Aging is invariably linked to bone loss in both genders across all ethnic backgrounds. This skeletal deterioration is a direct result of cell-autonomous and hormonal changes. Both longevity and bone mass are governed by strong genetic elements. More than 30 years of studies in twins have shown that there is a strong genetic component that influences individual peak bone mass,1, 2 whereas estimates from studies of long-lived families propose that lifespan has a 30% genetic influence.3–5 Common genetic factors for both longevity and bone mass may therefore exist, linking aging and longevity to the maintenance of adequate bone mass throughout life.
The well-conserved sirtuin family of longevity-associated genes has been shown to regulate lifespan in numerous lower organism models by initiating gene silencing.6–8 Silent Information Regulator T1 (SirT1) is a NAD-dependent class III histone deacetylase, the predominant mammalian member of the sirtuin family, and is strongly linked to cell survival and proliferation.9, 10 Existing evidence indicates that sirtuins may contribute to pathological processes known to be associated with aging such as neurodegeneration,11, 12 in which direct interactions between SirT1 and NF-κB repress activation in microglia.13, 14
In bone, the NF-κB family of signaling molecules maintains normal bone mass through strict control of osteoclast formation and function by a complex regulatory system of inhibitory proteins, primarily IκBα.15 Amplified NF-κB signaling ultimately leads to increased osteoclast differentiation and resorptive activity, a hallmark feature of aging bone.16–19 Thus, age-related bone loss is treated clinically by inhibiting osteoclastic resorption with bisphosphonates or direct inhibition of receptor activator of NF-κB ligand (RANKL). In addition, increased NF-κB activity has recently been shown to decrease mature osteoblast function in vivo, leading to impaired production and maturation of bone matrix, a second feature of aging bone.20
Because there are powerful genetic influences on both longevity and bone mass, we hypothesized that SirT1 could represent a genetic link between aging and bone remodeling, which may contribute to the increased osteoclast activity, decreased osteoblast function, and overall bone loss associated with advancing age. We addressed this hypothesis by analyzing the bone phenotype of Sirt1 conditional loss of function mouse models characterized by lack of Sirt1 in osteoblasts or osteoclasts.
Materials and Methods
All animals were purchased from Jackson Laboratory (Bar Harbor, ME, USA), and all procedures were approved by the Institutional Animal Care and Use Committee at Vanderbilt University Medical Center and conducted in accordance with AALAC guidelines. The SirT1−/−, IκBα−/− and NfkB reporter (NGL) mouse models were previously described.21–24SirT1Oc−/− mice and SirT1Ob−/− mice were generated at Vanderbilt University Medical Center, through the breeding of SirT1exon4 floxed mice22 with Lysozyme M-Cre mice25 or 2.3-kb type 1 collagen-cre mice.26
Gene, protein, and luciferase expression
Dissected long bones were stripped of soft tissue, snap-frozen, and ground under liquid nitrogen. Cell cultures were lysed directly from the well plate. RNA was isolated using the RNeasy Kit and QiaShredder homogenizer (Qiagen, Valencia, CA, USA). RT-PCR was performed using superscript III/platinum taq reagents and procedures (Invitrogen, Carlsbad, CA, USA). Gene expression was quantified by qPCR, using a 7300 ABI Realtime PCR system and validated Taqman probe/primers (Runx2: Mm00501584_m1, Sost: Mm00470479_m1, Osx/Sp7: Mm00504574_m1, Dkk1: Mm00438422_m1, Hprt1 Mm01545399_m1 (Applied Biosystems, Carlsbad, CA, USA). Western Blotting was performed using rabbit anti-mouse SirT1 (Upstate, Lake Placid, NY, USA), Lys 310 acetylated p65 and p65 antibodies (Cell Signaling Technology, Danvers, MA, USA). Luciferase values were assessed using BrightGlo detection reagents and protocol (Promega, San Luis Obispo, CA, USA).
Micro-computed tomography (µCT) imaging
The volume and architecture of trabecular bone in the distal femoral metaphysis was analyzed using a µCT40 (Scanco Medical AG, Bruttisellen, Switzerland) at an isotropic voxel size of 12 um, 55 kV, 114 µA, 500 projections per 180° rotation, and 300 msec integration time. The volume of interest comprised the entire medullary volume beginning 360 µm proximal to the growth plate and extending 1.24 mm with a border lying approximately 100 µm from the endocortical surface. Three-dimensional reconstructions were created by stacking the ROIs from each 2-dimensional slice, then applying a threshold and Gaussian noise filter specifically optimized for murine trabecular bone. Fractional bone volume (BV/TV) and architectural properties of trabecular bone were calculated using the manufacturer's software.
Bone histology and histomorphometry
Lumbar vertebrae and long bones were collected after death and fixed for 48 hours in 10% formalin (Fisher, Pittsburgh, PA, USA). Undecalcified vertebrae were processed and embedded in a methylmethacrylate-based resin (Sigma, St. Louis, MO, USA)27 and sectioned at 5 µm. Sections were deplastasized and stained by the Von Kossa procedure with a Van Gieson counterstain, hematoxylin and eosin (H&E), or using a post-coupling staining technique for tartrate-resistant acid phosphatase (TRAP), according to standard protocols. Bone volume and cellular distribution were quantified histomorphometrically using Osteomeasure quantification software (Osteometrics, Decatur, GA, USA).
In vivo bone resorption was determined by assessing the levels of excreted urinary deoxypyridinoline (DPD), normalized to creatinine levels (Quidel, San Diego, CA, USA). Urine samples were collected at the same time each day.
Cell culture, primary cell preparations, sorting and assays
Osteoclasts were cultured in vitro from precursor cells flushed from murine long bones or spleens. Cells within the bone marrow were flushed out using sterile α-MEM (Gibco, Carlsbad, CA, USA) and washed in PBS. The osteoclast precursor population was isolated by incubating the total cell population with anti-mouse CD11b conjugated magnetic beads and passage through a magnetized separator column (Miltinyl, Cambridge, MA, USA). Positively sorted cells were cultured for up to 10 days in α-MEM + 10% fetal calf serum (FCS; Hyclone, Pittsburgh, PA, USA) (including 10% penicillin/streptomycin and 5% L-glutamate [Gibco]), 25 ng/mL macrophage colony-stimulating factor (M-CSF), and 50 ng/mL RANKL (R&D Systems, Minneapolis, MN, USA). LPS (0.5 ug/mL) (Sigma), BMS 345541 (0.1 uM) (Sigma), and the sirtuin inhibitor sirtinol (10 uM) (Calbiochem, Billerica, MA, USA) were added as described. Calvarial osteoblasts were extracted by multiple collagenase/trypsine digestion from 4-day-old pups, and primary osteoblasts and the murine osteoblast cell line 2T3 were cultured in α-MEM + 10% FCS.
Statistically significant values were determined by the Mann-Whitney U-test or Student's t test with p values less than 0.05 considered significant. All data are presented as mean ± standard error.
SirT1 expression and NF-κB activity are inversely related in aging bone
The transcriptional activity of NF-κB is dysregulated in various aging tissues.28, 29 However, the factors underlying the loss of this fine-tuning of NF-κB activity with aging remain unknown. Although increased NF-κB activity is described in aging organs such as brain and kidney, the mRNA and protein expression of NF-κB, together with the main regulatory factors IκB and IKK, remained unaffected in these tissues.30 These observations suggested that an unknown molecule/mechanism, linked to aging and distinct from the canonical NF-κB pathway, exerts control over NF-κB signaling. Because SirT1 has been shown to repress NF-κB activity in the myeloid lineage,13 we hypothesized that the deacetylase activity of SirT1 in bone cells may regulate NF-κB activity and bone remodeling.
To address this question, we first analyzed the expression of SirT1 in bone cells. Whole bone samples, as well as primary osteoclasts, osteoblasts, and osteoblast cell lines, demonstrated clear expression of SirT1 mRNA (Fig. 1A). In addition, quantitative analysis of SirT1 mRNA levels (Fig. 1B) and NF-κB activity (Fig. 1C) in bone samples from young (2-month-old) and aging (16-month-old) mice demonstrated a clear inverse relationship. Although such an inverse relationship between SirT1 expression and NFkB activity in bone does not prove causality between these two outcomes, these results supported a possible connection between aging, SirT1 expression, NF-κB activity, and the concomitant bone loss seen during aging.
Osteoclast-specific SirT1 knockdown reduces bone volume
Inbred mice deficient in SirT1 (SirT1−/−) develop severe defects resulting in embryonic or early postnatal lethality, including reduced stature, craniofacial defects, and early blindness.21, 22, 31 These skeletal defects observed after global deletion of SirT1 strongly implicates this molecule in the normal regulation of bone development and possibly remodeling. However, the reduced number of surviving mice, widespread expression of SirT1, and severe metabolic abnormalities of whole body SirT1−/− mice suggest possible indirect effects of SirT1 deficiency on bone remodeling. Therefore, we examined the direct, bone cell-autonomous effect of SirT1 on NF-κB activity and bone remodeling using a conditional tissue-specific SirT1 deletion strategy.
To examine the direct role of SirT1 in cells of the osteoclast lineage, we generated mice lacking SirT1 in osteoclasts (SirT1Oc−/−) by crossing mice harboring loxP sites flanking the region encoding the catalytic domain of SirT1 (exon 4)22 with mice expressing cre-recombinase under the control of the lysozyme M promoter.25 This strategy results in the expression of nonfunctional SirT1 in the osteoclast lineage, starting in osteoclast precursors.32SirT1 deficiency in the osteoclast lineage induced a significant decrease in trabecular bone volume/tissue volume and trabecular thickness in the vertebrae of SirT1Oc−/− animals compared with wild-type (WT) littermates as measured by 2D histomorphometry (Fig. 2A–C) or 3D micro-computed tomographic analyses (Fig. 2D). Similar results were observed in long bones and between genders (Supplemental Table S1). The osteoclast number/bone perimeter ratio (NOc/BPm) in the SirT1Oc−/− vertebrae was significantly elevated (Fig. 2E), whereas the osteoblast number/bone perimeter ratio (NOb/BPm) and bone formation rate were not altered (Supplemental Table S1). These findings provide strong genetic evidence that SirT1 controls bone mass via its cell-autonomous activity in osteoclasts.
SirT1 deacetylation of NF-κB controls osteoclast formation
NF-κB signaling is a major mediator of osteoclast formation and bone resorption.16–18 Interestingly, SirT1 inhibits the activity of the NF-κB protein complex through the deacetylation of Lysine 310 on the p65 subunit.13 Although it has yet to be demonstrated whether NF-κB activity is dysregulated in aging osteoclasts specifically, the repression of NF-κB activity by SirT1 represents an attractive candidate regulatory mechanism through which normal bone resorption may be controlled throughout life and a novel potential therapeutic target.
To examine whether decreased levels of SirT1 releases NF-κB activity in cells of the osteoclast lineage and favors their differentiation, endogenous NF-κB activity, NF-κB target gene expression, and differentiation assays were performed. First, we used an ex vivo system to demonstrate the effect of SirT1 inhibition on NF-κB activity in activated monocyte/macrophage cells. In this system where osteoclastogenesis is activated to mimic what occurs within the bone microenvironment, the sirtuin inhibitor Sirtinol significantly increased NF-κB activity in LPS-stimulated CD11b+ osteoclast progenitor cells isolated from NF-κB reporter mice (Fig. 3A and Supplemental Fig. S1). Second, we examined the acetylation level of NF-κB in WT and SirT1−/− primary osteoclast cultures and observed that the acetylation level of lysine 310 of the p65 subunit of NF-κB was increased three- to fourfold in LPS-stimulated SirT1−/− osteoclast precursor cells versus WT controls (Fig. 3B), confirming that NF-κB is a target of SirT1 in osteoclasts. Third, quantitative-PCR analyses were performed to measure the expression of NF-κB target genes in osteoclasts. In agreement with the above-mentioned results, a significant rise in Tnfα and IL-1 levels in osteoclasts extracted from SirT1Oc −/− versus WT mice was observed (Fig. 3C, D). Lastly and most convincingly, osteoclast precursor cells isolated from the bone marrow of SirT1Oc−/− mice formed increased numbers of TRAP+ cells compared with cells extracted from WT littermates (Fig. 3E), confirming the functional relevance of the molecular changes observed in osteoclasts lacking SirT1. These data indicate that SirT1 activity in osteoclast progenitor cells inhibits their differentiation and suggest that it does so via the repression of NF-κB activity.
If hyperactivation of NF-κB signaling is responsible for increased osteoclast formation in the absence of SirT1, we reasoned that inhibiting NF-κB should rescue the effect of SirT1 knockout in these cells. To examine this hypothesis, a specific inhibitor of NF-κB (BMS 345541) was added to cultures of precursor cells treated with low-dose RANKL and M-CSF. In agreement with our hypothesis, BMS 345541 significantly inhibited the formation of TRAP+ cells from SirT1Oc−/− CD11b+ precursor cells at a dose that did not affect WT cells (Fig. 3E and Supplemental Fig. S2), strongly suggesting that lack of SirT1 promotes osteoclastogenesis via activation of NF-κB activity in osteoclasts. Together, these results suggest that decreased SirT1 leading to elevated NF-κB activity in osteoclast precursors makes them more responsive to basal osteoclastogenic stimuli. This system could be exacerbated by the elevated levels of RANKL reported in the aging bone environment19 and explain the enhanced sensitivity and increased differentiation rates reported in human osteoclast precursors cultured from aged bone samples.33
Osteoblast-specific SirT1 knockdown decreases bone volume
In addition to excessive resorption, age-related bone loss is characterized by a decrease in osteoblast formation and function.34 To determine whether SirT1 can directly control bone formation independent of bone resorption, we crossed SirT1 exon 4 floxed mice described above with mice expressing cre-recombinase under the control of the 2.3-kb type 1 collagen promoter26 to suppress SirT1 activity in cells of the osteoblast lineage (SirT1Ob−/−). Histomorphometric analysis of lumbar vertebrae revealed a significant decrease in trabecular bone volume (BV/TV) and trabecular thickness in SirT1Ob−/− mice compared with littermate controls (Fig. 4A–C). Micro-CT analyses corroborated these findings in femoral bones (Fig. 4D) and between genders (Supplemental Table S2). In addition, a 36% decrease in vertebral osteoblast numbers and surface (13.4%) was observed in SirT1Ob−/− animals compared with controls, with no significant change in osteoclast numbers (Fig. 4E and Supplemental Table S2).
SirT1 inhibition increases NF-κB activity to regulate osteoblasts
Recent studies documented an inhibitory role of NF-κB in osteoblasts.20 Therefore, we hypothesized that the phenotype of Sirt1Ob−/− mice may be mediated by dysregulation of NF-κB signaling in osteoblasts. In support of this hypothesis, alkaline phosphatase activity was suppressed by up to 46% in 2T3 osteoblasts treated by the SirT1 inhibitors sirtinol or splitomicin (Fig. 5A,B). Similar results were observed using osteoblasts differentiated from bone marrow stromal cells from SirT1Ob−/− mice compared with WT controls (Fig. 5C). Runx2 expression in calvarial osteoblast cultures differentiated for 10 days in osteogenic medium was not significantly affected by lack of SirT1, but Osx expression at day 10 and Ocn expression at day 20 was significantly reduced (Fig. 5D–H). Sost and Dkk1 expression in osteoblast cultures differentiated for 20 days was accordingly decreased (Fig. 5F,G) in contrast to what has been observed in bone marrow osteoprogenitor cultures from SirT1+/− mice.35 Pharmacological inhibition of SirT1 by sirtinol and splitomicin increased NF-κB activity in osteoblasts from NF-κB reporter mice, similar to that of the potent NF-κB activator, LPS, used here as a positive control (Fig. 5H). Moreover, NF-κB inhibitors rescued the decrease in alkaline phosphatase activity observed in SirT1Ob−/− osteoblasts to basal levels (Fig. 5I). These results strongly suggest that, similar to what was observed in osteoclasts, SirT1 promotes osteoblast differentiation via repression of NF-κB activity. These findings are consistent with recent studies documenting an inhibitory role of NF-κB in osteoblasts.20 In this study, genetic mutations to enhance NF-κB activity in osteoblasts only resulted in decreased bone mass comparable to that seen in the SirT1Ob−/− model.
Increased NF-κB activity uncouples osteoblasts and osteoclasts
The above data indicate that SirT1 represses NF-κB activity in both osteoclasts and osteoblasts and suggest that unrepressed NF-κB activity could affect bone mass in a similar manner to that seen with aging. To further examine the relevance of NF-κB signaling in bone homeostasis, we analyzed mice lacking one allele of the NF-κB inhibitory protein, IkBa, as an additional genetic model of unrepressed NF-κB activity.23 Similarly to SirT1 osteoblast and osteoclast-specific mutant mice, mice lacking one allele of IκBα (IκBα+/−) displayed a significant 50% decrease in BV/TV (Fig. 6A,B) and reduced trabecular thickness (Fig. 6C) compared with IκBα+/+ control littermates, as measured by histomorphometry. IκBα haplo-insufficiency led to a significant 40% increase in the number and surface of TRAP-positive, multinucleated osteoclasts, corresponding with increased osteoclast activity, as measured by urinary Dpd (Fig. 6D–F), as observed in mice lacking SirT1 specifically in osteoclasts. No concurrent increase in osteoblast numbers or bone formation rates (Fig. 6H,I) was observed. In fact, osteoblast numbers were decreased in IκBα+/− bones in vivo, consistent with an inhibitory role for NF-κB signaling in osteoblastic cells and in agreement with the osteoblast-specific ablation of SirT1. The results indicate that the coupling of formation to resorption was disrupted by increased NF-κB activity and is consistent with the known pathophysiology of age-related bone loss and correlates with our finding of increased NF-κB activity in aging murine bone.
The skeletal defects observed in mice lacking SirT1 specifically in osteoblasts or osteoclasts strongly implicate this molecule in the normal regulation of bone homeostasis. These results and the association of SirT1 with cellular survival and longevity, and its decrease in expression in normal aging bones suggest that SirT1 may be involved in the bone loss associated with aging. In support of this hypothesis, the decreased levels of SirT1 measured in aging bone negatively correlate with increased activity of NF-κB, a finding that is likely related to the close interaction and inhibitory effect of SirT1 on NF-κB activity. Future investigations aimed at maintaining SirT1 activity during aging will be necessary to validate this hypothesis.
The transcriptional activity of NF-κB is reported to be dysregulated in the aging brain and kidney,28, 29 despite the fact that the mRNA and protein expression of NF-κB proteins, along with the main regulatory factors IκB and IKK, remained unaffected in these tissues.30 These reports clearly indicate that an unknown molecule/mechanism regulating the function of NF-κB, linked to aging and distinct from the canonical NF-κB pathway, exerts control over NF-κB signaling. These studies provide evidence that the longevity-associated molecule SirT1 may represent this link between NF-κB activity, aging, and dysregulated osteoclast/osteoblast activity. Together, these investigations offer new insight into the endogenous control of NF-κB signaling in bone cells and potentially in other tissues where SirT1 is expressed and NF-κB signaling plays a crucial role (eg, T cells, B cells, neurons, and hepatocytes).22, 28
SirT1 expression decreases upon aging in various tissues, including bone, lung, heart, and fat. What controls this change in expression in aging tissues remains unclear. Mechanisms leading to the inhibition of SirT1 expression and/or activity might be in play. SirT1 functions as a sensor of energy and cellular stress, and its activity is regulated by multiple pathways, including AMPK signaling,36 cAMP/PKA signaling,37 and PPARγ.38 Changes in metabolic pathways associated with aging could hence potentially be involved,39 although this remains to be demonstrated.
It has been suggested that during age-related bone loss, resorbing osteoclasts are more active on a per cell basis.40 The consequent result of this is a greater removal of bone during a normal period of resorption. An elevation in NF-κB activity in hematopoietic stem cells resulting from decreased SirT1 levels would be expected to produce osteoclast precursor cells that are more responsive to basal endogenous levels of osteoclastogenic stimuli. Also, this system would be exacerbated by the elevated levels of RANKL reported in the aging bone environment19 and the enhanced sensitivity and increased differentiation rates reported in human osteoclast precursors cultured on old bone samples.33 To what extent the relative increase in IL1 and TNFα expression in SirT1−/− osteoclasts contributes to the observed increase in osteoclastogenesis and bone loss in SirT1Oc−/− mice remains to be determined. In addition, as TNFα and IL1 signal via the NF-κB pathway, one can speculate the existence of a possible feedforward loop created by SirT1 ablation in osteoclasts.
In addition to the well-documented effects on osteoclasts, NF-κB has recently been reported to influence cells of the osteoblast lineage. In this system, genetic mutations to enhance NF-κB activity in osteoblasts only resulted in decreased bone mass,20 comparable to that seen in the SirT1Ob−/− model. Supporting these data, SirT1 and enhanced SirT1 levels by the plant polyphenol resveratrol are necessary for the maintenance of stem cell populations41, 42 and commitment toward an osteoblast lineage over adipocyte formation.43–45 Moreover, resveratrol (and resveratrol analogues) can suppress NF-κB activity and osteoclast differentiation and protect against ovariectomy-induced bone loss.46–50 Also, activation of SirT1 by resveratrol may play a protective role in other musculoskeletal disorders such as arthritis51, 52 and metabolic bone disease.53, 54
Cohen-Kfir and colleagues recently reported decreased trabecular bone and osteoblastic cell differentiation from SirT1+/− mice.35 These observations were attributed to an increase in Sost expression by osteoblasts owing to hyperacetylation of histones at the Sost promoter in SirT1+/− osteoblasts.35 Although the current studies found that conditional deletion of SirT1 only in osteoblasts and osteocytes resulted in a similar skeletal phenotype, BMSCs from these mice expressed significantly lower levels of both Sost and Dkk1. These differences may be explained by the fact that SirT1 is specifically ablated in osteoblasts, whereas the Cohen-Kfir model is a global heterozygous deletion of SirT1 with the potential to impact all tissue types, including those regulating sex steroids, which may in turn impact normal skeletal remodeling. Alternatively, suppression of Sost and Dkk1 expression in SirT1Ob−/− osteoblasts may simply be a direct consequence of the impaired differentiation potential observed in these cultures, because both genes are known to be upregulated during late osteoblast differentiation.
Together, the phenotypic features of the cell-specific SirT1 deletion models in osteoblasts and osteoclasts (Figs. 2 and 4) recapitulate the alterations in bone cell abundance and function observed in mice characterized by increased NF-κB activity (Fig. 1) and in aging bone, where increased osteoclast number and activity drive rapid bone destruction quicker than it can be replaced by the decreased pool of bone-forming osteoblasts. These observations, supported by in vitro molecular analyses in both cell lineages, support the model whereby SirT1 in both the osteoclast and osteoblast lineages serves as a negative regulator of NF-κB activation and as a factor that favors the coupling of bone formation and resorption during bone turnover and possibly aging. Similar to the results of this latter report based on the use of SirT1+/− mice, we detected a more pronounced bone phenotype in females than in males in our mutant cKO SirT1 models, suggesting that gonadal hormones may interact with this newly identified SirT1-dependent regulatory pathway in skeletal tissues.
Although the exact role of sirtuins in mammalian aging is still under debate,55–57 the capacity of SirT1 to “fine-tune” intracellular signals, along with the ability to control metabolic functions and stress responses, is clear. Together, these findings indicate a significant and multimodal function of SirT1 in human health and disease, as illustrated by our results revealing a link between SirT1 and the formation, function, and coupling of osteoblasts and osteoclasts to maintain skeletal homeostasis and provide novel targets to improve bone mass.
All authors state that they have no conflicts of interest.
We are grateful to the American Federation for Aging Research (JE), Vanderbilt Department of Medicine (GM and FE), and the Vanderbilt Orthopaedic Institute Pilot Award (DP) for funding this work, and to Drs. Jack Martin, Claire Edwards, and Xiangli Yang for comments, assistance, and review of the manuscript.
Authors' roles: JRE, GRM, and FE conceived the study and experimental design. JRE, KO, MMM, STL, FE, NF, JSN, and DSP contributed to experimentation. JRE and DSP performed µCT and animal husbandry. LC and FEY provided control, IκB mutant, and NF-κB-reporter mouse models and expertise in NF-κB analysis. JRE, DP, and FE prepared the manuscript. All authors reviewed the manuscript before submission.