites. During neovasculogenesis in the adult, new vessels may arise from precursors or from preexisting vessels by branching. Both functions are stimulated by hypoxia via stabilization of the Von Hippel-Lindau transcription factor that drives the synthesis of VEGF. VEGF, along with angiopoeitin-1, also promotes the maturation of endothelial cells and the recruitment of pericytes to the vessels to promote their maturation. Genetic studies have established that low oxygen tension, which develops as the bone matrix expands, stimulates the expression of VEGF by osteoblasts and initiates a neovasculogenic program (313,314,315). VEGF may also act indirectly by its ability to stimulate the synthesis of Cyr61 by cells of the osteoblast lineage. This protein is deposited into the extracellular matrix where it promotes neovascularization by interacting with integrins on endothelial cells (316). Endothelial growth factor also plays a key role in angiogenesis by stimulating endothelial cell replication, survival, adhesion, and migration. Interestingly, both Wnt signaling and TGF-β act in concert with VEGF to promote neovascularization, and cultured endothelial cells respond to activation of Wnt/β-catenin signaling with increased proliferation and decreased apoptosis (317,318).

D. Bone hydration

Water represents at least 25% of the wet weight of bone (305) and confers to bone much of its unique strength and resilience by reducing bone stresses during dynamic loading and producing a 2.5-fold increase in ultimate strength (319). Conversely, fracture resistance of hard tissues is defective in the absence of water (320,321). Of the total volume of water, approximately 90% resides in the lacunar-canalicular system that surrounds the intricate network of osteocytes (former osteoblasts entombed within the mineral), bone vasculature, and bone marrow (304) (Fig. 7).

Recent work from our group has strongly suggested that angiogenesis is also critical for bone strength (322). Indeed, we have obtained evidence that the vascularity of the skeleton and perhaps, as a consequence, the osteocyte lacunar-canalicular fluid volume and thereby skeletal hydration are critical determinants of bone strength, independently of bone mass. Moreover, in full agreement with the evidence from humans, we found that aging in C57BL/6 mice is associated with an increase in the adrenal production of glucocorticoids and their conversion from inactive to active glucocorticoids, as evidenced by increased corticosterone levels, adrenal weight, and the expression of 11β-HSD1 (the enzyme that converts inactive to active glucocorticoids) in bone (322). Even more strikingly, both aging and exogenous glucocorticoids decrease bone interstitial fluid in the osteocyte lacunar-canalicular network, decrease the volume of the bone vasculature, suppress VEGF production and action, and also suppress angiogenesis. Furthermore, the effects of aging or glucocorticoid administration on bone angiogenesis, interstitial fluid, and vasculature volume are greatly attenuated or absent in the mice in which osteoblasts and osteocytes are protected from the actions of glucocorticoids by means of transgenic overexpression of 11β-HSD2 (322). Collectively, these findings strongly suggest that increased glucocorticoid production and sensitivity with increasing age contribute to the development of skeletal fragility by decreasing skeletal fluid flow and the volume of the vasculature in bone, and thereby skeletal hydration. The molecular underpinning of these effects is decreased angiogenesis resulting from decreased VEGF production by osteoblasts/osteocytes as well as decreased VEGF action. The possibility that these changes represent a chain of interconnected pathogenetic mechanisms is supported by anatomical evidence that the cellular processes of osteocytes are in fact in direct contact with the bone vasculature (Fig. 8) (185). As discussed earlier, similar to OS, glucocorticoids stimulate FoxO-mediated transcription and suppress β-catenin/TCF-mediated transcription in osteoblast progenitors (229). It is, therefore, quite plausible that the decreased production of VEGF by glucocorticoids is caused, at least in part, by increased OS and the corresponding FoxO-mediated inhibition of HIF-1α-induced VEGF expression. I submit that the elucidation of these molecular, cellular, and tissue changes represents an important breakthrough in understanding the phenomenon of increased skeletal fragility with age.

Figure 8.

The connection between osteocytes and blood vessels. Left, Low (upper panel) and high (lower panel) magnifications of electron microscopy images demonstrating reliefs of the osteocytes and their canalicular network following acid-etching of murine bone sections [courtesy of Lynda Bonewald from the University of Missouri and Kansas City Dental School]. Please note multiple attachments of the osteocyte processes to the vessels depicted in the center of these images. Right, Cartoon depicting the same connections. [From M. L. Knothe Tate et al.: Bone 22:107–117, 1998 (366). Reproduced with permission of the Journal © Elsevier.]

XI. Oxidative Stress and Nutrient-Dependent Deacetylases (Sirtuins) as Therapeutic Targets

During the last few years, a class of nutrient-sensing nicotinamide adenine dinucleotide-dependent protein deacetylases, termed sirtuins, has been shown to play a crucial role in the resistance against OS and the promotion of longevity (323,324,325,326,327). The prototype of this gene family in vertebrates is the histone deacetylase silent information regulator T-1 (SIRT-1). Sirtuins 1, 2, and 3 prolong life-span; improve energy expenditure, glucose production, and insulin sensitivity; and protect against high-fat diet-induced metabolic damage via several mechanisms, including the deacetylation of FoxOs and modulation of FoxO-mediated transcription (328). The importance of the latter mechanism has been recently highlighted by evidence that augmentation of FoxO3a-dependent MnSOD and catalase expression in mice by SIRT3 decreased cellular levels of ROS and thereby protected the mouse heart from cardiac hypertrophy (329).

Based on evidence that the polyphenol resveratrol (3,5,4′-trihydroxy-trans-stilbene), abundantly present in red wine, activates SIRTs and mimics the ability of dietary restriction to prolong lifespan in lower organisms, Pearson et al. (330) fed mice with resveratrol and found that it induces gene expression patterns in multiple tissues that parallel those induced by dietary restriction and every-other-day feeding. More important, resveratrol-fed elderly mice showed a marked reduction in signs of aging, including the preservation of BMD. Nonetheless, resveratrol did not influence the longevity of these mice. On the other hand, resveratrol increased longevity in mice fed with a high-fat diet (331). Furthermore, resveratrol prevented ovariectomy-induced bone loss and osteoblast apoptosis (332).

Consistent with the antiosteoporotic efficacy of resveratrol in rodents, several laboratories have recently shown that this polyphenol, probably acting via the activation of SIRT-1 and PPARγ suppression, decreases bone marrow adipogenesis whereas it promotes osteoblast progenitor proliferation and differentiation and prevents osteoclast formation in vitro (333,334,335). More importantly, Edwards et al. (336) have reported that SIRT-1 is required for the maintenance of normal bone mass because mice null for SIRT-1 have decreased bone mass associated with decreased osteoblast differentiation and increased osteoclast activity. These changes are associated with a 27% increase in osteoclast numbers and a 39% decrease in osteoblast numbers compared with wild-type animals. Furthermore, in agreement with evidence that endothelial NO synthase (eNOS) null mice exhibit a similar bone phenotype to that of the SIRT-1 null mice and SIRT-1 stimulates eNOS in other cell types (337), Edwards and co-workers (338) found that resveratrol enhances eNOS mRNA expression and increases the levels of the eNOS product nitric oxide (NO) in osteoblasts. In addition, NO donors enhance BMP-2 mRNA expression, osteoblast differentiation, and bone formation. On the other hand, eNOS-deficient osteoblast progenitors exhibit decreased BMP-2 transcription and osteoblast differentiation and fail to respond to resveratrol. This suggests that resveratrol increases SIRT-1 expression and activity, and in turn eNOS expression and activity, and by this mechanism, stimulates BMP-2 expression, osteoblast differentiation, and bone formation.

To follow up these observations, Edwards et al. (339) have generated mice with cell-specific deletion of SIRT-1 in osteoblasts, osteoclasts, or both cell types. SIRT-1-specific deletion in cells of the osteoblast lineage (using the 2.3 Col1α1 promoter) causes a decrease in bone volume accompanied by a decrease in osteoblast number and the rate of bone formation, but no change in osteoclast number. Similarly, deletion of SIRT-1 in osteoclast precursor cells, including monocytes and macrophages (using the Lysozyme M promoter), results in decreased bone volume; however unlike the osteoblast-specific deletion, the number of osteoclasts is increased compared with control mice. Crossing these models to generate an osteoblast/osteoclast-specific SIRT-1 knockout resulted in mice with up to 29% less bone volume than control animals. These findings strongly support the conclusion that SIRT-1 controls bone mass by independently regulating cells of both the osteoblast and osteoclast lineage and provide yet another molecular link between aging, osteoclast and osteoblast number, and bone loss.

A critical mediator of the effects of sirtuins on energy expenditure is PPARγ coactivator-1α (PGC-1α), a coactivator protein that interacts with several transcription factors, including FoxOs. Like FoxOs, PGC-1α promotes the transcription of antioxidant enzymes in response to oxidative stress (340,341,342,343). SIRT-1, PGC-1α, and FoxOs function together to accommodate metabolic adaptation to fluctuations in nutrient supply (344). Similar to FoxO-deficient mice, PGC-1α-deficient animals exhibit increased levels of ROS and reduced ability to withstand oxidative damage (345). Activation of PGC-1α/FoxO and subsequent reduction of energy expenditure contribute to decreased fat accumulation. Importantly, PGC-1α also promotes mitochondria biogenesis and thereby the production of ROS and OS, and ROS are feeding forward cAMP-responsive element-binding protein to increase PGC-1α expression (346). Caloric restriction, without malnutrition, represents a promising strategy for life extension and the delay in the onset of age-associated pathologies, and it is thought to work through the PGC-1α-induced mitochondria biogenesis program (347,348,349). Nonetheless, whereas the beneficial effects of caloric restriction on glucose homeostasis are well documented, the effect of caloric restriction on bone metabolism, if any, is minor and perhaps negative. Indeed, caloric restriction for 24 wk in young mice decreased cortical bone mass but had no effect on cancellous bone (350). And, caloric restriction in humans either had no effect on BMD or bone mineral content or led to a reduction of bone mass in the spine, total hip, and femoral neck (351,352).

More specific SIRT-1 agonists, such as SRT1720 or SIRT-1 activators unrelated to resveratrol (with 1000-fold higher potency), are very effective in reducing liver lipid accumulation, protecting from diet-induced obesity, and improving glucose and insulin homeostasis in mice (353,354,355). However, the effect of these new agents on bone is currently unknown.

In keeping with the theme that increased defense against OS may be an effective antiosteoporotic therapy, Jilka and co-workers from our group (unpublished data) have investigated the efficacy of intermittent PTH for old age osteoporosis by comparing its effects in young (age, 6 months) and old (age, 26 months) C57BL/6 mice. Similar to observations by others (356), they found that 4 wk of daily PTH administration (100 ng/g) is as effective in stimulating bone formation and increasing bone mass in old mice as it is in young mice. Moreover, after 4 wk of intermittent administration, PTH reduces the level of ROS and increases the level of GSH and GSR in the bone marrow, decreases the phosphorylation of p66^shc in vertebral lysates, and decreases the prevalence of osteoblast apoptosis in sections of vertebral cancellous bone. Moreover, PTH also attenuates the age-related increase in expression of the lipoxygenase Alox15, the enzyme known to generate ROS via the production of the prooxidant 4-HNE. As opposed to PTH, the antioxidant NAC suppresses osteoblast differentiation in vitro and reduces the number of osteoblast progenitors in vivo. Moreover, NAC suppresses Wnt-stimulated β-catenin/TCF-mediated transcription, whereas it is well established that PTH stimulates Wnt signaling by suppressing the Wnt antagonist sclerostin (357,358) and by activating LRP6 (96). The ability of antioxidants to prevent sex steroid deficiency, but not old age-induced bone loss, suggests that they are anticatabolic agents that also suppress osteoblast formation—a process that depends on signaling by low levels of ROS. PTH on the other hand can uniquely reverse the age-dependent decline in osteoblastogenesis because of its ability to selectively suppress OS, stimulate Wnt signaling, and attenuate osteoblast apoptosis, and thereby promote de novo bone formation.

XII. Summary and Conclusions

Epidemiological evidence indicates that in both women and men bone loss begins as early as the early part of the third decade—long before any change in sex steroid production. In addition, evidence primarily from mechanistic studies in rodents shows that advancing age, and specifically increased OS, is a fundamental mechanism of the loss of bone mass and strength. ROS are continuously generated as normal by-products of aerobic metabolism and can cause protein damage and DNA lesions leading to cell death, but they also serve as signaling molecules at lower concentrations to control cell proliferation and differentiation in a plethora of cell types. Consistent with this, mitochondria biogenesis and ROS play a very important role in the generation and survival of osteoclasts, osteoblasts, and osteocytes. Loss of estrogens or androgens decreases defense against OS, and this accounts for the increased bone resorption associated with acute sex steroid deficiency as well as the effects of either class of sex steroids on osteoclastogenesis and the apoptosis of osteoclasts and osteoblasts.

Genetic analysis of the role of the FoxO family of transcription factors in bone indicates that FoxO-dependent oxidative defense provides a mechanism to handle the oxygen free radicals constantly generated by the aerobic metabolism of osteoblasts and is therefore indispensable for bone mass homeostasis. FoxO activation in early mesenchymal progenitors also diverts ß-catenin away from Wnt signaling, leading to decreased osteoblastogenesis. The importance of this latter mechanism is supported by evidence that OS and FoxO activation are implicated not only in the pathogenesis of type 1 and 2 diabetes but also in the adverse effects of diabetes on bone formation. Similarly, attenuation of Wnt signaling by the activation of PPARγ by ligands generated from free fatty oxidation and a ROS/FoxO/PPARγ/β-catenin cascade contributes to the age-dependent decrease in bone formation, suggesting a mechanistic explanation for the link between atherosclerosis and osteoporosis. Lastly, increased glucocorticoid production and sensitivity with increasing age contribute to the suppression of bone formation. These age-dependent changes also contribute to development of skeletal fragility by decreasing skeletal fluid flow and the volume of the vasculature in bone, and thereby skeletal hydration. The molecular underpinning of these effects is decreased angiogenesis resulting from decreased VEGF production by osteoblasts/osteocytes as well as decreased VEGF action.

Elucidation of these mechanisms provides a paradigm shift from the “estrogen-centric” account of the pathogenesis of involutional osteoporosis to one in which age-related mechanisms intrinsic to bone and OS are protagonists and age-related changes in other organs and tissues, such as ovaries, accentuate them (Table 2). It also provides an understanding of osteoporosis that is well aligned with what is known about the biology of aging and the mechanisms of several other degenerative disorders—which often are present in the same patient and inexorably share molecular pathogenetic mechanisms related to aging per se. Elucidation of determinants of bone strength independent of bone mass not only provides fresh insights into the decline of bone strength with age and may explain the apparent mismatch between bone strength and bone mass that characterizes aging and glucocorticoid excess in animals and humans alike, but also suggests novel therapeutic opportunities for preventing it—in distinction from the current alternatives that aim at altering bone mass. Lastly, the evidence for the effectiveness of sirtuin ligands in both insulin resistance and osteoporosis in rodents (330) raises expectations that in the near future this and other classes of pathogenesis of aging-tailored drugs may provide a one-drug unified treatment for several degenerative diseases.

Table 2.

Pathogenetic mechanisms of involutional osteoporosis

1. OS

2. Decreased defense against OS, resulting from sex steroid deficiency

3. Increased lipid oxidation/PPARγ activation (partially through OS)

4. Endogenous hyperglucocorticoidism (partially through OS)

Almost 10 yr ago, I had reviewed in this journal the cellular and molecular mechanisms of osteoporosis and highlighted the overriding importance of osteoclast, osteoblast, and osteocyte numbers for bone homeostasis, as well as the paramount importance of the aberrations of the rate of birth and death of these cells in the pathogenesis of osteoporosis (1). In the present article, I have dealt with evidence accumulated in the interim that helps to unravel the complexities of the mechanisms whereby cell number can be dysregulated as a function of increasing age and how bone strength can decline independently from the decline of bone mass. At present, major advances in the field of OS and cell death indicate that enzymatic and transcriptional defenses preventing damage to intracellular components by free radicals (the subject of this article) are not enough to avoid cellular injury. A second line of defense, termed autophagy, is necessary for the repair and removal of damaged components of the cell in response to mild OS, via the ubiquitin/proteasome system and lysosomes (359,360,361). I strongly suspect that such mechanisms will be critical for the response of osteoclast and osteoblast precursors and their progeny to various levels of mild OS. Osteocytes, the bone cells with by far the most extensive lifespan, choreograph the process of bone remodeling by the short-lived osteoclasts and osteoblasts (362,363). Autophagy is likely to be critical for the ability of osteocytes to remain alive while sensing, and probably responding to, major changes in the ambient level of oxygen, at the same time as they deal with the ravages of aging, including the nuclear leakiness that accelerates with age in postmitotic cells (364,365). Understanding these events in the context of skeletal aging in the future will most likely dictate yet another revision of our current ideas about the pathogenesis of osteoporosis.