Pierre Marie, Inserm U606, Hopital Lariboisiere, 2 rue Ambroise Pare, 75475 Paris Cedex 10, France. Tel.: +33 1 49 95 63 89; fax: +33 1 49 95 84 52; e-mail: email@example.com
Human aging is associated with bone loss leading to bone fragility and increased risk of fractures. The cellular and molecular causes of age-related bone loss are current intensive topic of investigation with the aim of identifying new approaches to abolish its negative effects on the skeleton. Age-related osteoblast dysfunction is the main cause of age-related bone loss in both men and women beyond the fifth decade and results from two groups of pathogenic mechanisms: extrinsic mechanisms that are mediated by age-related changes in bone microenvironment including changes in levels of hormones and growth factors, and intrinsic mechanisms caused by the osteoblast cellular senescence. The aim of this review is to provide a summary of the intrinsic senescence mechanisms affecting osteoblastic functions and how they can be targeted to abolish age-related osteoblastic dysfunction and bone loss associated with aging.
During the adult life, the skeleton is continuously renewed and regenerated by a process called bone remodeling, which is in essence a bone regeneration mechanism aiming at the maintenance of bone mechanical integrity by removing old bone with high prevalence of fatigue microfractures and replacing it with new bone. Bone remodeling is characterized by a sequence of events occurring at bone surfaces, mediated by cellular units termed bone multicellular units comprised of a specialized group of cells, namely bone-resorbing osteoclasts and bone-forming osteoblasts and their precursors. Under steady-state conditions, the amount of bone removed by osteoclasts is equal to the amount of bone formed by osteoblasts, and a stable bone mass is maintained. Histomorphometric analysis of bone biopsies obtained from elderly men and women showed that age-related bone loss is caused by impaired bone formation compared to bone resorption. A decreased mean wall thickness (which is a histomorphometric measure of bone formation) in both trabecular and cortical bone indicates decreased osteoblastic bone-forming capacity (Brockstedt et al., 1993; Parfitt et al., 1995). Bone formation is dependent on the recruitment of sufficient number of osteoblasts as well as the activity of individual osteoblasts (Parfitt, 1991). Mature osteoblasts are assumed to be recruited mainly from a group of skeletal stem cells with osteogenic differentiation potential [referred to as skeletal, mesenchymal or stromal stem cells (MSC)]. In vitro, MSC are clonogenic and able to differentiate into osteoblasts, adipocytes, or chondrocytes (Owen, 1988). In vivo, MSC form a mixture of tissues including bone, cartilage, and hematopoietic-supporting stroma, and they exhibit capacity for self-renewal (Sacchetti et al., 2007). While the exact location of MSC in vivo is still debated, recent evidence suggests that some of these cells are pericytes located on the outer surface of blood vessels and sinusoids in the bone marrow (Sacchetti et al., 2007). Also, recent studies demonstrated the presence of a canopy of flat cells expressing osteoblastic markers covering active bone remodeling sites and associated with capillaries (Andersen et al., 2009), suggesting that osteoblastic cells may be recruited from a number of sources including pericytes and osteoprogenitor cells present in peripheral circulation or periosteum. Based on current understanding of the cellular events of bone formation phase of bone remodeling, several investigators have examined potential mechanisms causing age-related osteoblastic cell failure. As discussed further below, this review focuses on age-related processes that are broadly linked to impairment of proliferation, differentiation, and function of MSC or osteoblastic cells and grouped into separate topics.
Age-related skeletal stem cell atrophy
In bone, similar to other tissues, the availability of a sufficient number of stem cells is needed for continuous supply of mature, functionally competent cells that are required for normal tissue turnover and regeneration. Owing to the absence of specific markers, bone marrow MSC are usually identified by their ability to form colonies named colony-forming-unit-fibroblasts (CFU-f) that stain positive for alkaline phosphatase (AP) when cultured in vitro. Age-related changes in the number of bone marrow CFU-f have been reported in a number of species including rat and mice, with variable results, with some studies showing age-related decrease (Egrise et al., 1992; Quarto et al., 1995; Stolzing & Scutt, 2006), age-related increase (Xu et al., 1983), no evident age-related change (Brockbank et al., 1983) or, as recently demonstrated, a biphasic response, with an age-related increase in the number of osteogenic CFU-f from young to middle-aged mice (3–18 months) and then a rapid decrease (Zhang et al., 2008). Interestingly, in this study, the number of adipocytic colonies followed a similar pattern, suggesting a general atrophy of MSC. Human studies also revealed variable results with some studies showing age-related decrease in the number of CFU-f (Nishida et al., 1999; Stolzing et al., 2008; Kuznetsov et al., 2009) or no change (Oreffo et al., 1998; Stenderup et al., 2001). Also, the number of Stro-1+ (a marker of the clonogenic CFU-F) did not change with age (Stenderup et al., 2001; Zhou et al., 2008). Interestingly, the number of peripheral-blood-derived CFU-f was found to increase in patients with osteoporosis compared to age-matched controls (Dalle Carbonare et al., 2009). While the discrepancies in the results reported in human studies may be related to differences in the methods employed, it can be generally concluded that the reported age-related decline in the number of MSC occurred in early adulthood after the termination of skeletal growth, with no further change afterward. Thus, the contribution of decreased number of MSC to age-related impaired bone formation appears to be limited. Similar to MSC, small but statistically nonsignificant age-related changes in the number of stem cells were reported in the hematopoietic system (Sudo et al., 2000) and in skeletal muscle stem cells (Gibson & Schultz, 1983).
During serial long-term culture of normal human diploid cells and after the initial growth phase, cells experience a phase of delayed growth rate that ends in growth arrest. This phase is usually reached after 40–70 population doublings (PD) depending on the cell type and is termed the Hayflick limit (Hayflick, 1965). The limited lifespan reflected by the Hayflick limit suggests that cultured cells are controlled by intrinsic senescence-related cellular mechanisms that are relevant to the physiological aging of organisms (Campisi, 2001). Recent evidence demonstrated an age-related increase in the number of senescent cells in renewable tissues in vivo (Jeyapalan et al., 2007). Human osteoblasts (Kassem et al., 1997; Kveiborg et al., 2001) and human MSC (Stenderup et al., 2001; Stolzing et al., 2008) undergo replicative senescence related to the Hayflick limit during long-term culture. An age-related decline in the maximal lifespan from 30 to 40 PD in MSC cultures established from younger donors to 20 PD in MSC cultures of elderly donors was reported (Stenderup et al., 2003; Baxter et al., 2004). MSC established from elderly donors also exhibit increased expression of senescence markers (β-galactosidase) at early passage cultures, suggesting accumulation of senescent cells among the MSC cell populations in vivo (Stenderup et al., 2003; Zhou et al., 2008). These observations suggest that the age-related impaired proliferative capacity of MSC contributes to decreased bone formation in vivo.
Age-related decline in the functional lifespan of osteoblasts
The functional activity of osteoblasts during in vivo bone formation does not depend only on the number and rate of generation of osteoblastic cells but also on their functional lifespan (Manolagas, 2000). Apoptosis is a general mechanism utilized for regulating tissue regeneration, and it is the mechanism responsible for the termination of bone-forming activity of osteoblastic cells. It is estimated that around 60–90% of osteoblastic cells die by apoptosis at the end of bone remodeling phase in mice and probably in humans (Jilka et al., 2007). Age-related increase in cell apoptosis has been suggested to contribute to in vivo aging in some tissue compartments (Polster & Fiskum, 2004). Histomorphometric studies indicate that the age-related decrease in bone mass in mice is associated with increased osteoblastic and osteocytic apoptosis that may be caused by enhanced oxidative stress with age (Almeida et al., 2007). In one human study, no age-related changes in gene expression level of Bcl-2 (a marker for apoptosis) were detected in MSC obtained from 80 healthy subjects (Jiang et al., 2008), but in a recent study, Zhou et al. (2008) reported a age-related increase in the number of apoptotic cells in MSC cultures. These discrepancies may be related to the sensitivity of the method employed. Further studies are needed to demonstrate the causal role of age-related increase in apoptotic osteoblastic cells in mediating the age-related decrease in bone mass in humans.
Age-related decrease in osteoblast differentiation and function
Bone formation during bone remodeling requires the presence of a sufficient number of mature osteoblastic cells. Consequently, an impaired osteoblast differentiation from their precursors can lead to impaired bone formation. In rats, there is an accumulation of preosteoblastic cells and decreased number of mature osteoblasts with increasing age, suggesting that impaired osteoblast differentiation is a potential mechanism for age-related impaired bone formation (Roholl et al., 1994). Some in vitro studies have addressed this question by culturing osteoblastic cells from donors at different ages and examining production of osteoblastic markers as surrogate markers for their differentiation status. Some investigators found no age-related differences in basal levels of AP or osteocalcin expressed in osteoblastic cells derived from bone biopsies (Marie et al., 1989; Katzburg et al., 1999). On the other hand, Fedarko et al. (1992) found an age-dependent decrease in cellular production of collagen and decorin in osteoblastic cell cultures obtained from trabecular bone explants. Thus, impaired differentiation of MSC to osteoblasts may contribute to the age-related bone loss.
Age-related decreased osteoblastic response to calciotropic hormones and growth factors
Osteoblast cell proliferation and function are dependent on adequate response to growth factors and hormones known to control bone formation. In mice, aging has been reported to cause decrease in insulin-like growth factor (IGF)-I effects on osteoblast proliferation and differentiation owing to defective IGF-1 signaling (Cao et al., 2007). In humans, osteoblastic cells from aged donors exhibit decreased proliferative responses to growth hormone and platelet-derived growth factor compared with young donor cells (Pfeilschifter et al., 1993). Similarly, osteoblast cultures obtained from young subjects exhibited a better response to estradiol (Ankrom et al., 1998) and IGF-I (D’Avis et al., 1997) compared with cells obtained from older donors. On the other hand, there is an increasing recognition that the skeleton plays a role in energy metabolism and thus it is influenced by a number of hormones and factors that are gut-derived or involved in energy metabolism, e.g., serotonin, insulin, and leptin (Karsenty & Oury, 2010). For example, gut-derived serotonin has been demonstrated recently to control bone mass (Ducy & Karsenty, 2010). It is thus plausible that age-related changes in these molecules may contribute to age-related osteoblast dysfunction.
Age-related enhancement of adipocyte differentiation
In the bone marrow microenvironment, fat and bone co-exist and several histomorphometric studies have demonstrated that the observed decrease in trabecular bone volume with aging is associated with increased bone marrow adipocyte tissue volume. Also, advanced in vivo imaging technology showed that marrow fat is inversely correlated with bone mass (Griffith et al., 2006). The demonstration of the ability of skeletal stem cells (MSC) to differentiate to osteoblasts and adipocytes led to the hypothesis that enhanced adipocyte differentiation from MSC leads to reduction in the number of stem cells available for osteoblast differentiation and consequently bone formation (Rosen et al., 2009). This hypothesis, described as ‘the inverse relationship between adipocyte and osteoblast differentiation’, was supported by experimental evidence in MSC cultures (Bennett et al., 1991; Beresford et al., 1992). In addition to MSC, bone marrow contains a hierarchy of cells at different stages of differentiation. Recent evidence suggests that committed preosteoblastic and committed preadipocytic cells do exist (Post et al., 2008) and thus it is possible that fat and bone formation in the bone marrow microenvironment can be regulated at the level of stem cells or their committed precursors. Is aging associated with enhanced adipogenesis of MSC owing to intrinsic age-related alterations in programming? Studies in aged mice (Moerman et al., 2004) and in the senescence-accelerated mouse model (SAMP-6) (Kajkenova et al., 1997) revealed enhanced adipogenesis and impaired osteoblastogenesis in murine MSC cultures. On the other hand, one human study showed that the adipocyte-forming capacity of MSC does not change with donor age (Justesen et al., 2002). Similarly, no age-related change in the expression of mRNA levels of adipocyte and osteoblast differentiation markers was detected in humans (Justesen et al., 2002). Evidence from human studies suggests that age-related changes in bone microenvironment may play a role in directing MSC differentiation into osteoblasts. Sera (employed as surrogates for the microenvironment ‘seen’ by MSC) from elderly donors inhibited osteoblast differentiation (Abdallah et al., 2006) and enhanced adipogenesis of MSC (Stringer et al., 2007). Detailed understanding of the molecular pathways controlling the relationship between bone and fat in the bone marrow can potentially lead to novel approaches to abolish the age-related decrease in bone mass and increased bone marrow fat mass (Nuttall & Gimble, 2000).
Senescent osteoblastic cells create a defective microenvironment
Senescent cells exhibit a characteristic secretory phenotype involving inflammatory cytokines, growth factors, and matrix-degrading proteases that modulate and alter their local microenvironment. These events may contribute to aging of the organism (Campisi, 2005). The presence and the biological relevance of the age-related altered secretome to the aging skeleton are not known. However, some studies have demonstrated age-related changes in the concentration of osteoblast-active molecules such as IGF-I and TGF-β in the bone matrix (Bismar et al., 1999; Seck et al., 1999). Also, an osteoporotic phenotype was created in a normal healthy mouse by intramedullary injection of bone marrow cells obtained from the senescence-accelerated mouse prone 6 (SAMP6), a well-characterized model for accelerated aging, including an osteoporotic phenotype (Ueda et al., 2007). Owing to the limited engraftment of the transplanted cells, one explanation of these data could be a defective microenvironment in SAMP6 senescent and osteoporotic cells. A corollary of this suggestion is the rescue of the osteoporotic phenotype of SAMP6 mice by intramedullary injection of normal allogeneic bone marrow cells (Takada et al., 2006). In a recent study, circulating osteoblastic cells identified in peripheral blood from postmenopausal women were found to express several inflammatory and osteoclast-activating factors that may contribute to enhance bone resorption (Undale et al., 2010). It is clear that more detailed studies on the composition of the senescent skeletal microenvironment are required.
General molecular mechanisms mediating age-related osteoblast dysfunctions
The age-related impairment of osteoblastic functions can be viewed as a part of the general homeostatic failure experienced during aging and manifested in all organ systems, i.e., part of the generalized aging phenotype. Several general mechanisms causing intrinsic cellular defects that contribute to the senescent phenotype and osteoblast aging have been identified. These factors can be grouped into two defined groups: damage-related factors and maintenance (repair) factors. Disturbances in the balance between damage and maintenance factors increase the risk of cellular senescence and dysfunction. These senescence-related mechanisms affect both osteoblastic and osteoclastic cells and lead to bone loss (Fig. 1).
Telomerase deficiency and telomere shortening
One of the fundamental mechanisms responsible for age-related intrinsic cellular dysfunction is telomere shortening caused by incomplete replication of linear chromosomes by DNA polymerase (Campisi et al., 2001). Telomere length is maintained by telomerase, a ribonuclear protein complex consisting of an integral RNA (TERC), which serves as the telomeric template, and a catalytic subunit (TERT), which has reverse transcriptase activity, and associated protein components (Weinrich et al., 1997). The absence of telomerase activity in normal diploid somatic cells leads to telomere shortening and replicative senescence. Telomere shortening has therefore been implicated in the pathophysiology of several age-related diseases and premature aging syndromes (Campisi et al., 2001). Similar to other human diploid cells, MSC lack telomerase activity (Simonsen et al., 2002; Baxter et al., 2004) and exhibit telomere shortening in association with a replicative senescence phenotype in long-term culture (Simonsen et al., 2002; Baxter et al., 2004). ‘Re-telomerization’ of MSC through over-expression of the human TERT gene leads to elongation of telomeres, extension of lifespan, maintenance of ‘stemness’ characteristics, and enhanced bone forming in vitro and in vivo (Simonsen et al., 2002). The relevance of telomere shortening to in vivo aging is demonstrated by the presence of an osteoporotic phenotype in patients with excessive telomere shortening and telomere dysfunction (e.g., Werner’s syndrome and Dyskeratosis congenita) (Vulliamy et al., 2001; Crabbe et al., 2007). Furthermore, mice carrying the Werner mutation on a background of short telomeres (Wrn−/− Terc−/− deficient mouse) exhibit a low bone mass phenotype (Pignolo et al., 2008). In normal human populations, bone mass measured by DEXA exhibits a small but significant correlation with telomere length of peripheral blood leukocytes (Valdes et al., 2007). No difference in telomere length was however found in peripheral blood leukocytes in osteoporotic patients and age-matched controls (Kveiborg et al., 1999).
Telomere shortening seems to act upstream of other senescence-related mechanisms, and this leads to progenitor cell dysfunction. Cells with shortened telomeres exhibit enhanced DNA damage response resulting in up-regulation of expression of a number of cell cycle inhibitors and pro-apoptosis genes (e.g., p53) and consequently a decreased pool size of competent cells available for tissue regeneration (Song et al., 2009). Also, telomere shortening and absence of telomerase enzyme can lead to accumulation of senescent cells within the tissue and impaired cell responses to microenvironmental signals. This hypothesis suggests that somatic cell telomerase activation and telomere elongation can be one attractive option for ‘rejuvenation’ of osteoblastic cell populations and the enhancement of bone formation. One limitation of this approach is that long-term telomerization using gene modification in MSC can lead to genetic instability and tumor formation (Burns et al., 2005). Nevertheless, telomerase activity can be restored transiently to MSC by treatment with trichostatin A, which changes the epigenetic status of hTERT promoter (Serakinci et al., 2006). Thus, intermittent or transient telomerase activation using small chemical molecules or its conditional genetic activation may be a suitable approach for safe clinical intervention.
Age-related accumulation of oxidative damage
Damage to cellular components resulting from free radical production [reactive oxygen species (ROS)] has been proposed to cause aging (a process called ‘free radical hypothesis of aging’) (Holliday, 1995). Recently, a number of in vitro and in vivo studies have provided novel insights into the contribution of damage caused by free radicals to the impaired osteoblastic function and bone formation during aging (Manolagas, 2010). Age-related decrease in bone formation and osteoblast number is associated with increased ROS levels and decreased GSH reductase activity in bone marrow as well as increase in the phosphorylation levels of p66shc (an adapter protein that amplifies mitochondrial ROS generation and influence apoptosis) (Almeida et al., 2007). In addition, oxidative stress induces FoxO transcription factors known to activate genes involved in free radical scavenging as well as inhibiting Wnt signaling, leading to impaired bone formation (Ambrogini et al., 2010; Rached et al., 2010). The formation of oxidized lipids through ROS has been demonstrated to activate osteoblastic PPARγ signaling, leading to inhibition of bone formation and enhancing adipogenesis of skeletal stem cells (Almeida et al., 2009). These data demonstrate the important role of oxidative damage in age-related alterations in osteoblastogenesis.
Age-related impaired DNA repair and DNA damage responses
DNA is exposed to damage from both endogenous (e.g., ROS) and exogenous environmental factors (e.g., ionizing and UV radiation and environmental chemicals). If the rate of DNA damage exceeds the capacity of the cell to repair the damage, accumulation of DNA errors leads to somatic mutations owing to faulty repair or replication errors, potentially leading to senescence or cancer. In support of this hypothesis, human inherited diseases associated with faulty DNA repair and mouse models of defective DNA repair mechanisms exhibit premature aging and increased susceptibility for cancer (Hoeijmakers, 2009). Low bone mass and osteoporosis are characteristic features in these mouse models. For example, de Boer et al. (2002) created a mouse deficient in Xpd, a gene important for repairing a range of genetic lesions by nucleotide excision repair. The mouse exhibited premature aging phenotypes including osteopenia. p53 is one of the cellular responses to DNA damage, and its activation leads to cell cycle arrest, apoptosis, and senescence. A mouse model with enhanced activity of p53 revealed accelerated aging phenotype including bone loss (Tyner et al., 2002). The Ataxia-telangiectasia-mutated (ATM) gene encodes a Ser/Thr kinase, and its activation is one of the DNA double-strand break damage responses. It acts upstream of p53 and thus, in its absence, cells exhibit genomic instability, short telomeres, and premature senescence. A mouse model of ATM exhibits decreased bone mass owing to impaired bone formation and enhanced bone resorption (Rasheed et al., 2006). Thus, not only the degree of cellular damage, but also the balance between the damage and the repair mechanisms determines the cellular and tissue integrity, with functional consequences on bone formation and bone mass.
A large body of literature indicates that age-related impaired bone formation is the principal pathogenetic mechanism mediating age-related bone loss. The impaired bone formation results from age-related decreased osteoblast number and function as consequences of multitude intrinsic senescence-related mechanisms. A limited number of anabolic therapeutics are currently available for treating osteoporotic bone loss in aging adults. Targeting senescence-related osteoblastic dysfunction may create a novel class of ‘anti-aging’ therapeutics that may prove to be useful to promote bone formation and reduce bone loss associated with aging.